<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE article PUBLIC "-//NLM//DTD Journal Publishing with OASIS Tables v3.0 20080202//EN" "journalpub-oasis3.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:oasis="http://docs.oasis-open.org/ns/oasis-exchange/table" xml:lang="en" dtd-version="3.0" article-type="research-article">
  <front>
    <journal-meta><journal-id journal-id-type="publisher">CP</journal-id><journal-title-group>
    <journal-title>Climate of the Past</journal-title>
    <abbrev-journal-title abbrev-type="publisher">CP</abbrev-journal-title><abbrev-journal-title abbrev-type="nlm-ta">Clim. Past</abbrev-journal-title>
  </journal-title-group><issn pub-type="epub">1814-9332</issn><publisher>
    <publisher-name>Copernicus Publications</publisher-name>
    <publisher-loc>Göttingen, Germany</publisher-loc>
  </publisher></journal-meta>
    <article-meta>
      <article-id pub-id-type="doi">10.5194/cp-18-821-2022</article-id><title-group><article-title><italic>crestr</italic>: an R package to perform probabilistic climate
reconstructions from palaeoecological datasets</article-title><alt-title><italic>crestr</italic> R package for probabilistic climate reconstructions</alt-title>
      </title-group><?xmltex \runningtitle{\textit{crestr} R package for probabilistic climate reconstructions}?><?xmltex \runningauthor{M. Chevalier}?>
      <contrib-group>
        <contrib contrib-type="author" corresp="yes" rid="aff1 aff2">
          <name><surname>Chevalier</surname><given-names>Manuel</given-names></name>
          <email>chevalier.manuel@gmail.com</email>
        <ext-link>https://orcid.org/0000-0002-8183-9881</ext-link></contrib>
        <aff id="aff1"><label>1</label><institution>Institute of Geosciences, Sect. Meteorology, Rheinische Friedrich-Wilhelms-Universität Bonn, <?xmltex \hack{\break}?>Auf dem Hügel 20, 53121 Bonn, Germany</institution>
        </aff>
        <aff id="aff2"><label>2</label><institution>Institute of Earth Surface Dynamics, Geopolis, University of Lausanne, Lausanne, Switzerland</institution>
        </aff>
      </contrib-group>
      <author-notes><corresp id="corr1">Manuel Chevalier (chevalier.manuel@gmail.com)</corresp></author-notes><pub-date><day>19</day><month>April</month><year>2022</year></pub-date>
      
      <volume>18</volume>
      <issue>4</issue>
      <fpage>821</fpage><lpage>844</lpage>
      <history>
        <date date-type="received"><day>3</day><month>November</month><year>2021</year></date>
           <date date-type="rev-request"><day>2</day><month>December</month><year>2021</year></date>
           <date date-type="rev-recd"><day>10</day><month>February</month><year>2022</year></date>
           <date date-type="accepted"><day>17</day><month>March</month><year>2022</year></date>
      </history>
      <permissions>
        <copyright-statement>Copyright: © 2022 Manuel Chevalier</copyright-statement>
        <copyright-year>2022</copyright-year>
      <license license-type="open-access"><license-p>This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this licence, visit <ext-link ext-link-type="uri" xlink:href="https://creativecommons.org/licenses/by/4.0/">https://creativecommons.org/licenses/by/4.0/</ext-link></license-p></license></permissions><self-uri xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022.html">This article is available from https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022.html</self-uri><self-uri xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022.pdf">The full text article is available as a PDF file from https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022.pdf</self-uri>
      <abstract><title>Abstract</title>

      <p id="d1e93">Statistical climate reconstruction techniques are fundamental tools to
study past climate variability from fossil proxy data. In particular,
the methods based on probability density functions (or PDFs) can be used
in various environments and with different climate proxies because they
rely on elementary calibration data (i.e. modern geolocalised
presence data). However, the difficulty of accessing and curating these
calibration data and the complexity of interpreting probabilistic
results have often limited their use in palaeoclimatological studies.
Here, I introduce a new R package (<italic>crestr</italic>) to apply the PDF-based
method CREST (Climate REconstruction SofTware) on diverse
palaeoecological datasets and address these problems. <italic>crestr</italic>
includes a globally curated calibration dataset for six common climate
proxies (i.e. plants, beetles, chironomids, rodents,
foraminifera, and dinoflagellate cysts) associated with an extensive
range of climate variables (20 terrestrial and 19 marine variables) that
enables its use in most terrestrial and marine environments. Private
data collections can also be used instead of, or in combination with,
the provided calibration dataset. The package includes a suite of
graphical diagnostic tools to represent the data at each step of the
reconstruction process and provide insights into the effect of the
different modelling assumptions and external factors that underlie a
reconstruction. With this R package, the CREST method can now be used in
a scriptable environment and thus be more easily integrated with
existing workflows. It is hoped that <italic>crestr</italic> will be used to
produce the much-needed quantified climate reconstructions from the many
regions where they are currently lacking, despite the availability of
suitable fossil records. To support this development, the use of the
package is illustrated with a step-by-step replication of a 790 000-year-long mean annual temperature reconstruction based on a pollen record
from southeastern Africa.</p>
  </abstract>
    </article-meta>
  </front>
<body>
      

<sec id="Ch1.S1" sec-type="intro">
  <label>1</label><title>Introduction</title>
      <p id="d1e114">Fossil-based climate reconstruction techniques are commonly used to quantify past climates and shed light on the nature of the drivers of
climate change across space and time. Over the years, numerous
techniques of increasing complexity have been proposed, each one based
on a unique set of assumptions regarding the modelling of
(palaeo)ecological datasets and their translation into climate
reconstructions (e.g. <xref ref-type="bibr" rid="bib1.bibx2 bib1.bibx18" id="altparen.1"/>). In particular, many techniques focus on
modelling the relationships between proxy assemblages and climate from
collections of modern proxy samples. Of these techniques,
weighted averaging <xref ref-type="bibr" rid="bib1.bibx65" id="paren.2"><named-content content-type="pre">WA,</named-content></xref>,
weighted-averaging partial least square <xref ref-type="bibr" rid="bib1.bibx66" id="paren.3"><named-content content-type="pre">WA-PLS,</named-content></xref>,
and the modern analogue technique (MAT, <xref ref-type="bibr" rid="bib1.bibx49 bib1.bibx57" id="altparen.4"/>) have been the most widely employed because
of their conceptual simplicity, their demonstrated capacity to
reconstruct climate from various palaeoecological proxies (e.g.
fossil pollen, chironomids, foraminifera) and their accessibility via
multiple software tools. However, the limited availability of the
necessary calibration datasets beyond the Northern Hemisphere
extratropics has often hindered their application in many environments
and regions where quantified climate reconstructions are needed, despite
the existence of suitable fossil records <xref ref-type="bibr" rid="bib1.bibx18" id="paren.5"/>.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F1" specific-use="star"><?xmltex \currentcnt{1}?><?xmltex \def\figurename{Figure}?><label>Figure 1</label><caption><p id="d1e138"><bold>(a)</bold> Conceptual illustration of the differences between a modelling approach based on the full spread of the data with the probabilities
spread along the climate gradient (e.g. CREST; dark grey), and a
modelling approach focused on the estimation of the “most likely”
or “best” climate value with small statistical errors surrounding
it (e.g. MAT or WA-PLS; light grey). In both cases, the area under
the curve sums to one. The results of the two types of approaches
are illustrated in <bold>(b, c)</bold>, where two theoretical fossil assemblages (in blue and purple) are used to produce two independent reconstructions of the same climatic parameter for the same time interval. <bold>(b)</bold> The two reconstructions are derived from a method that only estimates the most likely climate value, resulting in “apparently” incompatible reconstructions. <bold>(c)</bold> The same fossil assemblages are analysed using an approach that estimates their complete uncertainty distributions. In this case, the blue reconstruction is broader, and the purple reconstruction becomes bimodal. When the full spread of these uncertainties is considered, the two reconstructions are not incompatible anymore, and a joint climate estimate (gold) can be derived from their overlapping sections (hashed polygons).</p></caption>
        <?xmltex \igopts{width=497.923228pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f01.png"/>

      </fig>

      <p id="d1e158">In contrast, the “indicator species” family of reconstruction techniques
uses modern proxy occurrences (i.e. collections of locations
where the studied proxy species can be observed in modern environments)
to estimate individual proxy–climate relationships
<xref ref-type="bibr" rid="bib1.bibx18" id="paren.6"/>. Because such occurrence data are generally
easier to obtain than modern proxy assemblages, this fundamental
difference implies that indicator species methods can contribute to
filling in the reconstruction gaps that exist at the global scale. The
CREST (Climate REconstruction SofTware) technique is a probabilistic
indicator species method initially developed to produce quantified
climate reconstructions from southern African pollen records
<xref ref-type="bibr" rid="bib1.bibx16" id="paren.7"/>. Derived from the original work of
<xref ref-type="bibr" rid="bib1.bibx51" id="text.8"/> – who first proposed to replace the
commonly used modern proxy assemblages with modern geolocalised
occurrence data to estimate probabilistic proxy–climate relationships
for palaeoclimatic studies – CREST estimates and combines probabilistic
proxy–climate relationships to reconstruct past climate parameters from
fossil proxy observations. Built on a private collection of modern plant
occurrences held by the South African National Botanical Institute
(SANBI), CREST was first employed to reconstruct diverse temperature,
precipitation and moisture-related variables for different time
intervals across the southern African drylands <xref ref-type="bibr" rid="bib1.bibx9 bib1.bibx8 bib1.bibx14 bib1.bibx15 bib1.bibx52 bib1.bibx21" id="paren.9"><named-content content-type="pre">see for instance</named-content></xref>.</p>
      <p id="d1e176">Since the assumptions of CREST do not restrict its use to southern
African pollen records, CREST was also integrated into a point-and-click
graphical user interface to enable its use by the broader community
<xref ref-type="bibr" rid="bib1.bibx16" id="paren.10"/>. However, the complexity of collating and
formatting the thousands of distinct occurrences required to estimate
reliable proxy–climate relationships limited its practical use. To
overcome this limitation, a global, multi-proxy calibration dataset
containing millions of modern occurrence data for plants, beetles,
chironomids, foraminifera, and diatoms was subsequently released
<xref ref-type="bibr" rid="bib1.bibx10" id="paren.11"/>, and this curated dataset contributed to creating
quantified climate reconstructions beyond southern Africa <xref ref-type="bibr" rid="bib1.bibx75 bib1.bibx48 bib1.bibx46" id="paren.12"><named-content content-type="pre">e.g.</named-content></xref>.
However, maintaining the compatibility of the graphical interface across
a range of constantly evolving operating platforms has been challenging.
This paper thus introduces a new multi-platform R package <italic>crestr</italic>
designed to replace the original interface. <italic>crestr</italic> integrates the
global calibration dataset and provides simple solutions to tailor it to
the users' specific needs. The package also proposes an array of
graphical diagnostic tools to represent the calibration and
reconstruction data at different pivotal steps of the reconstruction
process and thus facilitate data and result interpretations.</p>
      <p id="d1e196">In addition, the advantage of using CREST is not limited to its capacity
to produce quantified reconstructions in understudied regions. CREST is
equipped with some fundamental statistical features that make it
well-adapted to analysing extensive collections of palaeoecological
records from any region <xref ref-type="bibr" rid="bib1.bibx18" id="paren.13"/>. While techniques
such as MAT or WA-PLS are primarily designed to associate modern proxy
observations with their “most likely” or “mean” climate values only,
CREST estimates and weighs all the climate values that are compatible
with the observed fossil data. As such, the application of CREST yields
a probabilistic quantification of all the climate values that are
compatible with the studied data instead of simpler, less informative
“most likely” or “best” climate estimates. While the “best estimate”
approach might be optimal when a fossil assemblage is analysed in
complete isolation, the presence of independent – local or regional –
information (e.g. other reconstructions from the same core or
independent records) usually provides additional information that may
not always be consistent with the best estimate reconstructions (see
Fig. <xref ref-type="fig" rid="Ch1.F1"/>). In practice, joint solutions based on all
the available information often differ from best climate estimates.
Using methods such as CREST that can estimate the full range of climate
uncertainties associated with a proxy sample is thus critical to bring
reconstructions and even climate simulations together in a cohesive way.</p>
      <p id="d1e204">This article introduces the <italic>crestr</italic> R package and provides
good-practice recommendations to produce high-quality climate
reconstructions. The article is structured as follows: Sect. <xref ref-type="sec" rid="Ch1.S2"/> summarises the most important mathematics and
assumptions underlying the approach. Understanding all the details of
this section is not necessary to use the package. Then, Sect. <xref ref-type="sec" rid="Ch1.S3"/> describes the embedded calibration dataset, how it
was built and how the data are structured. Section <xref ref-type="sec" rid="Ch1.S4"/>
explains the philosophy and main elements of the package and describes
the format of the different input files required. Finally, Sect. <xref ref-type="sec" rid="Ch1.S5"/> documents a step-by-step guided tour of the package,
illustrating the successive stages of a CREST analysis and how to use
the diagnostic tools to reproduce a recently published pollen-based
temperature reconstruction <xref ref-type="bibr" rid="bib1.bibx19" id="paren.14"/>.</p>
</sec>
<sec id="Ch1.S2">
  <label>2</label><title>The CREST method</title>
      <p id="d1e230">The core process of a CREST reconstruction can be decomposed into two
successive stages: (1) modelling the proxy–climate responses of the
proxies observed in the fossil sequence by correlating modern occurrence
data with corresponding climate values (Fig. <xref ref-type="fig" rid="Ch1.F2"/>) and (2)
reconstructing past climate by combining these responses based on the
information provided by the fossil data. This section presents the most
important details of these two stages and details the parameters and
modelling assumptions that can be modified in <italic>crestr</italic>. Readers
interested in an in-depth description or discussion of the method and
its modelling assumptions are referred to <xref ref-type="bibr" rid="bib1.bibx16 bib1.bibx18" id="text.15"/>.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F2" specific-use="star"><?xmltex \currentcnt{2}?><?xmltex \def\figurename{Figure}?><label>Figure 2</label><caption><p id="d1e243">Conceptual representation of CREST illustrated with artificial pollen data. <bold>(a)</bold> Modern distribution of a climate variable to reconstruct (e.g. temperature). <bold>(b)</bold> Occurrences of four species part of the same pollen morphotype exhibiting marked preferences for the lowest values of that climate (e.g. dark/cold values) across the study area. <bold>(c)</bold> Combination of the four species PDFs (in colour) and the resulting pollen taxon PDF (in black). The histogram in the inset panel represents the proportion of the modern climate space (white) occupied by at least one of the four species (black), highlighting the higher chances of observing the taxon at the lower end of the climate gradient. <bold>(d)</bold> Taxon PDF derived from the combination of the four species PDFs and the type of synthetic statistics (e.g. optimum, mean, uncertainty range) that can be derived from it. While the shape of the species PDFs is constrained, taxon PDFs can be irregular, skewed and even multi-modal.</p></caption>
        <?xmltex \igopts{width=497.923228pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f02.png"/>

      </fig>

<sec id="Ch1.S2.SS1">
  <label>2.1</label><title>Nature of the required data</title>
      <p id="d1e271">Three types of data are required to reconstruct climate parameters with
the CREST method:</p>
      <p id="d1e274"><list list-type="bullet">
            <list-item>

      <p id="d1e279">a fossil proxy record with several taxa being co-recorded and
expressed as counts, percentages or binary presence/absence;</p>
            </list-item>
            <list-item>

      <p id="d1e285">modern presence-only occurrence data of species corresponding to the
taxa observed in the fossil record (i.e. collections of
geographical coordinates where the species are observed across a
user-defined study area; see Fig. <xref ref-type="fig" rid="Ch1.F2"/>b and Sect. <xref ref-type="sec" rid="Ch1.S3"/> for an example of a curated dataset);</p>
            </list-item>
            <list-item>

      <p id="d1e295">climatology(ies) of the variable(s) to reconstruct gridded at the same
resolution as the modern occurrences (Fig. <xref ref-type="fig" rid="Ch1.F2"/>a). All the
climate values observed in the study area define the <italic>climate space</italic>.</p>
            </list-item>
          </list></p>
</sec>
<sec id="Ch1.S2.SS2">
  <label>2.2</label><title>Modelling the proxy–climate relationships</title>
      <p id="d1e314">CREST takes into account that some fossil taxa can be identified at the
species level (e.g. plant macrofossils), while others are only
identified at a lower taxonomic resolution (e.g. fossil pollen
are commonly identified at the genus, sub-family, or even family level;
<xref ref-type="bibr" rid="bib1.bibx18" id="altparen.16"/>). The transformation of the information
contained in the modern observations of the biological climate proxies
into probabilistic climate responses is thus done in one or two steps
depending on the taxonomic resolution of the studied proxy. However,
determining a list of species that could have produced that fossil is
necessary when the observed fossil taxa are not identified at the
species level. The way to make this species to fossil proxy association
is described in Sect. <xref ref-type="sec" rid="Ch1.S4.SS3.SSS2"/>.</p>
      <p id="d1e322">The individual climate responses of all the species identified are
estimated as univariate probability density functions (PDFs) for every
climate variable. Because CREST aims to be applicable even in
data-sparse environments, the estimation of these responses is based on
simple assumptions that exclude using complex algorithms, such as those
described in, for instance, the recent review of <xref ref-type="bibr" rid="bib1.bibx67" id="text.17"/>. The
species' individual responses for climate variable <inline-formula><mml:math id="M1" display="inline"><mml:mi>c</mml:mi></mml:math></inline-formula> are derived
from the empirical mean (<inline-formula><mml:math id="M2" display="inline"><mml:mrow><mml:msub><mml:mover accent="true"><mml:mi>m</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula>) and associated variance
(<inline-formula><mml:math id="M3" display="inline"><mml:mrow><mml:msubsup><mml:mi>s</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow><mml:mn mathvariant="normal">2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula>) of all the <inline-formula><mml:math id="M4" display="inline"><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi>s</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula> climate values (<inline-formula><mml:math id="M5" display="inline"><mml:mrow><mml:msub><mml:mi>c</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula>,
where <inline-formula><mml:math id="M6" display="inline"><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn mathvariant="normal">1</mml:mn><mml:mo>,</mml:mo><mml:mi mathvariant="normal">…</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>N</mml:mi><mml:mi>s</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula>) where species <inline-formula><mml:math id="M7" display="inline"><mml:mi>s</mml:mi></mml:math></inline-formula> is observed:

                <disp-formula specific-use="gather" content-type="numbered"><mml:math id="M8" display="block"><mml:mtable displaystyle="true"><mml:mlabeledtr id="Ch1.E1"><mml:mtd><mml:mtext>1</mml:mtext></mml:mtd><mml:mtd><mml:mrow><mml:mstyle displaystyle="true" class="stylechange"/><mml:msub><mml:mover accent="true"><mml:mi>m</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:mfrac style="display"><mml:mn mathvariant="normal">1</mml:mn><mml:mrow><mml:munder><mml:mo movablelimits="false">∑</mml:mo><mml:mi>i</mml:mi></mml:munder><mml:mi>k</mml:mi><mml:mo>(</mml:mo><mml:msub><mml:mi>c</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>)</mml:mo></mml:mrow></mml:mfrac></mml:mstyle><mml:munderover><mml:mo movablelimits="false">∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi>s</mml:mi></mml:msub></mml:mrow></mml:munderover><mml:mi>k</mml:mi><mml:mo>(</mml:mo><mml:msub><mml:mi>c</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>)</mml:mo><mml:msub><mml:mi>c</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mtd></mml:mlabeledtr><mml:mlabeledtr id="Ch1.E2"><mml:mtd><mml:mtext>2</mml:mtext></mml:mtd><mml:mtd><mml:mrow><mml:mstyle class="stylechange" displaystyle="true"/><mml:msubsup><mml:mi>s</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow><mml:mn mathvariant="normal">2</mml:mn></mml:msubsup><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:mfrac style="display"><mml:mn mathvariant="normal">1</mml:mn><mml:mrow><mml:munder><mml:mo movablelimits="false">∑</mml:mo><mml:mi>i</mml:mi></mml:munder><mml:mi>k</mml:mi><mml:mo>(</mml:mo><mml:msub><mml:mi>c</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>)</mml:mo></mml:mrow></mml:mfrac></mml:mstyle><mml:munderover><mml:mo movablelimits="false">∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi>s</mml:mi></mml:msub></mml:mrow></mml:munderover><mml:mi>k</mml:mi><mml:mo>(</mml:mo><mml:msub><mml:mi>c</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>)</mml:mo><mml:mo mathsize="1.1em">(</mml:mo><mml:msub><mml:mi>c</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>-</mml:mo><mml:msubsup><mml:mover accent="true"><mml:mi>m</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow><mml:mn mathvariant="normal">2</mml:mn></mml:msubsup><mml:mo mathsize="1.1em">)</mml:mo><mml:mo>,</mml:mo></mml:mrow></mml:mtd></mml:mlabeledtr></mml:mtable></mml:math></disp-formula>

            where <inline-formula><mml:math id="M9" display="inline"><mml:mrow><mml:mi>k</mml:mi><mml:mo>(</mml:mo><mml:msub><mml:mi>c</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> is a weighting parameter that can be used to
account for the uneven distribution of climate in modern environments
<xref ref-type="bibr" rid="bib1.bibx51 bib1.bibx6" id="paren.18"/>. This correction takes
into account that extreme values are usually under-represented in the
climate space (see, for instance, the white inset histogram in Fig. <xref ref-type="fig" rid="Ch1.F2"/>c), which “pushes” the peak of the
<inline-formula><mml:math id="M10" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>sp</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>c</mml:mi><mml:mo>,</mml:mo><mml:mi>s</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> towards the mean climate observed
across the study area (i.e. towards the centre of the climate
space). It can also artificially shrink the range of the
reconstructions. Here, the weights are calculated by first sorting the
<inline-formula><mml:math id="M11" display="inline"><mml:mi>N</mml:mi></mml:math></inline-formula> climate values that compose the modern climate space into
bins of equal width (e.g. 2 <inline-formula><mml:math id="M12" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula> or 50 mm).
Then, each climate value <inline-formula><mml:math id="M13" display="inline"><mml:mrow><mml:msub><mml:mi>c</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula> (<inline-formula><mml:math id="M14" display="inline"><mml:mrow><mml:mi>j</mml:mi><mml:mo>=</mml:mo><mml:mn mathvariant="normal">1</mml:mn><mml:mo>,</mml:mo><mml:mi mathvariant="normal">…</mml:mi><mml:mo>,</mml:mo><mml:mi>N</mml:mi></mml:mrow></mml:math></inline-formula>) is given
a weight <inline-formula><mml:math id="M15" display="inline"><mml:mrow><mml:mi>k</mml:mi><mml:mo>(</mml:mo><mml:msub><mml:mi>c</mml:mi><mml:mi>j</mml:mi></mml:msub><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> defined as the inverse of the number of values that
belong to same bin <inline-formula><mml:math id="M16" display="inline"><mml:mrow><mml:msub><mml:mtext>bin</mml:mtext><mml:mrow><mml:msub><mml:mi>c</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula>:
            <disp-formula id="Ch1.E3" content-type="numbered"><label>3</label><mml:math id="M17" display="block"><mml:mrow><mml:mi>k</mml:mi><mml:mo>(</mml:mo><mml:msub><mml:mi>c</mml:mi><mml:mi>j</mml:mi></mml:msub><mml:mo>)</mml:mo><mml:mo>=</mml:mo><mml:msup><mml:mfenced close=")" open="("><mml:mrow><mml:mstyle displaystyle="true"><mml:mfrac style="display"><mml:mn mathvariant="normal">1</mml:mn><mml:mi>N</mml:mi></mml:mfrac></mml:mstyle><mml:munderover><mml:mo movablelimits="false">∑</mml:mo><mml:mrow><mml:mi>k</mml:mi><mml:mo>=</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow><mml:mi>N</mml:mi></mml:munderover><mml:msub><mml:mn mathvariant="bold">1</mml:mn><mml:mrow><mml:msub><mml:mi>c</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>∈</mml:mo><mml:msub><mml:mtext>bin</mml:mtext><mml:mrow><mml:msub><mml:mi>c</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:msub></mml:mrow></mml:msub></mml:mrow></mml:mfenced><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup><mml:mo>.</mml:mo></mml:mrow></mml:math></disp-formula></p>
      <p id="d1e837">With this correction, the most abundant climate values in the climate
space are down-weighted, and the rarer ones are up-weighted so that the
distribution of modern climate is overall more “balanced”. The two
parameters <inline-formula><mml:math id="M18" display="inline"><mml:mrow><mml:msub><mml:mover accent="true"><mml:mi>m</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula> and <inline-formula><mml:math id="M19" display="inline"><mml:mrow><mml:msub><mml:mi>s</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula> can be interpreted as
the climate preference and tolerance of the species, respectively. For a
reliable estimation, excluding species with few observations is
recommended. Different studies have shown that a threshold of a minimum
of <inline-formula><mml:math id="M20" display="inline"><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi>s</mml:mi></mml:msub><mml:mo>≥</mml:mo><mml:mn mathvariant="normal">20</mml:mn></mml:mrow></mml:math></inline-formula>–25 distinct occurrences usually leads to
reliable estimates <xref ref-type="bibr" rid="bib1.bibx16 bib1.bibx19" id="paren.19"><named-content content-type="pre">e.g.</named-content></xref>. However, this number can vary between
regions and proxies.</p>
      <p id="d1e895">Once estimated, <inline-formula><mml:math id="M21" display="inline"><mml:mrow><mml:msub><mml:mover accent="true"><mml:mi>m</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula> and <inline-formula><mml:math id="M22" display="inline"><mml:mrow><mml:msubsup><mml:mi>s</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow><mml:mn mathvariant="normal">2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> are used to
define a regular, unimodal <inline-formula><mml:math id="M23" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>sp</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> for species
<inline-formula><mml:math id="M24" display="inline"><mml:mi>s</mml:mi></mml:math></inline-formula> and climate variable <inline-formula><mml:math id="M25" display="inline"><mml:mi>c</mml:mi></mml:math></inline-formula>. Here, we assume that the shape of
these species responses should be unimodal and can be either normal:
            <disp-formula id="Ch1.E4" content-type="numbered"><label>4</label><mml:math id="M26" display="block"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>sp</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:mfrac style="display"><mml:mn mathvariant="normal">1</mml:mn><mml:msqrt><mml:mrow><mml:mn mathvariant="normal">2</mml:mn><mml:mi mathvariant="italic">π</mml:mi><mml:msubsup><mml:mi>s</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow><mml:mn mathvariant="normal">2</mml:mn></mml:msubsup></mml:mrow></mml:msqrt></mml:mfrac></mml:mstyle><mml:mi>exp⁡</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mo>-</mml:mo><mml:mstyle displaystyle="true"><mml:mfrac style="display"><mml:mrow><mml:mo>(</mml:mo><mml:mi>c</mml:mi><mml:mo>-</mml:mo><mml:msub><mml:mover accent="true"><mml:mi>m</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow></mml:msub><mml:msup><mml:mo>)</mml:mo><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow><mml:mrow><mml:mn mathvariant="normal">2</mml:mn><mml:msubsup><mml:mi>s</mml:mi><mml:mrow><mml:mi>c</mml:mi><mml:mo>,</mml:mo><mml:mi>s</mml:mi></mml:mrow><mml:mn mathvariant="normal">2</mml:mn></mml:msubsup></mml:mrow></mml:mfrac></mml:mstyle></mml:mrow></mml:mfenced></mml:mrow></mml:math></disp-formula>
          or log-normal if the variable is not defined for negative values
(e.g. precipitation variables):

                <disp-formula specific-use="align" content-type="numbered"><mml:math id="M27" display="block"><mml:mtable displaystyle="true"><mml:mlabeledtr id="Ch1.E5"><mml:mtd><mml:mtext>5</mml:mtext></mml:mtd><mml:mtd><mml:mstyle class="stylechange" displaystyle="true"/></mml:mtd><mml:mtd><mml:mrow><mml:mstyle displaystyle="true" class="stylechange"/><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>sp</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:mfrac style="display"><mml:mn mathvariant="normal">1</mml:mn><mml:msqrt><mml:mrow><mml:mn mathvariant="normal">2</mml:mn><mml:mi mathvariant="italic">π</mml:mi><mml:msup><mml:mi mathvariant="italic">σ</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup><mml:msup><mml:mi>c</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow></mml:msqrt></mml:mfrac></mml:mstyle><mml:mi>exp⁡</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mo>-</mml:mo><mml:mstyle displaystyle="true"><mml:mfrac style="display"><mml:mrow><mml:mo>(</mml:mo><mml:mtext>ln</mml:mtext><mml:mo>(</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo><mml:mo>-</mml:mo><mml:mi mathvariant="italic">μ</mml:mi><mml:msup><mml:mo>)</mml:mo><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow><mml:mrow><mml:mn mathvariant="normal">2</mml:mn><mml:msup><mml:mi mathvariant="italic">σ</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow></mml:mfrac></mml:mstyle></mml:mrow></mml:mfenced></mml:mrow></mml:mtd></mml:mlabeledtr><mml:mtr><mml:mtd><mml:mstyle displaystyle="true" class="stylechange"/></mml:mtd><mml:mtd><mml:mrow><mml:mstyle class="stylechange" displaystyle="true"/><mml:mtext>with</mml:mtext><mml:mfenced open="{" close=""><mml:mtable class="cases" columnspacing="1em" rowspacing="0.2ex" columnalign="left" framespacing="0em"><mml:mtr><mml:mtd><mml:mrow><mml:mi mathvariant="italic">μ</mml:mi><mml:mo>=</mml:mo><mml:mtext>ln</mml:mtext><mml:mo>(</mml:mo><mml:msub><mml:mover accent="true"><mml:mi>m</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow></mml:msub><mml:mo>)</mml:mo><mml:mo>-</mml:mo><mml:mstyle displaystyle="false"><mml:mfrac style="text"><mml:mn mathvariant="normal">1</mml:mn><mml:mn mathvariant="normal">2</mml:mn></mml:mfrac></mml:mstyle><mml:mtext>ln</mml:mtext><mml:mfenced close=")" open="("><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>+</mml:mo><mml:mstyle displaystyle="false"><mml:mfrac style="text"><mml:mrow><mml:msubsup><mml:mi>s</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow><mml:mn mathvariant="normal">2</mml:mn></mml:msubsup></mml:mrow><mml:mrow><mml:msubsup><mml:mover accent="true"><mml:mi>m</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow><mml:mn mathvariant="normal">2</mml:mn></mml:msubsup></mml:mrow></mml:mfrac></mml:mstyle></mml:mrow></mml:mfenced></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:msup><mml:mi mathvariant="italic">σ</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup><mml:mo>=</mml:mo><mml:mtext>ln</mml:mtext><mml:mfenced open="(" close=")"><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>+</mml:mo><mml:mstyle displaystyle="false"><mml:mfrac style="text"><mml:mrow><mml:msubsup><mml:mi>s</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow><mml:mn mathvariant="normal">2</mml:mn></mml:msubsup></mml:mrow><mml:mrow><mml:msubsup><mml:mover accent="true"><mml:mi>m</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mrow><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi></mml:mrow><mml:mn mathvariant="normal">2</mml:mn></mml:msubsup></mml:mrow></mml:mfrac></mml:mstyle></mml:mrow></mml:mfenced><mml:mo>.</mml:mo></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mfenced></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:math></disp-formula></p>
      <p id="d1e1286">For the fossil taxa <inline-formula><mml:math id="M28" display="inline"><mml:mi>t</mml:mi></mml:math></inline-formula> that are not identified at the species level,
the <inline-formula><mml:math id="M29" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>sp</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>c</mml:mi><mml:mo>,</mml:mo><mml:mi>s</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> of their <inline-formula><mml:math id="M30" display="inline"><mml:mrow><mml:mi>S</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> composing species
are combined together to meet the taxonomic resolution of the fossil
observation (hereafter <inline-formula><mml:math id="M31" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>tx</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula>):
            <disp-formula id="Ch1.E6" content-type="numbered"><label>6</label><mml:math id="M32" display="block"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>tx</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:mfrac style="display"><mml:mn mathvariant="normal">1</mml:mn><mml:mrow><mml:msub><mml:mo>∑</mml:mo><mml:mi>s</mml:mi></mml:msub><mml:msqrt><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi>s</mml:mi></mml:msub></mml:mrow></mml:msqrt></mml:mrow></mml:mfrac></mml:mstyle><mml:mfenced open="(" close=")"><mml:mrow><mml:munderover><mml:mo movablelimits="false">∑</mml:mo><mml:mrow><mml:mi>s</mml:mi><mml:mo>=</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow><mml:mrow><mml:mi>S</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:munderover><mml:msqrt><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi>s</mml:mi></mml:msub></mml:mrow></mml:msqrt><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>sp</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:mfenced><mml:mo>.</mml:mo></mml:mrow></mml:math></disp-formula></p>
      <p id="d1e1436">This linear combination ensures that all the climate values that support
the presence of at least one species have a non-null probability in the
<inline-formula><mml:math id="M33" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>tx</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula>. Contrary to the previous step, no
additional constraints are added here. The distribution of the
<inline-formula><mml:math id="M34" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>tx</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> can thus be asymmetrical and even
multimodal if different (groups of) composing species exhibit distinct
climate requirements. An additional option is to weigh the different
<inline-formula><mml:math id="M35" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>sp</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> by the square root of the number of
individual occurrences composing their distribution (<inline-formula><mml:math id="M36" display="inline"><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi>s</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula>).
Considering that it is more difficult to estimate robust parameters with
few points, this weighting gives more importance to the species with
more extensive geographical distributions today. Said differently, it
gives more weight to the species whose climate responses can be the most
reliably defined.</p>
      <p id="d1e1513">Finally, it is important to note that if a fossil taxon is taxonomically
resolved at the species level, its number of composing species <inline-formula><mml:math id="M37" display="inline"><mml:mrow><mml:mi>S</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula>
equals 1. The resulting <inline-formula><mml:math id="M38" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>tx</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> is thus
equivalent to its <inline-formula><mml:math id="M39" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>sp</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula>. A fossil sample can
thus be composed of a mix of taxa identified at the species level and
taxa identified at a lower taxonomical level without interfering with
the reconstruction algorithm.</p>
</sec>
<sec id="Ch1.S2.SS3">
  <label>2.3</label><title>Reconstructing climate</title>
      <p id="d1e1580">Climate <inline-formula><mml:math id="M40" display="inline"><mml:mi>c</mml:mi></mml:math></inline-formula> is reconstructed from fossil sample <inline-formula><mml:math id="M41" display="inline"><mml:mi>z</mml:mi></mml:math></inline-formula> (itself
associated with a unique age, depth or any other identifier) by
multiplying the <inline-formula><mml:math id="M42" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>tx</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> of the <inline-formula><mml:math id="M43" display="inline"><mml:mrow><mml:mi>T</mml:mi><mml:mo>(</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> selected
taxa:
            <disp-formula id="Ch1.E7" content-type="numbered"><label>7</label><mml:math id="M44" display="block"><mml:mrow><mml:mtext>RECON</mml:mtext><mml:mo>(</mml:mo><mml:mi>c</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo><mml:mo>=</mml:mo><mml:msup><mml:mfenced open="(" close=")"><mml:mrow><mml:munderover><mml:mo movablelimits="false">∏</mml:mo><mml:mrow><mml:mi>t</mml:mi><mml:mo>=</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow><mml:mrow><mml:mi>T</mml:mi><mml:mo>(</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:munderover><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>tx</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:msup><mml:mo>)</mml:mo><mml:mrow><mml:mi mathvariant="italic">ω</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:msup></mml:mrow></mml:mfenced><mml:mrow><mml:msup><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mo>∑</mml:mo><mml:mi>t</mml:mi></mml:msub><mml:mi mathvariant="italic">ω</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:mfenced><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:msup><mml:mspace linebreak="nobreak" width="0.125em"/><mml:mo>,</mml:mo></mml:mrow></mml:math></disp-formula>
          where <inline-formula><mml:math id="M45" display="inline"><mml:mrow><mml:mi mathvariant="italic">ω</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> is a positive value that is used to weigh taxon
<inline-formula><mml:math id="M46" display="inline"><mml:mi>t</mml:mi></mml:math></inline-formula> in sample <inline-formula><mml:math id="M47" display="inline"><mml:mi>z</mml:mi></mml:math></inline-formula>. For presence/absence observations, the weights
<inline-formula><mml:math id="M48" display="inline"><mml:mrow><mml:mi mathvariant="italic">ω</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> of taxon <inline-formula><mml:math id="M49" display="inline"><mml:mi>t</mml:mi></mml:math></inline-formula> will be either one (the taxon is
observed) or zero (the taxon is not observed and does not influence the
reconstruction). For compositional data, <inline-formula><mml:math id="M50" display="inline"><mml:mrow><mml:mi mathvariant="italic">ω</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> can be the
observed percentages (i.e. values between 0 and 100). However,
using raw percentages implies that the observed percentages are directly
proportional to the taxa composition across the catchment. This can give
considerable weight to abundant, ubiquitous taxa that do not necessarily
have a well-defined climate response and, in contrast, can strongly
limit the influence of rarer taxa with stricter climate preferences. An
empirical normalisation is proposed in CREST to account for the varying
production rates, distribution and preservation processes impacting the
relative proportions of taxa observed in the sediments
<xref ref-type="bibr" rid="bib1.bibx16" id="paren.20"/>. In this taxon-specific scaling method, the
percentages of each taxon are divided by their average percentage when
it is present (zeros are excluded from the calculation of the average):
            <disp-formula id="Ch1.E8" content-type="numbered"><label>8</label><mml:math id="M51" display="block"><mml:mrow><mml:mi mathvariant="italic">ω</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:mfrac style="display"><mml:mrow><mml:mi>O</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo></mml:mrow><mml:mrow><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mo>∑</mml:mo><mml:mi>z</mml:mi></mml:msub><mml:mi>O</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:mfenced><mml:mo>/</mml:mo><mml:mfenced close=")" open="("><mml:mrow><mml:msub><mml:mo>∑</mml:mo><mml:mi>z</mml:mi></mml:msub><mml:msub><mml:mn mathvariant="bold">1</mml:mn><mml:mrow><mml:mi>O</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo><mml:mo>&gt;</mml:mo><mml:mn mathvariant="normal">0</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:mfenced></mml:mrow></mml:mfrac></mml:mstyle><mml:mspace width="0.125em" linebreak="nobreak"/><mml:mo>,</mml:mo></mml:mrow></mml:math></disp-formula>
          where <inline-formula><mml:math id="M52" display="inline"><mml:mrow><mml:mi>O</mml:mi><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> represents the observed percentage of taxon <inline-formula><mml:math id="M53" display="inline"><mml:mi>t</mml:mi></mml:math></inline-formula> in
sample <inline-formula><mml:math id="M54" display="inline"><mml:mi>z</mml:mi></mml:math></inline-formula>. With this transformation, all the normalised percentages
vary on a standardised scale. The average presence is given a weight of
1, and values below and above this average presence threshold are
assumed to represent lower and higher abundance in the environment,
respectively. While the empirical nature of this solution makes it
imperfect and sensitive to the quality of the data, it nevertheless
enables using percentages to inform the reconstructions. The three
weighting options described here are available in <italic>crestr</italic>, but
users can also design their own weights to better account for the
specificity of their data.</p>
      <p id="d1e1934">Finally, the presence of each taxon in a sample is considered
independent from the others. It is thus possible to select a subset of
sensitive taxa to reconstruct a specific climate variable. In some
cases, identifying a subgroup of climate-sensitive taxa can help
disentangle the different climate signals represented by the
palaeoecological data and improve the quality of the reconstructions
<xref ref-type="bibr" rid="bib1.bibx14" id="paren.21"/>, even if it is not always necessary
<xref ref-type="bibr" rid="bib1.bibx19" id="paren.22"/>. These choices should be dictated by the
data themselves and the users' understanding of the studied proxy
system. <italic>crestr</italic> provides graphical tools to help identify the
possible climate sensitivities of the studied taxa across the study area
<xref ref-type="bibr" rid="bib1.bibx20" id="paren.23"/>.</p>
</sec>
</sec>
<sec id="Ch1.S3">
  <label>3</label><title>CREST calibration dataset</title>
      <p id="d1e1958">A multiproxy calibration dataset to estimate PDFs from a global
collection of presence-only occurrence data (hereafter proxy
“distributions”) was introduced in <xref ref-type="bibr" rid="bib1.bibx10" id="text.24"/>. These data
were obtained from the Global Biodiversity Information Facility (GBIF)
database, an online collection of geolocalised observations of
biological entities <xref ref-type="bibr" rid="bib1.bibx29" id="paren.25"/>. The calibration dataset
<xref ref-type="bibr" rid="bib1.bibx11" id="paren.26"><named-content content-type="pre">hereafter <italic>gbif4crest</italic>,</named-content></xref> contains the
species distributions of six common palaeoecological fossils: the five
taxa presented in the original version of the dataset – plants
<xref ref-type="bibr" rid="bib1.bibx41 bib1.bibx37 bib1.bibx40 bib1.bibx35 bib1.bibx36 bib1.bibx42 bib1.bibx38 bib1.bibx43 bib1.bibx39 bib1.bibx44 bib1.bibx45" id="paren.27"/> for fossil pollen and macrofossils,
chironomids <xref ref-type="bibr" rid="bib1.bibx31" id="paren.28"/>, beetles <xref ref-type="bibr" rid="bib1.bibx30" id="paren.29"/>,
diatoms <xref ref-type="bibr" rid="bib1.bibx32" id="paren.30"/> and foraminifera <xref ref-type="bibr" rid="bib1.bibx33" id="paren.31"/> –
to which rodents <xref ref-type="bibr" rid="bib1.bibx34" id="paren.32"/> were recently added (Fig. <xref ref-type="fig" rid="Ch1.F3"/>). These data were curated and stored in a relational
database to ensure the consistency of the data.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F3" specific-use="star"><?xmltex \currentcnt{3}?><?xmltex \def\figurename{Figure}?><label>Figure 3</label><caption><p id="d1e1998">Distribution and grid cell density of the six climate proxies available in the <italic>gbif4crest</italic> calibration database. The total number of unique species occurrences (N) is indicated for each proxy. The maps are based on the “Equal Earth” map projection to better account for the relative sizes of the different continents.</p></caption>
        <?xmltex \igopts{width=497.923228pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f03.png"/>

      </fig>

      <?xmltex \floatpos{t}?><fig id="Ch1.F4" specific-use="star"><?xmltex \currentcnt{4}?><?xmltex \def\figurename{Figure}?><label>Figure 4</label><caption><p id="d1e2012">Structure of the <italic>gbif4crest</italic> PostgreSQL database. By default, the package extracts data from the TAXA, DISTRIB-QDGC and DATA-QDGC tables. The DISTRIB table contains the raw occurrence data obtained from GBIF and can be used to curate the distribution data at a different spatial resolution. Note that the DATA-QDGC table spreads across two rows on this figure.</p></caption>
        <?xmltex \igopts{width=497.923228pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f04.png"/>

      </fig>

      <p id="d1e2025">The coordinates of all the presence records of these six common
palaeoecological fossil proxies were mapped onto a grid with a spatial
resolution of <inline-formula><mml:math id="M55" display="inline"><mml:mrow><mml:mn mathvariant="normal">0.25</mml:mn><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mo>×</mml:mo><mml:mn mathvariant="normal">0.25</mml:mn><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:mrow></mml:math></inline-formula> (hereafter QDGC for
quarter-degree grid cell). The QDGC spatial resolution is an empirical
trade-off between numerous factors, including the resolution of the
presence data, the quality of the data or the spatial representativity
of the studied proxy (see discussions in <xref ref-type="bibr" rid="bib1.bibx16" id="altparen.33"/>,
and <xref ref-type="bibr" rid="bib1.bibx10" id="altparen.34"/>). However, this trade-off may be suboptimal in
some situations. For that reason, <italic>crestr</italic> can also be used with
the raw GBIF data (stored in the <monospace>DISTRIB</monospace> table, Fig. <xref ref-type="fig" rid="Ch1.F4"/>) and even independent calibration datasets.</p>
      <p id="d1e2063">In the <italic>gbif4crest</italic> database, all the QDGC grid cells were
associated with a collection of terrestrial and oceanic environmental
variables that can be reconstructed (<xref ref-type="bibr" rid="bib1.bibx25 bib1.bibx76 bib1.bibx54 bib1.bibx77 bib1.bibx26 bib1.bibx27 bib1.bibx62" id="altparen.35"/>,
see details in Tables <xref ref-type="table" rid="Ch1.T1"/> and
<xref ref-type="table" rid="Ch1.T2"/>). Despite the diversity of variables
available, it is recommended to avoid serial reconstructions and, on the
contrary, to identify the few important variables for the studied
palaeoecological datasets a priori. The grid cells were also
associated with “non-reconstructible” environmental and geographical
descriptors that serve to tailor the calibration dataset to the users'
needs. These include the coordinates, the elevation and elevation
variability within the grid cell <xref ref-type="bibr" rid="bib1.bibx1" id="paren.36"/>, the country
(<uri>https://www.naturalearthdata.com</uri>, last access: September 2021) or ocean
(<uri>https://www.marineregions.org</uri>, last access: September 2021) names, as well as different levels of
ecological classification for the terrestrial <xref ref-type="bibr" rid="bib1.bibx56" id="paren.37"/> and
marine <xref ref-type="bibr" rid="bib1.bibx22" id="paren.38"/> realms.</p>

<?xmltex \floatpos{p}?><table-wrap id="Ch1.T1" specific-use="star"><?xmltex \currentcnt{1}?><label>Table 1</label><caption><p id="d1e2095">List of terrestrial variables available for reconstruction in the <italic>gbif4crest</italic> database. Each one can be selected in <italic>crestr</italic> using its associated code. Temp – temperature, Precip. – precipitation.</p></caption><oasis:table frame="topbot"><oasis:tgroup cols="3">
     <oasis:colspec colnum="1" colname="col1" align="left"/>
     <oasis:colspec colnum="2" colname="col2" align="left"/>
     <oasis:colspec colnum="3" colname="col3" align="left"/>
     <oasis:thead>
       <oasis:row rowsep="1">
         <oasis:entry namest="col1" nameend="col3" align="center">Terrestrial variables </oasis:entry>
       </oasis:row>
     </oasis:thead>
     <oasis:tbody>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">Code</oasis:entry>
         <oasis:entry colname="col2">Full name</oasis:entry>
         <oasis:entry colname="col3">Source</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio1</oasis:entry>
         <oasis:entry colname="col2">Mean annual temp. (<inline-formula><mml:math id="M56" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio2</oasis:entry>
         <oasis:entry colname="col2">Mean diurnal range (<inline-formula><mml:math id="M57" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio3</oasis:entry>
         <oasis:entry colname="col2">Isothermality (<inline-formula><mml:math id="M58" display="inline"><mml:mo lspace="0mm">×</mml:mo></mml:math></inline-formula>100) (unitless)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio4</oasis:entry>
         <oasis:entry colname="col2">Temp. seasonality  (standard deviation <inline-formula><mml:math id="M59" display="inline"><mml:mo>×</mml:mo></mml:math></inline-formula>100) (<inline-formula><mml:math id="M60" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio5</oasis:entry>
         <oasis:entry colname="col2">Max temp. of the warmest month (<inline-formula><mml:math id="M61" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio6</oasis:entry>
         <oasis:entry colname="col2">Min temp. of the coldest month (<inline-formula><mml:math id="M62" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio7</oasis:entry>
         <oasis:entry colname="col2">Temp. annual range (<inline-formula><mml:math id="M63" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio8</oasis:entry>
         <oasis:entry colname="col2">Mean temp. of the wettest quarter (<inline-formula><mml:math id="M64" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio9</oasis:entry>
         <oasis:entry colname="col2">Mean temp. of the driest quarter (<inline-formula><mml:math id="M65" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio10</oasis:entry>
         <oasis:entry colname="col2">Mean temp. of the warmest quarter (<inline-formula><mml:math id="M66" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio11</oasis:entry>
         <oasis:entry colname="col2">Mean temp. of the coldest quarter (<inline-formula><mml:math id="M67" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio12</oasis:entry>
         <oasis:entry colname="col2">Annual precip. (mm)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio13</oasis:entry>
         <oasis:entry colname="col2">Precip. of the wettest month (mm)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio14</oasis:entry>
         <oasis:entry colname="col2">Precip. of the driest month (mm)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio15</oasis:entry>
         <oasis:entry colname="col2">Precip. seasonality (coefficient of variation) (mm)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio16</oasis:entry>
         <oasis:entry colname="col2">Precip. of the wettest quarter (mm)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio17</oasis:entry>
         <oasis:entry colname="col2">Precip. of the driest quarter (mm)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio18</oasis:entry>
         <oasis:entry colname="col2">Precip. of the warmest quarter (mm)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">bio19</oasis:entry>
         <oasis:entry colname="col2">Precip. of the coldest quarter (mm)</oasis:entry>
         <oasis:entry colname="col3">Fick and Hijmans (2017)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">ai</oasis:entry>
         <oasis:entry colname="col2">Aridity index (unitless)</oasis:entry>
         <oasis:entry colname="col3">Zomer et al. (2008)</oasis:entry>
       </oasis:row>
     </oasis:tbody>
   </oasis:tgroup></oasis:table></table-wrap>

<?xmltex \floatpos{p}?><table-wrap id="Ch1.T2" specific-use="star"><?xmltex \currentcnt{2}?><label>Table 2</label><caption><p id="d1e2519">List of marine variables available for reconstruction in the <italic>gbif4crest</italic> database. Each one can be selected in <italic>crestr</italic> using its associated code.  SST – sea surface temperature, SSS – sea surface salinity.</p></caption><oasis:table frame="topbot"><oasis:tgroup cols="3">
     <oasis:colspec colnum="1" colname="col1" align="left"/>
     <oasis:colspec colnum="2" colname="col2" align="left"/>
     <oasis:colspec colnum="3" colname="col3" align="left"/>
     <oasis:thead>
       <oasis:row rowsep="1">
         <oasis:entry namest="col1" nameend="col3" align="center">Oceanic variables </oasis:entry>
       </oasis:row>
     </oasis:thead>
     <oasis:tbody>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">Code</oasis:entry>
         <oasis:entry colname="col2">Full name</oasis:entry>
         <oasis:entry colname="col3">Source</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">sst_ann</oasis:entry>
         <oasis:entry colname="col2">Mean annual SST (<inline-formula><mml:math id="M68" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Locarnini et al. (2018)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">sst_jfm</oasis:entry>
         <oasis:entry colname="col2">Mean winter SST (<inline-formula><mml:math id="M69" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Locarnini et al. (2018)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">sst_amj</oasis:entry>
         <oasis:entry colname="col2">Mean spring SST (<inline-formula><mml:math id="M70" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Locarnini et al. (2018)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">sst_jas</oasis:entry>
         <oasis:entry colname="col2">Mean summer SST (<inline-formula><mml:math id="M71" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Locarnini et al. (2018)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">sst_ond</oasis:entry>
         <oasis:entry colname="col2">Mean fall SST (<inline-formula><mml:math id="M72" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Locarnini et al. (2018)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">sss_ann</oasis:entry>
         <oasis:entry colname="col2">Mean annual SSS (PSU)</oasis:entry>
         <oasis:entry colname="col3">Zweng et al. (2018)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">sss_jfm</oasis:entry>
         <oasis:entry colname="col2">Mean winter SSS (PSU)</oasis:entry>
         <oasis:entry colname="col3">Zweng et al. (2018)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">sss_amj</oasis:entry>
         <oasis:entry colname="col2">Mean spring SSS (PSU)</oasis:entry>
         <oasis:entry colname="col3">Zweng et al. (2018)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">sss_jas</oasis:entry>
         <oasis:entry colname="col2">Mean summer SSS (PSU)</oasis:entry>
         <oasis:entry colname="col3">Zweng et al. (2018)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">sss_ond</oasis:entry>
         <oasis:entry colname="col2">Mean fall SSS (PSU)</oasis:entry>
         <oasis:entry colname="col3">Zweng et al. (2018)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">icec_ann</oasis:entry>
         <oasis:entry colname="col2">Mean annual sea ice concentration (%)</oasis:entry>
         <oasis:entry colname="col3">Reynolds et al. (2007)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">icec_jfm</oasis:entry>
         <oasis:entry colname="col2">Mean winter sea ice concentration (%)</oasis:entry>
         <oasis:entry colname="col3">Reynolds et al. (2007)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">icec_amj</oasis:entry>
         <oasis:entry colname="col2">Mean spring sea ice concentration (%)</oasis:entry>
         <oasis:entry colname="col3">Reynolds et al. (2007)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">icec_jas</oasis:entry>
         <oasis:entry colname="col2">Mean summer ice concentration (%)</oasis:entry>
         <oasis:entry colname="col3">Reynolds et al. (2007)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">icec_ond</oasis:entry>
         <oasis:entry colname="col2">Mean fall sea ice concentration (%)</oasis:entry>
         <oasis:entry colname="col3">Reynolds et al. (2007)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">diss_oxy</oasis:entry>
         <oasis:entry colname="col2">Dissolved oxygen concentration (<inline-formula><mml:math id="M73" display="inline"><mml:mrow class="unit"><mml:mi mathvariant="normal">µ</mml:mi><mml:mi mathvariant="normal">mol</mml:mi><mml:mspace width="0.125em" linebreak="nobreak"/><mml:msup><mml:mi mathvariant="normal">L</mml:mi><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Garcia et al. (2018a)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">nitrate</oasis:entry>
         <oasis:entry colname="col2">Nitrate concentration (<inline-formula><mml:math id="M74" display="inline"><mml:mrow class="unit"><mml:mi mathvariant="normal">µ</mml:mi><mml:mi mathvariant="normal">mol</mml:mi><mml:mspace width="0.125em" linebreak="nobreak"/><mml:msup><mml:mi mathvariant="normal">L</mml:mi><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Garcia et al. (2018b)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">phosphate</oasis:entry>
         <oasis:entry colname="col2">Phosphate concentration (<inline-formula><mml:math id="M75" display="inline"><mml:mrow class="unit"><mml:mi mathvariant="normal">µ</mml:mi><mml:mi mathvariant="normal">mol</mml:mi><mml:mspace linebreak="nobreak" width="0.125em"/><mml:msup><mml:mi mathvariant="normal">L</mml:mi><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Garcia et al. (2018b)</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">silicate</oasis:entry>
         <oasis:entry colname="col2">Silicate concentration (<inline-formula><mml:math id="M76" display="inline"><mml:mrow class="unit"><mml:mi mathvariant="normal">µ</mml:mi><mml:mi mathvariant="normal">mol</mml:mi><mml:mspace width="0.125em" linebreak="nobreak"/><mml:msup><mml:mi mathvariant="normal">L</mml:mi><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:math></inline-formula>)</oasis:entry>
         <oasis:entry colname="col3">Garcia et al. (2018b)</oasis:entry>
       </oasis:row>
     </oasis:tbody>
   </oasis:tgroup></oasis:table></table-wrap>

      <p id="d1e2930">In its current version (V2), the <italic>gbif4crest</italic> calibration dataset
contains about 25.3 million unique QDGC occurrence data for the six
climate proxies. Unfortunately, the density of data available varies
between proxies and regions (Fig. <xref ref-type="fig" rid="Ch1.F3"/>). Plant data largely
dominate the calibration dataset (<inline-formula><mml:math id="M77" display="inline"><mml:mrow><mml:mo>&gt;</mml:mo><mml:mn mathvariant="normal">22</mml:mn></mml:mrow></mml:math></inline-formula> million unique
occurrences) and allow for the use of <italic>crestr</italic> across all
landmasses. For the five other proxies, the calibration data are not as
extensive. However, these datasets are regularly updated by GBIF. For
example, the first version of the <italic>gbif4crest</italic> dataset released in 2018
contained about 17.5 million QDGC entries (<inline-formula><mml:math id="M78" display="inline"><mml:mrow><mml:mo>∼</mml:mo><mml:mn mathvariant="normal">44</mml:mn></mml:mrow></mml:math></inline-formula> %
increase). The range of “reconstructible” areas is thus rapidly
broadening <xref ref-type="bibr" rid="bib1.bibx10" id="paren.39"><named-content content-type="pre">see, for instance, the coverage of Russia by plant data
compared to the first version of the <italic>gbif4crest</italic> dataset presented
in</named-content></xref>. This database will be regularly updated, and
specific requests can also be made.</p>
      <p id="d1e2974">The <italic>gbif4crest</italic> database is composed of three main types of data:
taxonomic data (<monospace>TAXA</monospace> table in Fig. <xref ref-type="fig" rid="Ch1.F4"/>),
distribution data (<monospace>DISTRIB</monospace> (raw data) and
<monospace>DISTRIB_QDGC</monospace> (curated data) tables) and diverse geopolitical,
climatological, and environmental data (<monospace>DATA_QDGC</monospace> table). Its
structure is slightly different from the first version presented in
Chevalier (2019), with a grouping of all the separate QDGC tables in a
unique <monospace>DATA_QDGC</monospace> table to enable faster data extraction. The
<monospace>DISTRIB_QDGC</monospace> tables link the <monospace>TAXA</monospace> and
<monospace>DATA_QDGC</monospace> tables using the unique identifiers <monospace>taxonID</monospace>
and <monospace>locID</monospace>, respectively. Each occurrence's first and last
observation dates are also now included, along with the type of
observation reported by GBIF (see
<uri>https://rs.gbif.org/vocabulary/dwc/basis_of_record.xml</uri>, last access: September 2021) and the number
of observations <monospace>n_occ</monospace> reported between the first and the last
observation dates. The <monospace>DATA_QDGC</monospace> table was also entirely
recalculated using a new protocol that better accounts for coastal
margins, implying that some coastal climate values may differ, however
marginally, between the different versions of the <italic>gbif4crest</italic>
dataset.</p>
      <p id="d1e3026">Due to its large size (about 15 Gb), this database is not downloaded
when installing the package but can be accessed through different
routes. First, the data are stored in an open-access, cloud-based
PostgreSQL database that can be accessed via <italic>crestr</italic>. This
solution is the recommended option, as users without any SQL knowledge
can benefit from the package's interface to automatically query the
database simply by providing study-specific parameters (e.g. the
name of the taxa or geographical boundaries for the study area) to
import all the necessary data in the correct format to the R environment
(see Sect. <xref ref-type="sec" rid="Ch1.S5.SS2"/>). Second, more advanced
users can also directly query the database to extract and curate data
from the <monospace>DISTRIB</monospace> or <monospace>DISTRIB_QDGC</monospace> tables using SQL
requests and the <monospace>dbRequest()</monospace> function. Finally, the full
<italic>gbif4crest</italic> calibration dataset can also be downloaded as an
SQLite3 portable database file from <xref ref-type="bibr" rid="bib1.bibx11" id="text.40"/>.</p>
</sec>
<sec id="Ch1.S4">
  <label>4</label><?xmltex \opttitle{\texorpdfstring{The \textit{crestr} R
package}{The crestr R package}}?><title>The <italic>crestr</italic> R
package</title>
<sec id="Ch1.S4.SS1">
  <label>4.1</label><title>Philosophy of the package</title>
      <p id="d1e3069">The <italic>crestr</italic> package has been designed for two independent but
complementary modelling purposes. The probabilistic proxy–climate
responses can be used to quantitatively reconstruct climate from their
statistical combination, such as in <xref ref-type="bibr" rid="bib1.bibx19" id="text.41"/> or
<xref ref-type="bibr" rid="bib1.bibx14" id="text.42"/>, or they can be used in a more qualitative way to determine
the (relative) climate sensitivities of different taxa in a given area
to characterise past ecological changes, such as in
<xref ref-type="bibr" rid="bib1.bibx20" id="text.43"/> or <xref ref-type="bibr" rid="bib1.bibx59" id="text.44"/>. To simplify access to
the functionalities, default values are provided for all parameters to
enable a rapid generation of preliminary results that users can then use
as a starting point to adapt the model to their data. Different
publication-ready graphical diagnostic tools were designed to represent
the CREST data in a standardised way to guide users in this task and
avoid the typical “black box” criticism many complex statistical tools
face. These tools include plots of the calibration data, the estimated
climate responses, the reconstructions and more. These figures allow
looking at the data from different perspectives to help interpret the
results and possibly identify potential issues or biases in the selected
data and parameters. Such diagnostic tools are available for every stage
of the process, and, as exemplified in Sect. <xref ref-type="sec" rid="Ch1.S5"/>, they can
be generated with a single line of R code.</p>
</sec>
<sec id="Ch1.S4.SS2">
  <label>4.2</label><?xmltex \opttitle{\texorpdfstring{The central element: the \texttt{crestObj}
object}{The central element: the crestObj object}}?><title>The central element: the <monospace>crestObj</monospace>
object</title>

      <?xmltex \floatpos{t}?><fig id="Ch1.F5" specific-use="star"><?xmltex \currentcnt{5}?><?xmltex \def\figurename{Figure}?><label>Figure 5</label><caption><p id="d1e3104">Structure of a crestObj, here called “rcnstrctn”, with the five main sub-lists in colour. For simplicity, lists with many elements (“tax” or “clim”) are represented with double framed boxes. The unframed terminal nodes on the right-hand side of each branch are simple R objects, such as numbers, characters, vectors or data frames. The names of the functions that modify the objects in the sub-lists are indicated on the right.</p></caption>
          <?xmltex \igopts{width=483.69685pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f05.png"/>

        </fig>

      <p id="d1e3113">In <italic>crestr</italic>, all the CREST-related data are stored within a single
S3 object of the class <monospace>crestObj</monospace> that is first initialised by
either <monospace>crest.get_modern_data</monospace> or
<monospace>crest.set_modern_data</monospace> (see Sect. <xref ref-type="sec" rid="Ch1.S5.SS2"/> for details). Most package functions
will take a <monospace>crestObj</monospace> as their primary input and return an
updated version of that object. In practice, a <monospace>crestObj</monospace> is a
nested list that contains five sub-lists, each one grouping a specific
type of information, such as the calibration data, the fitted climate
responses, or the reconstructions (Fig. <xref ref-type="fig" rid="Ch1.F5"/>). Wrapper
functions have been implemented to manipulate and modify the information
contained in a <monospace>crestObj</monospace>, and users are never expected to
manually modify their <monospace>crestObj</monospace> – even if it is possible. The
five sub-lists contain the following information:</p>
      <p id="d1e3145"><list list-type="bullet">
            <list-item>

      <p id="d1e3150"><monospace>inputs</monospace> contains the input data (e.g. the
counts/percentages of the fossil proxy data, the ages of the samples
or the names of the fossil taxa).</p>
            </list-item>
            <list-item>

      <p id="d1e3158"><monospace>parameters</monospace> contains the parameters provided at the different
stages of the analysis (e.g. the tailoring of the
<italic>gbif4crest</italic> calibration dataset or the fitting and combination
of the PDFs; see Sect. <xref ref-type="sec" rid="Ch1.S5"/>).</p>
            </list-item>
            <list-item>

      <p id="d1e3171"><monospace>modelling</monospace> contains all the data related to the estimation of
the PDFs (e.g. the occurrence data (the “distributions”) used
to estimate the PDFs, the climate space of the study area, or the PDFs
themselves).</p>
            </list-item>
            <list-item>

      <p id="d1e3179"><monospace>reconstructions</monospace> contains all the results (e.g. best
estimates, synthetic error measurements, and the full distribution of
the reconstruction).</p>
            </list-item>
            <list-item>

      <p id="d1e3187"><monospace>misc</monospace> contains some additional metadata relative to the
reconstruction (e.g. the site location or, most importantly,
information related to the proxy-species associated process described
in Sect. <xref ref-type="sec" rid="Ch1.S4.SS3.SSS2"/>).</p>
            </list-item>
          </list></p>
</sec>
<sec id="Ch1.S4.SS3">
  <label>4.3</label><?xmltex \opttitle{\texorpdfstring{Input data for
\textit{crestr}}{Input data for crestr}}?><title>Input data for
<italic>crestr</italic></title>
      <p id="d1e3207">Five different input data files are compatible with <italic>crestr</italic>.
However, most applications will only require two files (the <monospace>df</monospace>
and <monospace>PSE</monospace> files, see below) to be created. More specific
applications may require up to four of these files. All the files can be
prepared outside the R environment and imported using standard R
functions. Examples files based on pseudo data accompany the “get
started” example provided in the package
(<uri>https://mchevalier2.github.io/crestr/articles/get-started.html</uri>, last access: February 2022). The
files used in the application to illustrate the package are available in the
Supplement to this paper.</p>
<sec id="Ch1.S4.SS3.SSS1">
  <label>4.3.1</label><?xmltex \opttitle{\texorpdfstring{The fossil data
(\texttt{df})}{The fossil data (df)}}?><title>The fossil data
(<monospace>df</monospace>)</title>
      <p id="d1e3233">The <monospace>df</monospace> data frame is required if <italic>crestr</italic> is used for
reconstructing climate and can be omitted if the objective is limited to
modelling the climate response(s) of different taxa. <monospace>df</monospace> is a
data frame with the samples entered as rows, with either the age, depth,
or sample ID as the first column and the fossil data in the subsequent
columns. <monospace>df</monospace> can contain raw counts, percentages,
presence/absence data (1s and 0s) or even relative weights to be used in
the reconstruction (see examples in Sect. <xref ref-type="sec" rid="Ch1.S5.SS6"/>).</p>
</sec>
<sec id="Ch1.S4.SS3.SSS2">
  <label>4.3.2</label><?xmltex \opttitle{\texorpdfstring{The proxy-species equivalency
(\texttt{PSE})
table}{The proxy-species equivalency (PSE) table}}?><title>The proxy-species equivalency
(<monospace>PSE</monospace>)
table</title>
      <p id="d1e3262">Creating a <monospace>PSE</monospace> data frame is required to extract distribution
data from the <italic>gbif4crest</italic> calibration dataset. It is used to group
individual species available in the <monospace>TAXA</monospace> table into their
corresponding fossil taxon. This step is important to estimate the
species responses (<inline-formula><mml:math id="M79" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>sp</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>s</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula>) and taxon responses
(<inline-formula><mml:math id="M80" display="inline"><mml:mrow><mml:msub><mml:mtext>PDF</mml:mtext><mml:mtext>tx</mml:mtext></mml:msub><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>c</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula>) described in Sect. <xref ref-type="sec" rid="Ch1.S2.SS2"/>. When all the fossil taxa are identified at
the species level, the <monospace>PSE</monospace> table is a simple data frame with
one row per taxon (such as, for instance, the row corresponding to
<italic>Elaeis guineensis</italic> in Table <xref ref-type="table" rid="Ch1.T3"/>). However, fossil taxa
are most often identified at a lower taxonomic resolution (sub-genus,
genus, sub-family, family). These varying levels of identification
should be encoded in the <monospace>PSE</monospace> file to link one or more (groups
of) species to their common fossil taxon name (i.e. group
together all the species that are likely to have produced the observed
fossil). Several species can be assigned to a taxon at once by limiting
the taxonomic description at the family or genus level (e.g.
<italic>Artemisia</italic> in Table <xref ref-type="table" rid="Ch1.T3"/>).</p>

<?xmltex \floatpos{t}?><table-wrap id="Ch1.T3" specific-use="star"><?xmltex \currentcnt{3}?><label>Table 3</label><caption><p id="d1e3339">Example classification of four pollen taxa from the example case study, each one with a different level of taxonomic resolution. The last column “Taxonomic resolution” is added here for explanatory purposes only and is not required in a real “PSE” table.</p></caption><oasis:table frame="topbot"><oasis:tgroup cols="6">
     <oasis:colspec colnum="1" colname="col1" align="left"/>
     <oasis:colspec colnum="2" colname="col2" align="left"/>
     <oasis:colspec colnum="3" colname="col3" align="left"/>
     <oasis:colspec colnum="4" colname="col4" align="left"/>
     <oasis:colspec colnum="5" colname="col5" align="left"/>
     <oasis:colspec colnum="6" colname="col6" align="left"/>
     <oasis:thead>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">Level</oasis:entry>
         <oasis:entry colname="col2">Family</oasis:entry>
         <oasis:entry colname="col3">Genus</oasis:entry>
         <oasis:entry colname="col4">Species</oasis:entry>
         <oasis:entry colname="col5"><italic>ProxyName</italic></oasis:entry>
         <oasis:entry colname="col6">Taxonomic resolution</oasis:entry>
       </oasis:row>
     </oasis:thead>
     <oasis:tbody>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">1</oasis:entry>
         <oasis:entry colname="col2">Asteraceae</oasis:entry>
         <oasis:entry colname="col3"/>
         <oasis:entry colname="col4"/>
         <oasis:entry colname="col5">Asteraceae undiff.</oasis:entry>
         <oasis:entry colname="col6">Family</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">2</oasis:entry>
         <oasis:entry colname="col2">Asteraceae</oasis:entry>
         <oasis:entry colname="col3"><italic>Stoebe</italic></oasis:entry>
         <oasis:entry colname="col4"/>
         <oasis:entry colname="col5"><italic>Stoebe</italic>-type</oasis:entry>
         <oasis:entry colname="col6">Subfamily</oasis:entry>
       </oasis:row>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">2</oasis:entry>
         <oasis:entry colname="col2">Asteraceae</oasis:entry>
         <oasis:entry colname="col3"><italic>Elytropappus</italic></oasis:entry>
         <oasis:entry colname="col4"/>
         <oasis:entry colname="col5"><italic>Stoebe</italic>-type</oasis:entry>
         <oasis:entry colname="col6">Subfamily</oasis:entry>
       </oasis:row>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">2</oasis:entry>
         <oasis:entry colname="col2">Asteraceae</oasis:entry>
         <oasis:entry colname="col3"><italic>Artemisia</italic></oasis:entry>
         <oasis:entry colname="col4"/>
         <oasis:entry colname="col5"><italic>Artemisia</italic></oasis:entry>
         <oasis:entry colname="col6">Genus level</oasis:entry>
       </oasis:row>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">3</oasis:entry>
         <oasis:entry colname="col2">Arecaceae</oasis:entry>
         <oasis:entry colname="col3"><italic>Elaeis</italic></oasis:entry>
         <oasis:entry colname="col4"><italic>Elaeis guineensis</italic></oasis:entry>
         <oasis:entry colname="col5"><italic>Elaeis guineensis</italic></oasis:entry>
         <oasis:entry colname="col6">Species</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">4</oasis:entry>
         <oasis:entry colname="col2"/>
         <oasis:entry colname="col3"/>
         <oasis:entry colname="col4"/>
         <oasis:entry colname="col5">Triletes spores</oasis:entry>
         <oasis:entry colname="col6">To be excluded</oasis:entry>
       </oasis:row>
     </oasis:tbody>
   </oasis:tgroup></oasis:table></table-wrap>

      <p id="d1e3525">A <monospace>PSE</monospace> file is composed of five columns (Table <xref ref-type="table" rid="Ch1.T3"/>).
The first one (<italic>Level</italic>) contains an integer that indicates the
level of taxonomic resolution of the row (1 for family, 2 for genus, 3
for species and 4 for taxa that should be excluded from the
reconstruction, e.g. “Triletes spores” in the example case
study). The fifth column, <italic>ProxyName</italic>, contains the name of the
taxon. All the taxa recorded in the <monospace>df</monospace> dataset should be listed
here, or they will be excluded from the study. Columns two to four
contain the taxonomic classification of that taxon as <italic>Family</italic>,
<italic>Genus</italic> and <italic>Species</italic>, respectively. For simplicity, a
pre-formatted version of the <monospace>PSE</monospace> table with the names of all
the taxa to study can be generated by <italic>crestr</italic> using the
<monospace>createPSE()</monospace> function that generates a spreadsheet with the
correct structure and with the <italic>Level</italic> and <italic>ProxyName</italic> columns
automatically filled in:</p>
      <p id="d1e3569"><preformat><![CDATA[list_of_taxa <- colnames(df)[-1]
createPSE(list_of_taxa)]]></preformat></p>
      <p id="d1e3573">The species–taxon association is performed in sequential steps by the
<monospace>crest.get_modern_data()</monospace> function (see Sect. <xref ref-type="sec" rid="Ch1.S5.SS2"/>). First, <italic>crestr</italic> classifies the
taxa with the lowest taxonomical resolution (i.e. when
<italic>Level</italic> is equal to one) and then increases the resolution
<italic>Level</italic> by <italic>Level</italic>. In the example in Table <xref ref-type="table" rid="Ch1.T3"/>,
different taxonomic resolution levels are provided for different plant
species belonging to the highly diverse Asteraceae family (the daisy
family). To distribute all the Asteraceae species observed across the
study area to their appropriate taxon, all the species are first
classified as “Asteraceae undiff.” (first row, <italic>Level</italic> <inline-formula><mml:math id="M81" display="inline"><mml:mo>=</mml:mo></mml:math></inline-formula> 1). In
subsequent steps, the classification of some of these Asteraceae species
is refined when reaching the better-resolved sub-groups
(<italic>Stoebe</italic>-type and <italic>Artemisia</italic> at <italic>Level</italic> <inline-formula><mml:math id="M82" display="inline"><mml:mo>=</mml:mo></mml:math></inline-formula> 2). At the
end of the process, the “Asteraceae undiff.” group only contains
Asteraceae species that grow in the study area but are not part of the
genera <italic>Stoebe</italic>, <italic>Elytropappus</italic> or <italic>Artemisia</italic>. The
latter are categorised separately as <italic>Stoebe</italic>-type or
<italic>Artemisia</italic>.</p>
      <p id="d1e3639">Additional taxa can also be added to the <monospace>PSE</monospace> file to exclude
species known not to be part of a group, even if the pollen grains
corresponding to these species have not been observed. For instance,
this “trick” could have been used to simplify the climate response of
the “Asteraceae undiff.” group by excluding more species from it. This
categorisation process can be time-consuming, as all the taxa must be
classified in a unique <monospace>PSE</monospace> table. This process will often
require a few iterations to be optimised. The results of the different
assignments are stored in the <monospace>crestObj</monospace> returned by the
<monospace>crest.get_modern_data()</monospace> function and can be evaluated by
checking <monospace>rcnstrctn$misc$taxa_notes</monospace>.</p>
</sec>
<sec id="Ch1.S4.SS3.SSS3">
  <label>4.3.3</label><?xmltex \opttitle{\texorpdfstring{The alternative modern calibration
dataset
(\texttt{distributions})}{The alternative modern calibration dataset (distributions)}}?><title>The alternative modern calibration
dataset
(<monospace>distributions</monospace>)</title>
      <p id="d1e3670">Users that prefer fitting proxy–climate responses from their own
calibration data instead of the proposed <italic>gbif4crest</italic> dataset
should prepare a <monospace>distributions</monospace> dataset following the specific
structure presented in Table <xref ref-type="table" rid="Ch1.T4"/>. The first two
columns should contain species names (or any unique identifiers) and the
corresponding proxy name. If more than one species correspond to one
taxon, the PDFs will be fitted in two steps, as explained in Sect. <xref ref-type="sec" rid="Ch1.S2"/>. The following two columns contain the
coordinates of the species occurrence data. Finally, the last columns
contain the climate values to be reconstructed. An optional column
called <monospace>weight</monospace> can be added to <monospace>distributions</monospace> in the
fifth position (i.e. between the coordinates and the climate
variables) if one wants to weigh the different observations. For
example, the (relative) abundance of taxa observed from modern proxy
assemblages can be used when fitting the PDFs to give more importance to
the observations where that abundance is highest. This could also be
used if accurate abundance data were available instead of presence-only
data. The weights take a value between 0 and 100.</p>

<?xmltex \floatpos{t}?><table-wrap id="Ch1.T4" specific-use="star"><?xmltex \currentcnt{4}?><label>Table 4</label><caption><p id="d1e3693">Template for the <italic>distributions</italic> data frame. The weights column, here indicated with an asterisk, is optional and can be omitted or its values all set to 1 to assign the same weight to each observation. The number of rows of the table should correspond to the number of unique occurrences available. The dots indicate that many more rows, and possibly columns, are expected in a real <italic>distributions</italic> table.</p></caption><oasis:table frame="topbot"><oasis:tgroup cols="8">
     <oasis:colspec colnum="1" colname="col1" align="left"/>
     <oasis:colspec colnum="2" colname="col2" align="left"/>
     <oasis:colspec colnum="3" colname="col3" align="right"/>
     <oasis:colspec colnum="4" colname="col4" align="right"/>
     <oasis:colspec colnum="5" colname="col5" align="right"/>
     <oasis:colspec colnum="6" colname="col6" align="right"/>
     <oasis:colspec colnum="7" colname="col7" align="right"/>
     <oasis:colspec colnum="8" colname="col8" align="right"/>
     <oasis:thead>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">Species name</oasis:entry>
         <oasis:entry colname="col2">Taxon name</oasis:entry>
         <oasis:entry colname="col3">Longitude</oasis:entry>
         <oasis:entry colname="col4">Latitude</oasis:entry>
         <oasis:entry colname="col5">Weight<inline-formula><mml:math id="M83" display="inline"><mml:msup><mml:mi/><mml:mo>*</mml:mo></mml:msup></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col6">clim_1</oasis:entry>
         <oasis:entry colname="col7"><inline-formula><mml:math id="M84" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col8">clim_n</oasis:entry>
       </oasis:row>
     </oasis:thead>
     <oasis:tbody>
       <oasis:row>
         <oasis:entry colname="col1"><italic>Stoebe plumosa</italic></oasis:entry>
         <oasis:entry colname="col2"><italic>Stoebe</italic>-type</oasis:entry>
         <oasis:entry colname="col3">18.875</oasis:entry>
         <oasis:entry colname="col4"><inline-formula><mml:math id="M85" display="inline"><mml:mo>-</mml:mo></mml:math></inline-formula>34.375</oasis:entry>
         <oasis:entry colname="col5">20</oasis:entry>
         <oasis:entry colname="col6">15.8</oasis:entry>
         <oasis:entry colname="col7"><inline-formula><mml:math id="M86" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col8">711</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1"><italic>Elytropappus rhinocerotis</italic></oasis:entry>
         <oasis:entry colname="col2"><italic>Stoebe</italic>-type</oasis:entry>
         <oasis:entry colname="col3">18.375</oasis:entry>
         <oasis:entry colname="col4"><inline-formula><mml:math id="M87" display="inline"><mml:mo>-</mml:mo></mml:math></inline-formula>33.625</oasis:entry>
         <oasis:entry colname="col5">32</oasis:entry>
         <oasis:entry colname="col6">16.9</oasis:entry>
         <oasis:entry colname="col7"><inline-formula><mml:math id="M88" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col8">477</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1"><inline-formula><mml:math id="M89" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col2"><inline-formula><mml:math id="M90" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col3"><inline-formula><mml:math id="M91" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col4"><inline-formula><mml:math id="M92" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col5"><inline-formula><mml:math id="M93" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col6"><inline-formula><mml:math id="M94" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col7"><inline-formula><mml:math id="M95" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col8"><inline-formula><mml:math id="M96" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1"><italic>Elaeis guineensis</italic></oasis:entry>
         <oasis:entry colname="col2"><italic>Elaeis guineensis</italic></oasis:entry>
         <oasis:entry colname="col3"><inline-formula><mml:math id="M97" display="inline"><mml:mo>-</mml:mo></mml:math></inline-formula>4.375</oasis:entry>
         <oasis:entry colname="col4">10.875</oasis:entry>
         <oasis:entry colname="col5">4</oasis:entry>
         <oasis:entry colname="col6">27.4</oasis:entry>
         <oasis:entry colname="col7"><inline-formula><mml:math id="M98" display="inline"><mml:mi mathvariant="normal">…</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col8">1020</oasis:entry>
       </oasis:row>
     </oasis:tbody>
   </oasis:tgroup></oasis:table></table-wrap>

</sec>
<sec id="Ch1.S4.SS3.SSS4">
  <label>4.3.4</label><?xmltex \opttitle{\texorpdfstring{The \texttt{climate\_ space} data
frame}{The climate\_ space data frame}}?><title>The <monospace>climate_space</monospace> data
frame</title>
      <p id="d1e3976">This data frame is only necessary if the users use a personal
calibration dataset (<monospace>distributions</monospace>) instead of the
<italic>gbif4crest</italic> dataset. This data frame enables (1) using the climate
space weighting option (Sect. <xref ref-type="sec" rid="Ch1.S2.SS2"/>) and (2)
including plots of modern climate in the different diagnostic tools. Its
structure is straightforward, with the first two columns containing
longitudes and latitudes and the subsequent columns the climate
variables to reconstruct. The spatial resolution and the ordering of the
climate variables should be identical to the <monospace>distributions</monospace>
table (Table <xref ref-type="table" rid="Ch1.T4"/>). However, the arrangement of the
rows is not important. Including a <monospace>climate_space</monospace> dataset is
recommended, even if it is not mandatory.</p><?xmltex \hack{\newpage}?>
</sec>
<sec id="Ch1.S4.SS3.SSS5">
  <label>4.3.5</label><?xmltex \opttitle{\texorpdfstring{The \texttt{selectedTaxa} data
frame}{The selectedTaxa data frame}}?><title>The <monospace>selectedTaxa</monospace> data
frame</title>
      <p id="d1e4009">The last data frame that may be used to inform the reconstruction is a
data frame of ones and zeros called <monospace>selectedTaxa</monospace>. This data
frame has as many rows and columns as there are taxa and climate
variables, respectively. Each entry, which should be either 1 or 0,
indicates if the taxon should be used to reconstruct the climate
variable (value <inline-formula><mml:math id="M99" display="inline"><mml:mrow><mml:mo>=</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:math></inline-formula>) or not (value <inline-formula><mml:math id="M100" display="inline"><mml:mrow><mml:mo>=</mml:mo><mml:mn mathvariant="normal">0</mml:mn></mml:mrow></mml:math></inline-formula>). If a <monospace>selectedTaxa</monospace> data
frame is not provided, a default data frame with all entries set to 1 is
added to the <monospace>crestObj</monospace> at initialisation. Users can then modify
this information at any point using the <monospace>includeTaxa()</monospace> and
<monospace>excludeTaxa()</monospace> built-in functions. The
<monospace>crest.get_modern_data()</monospace> function also modifies this data
frame by setting the value to <inline-formula><mml:math id="M101" display="inline"><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:math></inline-formula> when the <monospace>PSE</monospace> classification
failed for a taxon or when the amount of data in the study area is
insufficient to fit a reliable PDF (see the parameter description in
Sect. <xref ref-type="sec" rid="Ch1.S5.SS2"/>).</p>
</sec>
</sec>
<sec id="Ch1.S4.SS4">
  <label>4.4</label><title>Package dependencies</title>
      <p id="d1e4076">The <italic>crestr</italic> package is built in R <xref ref-type="bibr" rid="bib1.bibx60" id="paren.45"/> using the
<monospace>devtools</monospace> package <xref ref-type="bibr" rid="bib1.bibx72" id="paren.46"/>. <italic>crestr</italic> depends
on numerous packages: <monospace>clipr</monospace> <xref ref-type="bibr" rid="bib1.bibx53" id="paren.47"/>,
<monospace>DBI</monospace> <xref ref-type="bibr" rid="bib1.bibx61" id="paren.48"/>, <monospace>openxlsx</monospace> <xref ref-type="bibr" rid="bib1.bibx63" id="paren.49"/>,
<monospace>pals</monospace> <xref ref-type="bibr" rid="bib1.bibx74" id="paren.50"/>, <monospace>plot3D</monospace> <xref ref-type="bibr" rid="bib1.bibx64" id="paren.51"/>,
<monospace>plyr</monospace> <xref ref-type="bibr" rid="bib1.bibx68" id="paren.52"/>, <monospace>raster</monospace> <xref ref-type="bibr" rid="bib1.bibx47" id="paren.53"/>,
<monospace>rgdal</monospace> <xref ref-type="bibr" rid="bib1.bibx4" id="paren.54"/>, <monospace>rgeos</monospace> <xref ref-type="bibr" rid="bib1.bibx3" id="paren.55"/>,
<monospace>RPostgres</monospace> <xref ref-type="bibr" rid="bib1.bibx73" id="paren.56"/>, <monospace>scales</monospace>
<xref ref-type="bibr" rid="bib1.bibx71" id="paren.57"/>, <monospace>sp</monospace> <xref ref-type="bibr" rid="bib1.bibx58 bib1.bibx5" id="paren.58"/>,
<monospace>stringr</monospace> <xref ref-type="bibr" rid="bib1.bibx69" id="paren.59"/> and <monospace>viridis</monospace>
<xref ref-type="bibr" rid="bib1.bibx28" id="paren.60"/>. The dedicated documentation, tutorials and
application examples found at <uri>https://mchevalier2.github.io/crestr</uri> (last access: February 2022) were
generated and formatted by the package <monospace>pkgdown</monospace>
<xref ref-type="bibr" rid="bib1.bibx70" id="paren.61"/>.</p><?xmltex \hack{\newpage}?>
</sec>
</sec>
<sec id="Ch1.S5">
  <label>5</label><?xmltex \opttitle{\texorpdfstring{Step-by-step user guide for
\textit{crestr}}{Step-by-step user guide for crestr}}?><title>Step-by-step user guide for
<italic>crestr</italic></title>
<sec id="Ch1.S5.SS1">
  <label>5.1</label><title>Example application: pollen-based mean annual temperature
reconstructions from marine core
MD96-2048</title>
      <p id="d1e4214">To illustrate the different ways of using <italic>crestr</italic> and its
graphical diagnostic tools, I use pollen data recently analysed with the
original CREST software to reconstruct mean annual temperature (MAT)
from marine core MD96-2048. The core was retrieved off the coast of
South Africa and Mozambique near the mouth of the Limpopo River. The
terrestrial sediments are expected to come from the entire catchment of
the Limpopo River and the smaller local river catchments near the coast
<xref ref-type="bibr" rid="bib1.bibx24 bib1.bibx23 bib1.bibx7" id="paren.62"/>. The
MAT reconstruction is based on 181 fossil pollen samples and the
percentages of more than 150 terrestrial pollen taxa and covers the last
790 000 years.</p>
      <p id="d1e4223">As the catchment of these marine sediments is large, an extensive
calibration dataset covering all vegetation zones from tropical Africa
to the temperate southwestern tip of South Africa was designed to
prevent any artificial reduction in the possible range of variability of
the reconstruction. The glacial–interglacial trends and amplitude of the
MAT reconstructions were validated by comparing them with regional
temperature records and other global indicators of glacial–interglacial
temperature variability (e.g. Antarctic temperature or global
sea-level curves, see <xref ref-type="bibr" rid="bib1.bibx19" id="altparen.63"/>, for more details).
Here, I reproduce this reconstruction using the original
parameterisation of the CREST algorithm obtained from the original CREST
point-and-click interface to showcase how it can easily be replicated in
a few lines of code with the <italic>crestr</italic> package. Due to the update of
the climate data of the calibration dataset described in Sect. <xref ref-type="sec" rid="Ch1.S3"/>, marginal differences in the order of tenths of
degrees Celsius can, however, be observed between the original
publication and this reproduction. All the necessary datasets and R code
are available in the Supplement.</p>
</sec>
<sec id="Ch1.S5.SS2">
  <label>5.2</label><?xmltex \opttitle{\texorpdfstring{Formatting the calibration data in a
\texttt{crestObj}}{Formatting the calibration data in a crestObj}}?><title>Formatting the calibration data in a
<monospace>crestObj</monospace></title>
      <p id="d1e4245">As the <italic>gbif4crest</italic> dataset was used to fit the PDFs, the function
called <monospace>crest.get_modern_data()</monospace> was used to extract the
calibration data:</p>
      <p id="d1e4254"><preformat><![CDATA[rcnstrctn <- crest.get_modern_data(
  df = MD96_2048,
  pse = PSE,
  selectedTaxa = selectedTaxa,
  taxaType = 1,
  climate = c('bio1'),
  xmn = NA, xmx = NA,
  ymn = -35, ymx = NA,
  continents = 'Africa',
  countries = c('South Africa', 'Kenya',
    'Lesotho', 'eSwatini', 'Botswana',
    'Mozambique', 'Zimbabwe', 'Zambia',
    'Malawi', 'Tanzania', 'Namibia',
    'Uganda', 'Rwanda', 'Burundi'),
  realms = NA,
  biomes = NA,
  ecoregions = NA,
  minGridCells = 20,
  site_info = c(34.0167, -26.1667),
  site_name = 'MD96-2048',
  dbname = "gbif4crest_02",
  verbose = TRUE)]]></preformat></p>
      <p id="d1e4258">All the parameters of the function were defined by the characteristics
of the proxy (pollen), the climate to reconstruct (MAT/bio1) and the
definition of the study area (eastern and southern Africa). The three input
files (i.e. <monospace>df</monospace>, <monospace>PSE</monospace> and <monospace>selectedTaxa</monospace>;
see Sect. <xref ref-type="sec" rid="Ch1.S4.SS3"/>) required to realise this
reconstruction are reproduced from the published dataset
<xref ref-type="bibr" rid="bib1.bibx17" id="paren.64"/> and are available in the Supplement.
The following points describe the different parameters of the
<monospace>crest.get_modern_data()</monospace> function and how they relate to the
data extraction and modelling.
<list list-type="bullet"><list-item>
      <p id="d1e4281">The parameter <monospace>taxaType</monospace> is used to choose the type of proxy  used and takes a value between 1 to 6 for plants (i.e. pollen and plant macro remains), beetles, chironomids, foraminifers, diatoms and rodents, respectively.</p></list-item><list-item>
      <p id="d1e4288">The name(s) or code(s) of the <monospace>climate</monospace> variables to study  should be provided here (see Tables <xref ref-type="table" rid="Ch1.T1"/> and <xref ref-type="table" rid="Ch1.T2"/> or use the function  <monospace>accClimateVariables()</monospace> for a list of accepted names). Here,  “bio1” means mean annual temperature. More variables can be added if  necessary (e.g. <monospace>c('bio1', 'ai')</monospace>). However, serial reconstructions should be avoided, even if many variables are provided with this package. Careful interpretations of the fossil data should be made before selecting variables.</p></list-item><list-item>
      <p id="d1e4305">Geographical parameters can be provided to tailor the  <italic>gbif4crest</italic> dataset to the study area and data. These can  include minimum and maximum longitude and latitude (<monospace>xmn</monospace>,  <monospace>xmx</monospace>, <monospace>ymn</monospace>, <monospace>ymx</monospace>), continent, ocean or country  names (see <monospace>accCountryNames()</monospace> and <monospace>accBasinNames()</monospace> for  a list of accepted names), and also some ecological classifiers, such  as realms, biomes or ecoregions (see <monospace>accRealmNames()</monospace> for a  list of official names). Only the occurrences that respect all the  specified constraints will be returned.</p></list-item><list-item>
      <p id="d1e4334">To estimate reliable species PDFs, it is recommended to use at least  20 distinct occurrences for each species, even if different values can  be specified with the <monospace>minGridCell</monospace> parameter, depending on the  density of data available across the study area.</p></list-item><list-item>
      <p id="d1e4341">Optional information about the site, such as a name and coordinates,  can be provided and, where possible, this information will be  represented on the different graphical diagnostic tools created by  <italic>crestr</italic> (e.g. the location of the record is added to the  maps if the coordinates are provided).</p></list-item></list></p>
      <p id="d1e4348">The <monospace>crest.get_modern_data()</monospace> function reads all the data and
parameters, extracts the data from the cloud-based <italic>gbif4crest</italic>
database, processes the distribution data and returns everything as a
structured <monospace>crestObj</monospace> – here called <monospace>rcnstrctn</monospace> and whose
structure is displayed in Fig. <xref ref-type="fig" rid="Ch1.F5"/> – that will be read
and modified by all subsequent functions. Alternatively, the function
<monospace>crest.set_modern_data()</monospace> could be called instead of
<monospace>crest.get_modern_data()</monospace> to use personal calibration data
instead of the <italic>gbif4crest</italic> database. If the calibration data
needed for this study were available as a <monospace>distributions</monospace> and
<monospace>climate_space</monospace> data frames (Sect. <xref ref-type="sec" rid="Ch1.S4.SS3"/>),
similar results would be obtained with
<preformat><![CDATA[rcnstrctn <- crest.set_modern_data(
  df = MD96_2048,
  distributions = distributions,
  climate_space = climate_space,
  selectedTaxa = selectedTaxa,
  climate = c('bio1'),
  weight = FALSE,
  minGridCells = 20,
  site_info = c(34.0167, -26.1667),
  site_name = 'MD96-2048',
  verbose = TRUE)]]></preformat></p>
</sec>
<sec id="Ch1.S5.SS3">
  <label>5.3</label><title>Estimating the climate responses (the PDFs)</title>
      <p id="d1e4394">The probabilistic proxy–climate responses, i.e. the PDFs, are
estimated from the presence-only data using the
<monospace>crest.calibrate()</monospace> function:</p>
      <p id="d1e4400"><preformat><![CDATA[rcnstrctn <- crest.calibrate(
  rcnstrctn,
  shape = c('normal'),
  climateSpaceWeighting = TRUE,
  bin_width = c(2),
  npoints = 500,
  geoWeighting = TRUE,
  verbose = TRUE)]]></preformat></p>
      <p id="d1e4404">As described in Sect. <xref ref-type="sec" rid="Ch1.S2.SS2"/>, the different
parameters that control the reconstruction should be carefully
considered to produce reliable PDFs. These include specifying
<list list-type="bullet"><list-item>
      <p id="d1e4411">the <monospace>shape</monospace> of the species PDFs, which should be either  “normal” or “lognormal”, depending on the variable to reconstruct (see Sect. <xref ref-type="sec" rid="Ch1.S2.SS2"/>).</p></list-item><list-item>
      <p id="d1e4420">the width of the climate bins (<monospace>bin_width</monospace>) expressed in the  variables' units (e.g. 2 <inline-formula><mml:math id="M102" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula> or 50 mm) if the PDFs are corrected for an heterogeneous climate space (<monospace>climateSpaceWeighting</monospace> set to <monospace>TRUE</monospace>). Dividing the  total climate space in 15–25 bins often leads to good results, but  other values are possible.</p></list-item><list-item>
      <p id="d1e4445">the number of intervals required to divide the studied climate range  and fit the PDFs <monospace>npoints</monospace>. This will ultimately define the  climate resolution of the reconstructions. <italic>crestr</italic> runs faster  with lower values, but this can alter the reconstructions with visible  “jumps” between consecutive climate values (aliasing effect).</p></list-item><list-item>
      <p id="d1e4455">set <monospace>geoWeighting</monospace> to <monospace>TRUE</monospace> if the species PDFs of the  different composing species should be weighted according to the square root of the extent of their modern distribution (the  <inline-formula><mml:math id="M103" display="inline"><mml:msqrt><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi>s</mml:mi></mml:msub></mml:mrow></mml:msqrt></mml:math></inline-formula> in Eq. <xref ref-type="disp-formula" rid="Ch1.E6"/>).</p></list-item></list></p>
</sec>
<sec id="Ch1.S5.SS4">
  <label>5.4</label><title>Assessing the coherency of the climate space</title>
      <p id="d1e4487">In every study involving estimating relationships between biological
entities and environmental parameters, the first step is to ensure that
the defined study area and associated calibration dataset are coherent.
This includes ensuring that (1) all the essential taxa observed in the
past are present in the study area, and their distributions are not
truncated, (2) the climate values to reconstruct are likely to be covered
by present-day climate values (the reconstructions are bounded by the
lowest and highest values observed in the modern climate space) and (3)
there is no large sampling or representativity bias (e.g. along
country borders due to different sampling efforts). The “climateSpace”
graphical diagnostic tool (Fig. <xref ref-type="fig" rid="Ch1.F6"/>) was designed for
a rapid assessment of all these characteristics:</p>
      <p id="d1e4492"><preformat><![CDATA[plot_climateSpace(rcnstrctn, save=TRUE,
  filename='Figure 6.png',
  as.png=TRUE, png.res=600,
  width=6.9, height=4.4,
  y0=0.4,
  add_modern=TRUE)]]></preformat></p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F6" specific-use="star"><?xmltex \currentcnt{6}?><?xmltex \def\figurename{Figure}?><label>Figure 6</label><caption><p id="d1e4498">“climateSpace”
graphical diagnostic tool to evaluate the calibration dataset. The
map on the right <bold>(d)</bold> represents the density of unique species
occurrences in each grid cell, highlighting a certain bias towards
South Africa. The lower abundance of plant data available from
Angola and the Democratic Republic of Congo (data not shown) is the
reason why these two countries were excluded from the study
area. The map on the left <bold>(a)</bold> represents the studied climate variable (MAT) across the study area. The double histogram in the middle represents the distribution of MAT (the climate space in grey,  <bold>b</bold>), while the black histogram  <bold>(c)</bold> represents how the calibration data sample the climate space. Differences between the two histograms can be used to identify biases in the calibration dataset. Here, the small shift of the black histograms towards colder values (towards the left) is another way of seeing that more data are available from South Africa than other countries and reflects, in part, the regional patterns of biodiversity. If more variables had been selected in this study, additional rows would be added to the figure with a similar climate map and histograms and a scatterplot of the climate variables to highlight potential local or regional modern correlations.</p></caption>
          <?xmltex \igopts{width=497.923228pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f06.png"/>

        </fig>

      <p id="d1e4520">Ideally, the climate values sampled by the calibration data should be as
homogeneous as possible to ensure proper representation of all the
possible climate values, even if the extreme climate values will always
be under-represented compared to the median ones. However, deviations
from a theoretical equivalence between the observed climate distribution
and the climate values sampled by the calibration data are not
necessarily a bad characteristic. In our case study, the variability of
the sampling density represents actual patterns in regional species
diversity with the presence of several biodiversity hotspots across the
mountainous regions of eastern and southern Africa
<xref ref-type="bibr" rid="bib1.bibx55" id="paren.65"/>. This higher diversity in the colder areas
explains why the black histogram (i.e. the climate values
associated with occurrence data) in Fig. <xref ref-type="fig" rid="Ch1.F6"/> is
skewed towards the left compared to the grey histogram (i.e. the
distribution of the climate space in the study area). All these elements
should be checked and accounted for while designing the final
calibration dataset.</p>
      <p id="d1e4528">The “climateSpace” diagnostic figure is also practical for identifying
potential local or global correlations between different climate
variables and assessing the risk of confounding variables (i.e.
variables that are correlated with important variables but do not
directly impact the studied proxies; <xref ref-type="bibr" rid="bib1.bibx50 bib1.bibx18" id="altparen.66"/>). Any change to the parameters related to
the definition of the climate space (e.g. definition of the study
area, climate variables to reconstruct) will require re-running
<monospace>crest.get_modern_data()</monospace> or <monospace>crest.set_modern_data()</monospace>
with updated parameters and/or data.</p>
</sec>
<sec id="Ch1.S5.SS5">
  <label>5.5</label><title>Assessing the coherency of the climate responses</title>
      <p id="d1e4548">With the study area and climate space defined, the next step is to
search for taxa that show specific relationships with climate. While all
species eventually respond to all climate variables, they can be more
sensitive to one over another within a given region. The low taxonomic
resolution of some fossil proxies, such as pollen data, can also mask
strong species–climate relationships <xref ref-type="bibr" rid="bib1.bibx20" id="paren.67"/>. Looking
at each individual climate response(s) and assessing their significance
within the boundary conditions of the study is thus critical. The
“taxaCharacteristics” diagnostic plot (Fig. <xref ref-type="fig" rid="Ch1.F7"/>) was designed for this task:</p>
      <p id="d1e4556"><preformat><![CDATA[plot_taxaCharacteristics(rcnstrctn,
  taxanames='Ericaceae', save = TRUE,
  filename = 'Figure 7.png',
  as.png=TRUE, png.res=600,
  width=6.9, height=8.13,
  add_modern=TRUE)]]></preformat></p>

      <?xmltex \floatpos{p}?><fig id="Ch1.F7" specific-use="star"><?xmltex \currentcnt{7}?><?xmltex \def\figurename{Figure}?><label>Figure 7</label><caption><p id="d1e4562">“taxaCharacteristics”
graphical diagnostic tool to assess the sensitivity of taxa to
climate. In the top row <bold>(a, b)</bold>, the map <bold>(a)</bold>
represents the density of unique species occurrences per grid cell
derived from the modern calibration dataset, and the time series
<bold>(b)</bold> represents the variability of the taxon against time or
depth. The bottom row <bold>(c–e)</bold> is specific to each selected variable (only one row in this case). The map <bold>(c)</bold> represents the geographical distribution of the taxon (in white) against the climate background. The histogram <bold>(d)</bold> represents the climate space (in colour) and how this climate is sampled by the taxon (in black). Finally, the right plot <bold>(e)</bold> represents the climate response of the taxon (in black) and the response of all the composing species (in grey). Overall, this figure highlights that despite its high diversity, Ericaceae is primarily associated with the colder environments of the study area and its presence increases during glacial periods. The red diamond indicates the location of the studied record.</p></caption>
          <?xmltex \igopts{width=463.779921pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f07.png"/>

        </fig>

      <?xmltex \floatpos{t}?><fig id="Ch1.F8" specific-use="star"><?xmltex \currentcnt{8}?><?xmltex \def\figurename{Figure}?><label>Figure 8</label><caption><p id="d1e4596">“violinPDFs” graphical diagnostic tool to represent the PDFs of various taxa (here a random selection of 15 taxa). The taxa are sorted and colour-coded by their temperature optima (i.e. the temperature corresponding to the peak of their PDFs).</p></caption>
          <?xmltex \igopts{width=497.923228pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f08.png"/>

        </fig>

      <p id="d1e4605">One summary plot can be generated for every taxon to assess and
inter-compare their geographical distributions and climate responses. As
illustrated in Fig. <xref ref-type="fig" rid="Ch1.F7"/>, Ericaceae is
preferentially observed in the colder environments of the study area,
its higher percentages occur during glacial periods, and a coherent
response of all its composing species can be observed despite a high
diversity (141 species with at least 20 unique occurrences across the
study area). All these elements indicate that Ericaceae can be
considered a cold environment indicator in eastern and southern Africa.
Sensitivities to other variables can also be expected but are not
considered in this study. Similar sensitivity inferences can be made to
define a list of temperature-sensitive taxa to reconstruct MAT
<xref ref-type="bibr" rid="bib1.bibx14 bib1.bibx19" id="paren.68"/> or support qualitative
interpretations of palaeoecological datasets
<xref ref-type="bibr" rid="bib1.bibx20 bib1.bibx59" id="paren.69"/>. A complementary
diagnostic plot to assess climate sensitivities is the “violinPDFs”
(Fig. <xref ref-type="fig" rid="Ch1.F8"/>):</p>
      <p id="d1e4618"><preformat><![CDATA[tax <- sample(rcnstrctn$input$taxa.name,
              15, replace=FALSE)

plot_violinPDFs(rcnstrctn,
  taxanames = tax,
  save = TRUE,
  filename = 'Figure 8.png',
  as.png=TRUE, png.res=600,
  width = 6.9, height = 3,
  ylim=c(0,30))]]></preformat></p>
      <p id="d1e4622">The violin plot represents the PDFs of a selection of taxa on the same
plot, which helps compare the shape and spread of the different
responses. All the violins have the same area (all the probabilities sum
to 1), and the taxa are ranked by increasing values of their temperature
optima (i.e. the temperature corresponding to the peak of the
PDF). However, due to the possible multimodality of the PDFs and
differences in tolerance ranges, this ranking does not mean that a taxon
on the left always represents colder conditions than a taxon on the
right. This is illustrated in many ways in Fig. <xref ref-type="fig" rid="Ch1.F8"/>
with, for example, <italic>Cassia</italic>-type that is estimated to experience
warmer conditions than <italic>Coffea</italic>-type <inline-formula><mml:math id="M104" display="inline"><mml:mrow><mml:mo>∼</mml:mo><mml:mn mathvariant="normal">61</mml:mn></mml:mrow></mml:math></inline-formula> % of the
time (based on 100 000 random draws from their PDFs) despite having a
“colder” climate optimum, or with <italic>Diospyros</italic> that can tolerate
much warmer conditions than most taxa with warmer optima. This type of
representation can be beneficial to make more informed interpretations
of ecological changes from pollen diagrams
<xref ref-type="bibr" rid="bib1.bibx20 bib1.bibx59" id="paren.70"/>.</p>
</sec>
<sec id="Ch1.S5.SS6">
  <label>5.6</label><title>Reconstructing climate</title>
      <p id="d1e4658">Along with the <monospace>df</monospace> data frame provided to the
<monospace>crest.get_modern_data()</monospace> or <monospace>crest.set_modern_data()</monospace>
functions, a set of reconstruction parameters have to be chosen to
combine PDFs and estimate climate parameters. The <monospace>selectedTaxa</monospace>
data frame stored in the <monospace>crestObj</monospace> (see Fig. <xref ref-type="fig" rid="Ch1.F5"/>)
defines the taxa that will be used to reconstruct each climate variable
(by default, all taxa are included if sufficient data are available to
fit a PDF). This selection can be modified by using the
<monospace>includeTaxa()</monospace> and <monospace>excludeTaxa()</monospace> functions. For
instance, both Aizoaceae and Chenopodiaceae/Amaranthaceae were excluded
from the reconstruction because these taxa are not primarily sensitive
to temperature in southern Africa. This can be simply done as follows:
<preformat><![CDATA[rcnstrctn <- excludeTaxa(rcnstrctn,
  taxa=c('Aizoaceae',
     'Chenopodiaceae/Amaranthaceae'),
  climate=c('bio1'))]]></preformat></p>
      <p id="d1e4687">Climate reconstructions are then performed with the
<monospace>crest.reconstruct()</monospace> function:
<preformat><![CDATA[rcnstrctn <- crest.reconstruct(
  rcnstrctn,
  presenceThreshold = 0,
  taxWeight = "normalisation",
  verbose = TRUE)]]></preformat></p>
      <p id="d1e4695">A minimum “presence threshold” below which the taxa will always be
considered absent can be provided to reduce the noise of the fossil
dataset <xref ref-type="bibr" rid="bib1.bibx18" id="paren.71"><named-content content-type="pre">e.g. pollen percentages lower than 1 % or 2 % are commonly
excluded from reconstructions,</named-content></xref>. When
<monospace>presenceThreshold</monospace> is set to zero, as is the case here, all the
strictly positive pollen percentages are considered as actual presences
and used accordingly to reconstruct MAT. To weigh the taxa as described
in Eq. (<xref ref-type="disp-formula" rid="Ch1.E7"/>), four options are available in <italic>crestr</italic>:
<list list-type="bullet"><list-item>
      <p id="d1e4713">The data can be converted to presence/absence with all the values
above and below <monospace>presenceThreshold</monospace> being changed to ones and
zeros, respectively. This option is recommended for data such as
macrofossils for which relative abundances cannot be reliably
estimated.</p></list-item><list-item>
      <p id="d1e4720">The data can be converted to percentages to weigh the taxa according
to their relative abundance. This option is recommended for data where
reliable proportions can be estimated.</p></list-item><list-item>
      <p id="d1e4724">The data can be normalised following the method proposed by
<xref ref-type="bibr" rid="bib1.bibx16" id="text.72"/> and described here by Eq. (<xref ref-type="disp-formula" rid="Ch1.E8"/>). This option is recommended for
palaeoecological proxies where the observed percentages are not
proportional to their abundance in the environment, such as pollen
data.</p></list-item><list-item>
      <p id="d1e4733">The data can be directly weighted by the values provided in
<monospace>df</monospace>, which implies that users can define their own specific
weighting strategy (e.g. using the square-root transformation of the
pollen percentages).</p></list-item></list></p>
</sec>
<sec id="Ch1.S5.SS7">
  <label>5.7</label><title>Analysing and understanding the reconstruction(s)</title>
      <p id="d1e4747">Characterising the dominant factors that define a reconstruction can be
difficult. This section presents three different graphical diagnostic
tools that provide different perspectives on the reconstructed data and
help look inside the statistical black box. First, the full
probabilistic breadth of the reconstructions can be represented by using
the standard R <monospace>plot()</monospace> function (Fig. <xref ref-type="fig" rid="Ch1.F9"/>), which
has been adapted to plot data stored in a <monospace>crestObj</monospace>:
<preformat><![CDATA[plot(rcnstrctn,
  filename='Figure 9.png',
  save=TRUE,
  as.png=TRUE, png.res=600,
  width=6.9, height=3.54,
  ylim=c(14,23), uncertainties=1,
  simplify=FALSE,
  col=plot3D::gg2.col(200)[1:100],
  pt.cex=0.8, pt.lwd=1, pt.col='#2c2c2c')]]></preformat></p>

      <?xmltex \floatpos{p}?><fig id="Ch1.F9" specific-use="star"><?xmltex \currentcnt{9}?><?xmltex \def\figurename{Figure}?><label>Figure 9</label><caption><p id="d1e4762">MAT probabilistic reconstructions for marine core MD96-2048. The yellow–green–blue colour gradient represents the uncertainties associated with each sample. This reconstruction is identical to the reconstruction presented in Chevalier et al. (2021), and the reader is referred to this publication for an in-depth validation and discussion of these results.</p></caption>
          <?xmltex \igopts{width=497.923228pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f09.png"/>

        </fig>

      <?xmltex \floatpos{p}?><fig id="Ch1.F10" specific-use="star"><?xmltex \currentcnt{10}?><?xmltex \def\figurename{Figure}?><label>Figure 10</label><caption><p id="d1e4773">“combinedPDFs” graphical diagnostic tool that shows the combination of the PDFs (in colour) of all the taxa recorded in the sample dated at 64 kyr. The thickness of the lines is related to the weight of the taxa in the sample (absolute value indicated next to each taxon name). The black curve represents the MAT reconstruction, from which a “best” climate estimate can be estimated from the maximum of the curve (grey dashed line) and uncertainties derived by calculating the area under the curve.</p></caption>
          <?xmltex \igopts{width=497.923228pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f10.png"/>

        </fig>

      <p id="d1e4783">While the probabilistic representation of the data should be preferred
because it represents all the information available, a simpler version
of this plot with the climate optima and more common uncertainty ranges
expressed as colour bands can be obtained by specifying
<monospace>simplify=TRUE</monospace> (not shown).</p>
      <p id="d1e4789">The <monospace>plot_combinedPDFs()</monospace> function can then be used to identify
the taxa that are driving the reconstruction by focusing on the sample
level:</p>
      <p id="d1e4795"><preformat><![CDATA[plot_combinedPDFs(rcnstrctn,
    samples=3, only.present=TRUE,
    only.selected=TRUE, save = TRUE,
    filename = 'Figure 10.png',
    as.png=TRUE, png.res=600,
    width=6.9, height=3,
    xlim=c(7.5,32.5))]]></preformat></p>
      <p id="d1e4799">In this plot (Fig. <xref ref-type="fig" rid="Ch1.F10"/>), the PDFs of all the taxa
present in the sample and selected to reconstruct the climate variable
are represented along with the reconstruction. This type of plot can
help identify if a particular PDF is at odds with the general
assemblage, which usually indicates the possible presence of a
confounding factor. It is also helpful to visualise the full spread of
the uncertainties and, by extension, highlight that reconstructions can
be multimodal. While it is not the case in the example here,
multimodality can be the underlying cause of apparent noise in the
reconstructions with minor changes in the taxa composition or
percentages forcing the system to oscillate between two maxima and thus
“appear” noisy. This effect could also be seen from the reconstruction
plot where the full uncertainties would be represented
(<monospace>simplify=FALSE</monospace>).</p>
      <p id="d1e4807">Finally, a standard post-processing analysis of CREST reconstructions is
a form of leave-one-out (LOO) analysis that is done with the
<monospace>loo()</monospace> function:
<preformat><![CDATA[rcnstrctn <- loo(rcnstrctn,
  verbose = TRUE)]]></preformat></p>
      <p id="d1e4815">In the CREST context, a LOO analysis consists of repeatedly
“unselecting” one taxon at a time, running the reconstruction without
that taxon and measuring the associated reconstruction anomalies. In our
example, a positive anomaly means that the reconstruction without the
taxon is warmer and, by extension, that the taxon is a cold indicator
relative to a specific assemblage. A detailed analysis of the results
can contribute to a deeper understanding of which taxa are the most
important in driving the reconstructed climate signal. Taxa that exhibit
large LOO values indicate a strong influence on the reconstruction.
However, it does not necessarily mean that they are strong climate
indicators. Large LOO values can arise when the PDFs are biased by
unaccounted factors and are, as a result, at odds with the rest of the
PDFs. Such factors can, for instance, include an incomplete estimate of
the climate responses, which induces a bias or shift of the climate
preferences, or a sensitivity of the taxon to other climatic
(e.g. aridity instead of temperature) or non-climatic
(e.g. edaphic conditions) factors. It is usually preferable to
exclude such taxa from the reconstruction (e.g. Chenopodiaceae
and Aizoaceae in our example).</p>
      <p id="d1e4819">The LOO analysis is a powerful tool to understand the taxa that
primarily drive the reconstruction, and the LOO results can be
represented as a typical stratigraphic diagram, where each row
illustrates the effect of removing a taxon from the reconstruction (Fig. <xref ref-type="fig" rid="Ch1.F11"/>):
<preformat><![CDATA[tax <- rcnstrctn$inputs$taxa.name[6:25]
plot_loo(rcnstrctn,
    taxanames=tax,
    xlim=c(0, 340), save = TRUE,
    filename = 'Figure 11.png',
    as.png=TRUE, png.res=600,
    width=3.5, height=8,
    bar_width=3, col_neg='coral3',
    col_pos='darkcyan')]]></preformat></p>
      <p id="d1e4826">For example, the expected cold effect of including Ericaceae on the MAT
reconstructions from the MD96-2048 pollen record is immediately visible
in Fig. <xref ref-type="fig" rid="Ch1.F11"/>. Similarly, the effect of removing either
Ericaceae or Caryophyllaceae undiff. from the sample dated at 64 kyr
(represented in Fig. <xref ref-type="fig" rid="Ch1.F10"/>) is quite strong with a
warming of the reconstruction of 1.19 and 0.44 <inline-formula><mml:math id="M105" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>, respectively. This
type of plot can thus be used for both global and sample-specific
inferences about the drivers of the reconstruction. Depending on the
vegetation composition, some taxa can even sometimes be categorised as
cold indicators when the assemblage represents warm conditions or warm
indicators when the assemblage represents cold conditions (e.g.
<italic>Hypoestes</italic>/<italic>Dicliptera</italic>-type in Fig. <xref ref-type="fig" rid="Ch1.F11"/>).</p>

      <?xmltex \floatpos{p}?><fig id="Ch1.F11"><?xmltex \currentcnt{11}?><?xmltex \def\figurename{Figure}?><label>Figure 11</label><caption><p id="d1e4856">Leave-one-out (LOO) graphical diagnostic tool to illustrate the influence of different taxa on the reconstructions. Here, the results are only shown for a subset of the taxa observed in marine core MD96-2048 (only 20 out of the 171 available taxa are represented). The height of each bar represents the absolute effect (in <inline-formula><mml:math id="M106" display="inline"><mml:mrow class="unit"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:math></inline-formula>) of removing the taxon from the reconstruction, and the sign of this effect (increase or decrease of the reconstructed temperature) is colour-coded. Here, blue (cf. Ericaceae) and red (cf. Combretaceae-type) bars indicate that the taxon is a cold and warm indicator, respectively, in the sample.</p></caption>
          <?xmltex \igopts{width=227.622047pt}?><graphic xlink:href="https://cp.copernicus.org/articles/18/821/2022/cp-18-821-2022-f11.png"/>

        </fig>

</sec>
<sec id="Ch1.S5.SS8">
  <label>5.8</label><title>A wrapper function</title>
      <p id="d1e4885">To simplify the use of the package, the three stages of the
reconstruction process – data acquisition
(<monospace>crest.get_modern_data()</monospace> or
<monospace>crest.set_modern_data()</monospace> if the <italic>gbif4crest</italic> dataset is
not used), calibration (<monospace>crest.calibrate()</monospace>) and reconstruction
(<monospace>crest.reconstruct()</monospace>) – can be called in one line of code
using the wrapper function <monospace>crest()</monospace>. This function takes the
same parameters described in this “step-by-step” guide with the same
default values and may be more practical when reconstructing several
records in one run.</p>
</sec>
<sec id="Ch1.S5.SS9">
  <label>5.9</label><title>Exporting the reconstructions</title>
      <p id="d1e4915">All the data stored in the <monospace>crestObj</monospace> can be easily exported from
the R environment as spreadsheets and RData files using either
<monospace>export_pdfs()</monospace> to save the climate responses of the studied
taxa or <monospace>export()</monospace> to save the reconstructions and many
associated data in a publishable format. The latter also saves the
<monospace>crestObj</monospace> as an RData file for easy reloading and sharing of the
data:</p>
      <p id="d1e4930"><preformat><![CDATA[export(rcnstrctn,
  loc='path/to/folder',
  fullUncertainties = TRUE,
  loo = TRUE,
  weights = TRUE,
  pdfs = TRUE)]]></preformat></p>
</sec>
<sec id="Ch1.S5.SS10">
  <label>5.10</label><title>Citing building elements</title>
      <p id="d1e4943">Finally, all the reconstructions derived from <italic>crestr</italic> are built on
numerous independent research efforts, including data compilations,
modelling projects, statistical developments and software engineering.
To support the long-term growth and visibility of all these building
elements, it is crucial always to acknowledge them, even if their
processing is invisible to the users. The list of references that must
be cited for each use of <italic>crestr</italic> is automatically included in the
summary tab of the spreadsheet generated by the <monospace>export()</monospace>
function. In addition, the citation information can also be directly
obtained from R:</p>
      <p id="d1e4955"><preformat><![CDATA[cite_crest(rcnstrctn)]]></preformat></p>
      <p id="d1e4959">The function looks at the type of data used in the analysis (e.g.
the subset of the GBIF data were used and the climate variables) and
returns a corresponding list of references. For example, a simple way of
crediting all the contributors for the MAT reconstruction from marine
core MD96-2048 presented here could be the following: To create
this MAT reconstruction from the pollen record from marine core
MD96-2048 <xref ref-type="bibr" rid="bib1.bibx24" id="paren.73"/>, we employed the CREST method
<xref ref-type="bibr" rid="bib1.bibx16 bib1.bibx10" id="paren.74"/>. The PDFs were
estimated by combining the MAT field of <xref ref-type="bibr" rid="bib1.bibx25" id="text.75"/> and
plant occurrence data from GBIF <xref ref-type="bibr" rid="bib1.bibx42 bib1.bibx40 bib1.bibx36" id="paren.76"/>. The numerical
analyses were realised with the crestr R package v1.0.1 (this
reference).</p>
</sec>
</sec>
<sec id="Ch1.S6" sec-type="conclusions">
  <label>6</label><title>Perspectives and conclusion</title>
      <p id="d1e4983">CREST is a probabilistic framework designed to model proxy–climate
relationships from modern presence observations and to use these
relationships to reconstruct past climate <xref ref-type="bibr" rid="bib1.bibx16" id="paren.77"/>.
The method's mathematics and assumptions enable an easy application
everywhere, even in data-sparse regions. While developments to
complexify the method are possible, the current version of the algorithm
has proven to be reliable at producing high-quality reconstructions. The
first version of the <italic>crestr</italic> R package is thus dedicated to
implementing the original version of the algorithm. It also replaces the
original point-and-click CREST software, which was challenging to
maintain over time. A critical novelty of <italic>crestr</italic> is the inclusion
of the CREST algorithm in a programmatic environment. The benefits of
this transition are significant and include improved scriptability
(i.e. the possibility of analysing many records automatically and
sequentially), reproducibility (i.e. the capacity to reproduce an
analysis) and better inter-operability (i.e. R packages are
compatible with all computer systems). However, maintaining the highest
level of accessibility remained at the core of the development process
and is illustrated in the final product by the small number of functions
necessary to run a complete analysis and the suite of detailed graphical
diagnostic figures. In line with the objectives of the original
software, the <italic>crestr</italic> package is aimed at all researchers
interested in using CREST to reconstruct climate from palaeoecological
datasets, including those with limited coding expertise.</p>
      <p id="d1e4998">In addition, its broad applicability will allow taking advantage of the
recent growth of curated, open-access fossil datasets that offer
unprecedented opportunities to reconstruct climate from a wide range of
proxies, particularly in regions where quantified climate
reconstructions are urgently needed. This package will thus also
contribute to the current transition from single-site to multi-site
studies, which is necessary to better understand past climate dynamics.
However, it is essential to remember that running such techniques on
several datasets should always be done carefully, as many factors can
impact the reconstruction process. In practice, calibration datasets,
variable selection and reconstructions should be assessed, ideally
against independent evidence when they are available, even if there is
no single way to validate a reconstruction (but see
<xref ref-type="bibr" rid="bib1.bibx18" id="altparen.78"/>, for some discussions on the generic
principles).</p>
      <p id="d1e5004">Over time, the <italic>crestr</italic> package will be enriched with new
functionalities to facilitate reconstruction validation. The online
documentation will also be updated with diverse examples and tutorials
based on real applications and assessments
(<uri>https://mchevalier2.github.io/crestr/</uri>, last access: February 2022). Future package updates will
include Bayesian modules and propose more complex strategies to estimate
proxy–climate relationships from data-dense regions. Finally, bug
reports, feedback, and suggestions for newer functionalities and
graphical diagnostic tools are encouraged and can be transmitted to the
author directly or through GitHub's bug report portal
(<uri>https://github.com/mchevalier2/crestr/issues</uri>, last access: February 2022).</p>
</sec>

      
      </body>
    <back><notes notes-type="codedataavailability"><title>Code and data availability</title>

      <p id="d1e5020">The <italic>crestr</italic> package is currently accessible
from CRAN (<uri>https://CRAN.R-project.org/package=crestr</uri>, last access: 13 April 2022, <xref ref-type="bibr" rid="bib1.bibx13" id="altparen.79"/>). The version used in this paper (v1.0.1) is available on Zenodo (<ext-link xlink:href="https://doi.org/10.5281/zenodo.6458405" ext-link-type="DOI">10.5281/zenodo.6458405</ext-link>, <xref ref-type="bibr" rid="bib1.bibx12" id="altparen.80"/>) and future development versions will be available on GitHub (<uri>https://github.com/mchevalier2/crestr</uri>, last access: February 2022), where I
also welcome feedback and strongly encourage contributions via the issue
tracker (<uri>https://github.com/mchevalier2/crestr/issues</uri>, last access: 13 April 2022). The source data
used for the example application (pollen record from marine core
MD96-2048 and associated temperature reconstructions) can be accessed
from <ext-link xlink:href="https://doi.org/10.1594/PANGAEA.915923" ext-link-type="DOI">10.1594/PANGAEA.915923</ext-link> <xref ref-type="bibr" rid="bib1.bibx17" id="paren.81"/>. NOAA
High-Resolution SST data were provided by the NOAA/OAR/ESRL PSL,
Boulder, Colorado, USA, from their website at
<uri>https://psl.noaa.gov/data/gridded/data.noaa.oisst.v2.highres.html</uri> (last access: September 2021, <xref ref-type="bibr" rid="bib1.bibx62" id="altparen.82"/>).</p>
  </notes><app-group>
        <supplementary-material position="anchor"><p id="d1e5057">The supplement related to this article is available online at: <inline-supplementary-material xlink:href="https://doi.org/10.5194/cp-18-821-2022-supplement" xlink:title="zip">https://doi.org/10.5194/cp-18-821-2022-supplement</inline-supplementary-material>.</p></supplementary-material>
        </app-group><notes notes-type="competinginterests"><title>Competing interests</title>

      <p id="d1e5066">The author has declared that there are no competing interests.</p>
  </notes><notes notes-type="disclaimer"><title>Disclaimer</title>

      <p id="d1e5072">Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.</p>
  </notes><ack><title>Acknowledgements</title><p id="d1e5078">I thank Brian Chase, Yoshi Maezumi, Lynne Quick, Kale Sniderman and, last but not least, Mine Altinli for their invaluable
feedback on early versions of the package and manuscript. And
finally, I would also like to address a special thanks to the <italic>rticles</italic>
contributors for their Climate of the Past R Markdown template!</p></ack><notes notes-type="financialsupport"><title>Financial support</title>

      <p id="d1e5086">This research has been supported by the Swiss National Science
Foundation (grant no. CRSK-2_195875), with additional support from the
University of Lausanne. Manuel Chevalier is also supported by the German Federal
Ministry of Education and Research (BMBF) as a Research for
Sustainability initiative (FONA; <uri>https://www.fona.de/en</uri>, last access: 15 December 2021) through the PalMod Phase II project (grant no. FKZ:
01LP1926D).<?xmltex \hack{\newline}?><?xmltex \hack{\newline}?>This open-access publication was funded <?xmltex \notforhtml{\newline}?> by the University of Bonn.</p>
  </notes><notes notes-type="reviewstatement"><title>Review statement</title>

      <p id="d1e5100">This paper was edited by Denis-Didier Rousseau and reviewed by two anonymous referees.</p>
  </notes><ref-list>
    <title>References</title>

      <ref id="bib1.bibx1"><?xmltex \def\ref@label{{Amante and Eakins(2009)}}?><label>Amante and Eakins(2009)</label><?label ETOPO?><mixed-citation>Amante, C. and Eakins, B. W.: Etopo1 1 Arc-Minute Global Relief Model: Procedures, Data Sources and Analysis, NOAA Technical Memorandum NESDIS NGDC-24, National Geophysical Data Center, NOAA [data set], <ext-link xlink:href="https://doi.org/10.7289/V5C8276M" ext-link-type="DOI">10.7289/V5C8276M</ext-link>, 2009.</mixed-citation></ref>
      <ref id="bib1.bibx2"><?xmltex \def\ref@label{{Birks et~al.(2010)Birks, Heiri, Sepp{\"{a}}, and Bjune}}?><label>Birks et al.(2010)Birks, Heiri, Seppä, and Bjune</label><?label birks_etal_2010?><mixed-citation>Birks, H. J. B., Heiri, O., Seppä, H., and Bjune, A. E.: Strengths and weaknesses of quantitative climate reconstructions based on Late-Quaternary biological proxies, The Open Ecology Journal, 3, 68–110, <ext-link xlink:href="https://doi.org/10.2174/1874213001003020068" ext-link-type="DOI">10.2174/1874213001003020068</ext-link>, 2010.</mixed-citation></ref>
      <ref id="bib1.bibx3"><?xmltex \def\ref@label{{Bivand and Rundel(2020)}}?><label>Bivand and Rundel(2020)</label><?label Rpkg_rgeos?><mixed-citation>Bivand, R. and Rundel, C.: rgeos: Interface to Geometry Engine – Open Source (“GEOS”), r package version 0.5-5, <uri>https://CRAN.R-project.org/package=rgeos</uri> (last access: February 2022), 2020.</mixed-citation></ref>
      <ref id="bib1.bibx4"><?xmltex \def\ref@label{{Bivand et~al.(2021)Bivand, Keitt, and Rowlingson}}?><label>Bivand et al.(2021)Bivand, Keitt, and Rowlingson</label><?label Rpkg_rgdal?><mixed-citation>Bivand, R., Keitt, T., and Rowlingson, B.: rgdal: Bindings for the “Geospatial” Data Abstraction Library, r package version 1.5-23, <uri>https://CRAN.R-project.org/package=rgdal</uri> (last access: February 2022), 2021.</mixed-citation></ref>
      <ref id="bib1.bibx5"><?xmltex \def\ref@label{{Bivand et~al.(2013)Bivand, Pebesma, and Gomez-Rubio}}?><label>Bivand et al.(2013)Bivand, Pebesma, and Gomez-Rubio</label><?label Rpkg_sp2?><mixed-citation>Bivand, R. S., Pebesma, E., and Gomez-Rubio, V.: Applied spatial data analysis with R, 2nd edn., Springer, NY, ISBN: 978-1-4614-7617-7, <uri>https://asdar-book.org/</uri> (last access: February 2022), 2013.</mixed-citation></ref>
      <ref id="bib1.bibx6"><?xmltex \def\ref@label{{Bray et~al.(2006)Bray, Blockley, Coope, Dadswell, Elias, Lowe, and Pollard}}?><label>Bray et al.(2006)Bray, Blockley, Coope, Dadswell, Elias, Lowe, and Pollard</label><?label bray_etal_2006?><mixed-citation>Bray, P. J., Blockley, S. P., Coope, G. R., Dadswell, L. F., Elias, S. A., Lowe, J. J., and Pollard, A. M.: Refining mutual climatic range (MCR) quantitative estimates of palaeotemperature using ubiquity analysis, Quaternary Sci. Rev., 25, 1865–1876, <ext-link xlink:href="https://doi.org/10.1016/j.quascirev.2006.01.023" ext-link-type="DOI">10.1016/j.quascirev.2006.01.023</ext-link>, 2006.</mixed-citation></ref>
      <ref id="bib1.bibx7"><?xmltex \def\ref@label{{Casta{\~{n}}eda et~al.(2016)Casta{\~{n}}eda, Caley, Dupont, Kim, Malaiz{\'{e}}, and Schouten}}?><label>Castañeda et al.(2016)Castañeda, Caley, Dupont, Kim, Malaizé, and Schouten</label><?label Castaneda2016a?><mixed-citation>Castañeda, I. S., Caley, T., Dupont, L. M., Kim, J.-H., Malaizé, B., and Schouten, S.: Middle to Late Pleistocene vegetation and climate change in subtropical southern East Africa, Earth Planet. Sc. Lett., 450, 306–316, <ext-link xlink:href="https://doi.org/10.1016/j.epsl.2016.06.049" ext-link-type="DOI">10.1016/j.epsl.2016.06.049</ext-link>, 2016.</mixed-citation></ref>
      <ref id="bib1.bibx8"><?xmltex \def\ref@label{{Chase et~al.(2015a)Chase, Boom, Carr, Carr{\'{e}}, Chevalier, Meadows, Pedro, Stager, and Reimer}}?><label>Chase et al.(2015a)Chase, Boom, Carr, Carré, Chevalier, Meadows, Pedro, Stager, and Reimer</label><?label Chase_etal_2015b?><mixed-citation>Chase, B. M., Boom, A., Carr, A. S., Carré, M., Chevalier, M., Meadows, M. E., Pedro, J. B., Stager, J. C., and Reimer, P. J.: Evolving southwest African response to abrupt deglacial North Atlantic climate change events, Quaternary Sci. Rev., 121, 132–136, <ext-link xlink:href="https://doi.org/10.1016/j.quascirev.2015.05.023" ext-link-type="DOI">10.1016/j.quascirev.2015.05.023</ext-link>, 2015a.</mixed-citation></ref>
      <ref id="bib1.bibx9"><?xmltex \def\ref@label{{Chase et~al.(2015b)Chase, Lim, Chevalier, Boom, Carr, Meadows, and Reimer}}?><label>Chase et al.(2015b)Chase, Lim, Chevalier, Boom, Carr, Meadows, and Reimer</label><?label Chase_etal_2015a?><mixed-citation>
Chase, B. M., Lim, S., Chevalier, M., Boom, A., Carr, A. S., Meadows, M. E., and Reimer, P. J.: Influence of tropical easterlies in the southwestern Cape of Africa during the Holocene, Quaternary Sci. Rev., 107, 138–148, 2015b.</mixed-citation></ref>
      <ref id="bib1.bibx10"><?xmltex \def\ref@label{{Chevalier(2019)}}?><label>Chevalier(2019)</label><?label Chevalier2019?><mixed-citation>Chevalier, M.: Enabling possibilities to quantify past climate from fossil assemblages at a global scale, Global Planet. Change, 175, 27–35, <ext-link xlink:href="https://doi.org/10.1016/j.gloplacha.2019.01.016" ext-link-type="DOI">10.1016/j.gloplacha.2019.01.016</ext-link>, 2019.</mixed-citation></ref>
      <ref id="bib1.bibx11"><?xmltex \def\ref@label{{Chevalier(2020)}}?><label>Chevalier(2020)</label><?label Chevalier2018GBIF?><mixed-citation>Chevalier, M.: GBIF database for CREST, figshare, <ext-link xlink:href="https://doi.org/10.6084/m9.figshare.6743207" ext-link-type="DOI">10.6084/m9.figshare.6743207</ext-link>, 2020.</mixed-citation></ref>
      <ref id="bib1.bibx12"><?xmltex \def\ref@label{{Chevalier(2022a)Chevalier}}?><label>Chevalier(2022a)Chevalier</label><?label Chevalier_2022a?><mixed-citation>Chevalier, M.: <italic>crestr</italic>: v1.0.1 (v1.0.1), Zenodo [code], <ext-link xlink:href="https://doi.org/10.5281/zenodo.6458405" ext-link-type="DOI">10.5281/zenodo.6458405</ext-link>, 2022a.</mixed-citation></ref>
      <ref id="bib1.bibx13"><?xmltex \def\ref@label{{Chevalier(2022b)Chevalier}}?><label>Chevalier(2022b)Chevalier</label><?label Chevalier2022b?><mixed-citation>Chevalier, M.: <italic>crestr</italic>: an R package to perform probabilistic climate reconstructions from palaeoecological datasets, R package version 1.0.1, <uri>https://CRAN.R-project.org/package=crestr</uri> (last access: 13 April 2022),
2022b.</mixed-citation></ref>
      <ref id="bib1.bibx14"><?xmltex \def\ref@label{{Chevalier and Chase(2015)}}?><label>Chevalier and Chase(2015)</label><?label CC15?><mixed-citation>Chevalier, M. and Chase, B. M.: Southeast African records reveal a coherent shift from high- to low-latitude forcing mechanisms along the east African margin across last glacial–interglacial transition, Quaternary Sci. Rev., 125, 117–130, <ext-link xlink:href="https://doi.org/10.1016/j.quascirev.2015.07.009" ext-link-type="DOI">10.1016/j.quascirev.2015.07.009</ext-link>, 2015.</mixed-citation></ref>
      <ref id="bib1.bibx15"><?xmltex \def\ref@label{{Chevalier and Chase(2016)}}?><label>Chevalier and Chase(2016)</label><?label CC16?><mixed-citation>Chevalier, M. and Chase, B. M.: Determining the drivers of long-term aridity variability: a southern African case study, J. Quaternary Sci., 31, 143–151, <ext-link xlink:href="https://doi.org/10.1002/jqs.2850" ext-link-type="DOI">10.1002/jqs.2850</ext-link>, 2016.</mixed-citation></ref>
      <ref id="bib1.bibx16"><?xmltex \def\ref@label{{Chevalier et~al.(2014)}}?><label>Chevalier et al.(2014)</label><?label Chevalier_etal_2014?><mixed-citation>Chevalier, M., Cheddadi, R., and Chase, B. M.: CREST (Climate REconstruction SofTware): a probability density function (PDF)-based quantitative climate reconstruction method, Clim. Past, 10, 2081–2098, <ext-link xlink:href="https://doi.org/10.5194/cp-10-2081-2014" ext-link-type="DOI">10.5194/cp-10-2081-2014</ext-link>, 2014.</mixed-citation></ref>
      <ref id="bib1.bibx17"><?xmltex \def\ref@label{{Chevalier et~al.(2020a)}}?><label>Chevalier et al.(2020a)</label><?label chevalier2020_data?><mixed-citation>Chevalier, M., Chase, B. M., Quick, L. J., Dupont, L. M., and Johnson, T. C.: Mean Annual Temperature changes reconstructions from marine core MD96-2048, PANGAEA [data set], <ext-link xlink:href="https://doi.org/10.1594/PANGAEA.915923" ext-link-type="DOI">10.1594/PANGAEA.915923</ext-link>, 2020a.</mixed-citation></ref>
      <ref id="bib1.bibx18"><?xmltex \def\ref@label{{Chevalier et~al.(2020b)}}?><label>Chevalier et al.(2020b)</label><?label Chevalier_etal_2020?><mixed-citation>Chevalier, M., Davis, B. A. S., Heiri, O., Seppä, H., Chase, B. M., Gajewski, K., Lacourse, T., Telford, R. J., Finsinger, W., Guiot, J., Kühl, N., Maezumi, S. Y., Tipton, J. R., Carter, V. A., Brussel, T., Phelps, L. N., Dawson, A., Zanon, M., Vallé, F., Nolan, C., Mauri, A., de Vernal, A., Izumi, K., Holmström, L., Marsicek, J., Goring, S. J., Sommer, P. S., Chaput, M., and Kupriyanov, D.: Pollen-based climate reconstruction techniques for late Quaternary studies, Earth-Sci. Rev., 210, 103384, <ext-link xlink:href="https://doi.org/10.1016/j.earscirev.2020.103384" ext-link-type="DOI">10.1016/j.earscirev.2020.103384</ext-link>, 2020b.</mixed-citation></ref>
      <ref id="bib1.bibx19"><?xmltex \def\ref@label{{Chevalier et~al.(2021a)Chevalier, Chase, Quick, Dupont, and Johnson}}?><label>Chevalier et al.(2021a)Chevalier, Chase, Quick, Dupont, and Johnson</label><?label Chevalier2021Limpopo?><mixed-citation>Chevalier, M., Chase, B. M., Quick, L. J., Dupont, L. M., and Johnson, T. C.: Temperature change in subtropical southeastern Africa during the past 790 000 yr, Geology, 49, 71–75, <ext-link xlink:href="https://doi.org/10.1130/G47841.1" ext-link-type="DOI">10.1130/G47841.1</ext-link>, 2021a.</mixed-citation></ref>
      <ref id="bib1.bibx20"><?xmltex \def\ref@label{{Chevalier et~al.(2021b)Chevalier, Chase, Quick, and Scott}}?><label>Chevalier et al.(2021b)Chevalier, Chase, Quick, and Scott</label><?label Chevalier2021Atlas?><mixed-citation>Chevalier, M., Chase, B. M., Quick, L. J., and Scott, L.: An atlas of southern African pollen types and their climatic affinities, Palaeoecol. Afr., 35, 239–258, <ext-link xlink:href="https://doi.org/10.1201/9781003162766-15" ext-link-type="DOI">10.1201/9781003162766-15</ext-link>, 2021b.</mixed-citation></ref>
      <ref id="bib1.bibx21"><?xmltex \def\ref@label{{Cordova et~al.(2017)Cordova, Scott, Chase, and Chevalier}}?><label>Cordova et al.(2017)Cordova, Scott, Chase, and Chevalier</label><?label Cordova_etal_2017?><mixed-citation>Cordova, C. E., Scott, L., Chase, B. M., and Chevalier, M.: Late Pleistocene-Holocene vegetation and climate change in the Middle Kalahari, Lake Ngami, Botswana, Quaternary Sci. Rev., 171, 199–215, <ext-link xlink:href="https://doi.org/10.1016/j.quascirev.2017.06.036" ext-link-type="DOI">10.1016/j.quascirev.2017.06.036</ext-link>, 2017.</mixed-citation></ref>
      <ref id="bib1.bibx22"><?xmltex \def\ref@label{{Costello et~al.(2017)Costello, Tsai, Wong, Cheung, Basher, and Chaudhary}}?><label>Costello et al.(2017)Costello, Tsai, Wong, Cheung, Basher, and Chaudhary</label><?label Costello2017?><mixed-citation>Costello, M. J., Tsai, P., Wong, P. S., Cheung, A. K. L., Basher, Z., and Chaudhary, C.: Marine biogeographic realms and species endemicity, Nat. Commun., 8, 1–9, <ext-link xlink:href="https://doi.org/10.1038/s41467-017-01121-2" ext-link-type="DOI">10.1038/s41467-017-01121-2</ext-link>, 2017.</mixed-citation></ref>
      <ref id="bib1.bibx23"><?xmltex \def\ref@label{{Dupont et al.(2011)}}?><label>Dupont et al.(2011)</label><?label Dupont2011?><mixed-citation>Dupont, L. M., Caley, T., Kim, J.-H., Castañeda, I., Malaizé, B., and Giraudeau, J.: Glacial-interglacial vegetation dynamics in South Eastern Africa coupled to sea surface temperature variations in the Western Indian Ocean, Clim. Past, 7, 1209–1224, <ext-link xlink:href="https://doi.org/10.5194/cp-7-1209-2011" ext-link-type="DOI">10.5194/cp-7-1209-2011</ext-link>, 2011.</mixed-citation></ref>
      <ref id="bib1.bibx24"><?xmltex \def\ref@label{{Dupont et al.(2019)}}?><label>Dupont et al.(2019)</label><?label Dupont2019?><mixed-citation>Dupont, L. M., Caley, T., and Castañeda, I. S.: Effects of atmospheric CO2 variability of the past 800 kyr on the biomes of southeast Africa, Clim. Past, 15, 1083–1097, <ext-link xlink:href="https://doi.org/10.5194/cp-15-1083-2019" ext-link-type="DOI">10.5194/cp-15-1083-2019</ext-link>, 2019.</mixed-citation></ref>
      <ref id="bib1.bibx25"><?xmltex \def\ref@label{{Fick and Hijmans(2017)}}?><label>Fick and Hijmans(2017)</label><?label Fick_Hijmans_2017?><mixed-citation>Fick, S. E. and Hijmans, R. J.: WorldClim 2: new 1 km spatial resolution climate surfaces for global land areas, Int. J. Climatol., 37, 4302–4315, <ext-link xlink:href="https://doi.org/10.1002/joc.5086" ext-link-type="DOI">10.1002/joc.5086</ext-link>, 2017.</mixed-citation></ref>
      <ref id="bib1.bibx26"><?xmltex \def\ref@label{{Garcia et~al.(2019a)Garcia, Weathers, Paver, Smolyar, Boyer, Locarnini, Zweng, Mishonov, Baranova, Seidov, and Reagan}}?><label>Garcia et al.(2019a)Garcia, Weathers, Paver, Smolyar, Boyer, Locarnini, Zweng, Mishonov, Baranova, Seidov, and Reagan</label><?label WOA_2018_oxy?><mixed-citation>Garcia, H. E., Weathers, K., Paver, C., Smolyar, I., Boyer, T., Locarnini, M., Zweng, M., Mishonov, A., Baranova, O., Seidov, D., and Reagan, J.: World Ocean Atlas 2018, Vol. 3: Dissolved Oxygen, Apparent Oxygen Utilization, and Dissolved Oxygen Saturation, NOAA Atlas NESDIS 83, 38 pp., <uri>https://archimer.ifremer.fr/doc/00651/76337/</uri> (last access: September 2021), 2019a.</mixed-citation></ref>
      <ref id="bib1.bibx27"><?xmltex \def\ref@label{{Garcia et~al.(2019b)Garcia, Weathers, Paver, Smolyar, Boyer, Locarnini, Zweng, Mishonov, Baranova, Seidov, Reagan, and Others}}?><label>Garcia et al.(2019b)Garcia, Weathers, Paver, Smolyar, Boyer, Locarnini, Zweng, Mishonov, Baranova, Seidov, Reagan, and Others</label><?label WOA_2018_nutrients?><mixed-citation>Garcia, H. E., Weathers, K., Paver, C., Smolyar, I., Boyer, T., Locarnini, M., Zweng, M., Mishonov, A., Baranova, O., Seidov, D., and Reagan, J.: World Ocean Atlas 2018, Vol. 4: Dissolved Inorganic Nutrients (phosphate, nitrate and nitrate+nitrite, silicate), NOAA Atlas NESDIS 84, 35 pp., <uri>https://archimer.ifremer.fr/doc/00651/76336/</uri> (last access: September 2021), 2019b.</mixed-citation></ref>
      <ref id="bib1.bibx28"><?xmltex \def\ref@label{{{Garnier} et~al.(2021)}}?><label>Garnier et al.(2021)</label><?label Rpkg_viridis?><mixed-citation>Garnier, S., Ross, N., Rudis, boB, Filipovic-Pierucci, A., Galili, T., timelyportfolio, Greenwell, B., Sievert, C., Harris, D. J., and Chen, J. J. : sjmgarnier/viridis: viridis 0.6.0, Version v0.6.0pre, Zenodo [code], <ext-link xlink:href="https://doi.org/10.5281/zenodo.4679424" ext-link-type="DOI">10.5281/zenodo.4679424</ext-link>, 2021.</mixed-citation></ref>
      <ref id="bib1.bibx29"><?xmltex \def\ref@label{{GBIF(2018)}}?><label>GBIF(2018)</label><?label GBIFwebsite?><mixed-citation>GBIF: GBIF: Global Biodiversity Information Facility, <uri>https://www.gbif.org</uri> (last access: August 2021), 2018.</mixed-citation></ref>
      <ref id="bib1.bibx30"><?xmltex \def\ref@label{{GBIF(2020a)}}?><label>GBIF(2020a)</label><?label GBIF_beetles20?><mixed-citation>GBIF: Beetles occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.nteruy" ext-link-type="DOI">10.15468/dl.nteruy</ext-link>, 2020a.</mixed-citation></ref>
      <ref id="bib1.bibx31"><?xmltex \def\ref@label{{GBIF(2020b)}}?><label>GBIF(2020b)</label><?label GBIF_chironomids20?><mixed-citation>GBIF: Chironomids occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set], <ext-link xlink:href="https://doi.org/10.15468/dl.jv3wsh" ext-link-type="DOI">10.15468/dl.jv3wsh</ext-link>, 2020b.</mixed-citation></ref>
      <ref id="bib1.bibx32"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{c}})}}?><label>GBIF(2020c)</label><?label GBIF_diatoms20?><mixed-citation>GBIF: Diatoms occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.vfr257" ext-link-type="DOI">10.15468/dl.vfr257</ext-link>, 2020c.</mixed-citation></ref>
      <ref id="bib1.bibx33"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{d}})}}?><label>GBIF(2020d)</label><?label GBIF_forams20?><mixed-citation>GBIF: Foraminifera occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.692yg6" ext-link-type="DOI">10.15468/dl.692yg6</ext-link>, 2020d.</mixed-citation></ref>
      <ref id="bib1.bibx34"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{e}})}}?><label>GBIF(2020e)</label><?label GBIF_rodents20?><mixed-citation>GBIF: Rodentia occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.fscw6q" ext-link-type="DOI">10.15468/dl.fscw6q</ext-link>, 2020e.</mixed-citation></ref>
      <ref id="bib1.bibx35"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{f}})}}?><label>GBIF(2020f)</label><?label gbif_Cycadopsida20?><mixed-citation>GBIF: Cycadopsidae occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.sfjzxu" ext-link-type="DOI">10.15468/dl.sfjzxu</ext-link>, 2020f.</mixed-citation></ref>
      <ref id="bib1.bibx36"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{g}})}}?><label>GBIF(2020g)</label><?label gbif_Magnoliopsida20?><mixed-citation>GBIF: Magnoliopsida occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.ra49dt" ext-link-type="DOI">10.15468/dl.ra49dt</ext-link>, 2020g.</mixed-citation></ref>
      <ref id="bib1.bibx37"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{h}})}}?><label>GBIF(2020h)</label><?label gbif_antheceropsida20?><mixed-citation>GBIF: Anthocerotopsida occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.t9zenf" ext-link-type="DOI">10.15468/dl.t9zenf</ext-link>, 2020h.</mixed-citation></ref>
      <ref id="bib1.bibx38"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{i}})}}?><label>GBIF(2020i)</label><?label gbif_gingkoopsida20?><mixed-citation>GBIF: Gingkoopsidae occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.da9wz8" ext-link-type="DOI">10.15468/dl.da9wz8</ext-link>, 2020i.</mixed-citation></ref>
      <ref id="bib1.bibx39"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{j}})}}?><label>GBIF(2020j)</label><?label gbif_gnetopsida20?><mixed-citation>GBIF: Gnetopsidae occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.h2kjnc" ext-link-type="DOI">10.15468/dl.h2kjnc</ext-link>, 2020j.</mixed-citation></ref>
      <ref id="bib1.bibx40"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{k}})}}?><label>GBIF(2020k)</label><?label gbif_liliopsida20?><mixed-citation>GBIF: Liliopsida occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.axv3yd" ext-link-type="DOI">10.15468/dl.axv3yd</ext-link>, 2020k.</mixed-citation></ref>
      <ref id="bib1.bibx41"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{l}})}}?><label>GBIF(2020l)</label><?label gbif_lycopodiopsida20?><mixed-citation>GBIF: Lycopodiopsida occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.ydhyhz" ext-link-type="DOI">10.15468/dl.ydhyhz</ext-link>, 2020l.</mixed-citation></ref>
      <ref id="bib1.bibx42"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{m}})}}?><label>GBIF(2020m)</label><?label gbif_pinopsida20?><mixed-citation>GBIF: Pinopsidae occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/dl.x2r7pa" ext-link-type="DOI">10.15468/dl.x2r7pa</ext-link>, 2020m.</mixed-citation></ref>
      <ref id="bib1.bibx43"><?xmltex \def\ref@label{{GBIF(2020{\natexlab{n}})}}?><label>GBIF(2020n)</label><?label gbif_polypodiopsida20?><mixed-citation>GBIF: Polypodiopsida occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set], <ext-link xlink:href="https://doi.org/10.15468/dl.87tbp6" ext-link-type="DOI">10.15468/dl.87tbp6</ext-link>, 2020n.</mixed-citation></ref>
      <ref id="bib1.bibx44"><?xmltex \def\ref@label{{GBIF(2021a)}}?><label>GBIF(2021a)</label><?label gbif_bryophyta21?><mixed-citation>GBIF: Bryophyta occurrence data, downloaded on 2 August 2021, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/DL.WD527G" ext-link-type="DOI">10.15468/DL.WD527G</ext-link>, 2021a.</mixed-citation></ref>
      <ref id="bib1.bibx45"><?xmltex \def\ref@label{{GBIF(2021b)}}?><label>GBIF(2021b)</label><?label gbif_marchantiophyta21?><mixed-citation>GBIF: Marchantiophyta occurrence data, downloaded on 2 August 2021, The Global Biodiversity Information Facility (GBIF) [data set],  <ext-link xlink:href="https://doi.org/10.15468/DL.M2SSE4" ext-link-type="DOI">10.15468/DL.M2SSE4</ext-link>, 2021b.</mixed-citation></ref>
      <ref id="bib1.bibx46"><?xmltex \def\ref@label{{Gibson et~al.(2022)Gibson, McCoy, O'Keefe, {Nu{\~{n}}ez Ota{\~{n}}o}, Warny, and Pound}}?><label>Gibson et al.(2022)Gibson, McCoy, O'Keefe, Nuñez Otaño, Warny, and Pound</label><?label Gibson2022?><mixed-citation>Gibson, M. E., McCoy, J., O'Keefe, J. M. K., Nuñez Otaño, N. B., Warny, S., and Pound, M. J.: Reconstructing terrestrial paleoclimates: a comparison of the Co-existence Approach, Bayesian and probability reconstruction techniques using the UK Neogene, Paleoceanography and Paleoclimatology, 37, e2021PA004358, <ext-link xlink:href="https://doi.org/10.1029/2021PA004358" ext-link-type="DOI">10.1029/2021PA004358</ext-link>, 2022.</mixed-citation></ref>
      <ref id="bib1.bibx47"><?xmltex \def\ref@label{{Hijmans(2021)}}?><label>Hijmans(2021)</label><?label Rpkg_raster?><mixed-citation>Hijmans, R. J.: raster: Geographic Data Analysis and Modeling, r package version 3.4-13, <uri>https://CRAN.R-project.org/package=raster</uri> (last access: February 2022), 2021.</mixed-citation></ref>
      <ref id="bib1.bibx48"><?xmltex \def\ref@label{{Hui et~al.(2021)Hui, Zhou, Chevalier, Wei, Pan, and Chen}}?><label>Hui et al.(2021)Hui, Zhou, Chevalier, Wei, Pan, and Chen</label><?label Hui_etal_2021?><mixed-citation>Hui, Z., Zhou, X., Chevalier, M., Wei, X., Pan, Y., and Chen, Y.: Miocene East Asia summer monsoon precipitation variability and its possible driving forces, Palaeogeogr. Palaeocl., 581, 110609, <ext-link xlink:href="https://doi.org/10.1016/j.palaeo.2021.110609" ext-link-type="DOI">10.1016/j.palaeo.2021.110609</ext-link>, 2021.</mixed-citation></ref>
      <ref id="bib1.bibx49"><?xmltex \def\ref@label{{Hutson(1978)}}?><label>Hutson(1978)</label><?label Hutson1978?><mixed-citation>Hutson, W. H.: Application of transfer functions to Indian Ocean planktonic Foraminifera, Quaternary Res., 9, 87–112, <ext-link xlink:href="https://doi.org/10.1016/0033-5894(78)90084-4" ext-link-type="DOI">10.1016/0033-5894(78)90084-4</ext-link>, 1978.</mixed-citation></ref>
      <ref id="bib1.bibx50"><?xmltex \def\ref@label{{Juggins(2013)}}?><label>Juggins(2013)</label><?label Juggins2013?><mixed-citation>Juggins, S.: Quantitative reconstructions in palaeolimnology: new paradigm or sick science?, Quaternary Sci. Rev., 64, 20–32, <ext-link xlink:href="https://doi.org/10.1016/j.quascirev.2012.12.014" ext-link-type="DOI">10.1016/j.quascirev.2012.12.014</ext-link>, 2013.</mixed-citation></ref>
      <ref id="bib1.bibx51"><?xmltex \def\ref@label{{K{\"{u}}hl et~al.(2002)K{\"{u}}hl, Gebhardt, Litt, and Hense}}?><label>Kühl et al.(2002)Kühl, Gebhardt, Litt, and Hense</label><?label Kuhl_etal_2002?><mixed-citation>Kühl, N., Gebhardt, C., Litt, T., and Hense, A.: Probability Density Functions as Botanical-Climatological Transfer Functions for Climate Reconstruction, Quaternary Res., 58, 381–392, <ext-link xlink:href="https://doi.org/10.1006/qres.2002.2380" ext-link-type="DOI">10.1006/qres.2002.2380</ext-link>, 2002.</mixed-citation></ref>
      <ref id="bib1.bibx52"><?xmltex \def\ref@label{{Lim et~al.(2016)Lim, Chase, Chevalier, and Reimer}}?><label>Lim et al.(2016)Lim, Chase, Chevalier, and Reimer</label><?label Lim_etal_2016?><mixed-citation>Lim, S., Chase, B. M., Chevalier, M., and Reimer, P. J.: 50 000 years of climate in the Namib Desert, Pella, South Africa, Palaeogeogr. Palaeocl., 451, 197–209, <ext-link xlink:href="https://doi.org/10.1016/j.palaeo.2016.03.001" ext-link-type="DOI">10.1016/j.palaeo.2016.03.001</ext-link>, 2016.</mixed-citation></ref>
      <ref id="bib1.bibx53"><?xmltex \def\ref@label{{Lincoln(2020)}}?><label>Lincoln(2020)</label><?label Rpkg_clipr?><mixed-citation>Lincoln, M.: clipr: Read and Write from the System Clipboard, r package version 0.7.1, <uri>https://CRAN.R-project.org/package=clipr</uri> (last access: February 2022), 2020.</mixed-citation></ref>
      <ref id="bib1.bibx54"><?xmltex \def\ref@label{{Locarnini et~al.(2019)Locarnini, Mishonov, Baranova, Boyer, Zweng, Garcia, Reagan, Seidov, Weathers, Paver, Smolyar, and Others}}?><label>Locarnini et al.(2019)Locarnini, Mishonov, Baranova, Boyer, Zweng, Garcia, Reagan, Seidov, Weathers, Paver, Smolyar, and Others</label><?label WOA_2018_temp?><mixed-citation>Locarnini, R., Mishonov, A., Baranova, O., Boyer, T., Zweng, M., Garcia, H. E., Reagan, J., Seidov, D., Weathers, K., Paver, C., and Smolyar, I.: World ocean atlas 2018, Vol. 1: Temperature, NOAA Atlas NESDIS 81, 52pp., <uri>https://archimer.ifremer.fr/doc/00651/76338/</uri> (last access: September 2021), 2019.</mixed-citation></ref>
      <ref id="bib1.bibx55"><?xmltex \def\ref@label{{Myers et~al.(2000)Myers, Mittermeier, Mittermeier, da~Fonseca, and Kent}}?><label>Myers et al.(2000)Myers, Mittermeier, Mittermeier, da Fonseca, and Kent</label><?label Myers_etal_2000?><mixed-citation>Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B., and Kent, J.: Biodiversity hotspots for conservation priorities, Nature, 403, 853–858, <ext-link xlink:href="https://doi.org/10.1038/35002501" ext-link-type="DOI">10.1038/35002501</ext-link>, 2000.</mixed-citation></ref>
      <ref id="bib1.bibx56"><?xmltex \def\ref@label{{Olson et~al.(2001)Olson, Dinerstein, Wikramanayake, Burgess, Powell, Underwood, D'amico, Itoua, Strand, Morrison, Loucks, Allnutt, Ricketts, Kura, Lamoreux, Wettengel, Hedao, and Kassem}}?><label>Olson et al.(2001)Olson, Dinerstein, Wikramanayake, Burgess, Powell, Underwood, D'amico, Itoua, Strand, Morrison, Loucks, Allnutt, Ricketts, Kura, Lamoreux, Wettengel, Hedao, and Kassem</label><?label Olson2001?><mixed-citation>Olson, D. M., Dinerstein, E., Wikramanayake, E. D., Burgess, N. D., Powell, G. V. N., Underwood, E. C., D'amico, J. A., Itoua, I., Strand, H. E., Morrison, J. C., Loucks, C. J., Allnutt, T. F., Ricketts, T. H., Kura, Y., Lamoreux, J. F., Wettengel, W. W., Hedao, P., and Kassem, K. R.: Terrestrial Ecoregions of the World: A New Map of Life on Earth: A new global map of terrestrial ecoregions provides an innovative tool for conserving biodiversity, BioScience, 51, 933, <ext-link xlink:href="https://doi.org/10.1641/0006-3568(2001)051[0933:TEOTWA]2.0.CO;2" ext-link-type="DOI">10.1641/0006-3568(2001)051[0933:TEOTWA]2.0.CO;2</ext-link>, 2001.</mixed-citation></ref>
      <ref id="bib1.bibx57"><?xmltex \def\ref@label{{Overpeck et~al.(1985)Overpeck, {Webb III}, and Prentice}}?><label>Overpeck et al.(1985)Overpeck, Webb III, and Prentice</label><?label Overpeck_etal_1985?><mixed-citation>Overpeck, J. T., Webb III, T., and Prentice, I. C.: Quantitative interpretation of fossil pollen spectra: Dissimilarity coefficients and the method of modern analogs, Quaternary Res., 23, 87–108, <ext-link xlink:href="https://doi.org/10.1016/0033-5894(85)90074-2" ext-link-type="DOI">10.1016/0033-5894(85)90074-2</ext-link>, 1985.</mixed-citation></ref>
      <ref id="bib1.bibx58"><?xmltex \def\ref@label{{Pebesma and Bivand(2005)}}?><label>Pebesma and Bivand(2005)</label><?label Rpkg_sp1?><mixed-citation>Pebesma, E. J. and Bivand, R. S.: Classes and methods for spatial data in R, R News, 5, 9–13, <uri>https://CRAN.R-project.org/doc/Rnews/</uri> (last access: February 2022), 2005.</mixed-citation></ref>
      <ref id="bib1.bibx59"><?xmltex \def\ref@label{{Quick et~al.(2021)}}?><label>Quick et al.(2021)</label><?label Quick2021?><mixed-citation>Quick, L. J., Chase, B. M., Carr, A. S., Chevalier, M., Grobler, B. A., and Meadows, M. E.: A 25 000 year record of climate and vegetation change from the southwestern Cape coast, South Africa, Quaternary Res., 105, 1–18, <ext-link xlink:href="https://doi.org/10.1017/qua.2021.31" ext-link-type="DOI">10.1017/qua.2021.31</ext-link>, 2021.</mixed-citation></ref>
      <ref id="bib1.bibx60"><?xmltex \def\ref@label{{{R Core Team}(2020)}}?><label>R Core Team(2020)</label><?label R?><mixed-citation>R Core Team: R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, <uri>https://www.R-project.org/</uri> (last access: February 2022), 2020.</mixed-citation></ref>
      <ref id="bib1.bibx61"><?xmltex \def\ref@label{{{R Special Interest Group on Databases (R-SIG-DB)} et~al.(2021){R Special Interest Group on Databases (R-SIG-DB)}, Wickham, and M\"{u}ller}}?><label>R Special Interest Group on Databases (R-SIG-DB) et al.(2021)R Special Interest Group on Databases (R-SIG-DB), Wickham, and Müller</label><?label Rpkg_DBI?><mixed-citation>R Special Interest Group on Databases (R-SIG-DB), Wickham, H., and Müller, K.: DBI: R Database Interface, r package version 1.1.1, <uri>https://CRAN.R-project.org/package=DBI</uri> (last access: February 2022), 2021.</mixed-citation></ref>
      <ref id="bib1.bibx62"><?xmltex \def\ref@label{{Reynolds et~al.(2007)Reynolds, Smith, Liu, Chelton, Casey, and Schlax}}?><label>Reynolds et al.(2007)Reynolds, Smith, Liu, Chelton, Casey, and Schlax</label><?label Reynolds2007?><mixed-citation>Reynolds, R. W., Smith, T. M., Liu, C., Chelton, D. B., Casey, K. S., and Schlax, M. G.: Daily high-resolution-blended analyses for sea surface temperature, J. Climate, 20, 5473–5496, <ext-link xlink:href="https://doi.org/10.1175/2007JCLI1824.1" ext-link-type="DOI">10.1175/2007JCLI1824.1</ext-link>, 2007 (data available at: <uri>https://psl.noaa.gov/data/gridded/data.noaa.oisst.v2.highres.html</uri>, last access: September 2021).</mixed-citation></ref>
      <ref id="bib1.bibx63"><?xmltex \def\ref@label{{Schauberger and Walker(2020)}}?><label>Schauberger and Walker(2020)</label><?label Rpkg_openxlsx?><mixed-citation>Schauberger, P. and Walker, A.: openxlsx: Read, Write and Edit xlsx Files, r package version 4.2.3, <uri>https://CRAN.R-project.org/package=openxlsx</uri> (last access: ), 2020.</mixed-citation></ref>
      <ref id="bib1.bibx64"><?xmltex \def\ref@label{{Soetaert(2021)}}?><label>Soetaert(2021)</label><?label Rpkg_plot3D?><mixed-citation>Soetaert, K.: plot3D: Plotting Multi-Dimensional Data, r package version 1.4, <uri>https://CRAN.R-project.org/package=plot3D</uri> (last access: February 2022), 2021.</mixed-citation></ref>
      <ref id="bib1.bibx65"><?xmltex \def\ref@label{{ter Braak and van Dame(1989)}}?><label>ter Braak and van Dame(1989)</label><?label terBraak_vanDame_1989?><mixed-citation>ter Braak, C. J. F. and van Dame, H.: Inferring pH from diatoms: a comparison of old and new calibration methods, Hydrobiologia, 178, 209–223, <ext-link xlink:href="https://doi.org/10.1007/BF00006028" ext-link-type="DOI">10.1007/BF00006028</ext-link>, 1989.</mixed-citation></ref>
      <ref id="bib1.bibx66"><?xmltex \def\ref@label{{ter Braak and Juggins(1993)}}?><label>ter Braak and Juggins(1993)</label><?label TerBraak1993?><mixed-citation>ter Braak, C. J. F. and Juggins, S.: Weighted averaging partial least squares regression (WA-PLS): an improved method for reconstructing environmental variables from species assemblages, Hydrobiologia, 269, 485–502, <ext-link xlink:href="https://doi.org/10.1007/BF00028046" ext-link-type="DOI">10.1007/BF00028046</ext-link>, 1993.</mixed-citation></ref>
      <ref id="bib1.bibx67"><?xmltex \def\ref@label{{Valavi et~al.(2021)}}?><label>Valavi et al.(2021)</label><?label Valavi2021?><mixed-citation>Valavi, R., Guillera-Arroita, G., Lahoz-Monfort, J. J., and Elith, J.: Predictive performance of presence-only species distribution models: a benchmark study with reproducible code, Ecol. Monogr., 92, e01486, <ext-link xlink:href="https://doi.org/10.1002/ecm.1486" ext-link-type="DOI">10.1002/ecm.1486</ext-link>, 2021.</mixed-citation></ref>
      <ref id="bib1.bibx68"><?xmltex \def\ref@label{{Wickham(2011)}}?><label>Wickham(2011)</label><?label Rpkg_plyr?><mixed-citation>Wickham, H.: The Split-Apply-Combine Strategy for Data Analysis, J. Stat. Softw., 40, 1–29, <uri>http://www.jstatsoft.org/v40/i01/</uri> (last access: February 2022), 2011.
</mixed-citation></ref><?xmltex \hack{\newpage}?>
      <ref id="bib1.bibx69"><?xmltex \def\ref@label{{Wickham(2019)}}?><label>Wickham(2019)</label><?label Rpkg_stringr?><mixed-citation>Wickham, H.: stringr: Simple, Consistent Wrappers for Common String Operations, r package version 1.4.0, <uri>https://CRAN.R-project.org/package=stringr</uri> (last access: February 2022), 2019.</mixed-citation></ref>
      <ref id="bib1.bibx70"><?xmltex \def\ref@label{{Wickham and Hesselberth(2020)}}?><label>Wickham and Hesselberth(2020)</label><?label Rpkg_pkgdown?><mixed-citation>Wickham, H. and Hesselberth, J.: pkgdown: Make Static HTML Documentation for a Package, r package version 1.6.1, <uri>https://CRAN.R-project.org/package=pkgdown</uri> (last access: February 2022), 2020.</mixed-citation></ref>
      <ref id="bib1.bibx71"><?xmltex \def\ref@label{{Wickham and Seidel(2020)}}?><label>Wickham and Seidel(2020)</label><?label Rpkg_scales?><mixed-citation>Wickham, H. and Seidel, D.: scales: Scale Functions for Visualization, r package version 1.1.1, <uri>https://CRAN.R-project.org/package=scales</uri> (last access: February 2022), 2020.</mixed-citation></ref>
      <ref id="bib1.bibx72"><?xmltex \def\ref@label{{Wickham et~al.(2020)Wickham, Hester, and Chang}}?><label>Wickham et al.(2020)Wickham, Hester, and Chang</label><?label Rpkg_devtools?><mixed-citation>Wickham, H., Hester, J., and Chang, W.: devtools: Tools to Make Developing R Packages Easier, r package version 2.3.2, <uri>https://CRAN.R-project.org/package=devtools</uri> (last access: February 2022), 2020.</mixed-citation></ref>
      <ref id="bib1.bibx73"><?xmltex \def\ref@label{{Wickham et~al.(2021)Wickham, Ooms, and M\"{u}ller}}?><label>Wickham et al.(2021)Wickham, Ooms, and Müller</label><?label Rpkg_RPostgres?><mixed-citation>Wickham, H., Ooms, J., and Müller, K.: RPostgres: “Rcpp” Interface to “PostgreSQL”, r package version 1.3.3, <uri>https://CRAN.R-project.org/package=RPostgres</uri> (last access: February 2022), 2021.</mixed-citation></ref>
      <ref id="bib1.bibx74"><?xmltex \def\ref@label{{Wright(2021)}}?><label>Wright(2021)</label><?label Rpkg_pals?><mixed-citation>Wright, K.: pals: Color Palettes, Colormaps, and Tools to Evaluate Them, r package version 1.7, <uri>https://CRAN.R-project.org/package=pals</uri> (last access: February 2022), 2021.</mixed-citation></ref>
      <ref id="bib1.bibx75"><?xmltex \def\ref@label{{Yi et~al.(2020)Yi, Jun, Jo, Lee, Kim, Lee, Cao, and Lim}}?><label>Yi et al.(2020)Yi, Jun, Jo, Lee, Kim, Lee, Cao, and Lim</label><?label Yi_etal_2020?><mixed-citation>Yi, S., Jun, C. P., Jo, K.-n., Lee, H., Kim, M. S., Lee, S. D., Cao, X., and Lim, J.: Asynchronous multi-decadal time-scale series of biotic and abiotic responses to precipitation during the last 1300 years, Sci. Rep., 10, 1–10, <ext-link xlink:href="https://doi.org/10.1038/s41598-020-74994-x" ext-link-type="DOI">10.1038/s41598-020-74994-x</ext-link>, 2020.</mixed-citation></ref>
      <ref id="bib1.bibx76"><?xmltex \def\ref@label{{Zomer et~al.(2008)Zomer, Trabucco, Bossio, and Verchot}}?><label>Zomer et al.(2008)Zomer, Trabucco, Bossio, and Verchot</label><?label Zomer2008?><mixed-citation>Zomer, R. J., Trabucco, A., Bossio, D. A., and Verchot, L. V.: Climate change mitigation: A spatial analysis of global land suitability for clean development mechanism afforestation and reforestation, Agr. Ecosyst. Environ., 126, 67–80, <ext-link xlink:href="https://doi.org/10.1016/j.agee.2008.01.014" ext-link-type="DOI">10.1016/j.agee.2008.01.014</ext-link>, 2008.</mixed-citation></ref>
      <ref id="bib1.bibx77"><?xmltex \def\ref@label{{Zweng et~al.(2018)Zweng, Seidov, Boyer, Locarnini, Garcia, Mishonov, Baranova, Weathers, Paver, Smolyar, and Others}}?><label>Zweng et al.(2018)Zweng, Seidov, Boyer, Locarnini, Garcia, Mishonov, Baranova, Weathers, Paver, Smolyar, and Others</label><?label WOA_2018_salinity?><mixed-citation>Zweng, M., Seidov, D., Boyer, T., Locarnini, R., Garcia, H. E., Mishonov, A., Baranova, O., Weathers, K., Paver, C., and Smolyar, I.: World Ocean Atlas 2018, Vol. 2: Salinity, NOAA Atlas NESDIS 82, 50 pp., <uri>https://archimer.ifremer.fr/doc/00651/76339/</uri> (last access: September 2021), 2018.</mixed-citation></ref>

  </ref-list></back>
    <!--<article-title-html><i>crestr</i>: an R package to perform probabilistic climate reconstructions from palaeoecological datasets</article-title-html>
<abstract-html/>
<ref-html id="bib1.bib1"><label>Amante and Eakins(2009)</label><mixed-citation>
Amante, C. and Eakins, B. W.: Etopo1 1 Arc-Minute Global Relief Model: Procedures, Data Sources and Analysis, NOAA Technical Memorandum NESDIS NGDC-24, National Geophysical Data Center, NOAA [data set], <a href="https://doi.org/10.7289/V5C8276M" target="_blank">https://doi.org/10.7289/V5C8276M</a>, 2009.
</mixed-citation></ref-html>
<ref-html id="bib1.bib2"><label>Birks et al.(2010)Birks, Heiri, Seppä, and Bjune</label><mixed-citation>
Birks, H. J. B., Heiri, O., Seppä, H., and Bjune, A. E.: Strengths and weaknesses of quantitative climate reconstructions based on Late-Quaternary biological proxies, The Open Ecology Journal, 3, 68–110, <a href="https://doi.org/10.2174/1874213001003020068" target="_blank">https://doi.org/10.2174/1874213001003020068</a>, 2010.
</mixed-citation></ref-html>
<ref-html id="bib1.bib3"><label>Bivand and Rundel(2020)</label><mixed-citation>
Bivand, R. and Rundel, C.: rgeos: Interface to Geometry Engine – Open Source (“GEOS”), r package version 0.5-5, <a href="https://CRAN.R-project.org/package=rgeos" target="_blank"/> (last access: February 2022), 2020.
</mixed-citation></ref-html>
<ref-html id="bib1.bib4"><label>Bivand et al.(2021)Bivand, Keitt, and Rowlingson</label><mixed-citation>
Bivand, R., Keitt, T., and Rowlingson, B.: rgdal: Bindings for the “Geospatial” Data Abstraction Library, r package version 1.5-23, <a href="https://CRAN.R-project.org/package=rgdal" target="_blank"/> (last access: February 2022), 2021.
</mixed-citation></ref-html>
<ref-html id="bib1.bib5"><label>Bivand et al.(2013)Bivand, Pebesma, and Gomez-Rubio</label><mixed-citation>
Bivand, R. S., Pebesma, E., and Gomez-Rubio, V.: Applied spatial data analysis with R, 2nd edn., Springer, NY, ISBN: 978-1-4614-7617-7, <a href="https://asdar-book.org/" target="_blank"/> (last access: February 2022), 2013.
</mixed-citation></ref-html>
<ref-html id="bib1.bib6"><label>Bray et al.(2006)Bray, Blockley, Coope, Dadswell, Elias, Lowe, and Pollard</label><mixed-citation>
Bray, P. J., Blockley, S. P., Coope, G. R., Dadswell, L. F., Elias, S. A., Lowe, J. J., and Pollard, A. M.: Refining mutual climatic range (MCR) quantitative estimates of palaeotemperature using ubiquity analysis, Quaternary Sci. Rev., 25, 1865–1876, <a href="https://doi.org/10.1016/j.quascirev.2006.01.023" target="_blank">https://doi.org/10.1016/j.quascirev.2006.01.023</a>, 2006.
</mixed-citation></ref-html>
<ref-html id="bib1.bib7"><label>Castañeda et al.(2016)Castañeda, Caley, Dupont, Kim, Malaizé, and Schouten</label><mixed-citation>
Castañeda, I. S., Caley, T., Dupont, L. M., Kim, J.-H., Malaizé, B., and Schouten, S.: Middle to Late Pleistocene vegetation and climate change in subtropical southern East Africa, Earth Planet. Sc. Lett., 450, 306–316, <a href="https://doi.org/10.1016/j.epsl.2016.06.049" target="_blank">https://doi.org/10.1016/j.epsl.2016.06.049</a>, 2016.
</mixed-citation></ref-html>
<ref-html id="bib1.bib8"><label>Chase et al.(2015a)Chase, Boom, Carr, Carré, Chevalier, Meadows, Pedro, Stager, and Reimer</label><mixed-citation>
Chase, B. M., Boom, A., Carr, A. S., Carré, M., Chevalier, M., Meadows, M. E., Pedro, J. B., Stager, J. C., and Reimer, P. J.: Evolving southwest African response to abrupt deglacial North Atlantic climate change events, Quaternary Sci. Rev., 121, 132–136, <a href="https://doi.org/10.1016/j.quascirev.2015.05.023" target="_blank">https://doi.org/10.1016/j.quascirev.2015.05.023</a>, 2015a.
</mixed-citation></ref-html>
<ref-html id="bib1.bib9"><label>Chase et al.(2015b)Chase, Lim, Chevalier, Boom, Carr, Meadows, and Reimer</label><mixed-citation>
Chase, B. M., Lim, S., Chevalier, M., Boom, A., Carr, A. S., Meadows, M. E., and Reimer, P. J.: Influence of tropical easterlies in the southwestern Cape of Africa during the Holocene, Quaternary Sci. Rev., 107, 138–148, 2015b.
</mixed-citation></ref-html>
<ref-html id="bib1.bib10"><label>Chevalier(2019)</label><mixed-citation>
Chevalier, M.: Enabling possibilities to quantify past climate from fossil assemblages at a global scale, Global Planet. Change, 175, 27–35, <a href="https://doi.org/10.1016/j.gloplacha.2019.01.016" target="_blank">https://doi.org/10.1016/j.gloplacha.2019.01.016</a>, 2019.
</mixed-citation></ref-html>
<ref-html id="bib1.bib11"><label>Chevalier(2020)</label><mixed-citation>
Chevalier, M.: GBIF database for CREST, figshare, <a href="https://doi.org/10.6084/m9.figshare.6743207" target="_blank">https://doi.org/10.6084/m9.figshare.6743207</a>, 2020.
</mixed-citation></ref-html>
<ref-html id="bib1.bib12"><label>Chevalier(2022a)Chevalier</label><mixed-citation>
Chevalier, M.: <i>crestr</i>: v1.0.1 (v1.0.1), Zenodo [code], <a href="https://doi.org/10.5281/zenodo.6458405" target="_blank">https://doi.org/10.5281/zenodo.6458405</a>, 2022a.
</mixed-citation></ref-html>
<ref-html id="bib1.bib13"><label>Chevalier(2022b)Chevalier</label><mixed-citation>
Chevalier, M.: <i>crestr</i>: an R package to perform probabilistic climate reconstructions from palaeoecological datasets, R package version 1.0.1, <a href="https://CRAN.R-project.org/package=crestr" target="_blank"/> (last access: 13 April 2022),
2022b.
</mixed-citation></ref-html>
<ref-html id="bib1.bib14"><label>Chevalier and Chase(2015)</label><mixed-citation>
Chevalier, M. and Chase, B. M.: Southeast African records reveal a coherent shift from high- to low-latitude forcing mechanisms along the east African margin across last glacial–interglacial transition, Quaternary Sci. Rev., 125, 117–130, <a href="https://doi.org/10.1016/j.quascirev.2015.07.009" target="_blank">https://doi.org/10.1016/j.quascirev.2015.07.009</a>, 2015.
</mixed-citation></ref-html>
<ref-html id="bib1.bib15"><label>Chevalier and Chase(2016)</label><mixed-citation>
Chevalier, M. and Chase, B. M.: Determining the drivers of long-term aridity variability: a southern African case study, J. Quaternary Sci., 31, 143–151, <a href="https://doi.org/10.1002/jqs.2850" target="_blank">https://doi.org/10.1002/jqs.2850</a>, 2016.
</mixed-citation></ref-html>
<ref-html id="bib1.bib16"><label>Chevalier et al.(2014)</label><mixed-citation>
Chevalier, M., Cheddadi, R., and Chase, B. M.: CREST (Climate REconstruction SofTware): a probability density function (PDF)-based quantitative climate reconstruction method, Clim. Past, 10, 2081–2098, <a href="https://doi.org/10.5194/cp-10-2081-2014" target="_blank">https://doi.org/10.5194/cp-10-2081-2014</a>, 2014.
</mixed-citation></ref-html>
<ref-html id="bib1.bib17"><label>Chevalier et al.(2020a)</label><mixed-citation>
Chevalier, M., Chase, B. M., Quick, L. J., Dupont, L. M., and Johnson, T. C.: Mean Annual Temperature changes reconstructions from marine core MD96-2048, PANGAEA [data set], <a href="https://doi.org/10.1594/PANGAEA.915923" target="_blank">https://doi.org/10.1594/PANGAEA.915923</a>, 2020a.
</mixed-citation></ref-html>
<ref-html id="bib1.bib18"><label>Chevalier et al.(2020b)</label><mixed-citation>
Chevalier, M., Davis, B. A. S., Heiri, O., Seppä, H., Chase, B. M., Gajewski, K., Lacourse, T., Telford, R. J., Finsinger, W., Guiot, J., Kühl, N., Maezumi, S. Y., Tipton, J. R., Carter, V. A., Brussel, T., Phelps, L. N., Dawson, A., Zanon, M., Vallé, F., Nolan, C., Mauri, A., de Vernal, A., Izumi, K., Holmström, L., Marsicek, J., Goring, S. J., Sommer, P. S., Chaput, M., and Kupriyanov, D.: Pollen-based climate reconstruction techniques for late Quaternary studies, Earth-Sci. Rev., 210, 103384, <a href="https://doi.org/10.1016/j.earscirev.2020.103384" target="_blank">https://doi.org/10.1016/j.earscirev.2020.103384</a>, 2020b.
</mixed-citation></ref-html>
<ref-html id="bib1.bib19"><label>Chevalier et al.(2021a)Chevalier, Chase, Quick, Dupont, and Johnson</label><mixed-citation>
Chevalier, M., Chase, B. M., Quick, L. J., Dupont, L. M., and Johnson, T. C.: Temperature change in subtropical southeastern Africa during the past 790&thinsp;000 yr, Geology, 49, 71–75, <a href="https://doi.org/10.1130/G47841.1" target="_blank">https://doi.org/10.1130/G47841.1</a>, 2021a.
</mixed-citation></ref-html>
<ref-html id="bib1.bib20"><label>Chevalier et al.(2021b)Chevalier, Chase, Quick, and Scott</label><mixed-citation>
Chevalier, M., Chase, B. M., Quick, L. J., and Scott, L.: An atlas of southern African pollen types and their climatic affinities, Palaeoecol. Afr., 35, 239–258, <a href="https://doi.org/10.1201/9781003162766-15" target="_blank">https://doi.org/10.1201/9781003162766-15</a>, 2021b.
</mixed-citation></ref-html>
<ref-html id="bib1.bib21"><label>Cordova et al.(2017)Cordova, Scott, Chase, and Chevalier</label><mixed-citation>
Cordova, C. E., Scott, L., Chase, B. M., and Chevalier, M.: Late Pleistocene-Holocene vegetation and climate change in the Middle Kalahari, Lake Ngami, Botswana, Quaternary Sci. Rev., 171, 199–215, <a href="https://doi.org/10.1016/j.quascirev.2017.06.036" target="_blank">https://doi.org/10.1016/j.quascirev.2017.06.036</a>, 2017.
</mixed-citation></ref-html>
<ref-html id="bib1.bib22"><label>Costello et al.(2017)Costello, Tsai, Wong, Cheung, Basher, and Chaudhary</label><mixed-citation>
Costello, M. J., Tsai, P., Wong, P. S., Cheung, A. K. L., Basher, Z., and Chaudhary, C.: Marine biogeographic realms and species endemicity, Nat. Commun., 8, 1–9, <a href="https://doi.org/10.1038/s41467-017-01121-2" target="_blank">https://doi.org/10.1038/s41467-017-01121-2</a>, 2017.
</mixed-citation></ref-html>
<ref-html id="bib1.bib23"><label>Dupont et al.(2011)</label><mixed-citation>
Dupont, L. M., Caley, T., Kim, J.-H., Castañeda, I., Malaizé, B., and Giraudeau, J.: Glacial-interglacial vegetation dynamics in South Eastern Africa coupled to sea surface temperature variations in the Western Indian Ocean, Clim. Past, 7, 1209–1224, <a href="https://doi.org/10.5194/cp-7-1209-2011" target="_blank">https://doi.org/10.5194/cp-7-1209-2011</a>, 2011.
</mixed-citation></ref-html>
<ref-html id="bib1.bib24"><label>Dupont et al.(2019)</label><mixed-citation>
Dupont, L. M., Caley, T., and Castañeda, I. S.: Effects of atmospheric CO2 variability of the past 800 kyr on the biomes of southeast Africa, Clim. Past, 15, 1083–1097, <a href="https://doi.org/10.5194/cp-15-1083-2019" target="_blank">https://doi.org/10.5194/cp-15-1083-2019</a>, 2019.
</mixed-citation></ref-html>
<ref-html id="bib1.bib25"><label>Fick and Hijmans(2017)</label><mixed-citation>
Fick, S. E. and Hijmans, R. J.: WorldClim 2: new 1&thinsp;km spatial resolution climate surfaces for global land areas, Int. J. Climatol., 37, 4302–4315, <a href="https://doi.org/10.1002/joc.5086" target="_blank">https://doi.org/10.1002/joc.5086</a>, 2017.
</mixed-citation></ref-html>
<ref-html id="bib1.bib26"><label>Garcia et al.(2019a)Garcia, Weathers, Paver, Smolyar, Boyer, Locarnini, Zweng, Mishonov, Baranova, Seidov, and Reagan</label><mixed-citation>
Garcia, H. E., Weathers, K., Paver, C., Smolyar, I., Boyer, T., Locarnini, M., Zweng, M., Mishonov, A., Baranova, O., Seidov, D., and Reagan, J.: World Ocean Atlas 2018, Vol. 3: Dissolved Oxygen, Apparent Oxygen Utilization, and Dissolved Oxygen Saturation, NOAA Atlas NESDIS 83, 38 pp., <a href="https://archimer.ifremer.fr/doc/00651/76337/" target="_blank"/> (last access: September 2021), 2019a.
</mixed-citation></ref-html>
<ref-html id="bib1.bib27"><label>Garcia et al.(2019b)Garcia, Weathers, Paver, Smolyar, Boyer, Locarnini, Zweng, Mishonov, Baranova, Seidov, Reagan, and Others</label><mixed-citation>
Garcia, H. E., Weathers, K., Paver, C., Smolyar, I., Boyer, T., Locarnini, M., Zweng, M., Mishonov, A., Baranova, O., Seidov, D., and Reagan, J.: World Ocean Atlas 2018, Vol. 4: Dissolved Inorganic Nutrients (phosphate, nitrate and nitrate+nitrite, silicate), NOAA Atlas NESDIS 84, 35 pp., <a href="https://archimer.ifremer.fr/doc/00651/76336/" target="_blank"/> (last access: September 2021), 2019b.
</mixed-citation></ref-html>
<ref-html id="bib1.bib28"><label>Garnier et al.(2021)</label><mixed-citation>
Garnier, S., Ross, N., Rudis, boB, Filipovic-Pierucci, A., Galili, T., timelyportfolio, Greenwell, B., Sievert, C., Harris, D. J., and Chen, J. J. : sjmgarnier/viridis: viridis 0.6.0, Version v0.6.0pre, Zenodo [code], <a href="https://doi.org/10.5281/zenodo.4679424" target="_blank">https://doi.org/10.5281/zenodo.4679424</a>, 2021.
</mixed-citation></ref-html>
<ref-html id="bib1.bib29"><label>GBIF(2018)</label><mixed-citation>
GBIF: GBIF: Global Biodiversity Information Facility, <a href="https://www.gbif.org" target="_blank"/> (last access: August 2021), 2018.
</mixed-citation></ref-html>
<ref-html id="bib1.bib30"><label>GBIF(2020a)</label><mixed-citation>
GBIF: Beetles occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.nteruy" target="_blank">https://doi.org/10.15468/dl.nteruy</a>, 2020a.
</mixed-citation></ref-html>
<ref-html id="bib1.bib31"><label>GBIF(2020b)</label><mixed-citation>
GBIF: Chironomids occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set], <a href="https://doi.org/10.15468/dl.jv3wsh" target="_blank">https://doi.org/10.15468/dl.jv3wsh</a>, 2020b.
</mixed-citation></ref-html>
<ref-html id="bib1.bib32"><label>GBIF(2020c)</label><mixed-citation>
GBIF: Diatoms occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.vfr257" target="_blank">https://doi.org/10.15468/dl.vfr257</a>, 2020c.
</mixed-citation></ref-html>
<ref-html id="bib1.bib33"><label>GBIF(2020d)</label><mixed-citation>
GBIF: Foraminifera occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.692yg6" target="_blank">https://doi.org/10.15468/dl.692yg6</a>, 2020d.
</mixed-citation></ref-html>
<ref-html id="bib1.bib34"><label>GBIF(2020e)</label><mixed-citation>
GBIF: Rodentia occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.fscw6q" target="_blank">https://doi.org/10.15468/dl.fscw6q</a>, 2020e.
</mixed-citation></ref-html>
<ref-html id="bib1.bib35"><label>GBIF(2020f)</label><mixed-citation>
GBIF: Cycadopsidae occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.sfjzxu" target="_blank">https://doi.org/10.15468/dl.sfjzxu</a>, 2020f.
</mixed-citation></ref-html>
<ref-html id="bib1.bib36"><label>GBIF(2020g)</label><mixed-citation>
GBIF: Magnoliopsida occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.ra49dt" target="_blank">https://doi.org/10.15468/dl.ra49dt</a>, 2020g.
</mixed-citation></ref-html>
<ref-html id="bib1.bib37"><label>GBIF(2020h)</label><mixed-citation>
GBIF: Anthocerotopsida occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.t9zenf" target="_blank">https://doi.org/10.15468/dl.t9zenf</a>, 2020h.
</mixed-citation></ref-html>
<ref-html id="bib1.bib38"><label>GBIF(2020i)</label><mixed-citation>
GBIF: Gingkoopsidae occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.da9wz8" target="_blank">https://doi.org/10.15468/dl.da9wz8</a>, 2020i.
</mixed-citation></ref-html>
<ref-html id="bib1.bib39"><label>GBIF(2020j)</label><mixed-citation>
GBIF: Gnetopsidae occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.h2kjnc" target="_blank">https://doi.org/10.15468/dl.h2kjnc</a>, 2020j.
</mixed-citation></ref-html>
<ref-html id="bib1.bib40"><label>GBIF(2020k)</label><mixed-citation>
GBIF: Liliopsida occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.axv3yd" target="_blank">https://doi.org/10.15468/dl.axv3yd</a>, 2020k.
</mixed-citation></ref-html>
<ref-html id="bib1.bib41"><label>GBIF(2020l)</label><mixed-citation>
GBIF: Lycopodiopsida occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.ydhyhz" target="_blank">https://doi.org/10.15468/dl.ydhyhz</a>, 2020l.
</mixed-citation></ref-html>
<ref-html id="bib1.bib42"><label>GBIF(2020m)</label><mixed-citation>
GBIF: Pinopsidae occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/dl.x2r7pa" target="_blank">https://doi.org/10.15468/dl.x2r7pa</a>, 2020m.
</mixed-citation></ref-html>
<ref-html id="bib1.bib43"><label>GBIF(2020n)</label><mixed-citation>
GBIF: Polypodiopsida occurrence data, downloaded on 24 September 2020, The Global Biodiversity Information Facility (GBIF) [data set], <a href="https://doi.org/10.15468/dl.87tbp6" target="_blank">https://doi.org/10.15468/dl.87tbp6</a>, 2020n.
</mixed-citation></ref-html>
<ref-html id="bib1.bib44"><label>GBIF(2021a)</label><mixed-citation>
GBIF: Bryophyta occurrence data, downloaded on 2 August 2021, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/DL.WD527G" target="_blank">https://doi.org/10.15468/DL.WD527G</a>, 2021a.
</mixed-citation></ref-html>
<ref-html id="bib1.bib45"><label>GBIF(2021b)</label><mixed-citation>
GBIF: Marchantiophyta occurrence data, downloaded on 2 August 2021, The Global Biodiversity Information Facility (GBIF) [data set],  <a href="https://doi.org/10.15468/DL.M2SSE4" target="_blank">https://doi.org/10.15468/DL.M2SSE4</a>, 2021b.
</mixed-citation></ref-html>
<ref-html id="bib1.bib46"><label>Gibson et al.(2022)Gibson, McCoy, O'Keefe, Nuñez Otaño, Warny, and Pound</label><mixed-citation>
Gibson, M. E., McCoy, J., O'Keefe, J. M. K., Nuñez Otaño, N. B., Warny, S., and Pound, M. J.: Reconstructing terrestrial paleoclimates: a comparison of the Co-existence Approach, Bayesian and probability reconstruction techniques using the UK Neogene, Paleoceanography and Paleoclimatology, 37, e2021PA004358, <a href="https://doi.org/10.1029/2021PA004358" target="_blank">https://doi.org/10.1029/2021PA004358</a>, 2022.
</mixed-citation></ref-html>
<ref-html id="bib1.bib47"><label>Hijmans(2021)</label><mixed-citation>
Hijmans, R. J.: raster: Geographic Data Analysis and Modeling, r package version 3.4-13, <a href="https://CRAN.R-project.org/package=raster" target="_blank"/> (last access: February 2022), 2021.
</mixed-citation></ref-html>
<ref-html id="bib1.bib48"><label>Hui et al.(2021)Hui, Zhou, Chevalier, Wei, Pan, and Chen</label><mixed-citation>
Hui, Z., Zhou, X., Chevalier, M., Wei, X., Pan, Y., and Chen, Y.: Miocene East Asia summer monsoon precipitation variability and its possible driving forces, Palaeogeogr. Palaeocl., 581, 110609, <a href="https://doi.org/10.1016/j.palaeo.2021.110609" target="_blank">https://doi.org/10.1016/j.palaeo.2021.110609</a>, 2021.
</mixed-citation></ref-html>
<ref-html id="bib1.bib49"><label>Hutson(1978)</label><mixed-citation>
Hutson, W. H.: Application of transfer functions to Indian Ocean planktonic Foraminifera, Quaternary Res., 9, 87–112, <a href="https://doi.org/10.1016/0033-5894(78)90084-4" target="_blank">https://doi.org/10.1016/0033-5894(78)90084-4</a>, 1978.
</mixed-citation></ref-html>
<ref-html id="bib1.bib50"><label>Juggins(2013)</label><mixed-citation>
Juggins, S.: Quantitative reconstructions in palaeolimnology: new paradigm or sick science?, Quaternary Sci. Rev., 64, 20–32, <a href="https://doi.org/10.1016/j.quascirev.2012.12.014" target="_blank">https://doi.org/10.1016/j.quascirev.2012.12.014</a>, 2013.
</mixed-citation></ref-html>
<ref-html id="bib1.bib51"><label>Kühl et al.(2002)Kühl, Gebhardt, Litt, and Hense</label><mixed-citation>
Kühl, N., Gebhardt, C., Litt, T., and Hense, A.: Probability Density Functions as Botanical-Climatological Transfer Functions for Climate Reconstruction, Quaternary Res., 58, 381–392, <a href="https://doi.org/10.1006/qres.2002.2380" target="_blank">https://doi.org/10.1006/qres.2002.2380</a>, 2002.
</mixed-citation></ref-html>
<ref-html id="bib1.bib52"><label>Lim et al.(2016)Lim, Chase, Chevalier, and Reimer</label><mixed-citation>
Lim, S., Chase, B. M., Chevalier, M., and Reimer, P. J.: 50&thinsp;000 years of climate in the Namib Desert, Pella, South Africa, Palaeogeogr. Palaeocl., 451, 197–209, <a href="https://doi.org/10.1016/j.palaeo.2016.03.001" target="_blank">https://doi.org/10.1016/j.palaeo.2016.03.001</a>, 2016.
</mixed-citation></ref-html>
<ref-html id="bib1.bib53"><label>Lincoln(2020)</label><mixed-citation>
Lincoln, M.: clipr: Read and Write from the System Clipboard, r package version 0.7.1, <a href="https://CRAN.R-project.org/package=clipr" target="_blank"/> (last access: February 2022), 2020.
</mixed-citation></ref-html>
<ref-html id="bib1.bib54"><label>Locarnini et al.(2019)Locarnini, Mishonov, Baranova, Boyer, Zweng, Garcia, Reagan, Seidov, Weathers, Paver, Smolyar, and Others</label><mixed-citation>
Locarnini, R., Mishonov, A., Baranova, O., Boyer, T., Zweng, M., Garcia, H. E., Reagan, J., Seidov, D., Weathers, K., Paver, C., and Smolyar, I.: World ocean atlas 2018, Vol. 1: Temperature, NOAA Atlas NESDIS 81, 52pp., <a href="https://archimer.ifremer.fr/doc/00651/76338/" target="_blank"/> (last access: September 2021), 2019.
</mixed-citation></ref-html>
<ref-html id="bib1.bib55"><label>Myers et al.(2000)Myers, Mittermeier, Mittermeier, da Fonseca, and Kent</label><mixed-citation>
Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B., and Kent, J.: Biodiversity hotspots for conservation priorities, Nature, 403, 853–858, <a href="https://doi.org/10.1038/35002501" target="_blank">https://doi.org/10.1038/35002501</a>, 2000.
</mixed-citation></ref-html>
<ref-html id="bib1.bib56"><label>Olson et al.(2001)Olson, Dinerstein, Wikramanayake, Burgess, Powell, Underwood, D'amico, Itoua, Strand, Morrison, Loucks, Allnutt, Ricketts, Kura, Lamoreux, Wettengel, Hedao, and Kassem</label><mixed-citation>
Olson, D. M., Dinerstein, E., Wikramanayake, E. D., Burgess, N. D., Powell, G. V. N., Underwood, E. C., D'amico, J. A., Itoua, I., Strand, H. E., Morrison, J. C., Loucks, C. J., Allnutt, T. F., Ricketts, T. H., Kura, Y., Lamoreux, J. F., Wettengel, W. W., Hedao, P., and Kassem, K. R.: Terrestrial Ecoregions of the World: A New Map of Life on Earth: A new global map of terrestrial ecoregions provides an innovative tool for conserving biodiversity, BioScience, 51, 933, <a href="https://doi.org/10.1641/0006-3568(2001)051[0933:TEOTWA]2.0.CO;2" target="_blank">https://doi.org/10.1641/0006-3568(2001)051[0933:TEOTWA]2.0.CO;2</a>, 2001.
</mixed-citation></ref-html>
<ref-html id="bib1.bib57"><label>Overpeck et al.(1985)Overpeck, Webb III, and Prentice</label><mixed-citation>
Overpeck, J. T., Webb III, T., and Prentice, I. C.: Quantitative interpretation of fossil pollen spectra: Dissimilarity coefficients and the method of modern analogs, Quaternary Res., 23, 87–108, <a href="https://doi.org/10.1016/0033-5894(85)90074-2" target="_blank">https://doi.org/10.1016/0033-5894(85)90074-2</a>, 1985.
</mixed-citation></ref-html>
<ref-html id="bib1.bib58"><label>Pebesma and Bivand(2005)</label><mixed-citation>
Pebesma, E. J. and Bivand, R. S.: Classes and methods for spatial data in R, R News, 5, 9–13, <a href="https://CRAN.R-project.org/doc/Rnews/" target="_blank"/> (last access: February 2022), 2005.
</mixed-citation></ref-html>
<ref-html id="bib1.bib59"><label>Quick et al.(2021)</label><mixed-citation>
Quick, L. J., Chase, B. M., Carr, A. S., Chevalier, M., Grobler, B. A., and Meadows, M. E.: A 25&thinsp;000 year record of climate and vegetation change from the southwestern Cape coast, South Africa, Quaternary Res., 105, 1–18, <a href="https://doi.org/10.1017/qua.2021.31" target="_blank">https://doi.org/10.1017/qua.2021.31</a>, 2021.
</mixed-citation></ref-html>
<ref-html id="bib1.bib60"><label>R Core Team(2020)</label><mixed-citation>
R Core Team: R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, <a href="https://www.R-project.org/" target="_blank"/> (last access: February 2022), 2020.
</mixed-citation></ref-html>
<ref-html id="bib1.bib61"><label>R Special Interest Group on Databases (R-SIG-DB) et al.(2021)R Special Interest Group on Databases (R-SIG-DB), Wickham, and Müller</label><mixed-citation>
R Special Interest Group on Databases (R-SIG-DB), Wickham, H., and Müller, K.: DBI: R Database Interface, r package version 1.1.1, <a href="https://CRAN.R-project.org/package=DBI" target="_blank"/> (last access: February 2022), 2021.
</mixed-citation></ref-html>
<ref-html id="bib1.bib62"><label>Reynolds et al.(2007)Reynolds, Smith, Liu, Chelton, Casey, and Schlax</label><mixed-citation>
Reynolds, R. W., Smith, T. M., Liu, C., Chelton, D. B., Casey, K. S., and Schlax, M. G.: Daily high-resolution-blended analyses for sea surface temperature, J. Climate, 20, 5473–5496, <a href="https://doi.org/10.1175/2007JCLI1824.1" target="_blank">https://doi.org/10.1175/2007JCLI1824.1</a>, 2007 (data available at: <a href="https://psl.noaa.gov/data/gridded/data.noaa.oisst.v2.highres.html" target="_blank"/>, last access: September 2021).
</mixed-citation></ref-html>
<ref-html id="bib1.bib63"><label>Schauberger and Walker(2020)</label><mixed-citation>
Schauberger, P. and Walker, A.: openxlsx: Read, Write and Edit xlsx Files, r package version 4.2.3, <a href="https://CRAN.R-project.org/package=openxlsx" target="_blank"/> (last access: ), 2020.
</mixed-citation></ref-html>
<ref-html id="bib1.bib64"><label>Soetaert(2021)</label><mixed-citation>
Soetaert, K.: plot3D: Plotting Multi-Dimensional Data, r package version 1.4, <a href="https://CRAN.R-project.org/package=plot3D" target="_blank"/> (last access: February 2022), 2021.
</mixed-citation></ref-html>
<ref-html id="bib1.bib65"><label>ter Braak and van Dame(1989)</label><mixed-citation>
ter Braak, C. J. F. and van Dame, H.: Inferring pH from diatoms: a comparison of old and new calibration methods, Hydrobiologia, 178, 209–223, <a href="https://doi.org/10.1007/BF00006028" target="_blank">https://doi.org/10.1007/BF00006028</a>, 1989.
</mixed-citation></ref-html>
<ref-html id="bib1.bib66"><label>ter Braak and Juggins(1993)</label><mixed-citation>
ter Braak, C. J. F. and Juggins, S.: Weighted averaging partial least squares regression (WA-PLS): an improved method for reconstructing environmental variables from species assemblages, Hydrobiologia, 269, 485–502, <a href="https://doi.org/10.1007/BF00028046" target="_blank">https://doi.org/10.1007/BF00028046</a>, 1993.
</mixed-citation></ref-html>
<ref-html id="bib1.bib67"><label>Valavi et al.(2021)</label><mixed-citation>
Valavi, R., Guillera-Arroita, G., Lahoz-Monfort, J. J., and Elith, J.: Predictive performance of presence-only species distribution models: a benchmark study with reproducible code, Ecol. Monogr., 92, e01486, <a href="https://doi.org/10.1002/ecm.1486" target="_blank">https://doi.org/10.1002/ecm.1486</a>, 2021.
</mixed-citation></ref-html>
<ref-html id="bib1.bib68"><label>Wickham(2011)</label><mixed-citation>
Wickham, H.: The Split-Apply-Combine Strategy for Data Analysis, J. Stat. Softw., 40, 1–29, <a href="http://www.jstatsoft.org/v40/i01/" target="_blank"/> (last access: February 2022), 2011.

</mixed-citation></ref-html>
<ref-html id="bib1.bib69"><label>Wickham(2019)</label><mixed-citation>
Wickham, H.: stringr: Simple, Consistent Wrappers for Common String Operations, r package version 1.4.0, <a href="https://CRAN.R-project.org/package=stringr" target="_blank"/> (last access: February 2022), 2019.
</mixed-citation></ref-html>
<ref-html id="bib1.bib70"><label>Wickham and Hesselberth(2020)</label><mixed-citation>
Wickham, H. and Hesselberth, J.: pkgdown: Make Static HTML Documentation for a Package, r package version 1.6.1, <a href="https://CRAN.R-project.org/package=pkgdown" target="_blank"/> (last access: February 2022), 2020.
</mixed-citation></ref-html>
<ref-html id="bib1.bib71"><label>Wickham and Seidel(2020)</label><mixed-citation>
Wickham, H. and Seidel, D.: scales: Scale Functions for Visualization, r package version 1.1.1, <a href="https://CRAN.R-project.org/package=scales" target="_blank"/> (last access: February 2022), 2020.
</mixed-citation></ref-html>
<ref-html id="bib1.bib72"><label>Wickham et al.(2020)Wickham, Hester, and Chang</label><mixed-citation>
Wickham, H., Hester, J., and Chang, W.: devtools: Tools to Make Developing R Packages Easier, r package version 2.3.2, <a href="https://CRAN.R-project.org/package=devtools" target="_blank"/> (last access: February 2022), 2020.
</mixed-citation></ref-html>
<ref-html id="bib1.bib73"><label>Wickham et al.(2021)Wickham, Ooms, and Müller</label><mixed-citation>
Wickham, H., Ooms, J., and Müller, K.: RPostgres: “Rcpp” Interface to “PostgreSQL”, r package version 1.3.3, <a href="https://CRAN.R-project.org/package=RPostgres" target="_blank"/> (last access: February 2022), 2021.
</mixed-citation></ref-html>
<ref-html id="bib1.bib74"><label>Wright(2021)</label><mixed-citation>
Wright, K.: pals: Color Palettes, Colormaps, and Tools to Evaluate Them, r package version 1.7, <a href="https://CRAN.R-project.org/package=pals" target="_blank"/> (last access: February 2022), 2021.
</mixed-citation></ref-html>
<ref-html id="bib1.bib75"><label>Yi et al.(2020)Yi, Jun, Jo, Lee, Kim, Lee, Cao, and Lim</label><mixed-citation>
Yi, S., Jun, C. P., Jo, K.-n., Lee, H., Kim, M. S., Lee, S. D., Cao, X., and Lim, J.: Asynchronous multi-decadal time-scale series of biotic and abiotic responses to precipitation during the last 1300 years, Sci. Rep., 10, 1–10, <a href="https://doi.org/10.1038/s41598-020-74994-x" target="_blank">https://doi.org/10.1038/s41598-020-74994-x</a>, 2020.
</mixed-citation></ref-html>
<ref-html id="bib1.bib76"><label>Zomer et al.(2008)Zomer, Trabucco, Bossio, and Verchot</label><mixed-citation>
Zomer, R. J., Trabucco, A., Bossio, D. A., and Verchot, L. V.: Climate change mitigation: A spatial analysis of global land suitability for clean development mechanism afforestation and reforestation, Agr. Ecosyst. Environ., 126, 67–80, <a href="https://doi.org/10.1016/j.agee.2008.01.014" target="_blank">https://doi.org/10.1016/j.agee.2008.01.014</a>, 2008.
</mixed-citation></ref-html>
<ref-html id="bib1.bib77"><label>Zweng et al.(2018)Zweng, Seidov, Boyer, Locarnini, Garcia, Mishonov, Baranova, Weathers, Paver, Smolyar, and Others</label><mixed-citation>
Zweng, M., Seidov, D., Boyer, T., Locarnini, R., Garcia, H. E., Mishonov, A., Baranova, O., Weathers, K., Paver, C., and Smolyar, I.: World Ocean Atlas 2018, Vol. 2: Salinity, NOAA Atlas NESDIS 82, 50 pp., <a href="https://archimer.ifremer.fr/doc/00651/76339/" target="_blank"/> (last access: September 2021), 2018.
</mixed-citation></ref-html>--></article>
