Multi-decadal climate variability in southern Iberia during the mid- to late- Holocene

To assess the regional multi-decadal to multi-centennial climate variability at the southern Iberian Peninsula during the midto late10 Holocene transition multi-proxy records of two marine sediment cores were established for two sites in the Alboran Sea (ODP-161-976A) and the Gulf of Cadiz (GeoB5901-2). High-resolution records of organic geochemical proxies and planktic foraminiferal assemblages are used to decipher precipitation and vegetation changes as well as the sea surface conditions with respect to Sea Surface Temperature (SST) and marine primary productivity (MPP). n-Alkane records as a proxy for precipitation changes suggest a series of six distinct drought events at 5.4 ka BP, from ca. 5.1 ka BP to 4.9 ka BP, from 4.8 to 4.7 ka BP, at 4.6 ka BP, from 4.4 to 4.3 ka BP and, from 3.8 to 3.7 ka BP. Each drought event is 15 associated with a major vegetation change towards higher proportions of C4 vegetation. The drought events are further accompanied by annual and spring/ winter SST warming as well as decreasing MPP in the Alboran Sea. Altogether, the close correlation of the observed droughts with North Atlantic Oscillation (NAO)-like variability suggests changes in the atmospheric circulation as important driving mechanism of terrestrial and oceanic variability at southern Iberia and the Alboran Sea, respectively. Sea surface variability in the Gulf of Cadiz, instead, is intimately linked to the North Atlantic Bond Events. In particular, during Bond Events 3 and 4 a pronounced increase in seasonality is found. 20


Study area
The Iberian Peninsula is influenced by two major climatic regimes: the Atlantic and the Mediterranean regime (Lionello, 2012). Today, the Atlantic climate is typically marked by relatively cool annual air temperatures and evenly distributed precipitation throughout the year. In contrast, the Mediterranean climate is generally characterized by pronounced seasonal contrasts with a rainy winter season and a dry and hot summer season. In general, most of the precipitation at the Iberian Peninsula falls during the winter season ( Figure 1; Lionello, 2012). The 5 spatial pattern of winter precipitation also reflects the two climatic regimes (Figure 1). The Atlantic regime, which spans along the western and northern coasts, is characterized by high precipitation of more than 800 ml per year. The precipitation falling in the Atlantic climate is associated with the westerly wind belt at the Iberian Peninsula (Zorita et al., 1992) and is to a large extent controlled by the North Atlantic Oscillation (NAO) during winter Hurrell, 1995). During positive NAO conditions (i.e. a high pressure difference between the Azores High and the Icelandic Low) the North Atlantic storm track and associated precipitation is shifted towards northern Europe, while Iberia 10 experiences more dry conditions. On the other hand, during negative NAO conditions (i.e. a low pressure difference) more storms are directed towards the Iberian Peninsula, which experiences wetter winter conditions. Nonetheless, the central parts of the Iberian Peninsula as well as the eastern and southern coasts under influence of the Mediterranean climate remain relatively dry on average with precipitation less than 600 ml per year and also during winter during winter ( Figure 1).
The oceanic conditions at the Atlantic margin of the Iberian Peninsula are also characterized by pronounced seasonality. Coastal upwelling 15 along the western continental margin is responsible for relatively low SST during summer ( Figure 1; Haynes et al., 1993). This upwelling system eventually spreads into the Gulf of Cadiz (Haynes et al., 1993). Warm Atlantic water passes the Strait of Gibraltar and enters the Alboran Sea.
Within the Alboran Sea it circulates in two anti-cyclonic gyres (Lionello, 2012). At the western gyre relatively warm and fresh waters become upwelled (Sarhan et al., 2000). This upwelling is accompanied by an elevated MPP at the northern rim of the gyre (Minas et al., 1991).

Sediment cores and sampling
For this study two marine sediment cores from the Alboran Sea and the Gulf of Cadiz were analysed. Sediment core ODP-161-976A (36°12.32' N; 4°18.76' W; 1108 m water depth) was retrieved in the Alboran Sea during JOIDES Resolution cruise in 1995 (Zahn et al., 1999). To achieve multi-decadal resolution, the section from 103.0 cm to 145.0 cm of the working half was continuously sampled at 0.5 cm distances. An extended section from the archive half of this core (100.0 cm to 149.0 cm) was resampled on 0.5 cm resolution in the IODP Core Repository at MARUM 25 in Bremen (Germany) to obtain additional samples for AMS 14 C dating and planktic foraminiferal assemblage analysis. Due to an earlier Clim. Past Discuss., https://doi.org /10.5194/cp-2018-158 Manuscript under review for journal Clim. Past Discussion started: 19 November 2018 c Author(s) 2018. CC BY 4.0 License. sampling of the archive half, we were just able to sample every second centimetre. Sediment core GeoB5901-2 (36°22.80' N; 7°04.28' W; 574 m water depth) was retrieved during METEOR cruise in 1999 in the Gulf of Cadiz (Schott et al., 2000). This sediment core was sampled on 0.5 cm resolution from 1.0 cm to 50.0 cm in the GeoB Core Repository at MARUM in Bremen (Germany). During this study only the section from 12.5 cm to 29.5 cm was analysed. While most of the samples from ODP-161-976A (the ones sampled in 2015) were already available freezedried and used for organic geochemical analyses exclusively, samples of sediment core GeoB5901-2 as well as the resampled section from 5 sediment core ODP-161-976A were subsampled for organic geochemical and foraminiferal analyses. The samples used for organic geochemical analysis were subsequently freeze-dried.

Age model
Age models of both cores are based on existing and new AMS 14 C dates, all measured at Leibniz Laboratory at Kiel University (Table 1).
Published data of cores ODP-161-976A and GeoB5901-2 were taken from Combourieu Nebout et al. (2002) and Kim et al. (2004), respectively. 10 In addition, we measured six new AMS 14 C dates on monospecific planktic foraminiferal samples of Globigerinoides ruber white + pink or Globigerina bulloides larger than 150 µm from sediment core ODP-161-976A as well as seven new dates from sediment core GeoB5901-2. AMS 14 C data were calibrated and existing data recalibrated using the marine13 calibration curve and Calib7.0.4 software (Stuiver and Reimer, 1993) applying a global marine reservoir correction of 400 years. In the case of sediment core ODP-161-976A the initial age model of Jiménez-Amat and Zahn (2015) through the additional age control points within the studied time period was shifted massively. In this core, for the 15 double-dated samples yielding slightly differing 14 C ages at the same sample depths (120.25 cm and 132.75 cm) the age with the lowest analytical error was chosen. Two additional dates (at 116.25 cm and 124.75 cm) have been oppositely excluded, because they resulted in the same age as the sample at 120.25 cm depth (Table 1). In order to achieve the final age-depth models for both sediment cores, we linearly interpolated between the two closest AMS 14 C age control points. Accordingly, sedimentation rates range between 10.8 and 47.0 cm/kyr in core ODP-161-976A and between 4.9 and 68.6 cm/kyr in core GeoB5901-2, respectively and at 0.5 sample intervals enabling thus decadal to multi-decadal resolution. 20 The studied section of sediment core ODP-161-976A dates between 5.40 to 2.92 ka BP exhibiting a temporal resolution between 10 to 60 years per sample with continuous high resolution between 5.40 and 4.62 ka BP. The analysed section of sediment core GeoB5901-2 dates between 3.07 ka BP and 5.33 ka BP resulting in a temporal resolution between 20 and 95 years per sample. Age models and sedimentation rates for both sediment cores are shown in Figure 2.

Organic geochemical analysis and calculations
Lipids were extracted from the freeze-dried and finely ground sediment samples with an Accelerated Solvent Extractor (ASE-200, Dionex) at 100 bar and 100 °C using a 9:1 (v/v) mixture of dichloromethane (DCM) and methanol (MeOH). After extraction samples were de-sulphured by stirring for 30 minutes with activated copper. The de-sulphured lipids were subsequently separated by silica gel column chromatography using activated silica gel (450 °C for 4 h) into neutral (hexane) and polar (DCM) fractions containing n-alkanes and alkenones, respectively. 5 The neutral fraction was further separated using silver-nitrate (AgNO3) coated silica gel. Samples were then left at room temperature for approximately 24 hours for homogenisation.
Afterwards, n-alkanes were analysed by gas chromatography (GC) using an Agilent 6890N gas chromatograph equipped with a Restek XTI-5 capillary column (30 m x 320 µm x 0.25 µm) and a Flame Ionization Detector (FID) at Christian-Albrechts University Kiel. n-Alkanes were identified by comparison of their retention times with an external standard containing a series of n-alkane homologues of known concentration. 10 On this basis, n-alkanes were also quantified using the FID peak areas calibrated against the external standard. For environmental interpretation the sum of terrestrial sourced, odd n-alkane homologues C27 to C33 is used. The mean analytical error (2σ) is 7.0 ng/ g sediment based on replicate analyses (n = 62).
Various studies used ratios among individual n-alkane homologues as environmental sensitive parameter, for example the Norm33 ratio (e.g. (Herrmann et al., 2016). The Norm33 ratio was calculated by the following equation: 15 where CX is the peak area of the n-alkane with x carbon atoms in the chromatogram. The mean analytical error (2σ) is 0.004 based on replicate analyses (n = 62).
Alkenones were analysed on a multi-dimensional, double gas column chromatography (MD-GC) set up with two Agilent 6890 gas chromatographs. The compounds (C37:2 and C37:3) were quantified by calibration to an external standard. The alkenone concentration is derived 20 from the sum of the C37:2 and C37:3 isomers. The alkenone unsaturation index (U 37 K' ) was obtained by using the peak areas of the aforementioned compounds applying the equation of Prahl and Wakeham (1987): where CX represents the respective peak area in the chromatogram. The U 37 K' index was subsequently transferred into annual mean SST using the calibration of Müller et al. (1998): 25 The laboratory internal analytical error is approximately 0.12 °C, while the error of the calibration is 1.0 °C.

Planktic foraminiferal analysis and Modern Analogue Technique
For foraminiferal analysis sediment samples of approximately 10 ccm were washed over 63 μm sieves, dried at 40 °C and, subsequently dry sieved. Planktic foraminifera assemblages were analysed in the size fractions >150 μm enabling the application of commonly used transfer techniques for SST reconstructions based on relative abundances of 26 taxonomic categories within the assemblage, following concepts by Pflaumann et al. (1996). For reliable assemblage counts samples were dry split into aliquots of at least 300 specimens with a Kiel dry sample 5 splitter. For SST estimates we used SIMMAX non-distance-weighted modern analogue technique, using a similarity index of >0.963 and based on 10 closest analogues (Pflaumann et al., 2003). A calibration data set was used combining 1066 core top assemblages from both, the North Atlantic as compiled by the MARGO project (Hayes et al., 2005;Kucera et al., 2005a;Kucera et al., 2005b) and for the Mediterranean . Modern temperature at 10 m water depth present-day SST was retrieved from the World Ocean Atlas 1998 .
Seasonal temperatures are averaged for northern hemisphere summer (July to September) and for winter (December to February). This method 10 as applied on core top samples yields an accuracy of better than 1 °C standard deviations for both, the Atlantic and the Mediterranean for summer and winter seasons (Hayes et al., 2005;Pflaumann et al., 2003;Salgueiro et al., 2014).

Proxy restrictions
Terrestrial n-alkanes derived from higher plant leaf waxes are commonly transported into the marine realm via aeolian and riverine transport (Bird et al., 1995;Conte and Weber, 2002;Schreuder et al., 2018). The fraction of the riverine input in coastal areas is usually much higher 15 compared to the aeolian input. Therefore, terrestrial n-alkane concentrations have already been successfully applied to study the riverine input in coastal settings in the Mediterranean (Abrantes et al., 2017;Cortina et al., 2016;Jalali et al., 2016;Jalali et al., 2017). More specifically, Rodrigo-Gámiz et al. (2015) have shown from radiogenic isotopes that in the Alboran Sea the riverine dominates the aeolian input during the studied time period. Consequently, the n-alkane concentrations analysed during this study are considered to primarily reflect the river discharge.
The Norm33 ratio or the proportion of the C33 n-alkane homologue, respectively, is a function of a change in C4 plant distribution according to 20 air temperature and/ or precipitation change (Bush and McInerney, 2013;Herrmann et al., 2016;Leider et al., 2013;Rommerskirchen et al., 2006;Vogts et al., 2009;Vogts et al., 2012). Because of the dominant riverine transport mechanism, it is assumed that both proxies integrate spatially over the river catchment areas shown in Figure 1 and, thus, indicate climate conditions on a sub-regional level. The dominant main catchment for sediment core GeoB5901-2 is the one of the Guadalquivir river draining into the Gulf of Cadiz. The smaller mountainous rivers draining the southern Sierra Nevada area are considered as more relevant for sediment core ODP-161-976A. A delivery of material from the 25 Guadalquivir with the inflow of the Atlantic water through the Strait of Gibraltar might also be possible, though. Since the river discharge and Clim. Past Discuss., https://doi.org /10.5194/cp-2018-158 Manuscript under review for journal Clim. Past Discussion started: 19 November 2018 c Author(s) 2018. CC BY 4.0 License. also the plant growing season is strongly coupled to precipitation and the rainy season at the Iberian Peninsula is in winter, both proxies are probably biased towards the winter season ( Figure 1; Lionello, 2012).

Results
While the reconstructions from ODP-161-976A cover a period from 5.4 to 2.9 ka BP on high-resolution, the organic geochemical data from GeoB5901-2 covers a slightly shorter period from 5.3 to 3.0 ka BP also on generally lower resolution. The planktic foraminiferal data from 5 GeoB5901-2 are shown for the period from 6.0 to 2.6 ka BP.
The terrestrial n-alkane concentration vary between 67 and 714 ng/g sediment in ODP-161-976A and between 9 and 535 ng/g sediment in GeoB5901-2 ( Figure 3). While in ODP-161-976A n-alkane concentrations appear generally stable and no trends are recognizable, the data from GeoB5901-2 reveal an increasing trend towards younger ages. In the later core several distinct and brief changes towards concentrations below 100 ng/g sediment are observed from 5.1 to 4.7, from 4.4 to 4.3 and, from 3.8 to 3.7 ka BP. In ODP-161-976A minima of similar amplitude 10 occur at 5.4, at 4.9, from 4.8 to 4.7, at ca. 4.6 as well as at 3.7 ka BP and, thus, are generally contemporaneous. High and overall stable concentrations above 500 ng/g sediment are observed between 4.3 and 3.8 ka BP in ODP-161-976A. The Norm33 of both sediment cores show no trends varying between 0.29 and 0.49 ( Figure 3). Norm33 maxima in sediment core ODP-161-976A range from 5.4 to 5.3, from 5.0 to 4.9 and, from 3.0 to 2.9 ka BP. In addition, short-lived maxima occur at ca. 4.8, at ca. 4.6 and, at 4.2 ka BP. In GeoB5901-2 Norm33 maxima occur at 5.0, 4.7, 4.3 and, 3.7 ka BP. 15 The alkenone derived SST also exhibits relatively stable levels in both sediment cores with annual temperatures varying between 18.9 °C and 20.0 °C in ODP-161-976A and higher annual temperatures in GeoB5901-2 (20.0 °C to 22.7 °C; Figure 3). Notably, the ODP-161-976A record shows higher variability between 5.4 and 4.6 ka BP with maximal amplitudes of 1.0 °C varying around the mean of 19.4 °C. After 4.6 ka BP the mean annual SST slightly cools to very stable means of 19.1 °C. This stability is just interrupted by three short-lived maxima at 4.4, 3.7 and, 3.3 ka BP. warmings between 1-2 °C amplitude are observed at 6.0, from 5.9 to 5.6, from 5.5 to 5.4, at 4.7 and, at 4.1 ka BP. The summer warmings are accompanied by coolings of similar amplitude during winter. The seasonality slightly decreases towards younger ages from SST differences of 6.4 °C to 4.7 °C. During the summer warmings and winter coolings, respectively, the seasonal difference increases up to 9.3 °C.
Altogether, the data of both sediment cores allow the detailed comparison of seasonal climate variability in the terrestrial and marine realm during the mid-to late-Holocene in southern Iberia.

Terrestrial and marine environmental conditions in southern Iberia 10
The terrestrial n-alkane concentration record of GeoB5901-2 from the Gulf of Cadiz generally well matches the high resolution data of ODP-161-976A from the Alboran Sea (Figure 3). A notable difference appears at around 5.0 ka BP, when the GeoB5901-2 data records low concentrations since 5.1 ka BP, while low concentrations in ODP-161-976A appear not until 4.9 ka BP. This might be the result of dating uncertainties and/or due to the different river catchments with the ODP-161-976A data reflecting more the coastal Mediterranean climate, while the GeoB5901-2 data indicates more the climate variability of the hinterland and the northern Sierra Nevada. Apart from that time lag the 15 amplitude of changes towards lower concentrations are remarkably similar, suggesting that both archives jointly robustly record the wider regional climatic conditions of the southern Iberian Peninsula.
Since the concentration of terrestrial n-alkanes in marine sediment cores strongly depend on the amount of river discharge, we interpret the periods of very low concentrations in the studied sediment cores as dry periods. Accordingly, drought episodes occurred in southern Iberia at 5.4, from ca. 5.1 to 4.9, from 4.8 to 4.7, at around 4.6, from 4.4 to 4.3 and, from 3.8 to 3.7 ka BP. All of these droughts are already well 20 documented within dating uncertainties in various records from the area (Figure 4). A widespread trend towards drier conditions since ca. 5.5 ka BP is not only described from pollen sequences in southern Iberia (Fletcher and Sánchez Goñi, 2008;Jalut et al., 2009) but is also well in phase with the end of the African Humid Period (deMenocal et al., 2000). This transition appears to be marked by an aridity event at ca. 5.5 ka BP as evidenced by speleothem data from El Refugio Cave and Grotte de Piste (Walczak et al., 2015;Wassenburg et al., 2016) and high charcoal concentration in Cabo de Gata between ca. 5.4 and 5.3 ka BP, respectively. Also, a contemporaneous remarkable decrease in Mediterranean BP in Lake Medina (SW Iberia). The drought events between ca. 5.1 and 4.9, from 4.8 to 4.7 and, at ca. 4.6 ka BP observed in this study are corroborated by several regional speleothem records Walczak et al., 2015;Wassenburg et al., 2016) as well as a charcoal record from Cabo de Gata (Burjachs and Expósito, 2015). A dramatic forest decline occurred between 5.0 and 4.5 ka BP in SE Iberia . Moderate forest declines are found in pollen records from the Alboran Sea and Elx sequence (Burjachs and Expósito, 2015;Fletcher and Sánchez Goñi, 2008). The drought event from 4.4 to 4.3 ka BP broadly coincides with the 4.2 ka event observed speleothem records 5 from the Eastern and Central Mediterranean Zanchetta et al., 2016). According to Burjachs and Expósito (2015)    and, speleothem data Walczak 10 et al., 2015;Wassenburg et al., 2016). The latest drought found in this study between 3.75 and 3.73 ka BP, although just lasting ca. 20 years, is described in pollen data from Elx and Villaverde (Burjachs and Expósito, 2015;Carrión et al., 2016), while pollen data from the Alboran Sea indicate a moderate forest decline (Fletcher and Sánchez Goñi, 2008). Additionally, indications for a severe dry event around 3.7 ka BP again come from speleothem data Walczak et al., 2015).
Notably, in our records all observed drought episodes are paralleled by Norm33 maxima (Figure 4). These maxima indicate a shift towards a 15 higher abundance of C4 plants, which are much more adapted to drier (and warmer) conditions (e.g. Bush and McInerney, 2013). Moreover, our data gains support from Sierra Nevada bog sediments Borreguil de la Virgen and Borreguil de la Caldera (Figure 4; García-Alix et al., 2018), which also indicate an increase in C4 plant abundance during the droughts. In addition, pollen data from Elx and Padul record an increase in C4 grasses (Poaceae) at ca. 5.4, 5.0, 4.8, 4.4 and, 3.7 ka BP (Burjachs pers. comm., 2018;Ramos-Román et al., 2018b). Accordingly, the droughts caused a dramatic vegetation shift although they occurred relatively fast within decades or even within years. This further implies that 20 the vegetation in southern Iberia at that time was very sensitive to (winter) precipitation changes.
In contrast to the terrestrial variability our marine reconstructions suggest fairly stable conditions between 5.4 and 2.9 ka BP in the Alboran Sea and the Gulf of Cadiz (Figure 3). Alkenone derived annual mean SSTs from sediment core ODP-161-976A suggest temperatures around 19.2 °C, which are just slightly exceeding the modern annual SST of 18.0 °C (Locarnini et al., 2013). An overall cooling trend over the course of the Holocene in response to decreasing insolation is known on regional as well as on global scale (e.g. Cacho et al., 2001;NGRIP-Members, 2004), 25 thus corroborating slightly warmer SSTs during the mid-and late-Holocene. Moreover, annual mean SSTs around 19 °C at that time are also found by other studies from the Alboran Sea (Cacho et al., 2001;Martrat et al., 2014;Pérez-Folgado et al., 2003;Rodrigo-Gámiz et al., 2014 Locarnini et al., 2013). Winter SSTs around 14.5 °C corroborating our results are also found by Pérez-Folgado et al. (2003), while they found warmer SSTs by about 2 °C during summer. On the other hand, our summer SST reconstructions confirm recent reconstructions from Rodrigo-Gámiz et al. (2014), who based on the long-chain diol index found summer SSTs around 22 °C. Some more subtle cooling events with SST declines of up to 1 °C occurred at 4.9, 4.6, 4.3, 3.6 and, 3.1 ka BP during summer and are accompanied by warming events of similar magnitude during winter. Consequently, 5 seasonality diminished during these periods. Unfortunately, the low temporal resolution of available reference cores from the area hampers the correlation of such brief SST events. Detailed comparison between our SST data and the MPP data reveals a correlation between warmer annual mean SSTs and decreased MPP at around 5.4, 5.2, 4.9, 4.7, 4.6, 4.4, 3.7 and, 3.0 ka BP ( Figure 6). Comparison with the SSTs events observed in the seasonal reconstructions is difficult due to the lower temporal resolution. Nonetheless, it appears that also spring SST increases during the MPP minima. Notably, the warming event from ca. 3.3 to 3.1 ka BP is not well reflected in the MPP data. Moreover, maximum MPP 10 between 4.11 and 3.88 ka BP can be related with stable and relatively low annual and spring SSTs.
In the Gulf of Cadiz annual mean SSTs vary between 20.0 to 22.7 °C and are much warmer compared to recent temperatures (18.7 °C;Locarnini et al., 2013). Previous analyses of Kim et al. (2004) on the same sediment core yielded mean annual SSTs between 19 and 20 °C using a different SST calibration of Prahl et al. (1988), which would result in slightly cooler SSTs for our data as well. Annual mean SSTs around 20 °C in Gulf of Cadiz were also found by Cacho et al. (2001). The here reconstructed annual mean SSTs appear actually more close to summer conditions 15 ( Figure 3). From the foraminiferal analysis we reconstruct summer SSTs around 22.1 °C, which compare well with modern summer SSTs of 21.7 °C at the location (Locarnini et al., 2013). Supportingly, Salgueiro et al. (2014) found summer SSTs around ca. 23 °C in the Gulf of Cadiz.
Also the reconstructed mean winter temperatures of 16.4 °C are in good agreement with the modern SSTs of 16.0 °C (Locarnini et al., 2013).
During summer warming events of 1-2 °C amplitude are observed at 5.99, from 5.86 to 5.64, from 5.49 to 5.42, at 4.65 and, at 4.11 ka BP and these events are accompanied by cooling events during winter of similar amplitude. These events, notably, differ from the warm period found 20 in the alkenone derived SST from 4.26 to 3.79 ka BP.

Possible drivers of terrestrial and marine climate variability
During modern times the NAO is responsible for much of the winter precipitation at the Iberian Peninsula (Hurrell, 1995;Zorita et al., 1992).
Many studies show that NAO-like variability was also evident during the Holocene (Abrantes et al., 2017;Deininger et al., 2017;Olsen et al., 2012). We here compare our drought events to a recently published North Atlantic storminess record from Filsø, Denmark ( Figure 5; Goslin et 25 al., 2018). Periods of increased storminess are indicating positive NAO-like conditions, which compare well with the drought events observed in this study. Accordingly, we conclude that the droughts observed between 5.4 and 2.9 ka BP in southern Iberia are very likely caused by Clim. Past Discuss., https://doi.org /10.5194/cp-2018-158 Manuscript under review for journal Clim. Past Discussion started: 19 November 2018 c Author(s) 2018. CC BY 4.0 License. positive NAO-like conditions. It has to be cautioned, though, that other atmospheric circulation patterns like the Scandinavian and the East Atlantic pattern might also contribute to the observed precipitation changes (Abrantes et al., 2017;Hernández et al., 2015). It has also been suggested that NAO-like variability in the past is responsible for oceanic changes such as the position of the Azores front (Goslin et al., 2018;Repschläger et al., 2017) and increased upwelling as well as SST changes at the western Iberian margin (Abrantes et al., 2017). In more detail, Abrantes et al. (2017) suggest that warm winters and cool summers during the Medieval Climate Anomaly are related to positive NAO-like 5 conditions at the western Iberian margin. During the mid-to late-Holocene we cannot confirm a general link between the atmospheric and the oceanic conditions in the Gulf of Cadiz. While during the drought event between 4.4 and 4.3 ka BP our seasonal SSTs suggest no change, the alkenone-derived SST indicate a pronounced warming. On the other hand, during the drought event from 3.8 to 3.7 ka BP the Gulf of Cadiz experienced a sudden cooling in the annual mean SST, which lasted ca. 500 years. But in general, we observe two notable differences between the oceanic conditions in the Gulf of Cadiz and the Alboran Sea: (1) the summer/winter SST variability in the Gulf of Cadiz is opposite resulting 10 in a greater seasonality, while seasonality decreases during summer/winter variability in the Alboran Sea and (2)  Skagerrak and the northern North Atlantic, respectively. For the Subtropics our record from the Gulf of Cadiz is the first indicating increased seasonality from SST data associated with Bond Events. The Alboran Sea, instead, is not related to the SST development in the North Atlantic (i.e. Bond Events). But, the annual as well as seasonal SST development relate to the MPP, where higher annual mean temperatures are related to decreased productivity ( Figure 6). Also, warmings during spring, which is the main blooming season of coccoliths in the Alboran Sea (Bosc 20 et al., 2004), relate to MPP minima. These changes are most likely caused by dynamics of the Western Alboran Gyre. Ausín et al. (2015) also proposed a coupling of a stable water column (i.e. higher SST and decreased upwelling) and low productivity in the Alboran Sea with dry conditions in southern Iberia. Indeed, our data corroborates the presence of drier conditions during SST increases in the annual mean and spring temperatures and low productivity in the Alboran Sea. Following Ausín et al. (2015)  Clim. Past Discuss., https://doi.org /10.5194/cp-2018-158 Manuscript under review for journal Clim. Past Discussion started: 19 November 2018 c Author(s) 2018. CC BY 4.0 License.

Conclusion
Our terrestrial proxies of both sediment cores agree well, revealing four drought events in southern Iberia from ca. 5.1 to 4.9, from 4.8 to 4.7, from 4.4 to 4.3 and, from 3.8 to 3.7 ka BP. The 4.2 ka event might be reflected in the drought occurring from 4.4 to 4.3 ka BP. Additional droughts are observed in the high-resolution core from the Alboran Sea at 5.4 and at 4.6 ka BP. All drought events occurred very fast within decades or even within years and are accompanied by a shift towards more C4 vegetation. The droughts observed in southern Iberia are closely 5 correlating with NAO-like variability suggesting that the atmospheric circulation was an important driver of terrestrial variability in southern Iberia. Moreover, the oceanic variability observed in the Alboran Sea with higher annual and seasonal (winter and spring) SSTs and lower MPP seems to be driven by NAO-like variability as well. According to Ausín et al. (2015) positive NAO-like conditions alter the gyre circulation by limiting the Atlantic inflow through the Strait of Gibraltar, which results in a stable water column and decreased upwelling. This oceanic variability in the Alboran Sea is coinciding with droughts in southern Iberia. Contrastingly, oceanic variability in the Gulf of Cadiz shows no 10 relation to the droughts or NAO-like variability, respectively. In fact, the Gulf of Cadiz reveals an opposite pattern compared to the Alboran Sea with summer warming and winter cooling. Also, the observed SST events in the Gulf of Cadiz and the Alboran Sea do not match temporarily implying different driving mechanisms. We found that the variability in the Gulf of Cadiz is related to Bond Events 3 and 4 in particular. Bond Event 2 is not resolved in our data. During Bond Events 3 and 4 we found a sharp increase in seasonality (i.e. summer warming and winter cooling) not described for this area before. 15 Consequently, we found that climate variability at the southern Iberian Peninsula during the mid-to late-Holocene strongly depends on North Atlantic forcing. While a North Atlantic atmospheric variability similar to modern day NAO is responsible for terrestrial as well as oceanic changes in the Alboran Sea, oceanic variability in the Gulf of Cadiz is more related to the marine forcing in the North Atlantic (i.e. Bond Events).
Data availability. The data reported in this paper are archived in Pangaea (www.pangaea.de). 20 Author contribution. JS, the main author of this study (with contributions from all co-authors), performed the geochemical analyses underlying the precipitation, vegetation and, alkenones temperature reconstructions. JS also established the new age models for the two sediment cores.
Planktic foraminiferal analyses underlying the seasonal SST reconstructions were provided by MW. ES provided the compiled calibration data sets for the SIMMAX MAT analyses and computed SIMMAX SST. TB and RS supported the processing of the biomarker in the lab at the 25 Institute of Geosciences (CAU Kiel) with the alkenone evaluation.