Sea-level fluctuations relative to the modern sea level are well constrained for the last 0.5 Myr (e.g. Imbrie et al., 1984; Martinson et al., 1987; Bassinot et al., 1994; Chappell et al., 1996; Dickinson, 2001; Siddall et al., 2003; Rabineau et al., 2005; Lambeck and Purcell, 2005; Lisiecki and Raymo, 2005; Antonioli et al., 2007). When these fluctuations are used in conjunction with dating techniques, they provide a powerful tool for interpreting coastal geomorphology and assessing vertical deformation from marine and marginal marine deposits through the middle and late Quaternary (e.g. Pirazzoli et al., 1996; Rabineau et al., 2005; Antonioli et al., 2007; Mouslopoulou et al., 2015a). While there is little debate about the role of tectonic uplift in generating the topographic relief required for the processes of erosion and deposition, uncertainty still exists as to the relative significance of tectonic, eustatic and climatic contributions to deposition and incision of fans of the Quaternary age and their variation through time (e.g. Waters et al., 2010).

Alluvial fans are excellent proxies for Quaternary landscape evolution in a climate such as the Mediterranean and their study could potentially place some constraints on the factors that impacted the landscape during its formation and evolution (e.g. Pope et al., 2008, 2016; Zacharias et al., 2009). Overall, alluvial fan deposition is influenced by a rising or relatively high sea level, by catchment size and sediment supply, by major changes in climatic conditions (such as high rainfall and/or short, intense storms), and by vegetation coverage in the catchment area. These factors regulate stream carrying capacity and sediment supply (e.g. Bull, 1990; Pope et al., 2016) and are responsible for whether deposition or river entrenchment processes predominate at any one time. Alluvial fan surface abandonment and river entrenchment is favoured by eustatic sea-level fall or tectonic uplift (or a combination of both), reduction in sediment supply due to climatic amelioration, densification of catchment vegetation, or reduced rainfall (e.g. Pope et al., 2008, 2016; Waters et al., 2010). Fan aggradation is encouraged by factors such as rising base level (sea level for Domata), sediment supply, increased stream carrying capacity (rainfall and/or temporal rainfall distribution) and reduction in catchment vegetation cover.

For example, in southwest Crete (eastern Mediterranean), Nemec and Postma (1993) and Pope et al. (2008, 2016) studied a fan system and showed that fan deposition was associated with all last glacial stadial and interstadial conditions and that fan entrenchment was governed by the major climatic transitions between MIS5/4 and MIS2/1. Conversely, Tiberti et al. (2014) and Mouslopoulou et al. (2015b) have shown that tectonic uplift on western Crete is significant, reaching rates of 7–8 mmyr-1 during the late Quaternary. Despite this progress in understanding, we are, however, still unable to precisely appreciate the interplay between, and the relative importance of, climate and tectonics during the late Quaternary in the Mediterranean. How do severe uplift rates recorded along the forearc of the Hellenic margin (where Crete lies) reconcile with periods of prevalence of eustatic processes? Here we capitalise on a well-preserved alluvial fan system at Domata in southern Crete (Fig. 1) to study the late Quaternary (∼50 kyr BP) interplay between sea-level fluctuations and tectonics. To our knowledge, the site at Domata is unique on the island of Crete as each alluvial fan-building episode has been followed by distinct episodes of alluvial incision and subsequent marine trimming (Figs. 2 and 3). Thus, at this site, we have the opportunity to test the idea proposed by Pope et al. (2016) that eustasy largely controls the landscape evolution in southwest Crete.

Using luminescence dating together with the Siddall et al. (2003) sea-level curve, we find that the alluvial fan system at Domata was consecutively affected by (1) sea-level fluctuations, triggering building of the fans and subsequent river and marine incision between ∼45 and 20 kyr BP, during a period of minimal tectonic activity, and (2) intense tectonic uplift between ∼20 ka and present, at rates that exceeded those of the rising sea level, resulting in the preservation of the entire fan sequence. These findings are in accord with Pope et al. (2016) in showing that regional tectonics did not necessarily play a key role in fan incision in southern Crete.

The Mediterranean island of Crete is a mountainous and elongate landmass (∼260 km long from west to east, 60 km wide from north to south) that lies within the uplifted forearc section of the Hellenic subduction margin, the most seismically active region in Europe (Fig. 1). The total relative convergence rate between the subducting African plate and the overriding Eurasian plate is ∼35–40 mmyr-1 (Reilinger et al., 2010; Fig. 1). The subduction trench lies ∼225 km to the south of Crete (e.g. Ryan et al., 1970; Le Pichon and Angelier, 1979) while the north-dipping subduction interface lies at a depth of ∼40 to 65 km beneath Crete (Papazachos et al., 2006; Vernant et al., 2014), with the projection of the down-dip end of the locked zone aligning with the southern coastline of Crete, where the study area is located (Fig. 1) (Meier et al., 2007). The Hellenic Trough, a major bathymetric and tectonic feature within the forearc, lies south of Crete and includes three secondary features, which, from west to east, are named as the Ptolemy, Pliny and Strabo troughs (Fig. 1).

Crete has been characterised by a complex history of vertical movements during the Cenozoic (e.g. Peters et al., 1985). Onshore sediments record a period of subsidence and basin development through the middle and late Miocene (Serravallian to Messinian) with a change to rapid uplift in the early Pliocene (Zanclean), followed by slower long-term uplift that continues to the present day (e.g. Le Pichon and Angelier, 1981; Angelier et al., 1982; Meulenkamp et al., 1994; Zachariasse et al., 2008; Roberts et al., 2013; Gallen et al., 2014).

Late Quaternary tectonic uplift on Crete is uniform but transient (Tiberti et al., 2014; Mouslopoulou et al., 2015b). Using dated palaeo-shorelines and numerical models, Tiberti et al. (2014) and Mouslopoulou et al. (2015b) show that the island of Crete has experienced, during the last 20 thousand years, periods of severe fluctuation in its vertical deformation (at rates of up to 8 mmyr-1), while in the preceding ∼30 thousand years, vertical movement on Crete was either minimal or reversed (subsidence of 0–3 mmyr-1; Tiberti et al., 2014). High uplift rates (∼7–8 mmyr-1) have also been documented on western Crete since 2 kyr BP, in response to co-seismic uplift (that locally reached up to 10 m) (Pirazzoli et al., 1996; Shaw et al., 2008; Mouslopoulou et al., 2015a). Uplift rate transients on Crete are thought to result from non-uniform stress accumulation and release on upper-plate reverse faults in the overriding plate (Shaw et al., 2008; Stiros, 2010; Tiberti et al., 2014; Mouslopoulou et al., 2015b).

Historical and archaeological records have also been used to link uplift in western Crete to earthquakes (Pirazzoli et al., 1982, 1996; Stiros, 2001; Papazachos and Papazachou, 2003; Papadimitriou and Karakostas, 2008; Shaw et al., 2008; Stefanakis, 2010; Strasser et al., 2011). In particular, historic accounts of a major earthquake in Crete in ∼ AD 365 are approximately coincident with historic documents recording tsunami inundation of parts of the Libyan and Egyptian coastlines, particularly Alexandria (Ammianus Marcellinus, translated by C. D. Yonge, 1862; Polonia et al., 2013). A gently tilted palaeo-shoreline (tidal notch) can be followed along the western shoreline of Crete for ∼150 km from the area of maximum uplift near the southwest tip (Elafonisi) to as far east as Agios Georgios (Fig. 1). At Domata, our study site, this notch is at 6 m a.s.l. A number of studies attribute the timing of this prominent palaeo-shoreline to the historic AD 365 earthquake (e.g. Pirazzoli et al., 1982, 1996; Stiros, 2001; Shaw et al., 2008).

At Domata a unique sequence of two juxtaposed generations of alluvial fans is documented, each truncated by different episodes of river and marine incision (see Figs. 2, 3 and 4). The discussion that follows gives a detailed account of the materials and the geometry of the alluvial fan system at Domata, establishes the stratigraphic relationships of its key geomorphic features and provides the chronological framework (sequence of events) within which the established stratigraphic relationships developed.

According to our luminescence age range and the relative location of the fan surfaces below the marine terrace (see Fig. 2) of inferred MIS5 age, the majority of the geomorphic landforms at Domata we discuss formed during MIS 3 (29–57 kyr) (Table 1 and Figs. 2 and 7). The sequence of events that resulted in the development of the landscape at Domata, as dictated by the analysis of the geomorphic landforms and the superposition of the luminescence dating onto the well-established sea-level curve of Siddall et al. (2003)

The shape of the sea-level fluctuations varies slightly globally. Thus, in this work we designate the empirical, high-resolution sea-level curve of Siddall et al. (2003) from the Red Sea that covers the last 128 kyr. Siddall et al. (2003) estimate the error in their sea-level curve at ∼12 m, thus, all sea-level elevations discussed subsequently are subject to this uncertainty.

The initiation of deposition of the upper fan (Fig. 10a) has occurred post ∼50 kyr BP and prior to 45 kyr (RB-1 sample; within MIS 3 of Lisiecki and Raymo, 2005; Table 1); this coincides with a period of elevated sea level (ca. -70 to -80 m according to the sea-level curve of Siddall et al. (2003); Fig. 9). We argue that upper-fan deposition cannot have started as early as 53.4 kyr (mean of RB-1 sample; Table 1) because there is no geomorphic evidence (e.g. marine cliffs) representing the two subsequent high sea-level stands at ca. 45 and 39 kyr. Thus, the deposition of the upper fan post-dates 53 kyr and is completed by ∼45 kyr BP, when the sea level rose to reach ca. -72 m (Siddall et al., 2003). The period that follows, between 45 and 41 kyr, reflects the start of a cooling climatic period (possibly with an increase in sediment supply) and a falling sea level to -87 m (Fig. 9). During this period, alluvial fan entrenchment resulted in fan surface abandonment and development of the alluvial cliff in the upper-fan deposits (Fig. 3), and the upper fan was eventually abandoned (Fig. 10b). This was followed, at ∼39 kyr BP, by marine transgression resulting in marine trimming of the upper-fan surface, the alluvial entrenchment cliff and development of a sub-horizontal marine abrasion surface in early upper-fan deposits at about the sea level of the time (ca. -70 m) (see yellow dashed line on the river cliff in Figs. 3, 9 and 10c). This allowed up to 10 000 years for upper-fan deposition and incision before the fan is trimmed by the sea during the high sea-level peak at ca. 39 kyr.

Lower-fan deposition commenced (Fig. 10d) soon after marine trimming of the upper fan (Fig. 9). The relatively high sea level at 37 kyr (ca. -75 m; see Fig. 9) promoted fan deposition, and the possibly deteriorating climatic conditions, involving episodically increased river flow and/or carrying capacity and diminishing vegetation density in the upper catchment, resulted in increased sediment supply. Lower-fan deposits lapped against alluvial entrenchment and marine cliffs cut previously in upper-fan deposits by alluvial incision along the Klados River and by the marine trimming sub-parallel to the modern shoreline (Fig. 10d). Lower-fan surface abandonment and river entrenchment through incision commenced at ∼36 kyr (LF-2a/b sample; Table 1). We argue that the lower-fan surface was abandoned (Fig. 10e) sometime between ∼36 kyr and 29 kyr due to river entrenchment resulting from a rapidly falling sea level (that continued until ∼18 kyr) (Fig. 9).

Marine trimming of the lower-fan surface and deposits cannot have occurred between deposition and the last glacial maximum (∼18 kyr), as the sea level progressively declined during that period. Following 18 kyr, the sea level rose rapidly by ∼100 m in less than 10 kyr (Fig. 9). The marine trimming of the lower-fan surface and deposits occurred during the Holocene high sea-level stand. This is expected to have commenced as the sea level approached roughly the present level ca. 4–5 kyr ago (Fig. 10f). Between 18 and 5 kyr BP, while the sea level was rising fast, tectonic uplift at Domata must have outpaced the rising sea level, protecting the entire sequence from marine inundation and destruction. Immediately prior to the co-seismic uplift that affected western Crete at AD 365 (Pirazzoli et al., 1982), the foot of the lower marine cliff would have been within the intertidal zone. Today, due to the 6 m of uplift associated with that earthquake, the prominent stranded palaeo-shoreline and the foot of the marine-trimmed cliff are at approximately similar levels and the sea cliff may be isolated from further trimming (Figs. 6b and 10g). We interpret a low terrace riser at about this elevation near the mouth of the Klados River (see lower white dashed line in Fig. 6b) to be a relict from the river channel of that time, and in the lower ca. 100 m of its course the river has downcut in response to that first-millennium earthquake uplift.

In summary, deposition of the upper and lower fan was controlled by a marine base level and, in both cases, fan incision resulted due to the falling sea level. The deposition of the upper fan was largely completed by ∼45 kyr BP, during a period of relatively high sea level (ca. -70 m), and fan incision resulting in surface abandonment occurred between ca. 45 and 40 kyr. Marine trimming of the upper-fan deposits occurred during a sea-level high (ca. -72 m) at ca. 39 kyr. Lower-fan deposition was initiated soon afterwards and fan surface abandonment occurred between 36 and 29 kyr. The age of the sandy unit that mantles both the upper- and the lower-fan surfaces is ∼25 kyr (UF-2; Fig. 9 and Table 1), post-dating both fans as expected by its stratigraphic relationship with respect to the fans (silty sand that mantles both the lower and upper fans).

The fan sequence at Domata provides a unique opportunity to link terrestrial deposition with sea-level fluctuations and vertical tectonics on southwestern Crete. Geomorphic analysis combined with dating shows that the development of the fan sequence can be accounted for by eustatic changes coupled with vertical tectonics. The latter can be rationalised if we consider that the landforms at Domata were formed 70 to 90 m (within the known error margins) below the current sea level, implying that, unless tectonic uplift was significant, the entire sequence would have been inundated, and thus modified or destroyed, by the rising sea level during the last 20 kyr (Siddall et al., 2003). A requirement of its preservation is that since its deposition, the tectonic uplift rate has outpaced the rising sea level. Thus, a question that arises concerns the relationship between the rate of tectonic uplift and rising sea level at Domata. Dating key terrestrial and marginal marine geomorphological features at Domata provided an average uplift rate for this part of Crete that can be compared to the rate of rising sea level and also to other rates of tectonic uplift on western Crete (which have been independently derived). It also provides a means of testing the main finding of Pope et al. (2016) that, during the late Quaternary, the landscape at the nearby site of Sfakia mainly responded to sea-level and climatic changes.

Our data show that the marine trimming episode at ca. 39 kyr left a coastal cliff and cut an erosional intertidal abrasion surface in the upper-fan deposits (see yellow dashed line in Fig. 3). This surface provides a good datum upon which to estimate subsequent uplift. Indeed, a total uplift of 86 m (± the 12 m error margin of the sea-level curve) is required to elevate this marine abrasion surface to its current altitude of 14 ma.s.l. Thus, the minimum uplift rate required to accomplish this is ca. 2.2 mmyr-1. Indeed, the plot in Fig. 11 shows that with an average rate of ∼2.2 mmyr-1 since formation (black line), the fan sequence would have escaped the destructive interaction with the wave zone and, therefore, modification due to erosion (e.g. the black line of uniform uplift rate does not intersect the sea-level curve during the last 39 kyr). Independent support for similar uplift rates comes from published radiocarbon ages on beachrock materials that mantle marine palaeo-shorelines in nearby localities: a calibrated radiocarbon age of 36 790–38 694 yr BP from reworked rhodoliths in beachrock at an elevation of 10.5 m at Sougia, 9 km to the west of Domata, is within a few thousand years of the proposed timing of marine trimming of the upper fan and yields an average uplift rate of 2.4 mmyr-1 (Mouslopoulou et al., 2015a). Similarly, beachrock on a marine terrace at 17 m elevation at Palaiochora, 20 km west of Domata, yields a calibrated radiocarbon age of 36 682–38 732 yr BP, producing an average uplift rate of 2.5 mmyr-1 (Mouslopoulou et al., 2015a). Comparable uplift rates (1.8–2.7 mmyr-1) have been independently recorded for the last ∼40 000 years at numerous localities on western Crete by Shaw et al. (2008), Strasser et al. (2011) and Tiberti et al. (2014). Thus, the preservation and subaerial exposure of the landscape at Domata is due to the sufficient tectonic uplift that southern Crete experienced during the late Quaternary.

In order to quantify the relative contribution of tectonics and eustasy to the formation of the landscape at Domata, here we compare published information on incremental uplift rates calculated by Tiberti et al. (2014) and Mouslopoulou et al. (2015b) for western Crete over the last ∼40 kyr with the uplift rate calculated for Domata over the last 39 kyr (ca. 2.2 mmyr-1; this study). Comparison shows that during the time period over which the key features at Domata formed (MIS 3), no significant uplift was accommodated on Crete as the region was experiencing a tectonically quiet period with no uplift (Mouslopoulou et al., 2015b) or even gentle subsidence between ∼20 and 45 kyr (Tiberti et al., 2014). Thus, the shaping of the landscape at Domata during MIS 3 must have been largely achieved by sea-level fluctuations. This comparison also suggests that most of the ∼80 m of uplift (subtracting 6 m of late Holocene co-seismic uplift) has been accumulated sometime between 5 and 20 kyr BP, resulting in an average uplift rate of 5.3 mmyr-1 (Fig. 11; dashed red line). However, this uplift rate would have been insufficient to outpace the rising sea level between 8 and 12 kyr BP (see dashed red line intersecting the sea-level curve in Fig. 11) and thus the fan sequence would have been inundated and modified or destroyed by the rising sea level. This, in turn, implies that the uplift rate at Domata was higher than 5.3 mmyr-1. We favour a scenario in which uplift was mostly accommodated by about 9 kyr BP, at an average rate of ∼7 mmyr-1 (see solid red line in Fig. 11 that does not intersect the sea-level curve). Comparable uplift rates (7–8 mmyr-1) have been independently recorded at numerous localities on western and eastern Crete for the last 20 000 years by Tiberti et al. (2014) and Mouslopoulou et al. (2015b) and they result from transient earthquake slip on upper-plate faults that splay off the plate interface and extend beneath Crete at relatively steep angles (Mouslopoulou et al., 2015b). Such transient rates have been observed at several other margins globally (Mouslopoulou et al., 2016, and references therein).

Thus, the development and evolution (∼50–20 kyr BP) of the suite of geomorphic features at Domata can be largely explained by eustatic sea-level fluctuations and sedimentation variations controlled by climatic conditions, without the requirement for significant vertical movements. This conclusion largely supports the main finding of Pope et al. (2016), who dated a fan sequence at Sfakia (ca. 25 km to the east of Domata). However, it is the subsequent tectonic uplift that preserved and subaerially exposed the coastal geomorphic features.

Intriguing conclusions of the Pope et al. (2016) high-resolution dating work also include that three sometimes overlapping phases of fan deposition since the last interglacial are separated by two phases of fan entrenchment, the first close to the MIS 5–4 (ca. 70 kyr) boundary, the other close to the MIS 2–1 boundary (ca. 14 kyr), triggered by major climatic changes. Fan deposition at Sfakia has to a large degree persisted through stadial and interstadial periods during the last 125 kyr. Periods of entrenchment at Sfakia do not appear to correlate with the two entrenchment periods at Domata. The Sfakia fan is somewhat different from the Domata fan in catchment size (ca. 28 km2 compared with ca. 11 km2), fan size (5.3 km2 compared with 0.1 km2), the presence of more than one feeder channel at Sfakia, and in the nature of deposits (primarily clast-supported gravels at Sfakia compared with primarily matrix-supported gravels at Domata). Whether these differences are responsible for differences in depositional and entrenchment histories and in preservation of marine cliffs at Sfakia is uncertain. However, one conclusion of the work of Pope et al. (2016) compatible with our own is recognition of the importance of base-level (sea-level) changes in the process of incision.

Alluvial fans often provide a useful index with which to decode the information recorded on the landscape in complex tectonic settings, such as those of the eastern Mediterranean. Herein we use analysis of geomorphic landforms and luminescence dating on an alluvial fan system with two separate periods of depositional activity on Crete, an island straddling the forearc of the Hellenic subduction margin, to constrain its vertical deformation and discuss the contributing factors responsible for its landscape evolution. Our interpretations suggest that sea-level fluctuations in response to varying climatic conditions formed the landscape at Domata during MIS 3 (∼57–29 kyr BP). It is, however, because of the fast tectonic uplift that Crete experienced during the subsequent ∼20 thousand years that the entire alluvial sequence escaped destruction and/or modification due to marine inundation and is subaerially preserved today. Thus, both eustasy and tectonism impacted the formation and preservation of the landscape at Domata, but over temporally distinct time periods.