Modern environment in Celestun Lagoon
Distributions of flora and fauna
Foraminiferal and geochemical analyses in surface sediments and water along Celestun Lagoon reveal differences between the northern and southern lagoon regions. The Agglutinated-Haynesina assemblage occurs only in the northern lagoon, whereas the Bolivina-Hanzawaia-Rosalina assemblage occurs only in the southern lagoon (Fig. 4b). The Ammonia-Elphidium assemblage is common in both regions. The Trichohyalus and Archaias-Laevipeneroplis assemblages were not observed in the modern lagoon. Maximum species richness (15) and diversity (1.96) both occur in the middle of the lagoon, about 13 km from the southern opening, where assemblages overlap (Fig. 4b) and the habitat transitions from a protected low-salinity lagoon to a higher-energy marine-influenced lagoon. The assemblage distribution aligns with gradients in salinity, temperature, δ13C, and C:N of OM (Fig. 6b). Strong association (clustering) of foraminifera assemblages with specific salinity and carbon isotope ranges suggests that the distribution of benthic foraminifera in Celestun Lagoon is controlled in part by salinity, vegetation, and, to a lesser extent, temperature. PERMANOVA results support significant relationships among these variables (Table 4).
Clusters dominated by the Agglutinated-Haynesina assemblage represent brackish salinity and shallow lagoon mangroves, and clusters dominated by the Bolivina-Hanzawaia-Rosalina assemblage represent inner marine shelf and seagrass habitat, which are the environments in which species of these assemblages have been found around the Caribbean (Table 3). Low wave energy and muddy substrate in the enclosed northern lagoon also promote the abundance of agglutinated foraminifera like Miliammina fusca (Debenay and Guillou 2002). Lagoon water temperature is weakly correlated with cluster composition (Table 4, Fig. 6b), but does not exhibit large changes in mean temperature among sites sampled (29.2 ± 2.6 °C, Table 1). Salinity exhibits the greatest variability in the lagoon. Although Trichohyalus and Archaias-Laevipeneroplis were not observed in modern Celestun, we infer their habitats based on modern occurrence at other sites in the Gulf of Mexico and Caribbean. Trichohyalus aguayoi occurs in inland mangrove ponds in Bermuda (Javaux and Scott 2003) and groundwater-flooded blue holes in the Bahamas (van Hengstum et al. 2020), and thus represents low-salinity, isolated mangrove ponds with minimal ocean influence (Table 3). Archaias angulatus and Laevipeneroplis proteus (also known as Peneroplis proteus) inhabit marine seagrass of the genus Thalassia off the coasts of Belize and Florida (Hallock and Peebles 1993; Sen Gupta 1999). Thus, the Archaias-Laevipeneroplis assemblage represents marine-salinity, shallow-shelf environments with Thalassia seagrass vegetation (Table 3).
In the marine-brackish conditions of southern Celestun Lagoon, euryhaline Ammonia and Elphidium were found alongside epiphytic Quinqueloculina and Rosalina taxa, suggesting seagrass vegetation may contribute to assemblage diversity, but alone is insufficient in determining modern assemblage distribution. The importance of salinity is further demonstrated by benthic foraminifera observed 10 km south of Celestun Lagoon, in Isla Arena, which has similar sedimentology, geometry, and vegetation as Celestun, but higher salinity because of greater seawater input (Lowery and Rankey 2017). As a result of this higher salinity, the Isla Arena assemblages are dominated by miliolid and peneroplid foraminifera that prefer near-marine salinity.
Variable salinity in coastal environments restricts foraminifera occurrence to only those that can withstand a large and fluctuating salinity range (Sen Gupta 1999). Such conditions occur in the northern lagoon and result in low species richness (low S values, Figs. 4b, 6b). The pattern of assemblage change along a salinity gradient is found in other Caribbean and Atlantic coastal systems with large salinity ranges. On the mangrove forest coast of French Guiana, Debenay et al. (2002) noted that seasonal salinity oscillations from 20 to 35 psu are accompanied by assemblage compositional changes from Miliammina-dominated (agglutinated) to Ammonia- and Elphidium-dominated. A narrow estuary of Brazil exhibited shallow protected brackish-water assemblages of agglutinated Miliammina and Ammotium taxa and seaward marine-influenced assemblages dominated by A. tepida, E. gunteri, and Bolivina taxa (Debenay et al. 1998), similar to the southern opening of Celestun Lagoon. Collectively, these studies point to salinity as being the most important environmental variable controlling foraminifera distribution in Celestun Lagoon. Based on these associations, we interpret changes in downcore assemblages as representing changes in salinity in the northern lagoon over time.
We acknowledge that analyzing the > 125-µm size fraction may exclude some foraminifera species or smaller adult specimens of the species we identified (Table 3, Murray 2006). However, benthic foraminifera at our site and other coastal Yucatan Peninsula sites are characterized by abundant large (> 250 µm) specimens (Martinez et al. 2018), suggesting we observed most taxa present. Based on the number of taxa identified (n = 26, Table 3), the number of specimens counted (n = 27,000), and the striking differences in assemblage composition observed in the > 125-µm fraction (Fig. 4), we consider the statistical relationships noted between environmental parameters and assemblage composition (Table 4) to be real patterns for the modern system, despite exclusion of small foraminifera. Furthermore, δ13C and C:N of bulk sediment OM are independent of foraminifera size fractions and are proxies for mangrove and seagrass vegetation that are also distributed along the lagoon salinity gradient (Gonneea et al. 2004; Khan et al. 2017; Carnero-Bravo et al. 2018), supporting interpretation of environments inferred from assemblages. It is possible that drying the samples prior to counting may affect preservation of agglutinated taxa (Murray and Alve 1999) and bias assemblage composition. However, the hyaline foraminifera Haynesina germanica observed in the low-energy muddy substrate preferred by agglutinated taxa provides an additional constraint on the distribution of the Agglutinated-Haynesina assemblage in space and time. We thus are confident in using assemblages defined in the modern environment to infer downcore, paleoenvironmental conditions, specifically changes in salinity.
Lagoon salinity changes—geomorphic processes and climate changes
The salinity in a restricted lagoon such as Celestun Lagoon can vary as a consequence of two distinct processes that affect the relative contribution of seawater and fresh/brackish groundwater inputs to the lagoon. The first is physical (geomorphic) changes to the coastline (i.e., lagoon restriction or sea level changes) that determine the amount of seawater input to the lagoon, and the second is climate changes that impact regional and local precipitation, which control the amount of groundwater input.
Coastal lagoon morphology in the Yucatan Peninsula was studied by Brady (1972), which remains the most thorough compilation of epikarst sedimentological processes for the peninsula. Brady indicates that modern barrier islands evolved from Pleistocene aeolian dunes on the continental shelf. Once formed, dunes were submerged by rising seas and provide shallow sites to initiate submarine barrier island formation and subaerial aeolian accumulation. Island growth may be episodic, particularly when hurricane overwash adds or removes sediments, but mangrove encroachment rapidly stabilizes island shorelines and facilitates additional shoreline sediment accumulation. Importantly, sediment cores of Nichupte and Holbox Lagoons (Fig. 1c) reflect sharp contacts between basal mangrove peat and overlying carbonate mud, indicating sea level rise and lagoon formation, but are subsequently characterized by faunal shifts from inner-shelf miliolids and clams to restricted lagoon Elphidium and encrusting algae, along with a fining grain size (Brady 1972). Faunal change was attributed to barrier island growth, but constraints on accumulation rate were poor.
More recent work indicates that longshore sediment transport along the Yucatan Peninsula coastline can build barrier island complexes within decades. Based on the observed net 200 m per 33 years of barrier island progradation offshore near Celestun Lagoon (Fig. 1b, Lowery and Rankey 2017), a 600 m barrier can develop within a century, and around 30 km can accumulate in 5000 years. Volumetric rates of sediment transport (35,000 m3 yr−1) along the northwest Yucatan Peninsula coast are sufficient to create the estimated volume of sediment of the Celestun west bank (5.3 × 107 m2 × 3 m elevation) on this timescale (Appendini et al. 2012). These rates of barrier island development are consistent with gradual accretion of the west Celestun Lagoon bank, resulting in an increase in lagoon length over time and a parallel gradual decrease in the mixing and input of seawater into the northern lagoon. Specifically, sediment deposition at the Celestun coast has resulted from both a Pleistocene dune high point, as indicated by Brady (1972), and a change of shoreline orientation relative to the west-flowing Yucatan Current. Cuevas et al. (2013) noted that offshore dunes tend to accumulate at abrupt changes in shoreline orientation on the north coast of the Yucatan Peninsula, and both the northwest (Celestun) and northeast (Holbox) corners of the Peninsula exhibit extensive barrier-dune complexes.
We note similar barrier-dune complexes from Google Earth™ near lagoons of Holbox, Ria Lagartos, Dzilam de Bravo, and Progreso, on the north coast of the Yucatan Peninsula, and near Isla Arena and Terminos on the west coast (Fig. 1c). In Terminos Lagoon, the dune-barrier complex called Isla de Carmen consists of 38 km of shelly coquina (Phleger and Ayala-Castanares 1971). Based on the progradation rate determined by Lowery and Rankey (2017), Isla de Carmen took 6300 years to accumulate, consistent with the island’s estimated age of 7000 years based on Holocene sea level terraces (Phleger and Ayala-Castanares 1971). These studies all indicate that the rate and scale of longshore transport and barrier island formation observed today were similar over the Late Holocene.
Although the gradual restriction of seawater into Celestun Lagoon will result in a gradual decrease in salinity, variability superimposed on this trend may be linked to climate-induced changes in groundwater input. Specifically, an increase in the salinity would suggest lower precipitation and possible drought conditions.
Reconstruction of the northern lagoon environmental setting
Based on comparison between observations from core-top sediments along the present-day lagoon transect and downcore data from core 4A in the northern lagoon, we suggest that the northern lagoon environment has experienced four phases of development, characterized by distinct environmental settings controlled by climate, sea-level rise, and coastal geomorphic change (Fig. 7). Based on general agreement between cores 4A and UL-5 (ESM), these phases are representative of the entire northern lagoon.
Phase 1 (5300–5100 BP)
Between 5300 and 5100 BP, the northern area of Celestun Lagoon was within a mangrove forest that was isolated from ocean influence. The peat δ13Corg value of − 28.26‰ indicates mangroves and other terrestrial vegetation (Gonneea et al. 2004; Young et al. 2005; Lamb et al. 2006; Tamalavage et al. 2018) (Fig. 2a), and peat accumulation suggests the absence of oxidants like dissolved oxygen and marine sulfate, consistent with a water-logged freshwater swamp setting. The oligohaline assemblage Trichohyalus observed during this interval is found today in isolated, brackish, groundwater-fed pond environments in Bermuda, supporting interpretation of minimal marine influence (Javaux and Scott 2003). Regional sea level at that time is estimated to have been 2–3 m lower than today (Khan et al. 2017), and a bathymetric slope of − 0.4 m km−1 off the northwest Campeche Bank near Celestun (Appendini et al. 2012) would place the coastline 5 km to the northwest of the current location of Celestun Lagoon, consistent with reduced marine influence. Groundwater was likely provided via springs in the permeable Ring of Cenotes because this Cretaceous-age feature existed prior to Holocene coastal development of the Yucatan Peninsula (Perry et al. 2002). The environment may have resembled the “Petenes” that are present south of Celestun Lagoon today, a periodically flooded black mangrove forest characterized by other marsh and subtropical vegetation, diffuse groundwater inputs, organic-rich sediments, and gastropods (Lowery and Rankey 2017). Based on the low-salinity habitat of Trichohyalus in Bermuda (Javaux and Scott 2003) and modern groundwater discharge salinity values (Herrera-Silveira 1996), salinity was likely less than 10 psu in these mangrove ponds and probably closer to 3 psu, the value of modern groundwater at this site (Young et al. 2008).
We know of no published climate records that span the period between 5300–5100 BP in the northwest Yucatan Peninsula. However, pollen records suggest mesic (semi-moist) tropical forests in the southwest and northeast areas of the Yucatan Peninsula and in Haiti ca. 5200–4800 BP (Hodell et al. 1991; Higuera-Gundy et al. 1999; Aragón-Moreno et al. 2012; Vela-Pelaez et al. 2018). Regionally wet conditions would provide groundwater to inland mangrove ponds to support Trichohyalus assemblages whose presence in this environment distinguishes this phase from subsequent, more saline conditions in the lagoon.
Phase 2 (5100–3000 BP)
This phase of over 2000 years is characterized by both changes in foraminifera assemblages and sources of organic matter to the lagoon. Between 5100 and 4900 BP, the peat in the core transitions to carbonate mud (Fig. 2a) as δ13Corg increases from − 28.26 to − 20.21‰ and C:N decreases from 25 to 11 (Fig. 4a). The Trichohyalus assemblage disappears from Celestun Lagoon, and at 4900 BP the Archaias-Laevipeneroplis assemblage appears alongside the Ammonia-Elphidium assemblage (Fig. 4a). The δ13Corg shift to less negative values indicates an influx of marine particulate matter and seagrass vegetation (Gonneea et al. 2004), corroborated by the Archaias-Laevipeneroplis assemblage which inhabits marine Thalassia seagrass beds (Table 3, Sen Gupta 1999). These data suggest that coastal marine conditions at Celestun Lagoon commenced at this time, probably caused by sea-level rise ca. 5100 BP (Fig. 8). On the Caribbean side of the Yucatan Peninsula, in Puerto Morelos, mangrove peat layers also date to 5000 years ago (Islebe and Sánchez 2002), and coastal caves in Belize and Quintana Roo (Yucatan Peninsula east coast) exhibit basal peats dated to 6000–4000 years ago, with mixed pollen and marine microfossil assemblages suggesting a transition from terrestrial to marine-influenced coastal zones (Brady 1972; Polk et al. 2007; Gabriel et al. 2009; van Hengstum et al. 2010; Collins et al. 2015). A composite sea-level curve indicates that at 5000 BP, sea level was 2.5 m lower than today along the north coast of the Yucatan Peninsula (Khan et al. 2017), consistent with the 2.6 m depth at which peat is found at Celestun (Fig. 2a).
Rising sea level may have been coincident with drought in the northwest Yucatan Peninsula, as seen from records in Lakes Chichancanab and Punta Laguna (Fig. 8; Hodell et al. 1995; Curtis et al. 1998). The Ammonia-Elphidium assemblage at Celestun suggests brackish groundwater discharge was still present, but limited, otherwise the brackish-intolerant Archaias angulatus would not be observed (Hart and Kaesler 1986). This is also consistent with palynological and sedimentary changes of riverine soils in the south Yucatan Peninsula that suggest the onset of droughts ca. 4800 BP, which would have impacted the catchment area draining to Celestun Lagoon (Bauer-Gottwein et al. 2011; Aragón-Moreno et al. 2018; Vela-Pelaez et al. 2018). These events occur within chronological uncertainty of the appearance of Archaias-Laevipeneroplis (4900 BP) and may also explain the short duration (century-scale drought event) of this assemblage in the lagoon, which would have disappeared once average precipitation had resumed.
By 4500 BP, the Archaias-Laevipeneroplis assemblage was replaced by the Bolivina-Hanzawaia-Rosalina assemblage with increased abundance of the Ammonia-Elphidium assemblage, while δ13C remained between − 19 and − 21‰ (Fig. 4a), indicating persistent seagrass and marine vegetation from 4500 to 3500 BP. The foraminifera community of that time interval greatly resembles the modern assemblage in the southern part of the lagoon (Fig. 4b), suggesting that salinity was likely < 30 psu, which is inhospitable to Archaias (Hallock and Peebles 1993), but ideal for the other taxa (Table 3). This represents a decrease from a marine salinity supporting Thalassia seagrass and Archaias-Laevipeneroplis at 4800 BP, to salinities of < 30 psu, likely caused by increased groundwater discharge or reduced contribution of seawater. Dry conditions persisted from 4600 to 4400 BP in Belize (Akers et al. 2016), at 4500 BP in the south-central Yucatan Peninsula (Curtis et al. 1998; Wahl et al. 2014), from 4700 to 3600 BP in the southwest Yucatan Peninsula (Islebe et al. 2019), though conditions in the north-central Yucatan Peninsula at that time are unclear. Hence, precipitation-induced increase in groundwater discharge in Celestun Lagoon at 4500 BP is at odds with available climate data (Fig. 8).
It is likely that formation of barrier islands and spits around the northern lagoon, where springs are located (Fig. 1a, b), reduced mixing between groundwater and seawater, resulting in the observed decrease in salinity (Fig. 7). Highest species richness and diversity occurred between 4500 and 3800 BP (S = 17, H = 1.97, Fig. 4a), similar in magnitude to, and comprised of the same taxa, as the modern central lagoon assemblages (Fig. 4b). Indeed, high diversity is expected of foraminifera that colonize the diverse habitats typical of such an environment (fringing mangrove, brackish-tolerant seagrass, semi-protected lagoon, and unrestricted seawater circulation). West of Celestun Lagoon, across the Gulf of Mexico (Fig. 1c), La Mancha Lagoon in Veracruz, Mexico, also exhibits a rapid change in foraminifera composition between 6500 and 4500 BP, attributed to longshore transport and barrier island formation (Arellano-Torres et al. 2019). Initiation of coastal dunes and barrier island formation across the Gulf of Mexico, as sea-level rose, is consistent with apparent freshening and maximum diversity in Celestun Lagoon, despite regionally dry conditions (i.e., the restriction of seawater mixing is more effective in controlling the salinity than the reduction in groundwater input caused by drought). A more recent example (1000 BP) of this process was observed in sediment cores from Nichupte Lagoon on the eastern Yucatan Peninsula (Fig. 1c), where assemblages shifted from an Archaias assemblage to an Ammonia-Elphidium assemblage as barrier islands isolated the lagoon and reduced mixing with seawater (Hart and Kaesler 1986).
From 4500 to 3600 BP, diversity decreased as marine taxa progressively disappeared, while mean δ13C (− 19.5‰) and C:N (8.2) changed little (Fig. 4a). the Ammonia-Elphidium assemblage became dominant by 3400 BP, while the Bolivina-Hanzawaia-Rosalina assemblage decreased in abundance, indicating salinity ranged from 20–25 psu based on modern lagoon distributions (Table 3, Fig. 4b). We note, however, that sea level rise during that time (Fig. 8) should have resulted in greater marine influence, not less. Hence, we interpret the shift towards brackish taxa as indicative of greater influence of groundwater discharge, which may have been caused by increasingly moist climate between 4500 and 3400 BP.
The lower salinity could also be attributed to reduced contribution of seawater related to sand-bar buildup, however rapid fluctuation between dominant assemblages suggests an unstable environment, more likely reflecting climate change rather than more gradual and unidirectional barrier island accumulation. Between 3600 and 3400 BP (Fig. 4a), assemblages oscillated between Bolivina-Hanzawaia-Rosalina and Ammonia-Elphidium, accompanied by large changes in S and H. Specifically, because salinity correlates best with assemblage distribution (Table 4), shifting relative abundances between these two assemblages suggest fluctuating lagoon salinity, and is not consistent with the generally unidirectional formation of barrier islands. Instead, the lagoon geometry at that time was such that relatively small changes in absolute groundwater discharge greatly affected the salinity in the lagoon, thereby impacting foraminifera assemblages (Fig. 7). These relatively short-term salinity fluctuations were likely related to climate changes, such as variability in the ITCZ position and regional precipitation, as manifested by high variability in other climate records during this period (Fig. 8; Hughen et al. 1996; Douglas et al. 2016b; Aragón-Moreno et al. 2018). A wet period ca. 3500 BP is supported by negative oxygen isotope excursions in lake records from Lake Salpeten and mesic forest pollen from Lake Silvituc in the southwest peninsula (Wahl et al. 2006; Douglas et al. 2015; Vela-Pelaez et al. 2018). However, dry conditions are indicated at the same time by the oxygen records from Lakes Chichancanab and Puerto Arturo in the central peninsula (Hodell et al. 1991; Wahl et al. 2014). Indeed, at 3500 BP our data best match the Fe/Ca data (a runoff proxy) of La Mancha Lagoon, west of Celestun Lagoon (Figs. 1c, 8), across the Gulf of Mexico, demonstrating the regional link between climate and assemblage variability in the Gulf of Mexico southern coasts.
In Celestun Lagoon, from 3400 to 3000 BP, the Ammonia-Elphidium assemblage became dominant and Bolivina-Hanzawaia-Rosalina decreased in prevalence, indicating a reduction in salinity, while a decrease in δ13C suggests more mangrove-derived carbon in sediments (Fig. 4a). It is difficult to say whether these salinity and carbon trends are the result of barrier island growth or climate. Mangrove forest cover in particular may respond to both wetter climate and/or to additional shoreline to colonize. K/Ca ratios in nearby Los Petenes and Fe/Ca ratios in La Mancha Lagoon suggest reduced erosion and hence dry conditions across the southern Gulf of Mexico (Roy et al. 2017; Arellano-Torres et al. 2019), whereas δ18O records from Punta Laguna and δD records from Lake Salpeten imply wet conditions over the east and south Yucatan Peninsula around 3000 BP (Fig. 8; Curtis et al. 1998; Douglas et al. 2015). The general divide between dry northwest and wet southeast among records in the region may suggest that a specific atmospheric circulation pattern drove precipitation at that time. Indeed, previous studies indicate changes in the positions of the ITCZ, North Atlantic Subtropical High, and descending limb of Walker circulation (Metcalfe et al. 2015; Bhattacharya et al. 2017).
Overall, Phase 2 is distinguished by the apparent faunal succession from coastal marine taxa to brackish lagoon taxa (Fig. 4a), despite a sea-level rise of 1 m (Khan et al. 2017), with occasional oscillations between dominant assemblages during periods of climate variability (Fig. 8). Close agreement among observed sediment transport rates (Appendini et al. 2012), calculated barrier island aggradation time (Lowery and Rankey 2017), and transition between dominant foraminifera assemblages (Fig. 4a), suggests coastal geomorphology continued to influence the hydrogeology and salinity of Celestun Lagoon during Phase 2, with assemblages exhibiting variable sensitivity to changes in groundwater discharge induced by climate at that time (Fig. 7).
Phase 3 (3000–1700 BP)
Between 3000 and 1700 BP, species richness decreased and assemblages were dominated by the near-exclusive presence of Ammonia-Elphidium (Fig. 4a). Bulk sediment δ13Corg decreased from − 19.51 to − 23.27‰ by 1700 BP, indicating vegetation shifted from seagrass/phytoplankton to a greater proportion of mangroves. The decrease in species richness indicates a more homogenous setting (fewer habitats) compared to Phase 2 and was probably similar to the modern northern lagoon (Fig. 4b). Celestun Lagoon salinity during this phase was likely15-25 psu, indicated by the Ammonia-Elphidium assemblage and decrease in species richness (Fig. 4b). This trend is consistent with the ongoing barrier island buildup that reduced seawater input. However, at 2500 BP (105–106 cm), Quinqueloculina bicarinata and Hanzawaia concentrica appear briefly and Ammonia-Elphidium abundance decrease, suggesting a reduction in groundwater and more marine influence in the lagoon (Fig. 4a). This change corresponds to drought conditions noted in Lake Chichancanab (Central Yucatan Peninsula), and Macal Chasm (Belize) speleothem records ca. 2600 BP (Hodell et al. 1991; Akers et al. 2016). The generally drier conditions on the Yucatan Peninsula from 3000 to 2000 BP (Douglas et al. 2015) also match the timing of increased marine influence in Celestun Lagoon, as suggested by slight carbon isotope changes from -21.7 to -21.5‰ (indicating greater input of marine particulate matter). After 2000 BP, Ammonia-Elphidium returned, and the conclusion of this phase is the onset of the brackish, semi-protected lagoon environmental conditions that characterize the present-day northern lagoon.
Phase 4 (1700 BP-present)
The last 1700 years of northern Celestun Lagoon history are characterized by the appearance of the Agglutinated-Haynesina assemblage, low species richness (S = 7), δ13Corg decrease from − 22.97 to − 24.22‰, and variable C:N (mean = 7.5). Elphidium galvestonense disappeared, but other taxa of the Ammonia-Elphidium assemblage (Fig. 5c) remained, alongside increased abundance of Q. laevigata. These conditions persist to the present, thus representing the onset of modern conditions in the northern lagoon beginning at 1700 BP. The Agglutinated-Haynesina assemblage indicates increased muddy substrate, which is observed in the protected mangrove lagoon today and is consistent with mangrove forest expansion, indicated by decreasing carbon isotope values (Fig. 4a). Mangrove expansion around Celestun Lagoon is consistent with expansion of mangrove forest documented elsewhere in the Caribbean (Peros et al. 2007). Though barrier islands provide more land for mangrove colonization, the effect is local, so a regional expansion of mangroves suggests regional climate influence. Onset of moister climate ca. 1700 BP is consistent with observation in the Chichancanab and Salpeten lake records (Fig. 8; Hodell et al. 1995; Douglas et al. 2015).
The age-depth model indicates that the coarse bioclastic layer at 35 cm depth (Fig. 2b) begins at 1300 BP, or about 700 Common Era (CE). In core 4A this layer consists of an increase in absolute abundance of foraminifera in the Ammonia-Elphidium assemblage and a pronounced decrease in agglutinated taxa and H and E indices (Fig. 4b). In the southern lagoon, abundant bivalves and fragments are noted. This interval coincides with the well-documented Maya droughts (Douglas et al. 2016b), which implies decreased groundwater discharge in Celestun Lagoon. Although an increase in the polyhaline assemblage Ammonia-Elphidium could be a response to increased salinity, the lack of fine mud across the cores would imply that authigenic carbonate precipitation was reduced in the lagoon at the same time. If included in the age-depth model, radiocarbon sampled at 31 cm (Table 2, 1485 14C yr) causes a rapid change in slope in the age-depth regression line that results in abrupt decrease in sedimentation to 0.2 mm y−1. This slowed sedimentation rate is not consistent with the 1.5 mm y−1 rate determined by 210Pbexcess (Fig. 3) or rising sea level at this coastal setting (Fig. 8, Khan et al. 2017). We thus interpret this lagoon-wide layer to be caused by storm erosion or a tsunami. Hurricanes are known to sort and transport shelf foraminifera into coastal systems (Murray 2006), and identified features of this coarse layer, including the lack of fine sediments or agglutinated taxa, are consistent with such an event. The rapid increase in absolute abundance of Ammonia and Elphidium, which occurs in hurricane overwash deposits on the Atlantic coast of the United States (Collins et al. 1999), also supports this interpretation. The presence of an erosive shell layer representing the missing interval, 1300 to 500 BP, provides some paleoclimate information, as other studies suggest increased hurricane activity from 800 to 1350 AD (1150–600 BP) in the Caribbean (Peros 2015), although our age model uncertainty is high near the shell layer at 35 cm (Fig. 3).
Alternatively, the bioclastic layer could represent a tsunami deposit, likely from submarine continental slope failure (Horrillo et al. 2013). Holocene sea-level rise has been suggested as a cause of slope failure from water overburden pressure (Smith et al. 2013), and in the Gulf of Mexico, sea level has risen 9 m during the last 7.8 ka (Blum et al. 2002). Continental slopes may also fail because of additional sediment load, which would be expected from increased continental weathering and riverine transport to coasts. Fe/Ca and K/Rb ratios in sediments of La Mancha Lagoon (Fig. 1c), indicate increased detrital input and coastal weathering at 1000 BP (Arellano-Torres et al. 2019). The timing is coincident with the high-energy shell deposit in Celestun Lagoon (Figs. 2, 3), and evidence of a tsunami deposit might be recognized in sediment cores from other lagoons around the southern Gulf of Mexico. Brady (1972) noted an erosional contact with increased grain size at 20 cm depth in Holbox Lagoon (Fig. 1c). South of Celestun in Los Petenes (Fig. 1c), sediments transition from peat at 38 cm to silt between 35 and 15 cm and are dated around 700 BP (1300 CE) with a spike in K/Ca that indicates rapid erosion inland (Roy et al. 2017; Fig. 8). Rapid erosion from rainfall is unlikely during the Terminal Classic Maya mega-droughts of this time (Hodell et al. 1995; Douglas et al. 2016b), further implying a catastrophic depositional event such as a tsunami inundation and subsequent outwash. Lastly, an anomalous berm of unsorted boulders and sand on the Caribbean coast of the Yucatan Peninsula has been interpreted as a potential tsunami deposit with a radiocarbon age of 1500 BP (Shaw and Benson 2015). At our site, the occurrence of a single (anomalous) shell layer with an upper age limit of 1300 BP suggests that the causal event was relatively uncommon, further supporting a tsunami as opposed to hurricanes, which are common in the region. Additional research is required to better characterize the tsunami history of the Gulf of Mexico, but a tsunami remains a plausible cause of the shell layer at 35 cm.
From 500 BP to present, the Ammonia-Elphidium and Agglutinated-Haynesina assemblages dominate, with a notable increase in A. tepida at 150 BP, or about 1800 CE (Fig. 4a). The only other occurrence of abundant A. tepida was during Phase 1, at 5200 BP. Thin-walled A. tepida thrive in low-salinity, protected conditions (Table 3, Murray 2006). Barrier island aggradation is unlikely to have altered the entire lagoon geometry substantially over just the last few hundred years, so the increase in A. tepida may be a response to low-salinity conditions, induced by wet climate. The nearest speleothem-derived oxygen isotope record, from Tecoh Cave (100 km west of Celestun Lagoon), shows three century-scale pulses of relatively wet conditions occurred during the Medieval Warm Period and Little Ice Age (Medina-Elizalde et al. 2010), and a wet interval is also noted in Belize speleothems (Kennett et al. 2012). A diatom record from Cenote San Jose Chulchaca, 22 km east of Celestun Lagoon indicates a decrease in salinity over the last 800 years (Whitmore et al. 1996; Hodell et al. 2005). However, Whitmore et al. (1996) did not note the same trend in Cenote Sayaucil (north-central Yucatan lowlands). Furthermore, the pollen record from Lake San Jose suggests a decline of forest taxa that depend on high rainfall at the same time the cenote was freshening (Hodell et al. 2005), a trend corroborated by pollen and microfossil observations from Aguada X’caamal (75 km southeast of Celestun Lagoon) that imply drying in the northwest region during the Little Ice Age (600 to 100 BP) (Hodell et al. 2005). Together these records suggest both increased precipitation and increased evaporation for the northwest Yucatan Peninsula, which today has the highest evapotranspiration potential of the entire region (Bauer-Gottwein et al. 2011). Phase 4 of Celestun Lagoon development is thus characterized by expanding mangrove forests during a period of variable precipitation. Ammonia-Elphidium and Agglutinated-Haynesina assemblages dominate foraminifera composition of the northern semi-protected habitat with a wide range of salinity, features observed in the modern northern lagoon today.