Late glacial–Younger Dryas climate in interior Alaska as inferred from the isotope values of land snail shells

Abstract

The isotope values of fossil snail shells can be important archives of climate. Here, we present the first carbon (δ¹³C) and oxygen (δ¹⁸O) isotope values of snail shells in interior Alaska to explore changes in vegetation and humidity through the late-glacial period. Snail shell δ¹³C values were relatively consistent through the late glacial. However, late-glacial shell δ¹³C values are 2.8‰ higher than those of modern shells. This offset is best explained by the Suess effect and changes in the δ¹³C values of snail diet. Snail shell δ¹⁸O values varied through the late glacial, which can be partially explained by changes in relative humidity (RH). RH during the snail growing period was modeled based on a published flux balance model. Results suggest a dry period toward the beginning of the Bølling–Allerød (~14 ka) followed by two distinct stages of the Younger Dryas, a wetter stage in the early Younger Dryas from 12.9 to 12.3 ka, and subsequent drier stage in the late Younger Dryas between 12.3 and 11.7 ka. The results show that land snail isotopes in high-latitude regions may be used as a supplementary paleoclimate proxy to help clarify complex climate histories, such as those of interior Alaska during the Younger Dryas.

INTRODUCTION

The late Pleistocene was characterized by dramatic changes in climate. This period covers both glacial (Marine Isotope Stages [MIS] 4 and 2) and interglacial periods (MIS 5 and 3). The late-glacial period (~16–12 ka; ka = thousands of years before present [BP]; 0 years BP = AD 1950) marks the transition from the last glacial episode (115–11.7 ka) to the interglacial Holocene (past 11.7 ka) (Walker et al., 2009). With the onset of the warming Holocene epoch, organisms had to adapt, migrate, or go extinct. The late-glacial transitional period was characterized by cold stadial and warm interstadial phases, as seen in Greenland ice core records (Seierstad et al., 2014). Two well-studied stadial episodes during the late glacial are the so-called Oldest Dryas (16–15 ka) and Younger Dryas (12.9–11.7 ka) (Broecker et al., 2010).

The Younger Dryas (12.9–11.7 ka) was a reversal of a warming period (Bølling–Allerød) (14.7–12.9 ka) and a temporary return to glacial like conditions (Broecker et al., 2010). Climate during the Younger Dryas was complex and varied spatially (Kaufman et al., 2010). Evidence for the Younger Dryas is strongest in the North Atlantic region but is observable in many areas across the world. Emerging evidence from sediment cores in the North Atlantic, western Europe, east Asia, and western North America suggest the Younger Dryas can be grouped into two distinct stages (Brauer et al., 1999; Bakke et al., 2009; Kaufman et al., 2010; Schlolaut et al., 2017; Pigati and Springer, 2022). The first stage, the early Younger Dryas (~12.9–12.3 ka), is characterized by stable conditions due to a weakened Atlantic Meridional Overturning Circulation (AMOC), and the second stage, the late Younger Dryas (12.3–11.7 ka), is defined by variable conditions as the AMOC began to strengthen (Pigati and Springer, 2022). The AMOC is an important component of the Earth's climate system that transports warm water from the tropics to the North Atlantic Ocean, where it cools and sinks before returning southward. The AMOC helps regulate temperatures and influence weather patterns globally. As noted, at the start of the Younger Dryas, the AMOC weakened, leading to a decrease in the amount of warm water transported to the North Atlantic Ocean. This reduction in the poleward transport of warm ocean water likely played a role in triggering a cold period lasting from 12.9 to 11.7 ka and had significant impacts on global climate and ecosystems (Alley et al., 1993). Understanding the behavior of climatic responses to the AMOC is critical for forecasting future climate changes, as recent observations suggest the AMOC has been gradually weakening over the past several decades (Boers, 2021).

While the effects of climate change are often manifested globally, polar regions have exhibited greater vulnerability to current rapid global warming (Stewart et al., 2013). Changes in Arctic climate have global repercussions that are forecast to intensify in the future. Alaska specifically is warming three to four times as fast as the rest of the United States (Taylor et al., 2017; Walsh and Brettschneider, 2019). Since the 1950s, Alaska's average annual air temperatures have increased by 1.5°C, and average winter temperatures have increased by 4°C (Walsh and Brettschneider, 2019). Paleoclimate records from climate-sensitive regions are thus critical for developing a better understanding of the biological impacts of climate change at both organismal and ecosystem scales.

Despite the significance of these regions in past and present climate change, oxygen and carbon isotopic paleoclimate records in interior Alaska are rare. A few studies have looked at δ¹⁸O records of diatom silica from Alaska lake sediments (Hu and Shemesh, 2003; Kaufman et al., 2012). Nevertheless, using other terrestrial climate proxies is essential to deepen our understanding of climate transitions in the past, which also provides clues for future climate change. Here, we investigate the utility of land snails as a new paleoclimate proxy for interior Alaska across a range of sedimentary and archaeological settings, to develop a better understanding of late-glacial climatic conditions in interior Alaska.

Land snails are useful indicators of the environment due to their widespread occurrence and abundance and the persistence of their shells after they die. In North America alone, there are around 1200 species of land snails (Nekola, 2014). While the greatest variety of species is found at tropical latitudes, land snails can be found in all terrestrial environments except extremely dry deserts (e.g., Yanes et al., 2019). Their shells are made of aragonite (CaCO₃) (Balakrishnan and Yapp, 2004) and can commonly be found in Quaternary sediment archives worldwide (Goodfriend, 1992).

Land snail shell δ¹³C has been studied in both natural (Goodfriend and Ellis, 2000; Yanes et al., 2008) and laboratory settings (Stott, 2002; Metref et al., 2003; Zhang et al., 2018). A model developed by Goodfriend and Hood (1983) proposed that land snail carbon is derived from three sources: (1) snail diet, (2) atmospheric CO₂, and (3) ingested carbonates (including limestone and recycled snail shell). However, recent studies have found that ingested carbonates do not seem to affect the shell δ¹³C of small (<5 mm) Succineidae shells (Pigati et al., 2004, 2013). Hence, stable carbon isotope values of land snail shells are most often used to infer the carbon isotope compositions of local vegetation and, where present, calculate the relative proportions of C₃, C₄, and CAM plants in a given setting (Goodfriend and Stipp, 1983; Goodfriend, 1990; Yanes et al., 2008, 2009).

While snail shell δ¹⁸O compositions are most commonly used to reconstruct δ¹⁸O values from precipitation experienced during a snail's active period (Goodfriend, 1991; Zanchetta et al., 2005; Yanes et al., 2011), Balakrishnan and Yapp (2004) showed that the oxygen isotope composition of land snail shells is also controlled by relative humidity (RH) and temperature, as well as by the δ¹⁸O composition of local precipitation and water vapor. Because land snails are mainly active during or after rain events (Ward and Slotow, 1992), most of the water absorbed by a snail and in a snail's body fluid is derived from rainwater (Riddle, 1983). The flux balance mixing model by Balakrishnan and Yapp (2004) demonstrates that shells are precipitated in oxygen isotopic equilibrium with the snail body fluid. While snails are forming their shells, the body fluid and shell experience ¹⁸O enrichment due to evaporation, by about 2‰ on average, compared with that of the source precipitation (Lécolle, 1985; Goodfriend, 1992; Balakrishnan and Yapp, 2004).

Here, we report the first oxygen and carbon isotope values of fossil land snail shells from interior Alaska, retrieved from an array of Quaternary depositional settings. The shells are constrained chronologically by stratigraphy and/or radiocarbon dating. The stable isotope results are interpreted in the context of a published flux balance mixing model and compared with other published climate proxy records to elucidate the complex late-glacial climate history of interior Alaska.

BACKGROUND

Modern climate and environment in interior Alaska

Crucial to interpretations of stable isotopic compositions of fossil land snails is an understanding of how modern snails are conditioned to today's climate. Nield et al. (2022) found that the δ¹⁸O values of modern Succinea strigata and Euconulus fulvus shells are correlated with temperature and precipitation δ¹⁸O between April and October; they referred to this as the “extended growing season.” Interior Alaska can be described as having a continental climate that is largely driven by the distance to oceanic moisture and the precipitation shadows formed by both the Brooks Range to the north and the Alaska Range to the south (Pewe, 1975). This climatic setting supports a boreal forest biome in most of interior Alaska. Typical trees in the Alaskan boreal forest include white spruce (Picea glauca), black spruce (Picea mariana), larch (Larix), balsam poplar (Populus balsamifera), aspen (Populus tremuloides), Alaska birch (Betula neoalaskana), willow (Salix polaris), and alder (Alnus) (Viereck and Little, 1994). Interior Alaska receives, on average, 305 mm of rain per year and 1550 mm of snow per year (statesummaries.sncics.org, accessed January 2023). Precipitation in interior Alaska is derived mainly from North Pacific Ocean moisture (Abbott et al., 2000). Based on modeled outputs (IAEA/WMO, 2015; Bowen et al., 2005), the average δ¹⁸O value for precipitation during the snail extended growing season is −17.1‰ (Bowen, 2020). Average extended summer RH is 60% (www.ncdc.noaa.gov, accessed August 2020). Average high and low temperatures during the extended growing season are 14°C and 2°C, respectively (www.usclimatedata.org, accessed August 2020).

Late-glacial climate and environment in interior Alaska

Based on interpretations of lake and peat pollen records throughout Alaska, climate conditions during the Younger Dryas were spatially complex (Kokorowski et al., 2008). Nevertheless, despite high spatial variability, these proxy records consistently indicate that Younger Dryas temperature changes in interior Alaska were muted compared with the North Atlantic region (Graf and Bigelow, 2011). Most records from interior Alaska do not record a sharp temperature reversal at the onset of the Younger Dryas, unlike those that can be seen in Greenland ice core records (Alley et al., 1993). Pollen records from Windmill Lake (interior Alaska) do not show evidence for RH changes during the Younger Dryas (Bigelow and Edwards, 2001). In contrast, these records indicate cool and dry conditions during the Bølling–Allerød that were replaced by high lake levels at the onset of the Holocene (Bigelow and Edwards, 2001). Similarly, using branched glycerol dialkyl glycerol tetraethers (brGDGT) from loess–paleosol sequences as a temperature proxy, Kielhofer and colleagues (2023) found no dramatic temperature change at the onset or during the Younger Dryas.

Shifts in microfloral fossils indicate mild changes in climate during this time as well. During the late glacial (~16–11.7 ka), interior Alaska was characterized by herb-rich tundra vegetation (Bigelow and Powers, 2001; Anderson, 2003). In the Tanana River valley, vegetation shifted from herb tundra to shrub tundra dominated by birch around ~14 ka. This switch from herb to birch tundra points to a warming air temperature and increased precipitation at ~14 ka (Anderson, 2003). Land snails were collected from deposits dated to these time periods in order to test these hypotheses about paleoclimate shifts in interior Alaska.

MATERIALS AND METHODS

Modern land snail assemblages

Modern snails were collected live from seven locations in the boreal forest in interior Alaska for two summer seasons, 2015 and 2022 (Fig. 1). Sampling took place for 1 hour at each field location where all living and recently dead snails found were collected within a 20 m radius. Field locations were chosen based primarily on their accessibility; however, all localities are representative of a modern, undisturbed boreal forest. Sampled modern sites in interior Alaska contained up to five land snail species, including S. strigata, E. fulvus, Columella simplex, Discus whitneyi, and Punctum sp. However, in this study, isotopic analyses were conducted on the most abundant and widespread genus, Succinea, to taxonomically standardize our comparisons through time, thus removing isotopic variability associated with different species’ vital rates, body sizes, or ecologies (Yanes et al., 2017). The family Succineidae exhibits a broad spatial distribution, with shells present on almost every continent. Succineids are associated with highly humid swamp and flooded areas and typically live for 1–2 yr (Oerstan, 2010). However, recent field analysis has observed succineids in drier areas as well, and they are frequently recovered from Quaternary loess deposits of North America (Pigati et al., 2010, 2013) and Europe (Moine et al., 2005).

Figure 1. (A) Quaternary depositional settings (circles) where fossil shells and modern snails were sampled (this study). Studied sites include permafrost soils (Cold Regions Research and Engineering Laboratory [CRREL] tunnel), loess deposits (Dalton highway), and archaeological sites (Mead, Bachner, North Gerstle, and Upward Sun River [USRS]). Published paleoclimate proxy sites (white stars), including marine cores, lake cores, and paleosols (see text). (B) Map of Greenland showing locations of two previously published paleoclimate proxies from ice cores; North Greenland Ice Core Project (NGRIP; north location) and North Greenland Eemian Ice Drilling (NEEM; south location). Locations of paleoclimate proxies in the published literature: (1) Grebmeier et al., 1990; (2) Hu et al., 2002; (3) Hu et al., 2006; (4) Kaufman et al., 2012; (5) Kaufman et al., 2009; (6) Jones et al., 2009; (7) Kurek et al., 2009; (8) Abbott et al., 2000; (9) Kielhofer et al., 2023; (10) NGRIP Members, 2004; (11) Masson-Delmotte et al., 2015.

Late-glacial snail assemblages

Six sites dating to the late-glacial period containing pristine fossil snail shells were investigated in this study (Figs. 1 and 2). In sharp contrast to modern assemblages, most late glacial–age sites contained only one genus, Succinea (Fig. 3). However, the permafrost tunnel site, described below, contained three genera, including Succinea, Columella, and Euconulus. A total of 121 fossil shells were selected for isotope analysis (Table 1). Snail shells collected from all paleontological sites were handpicked and sieved from bulk sediment.

Figure 2. Photographs where late-glacial snail shells were collected. (A) Mead archaeological site. (B) Cold Regions Research and Engineering Laboratory (CRREL) permafrost research tunnel. (C) Upward Sun River archaeological site. (D) Yukon-Tanana Upland loess deposits at Dalton highway (adapted from Muhs et al., 2018).

Figure 3. Photographs of modern and fossil Succinea specimens. Pictures depict (A and B) late-glacial (~14,000 cal yr BP) specimens retrieved from the Cold Regions Research and Engineering Laboratory (CRREL) permafrost tunnel and (C and D) living Succinea specimens found in the forest near Fairbanks, Alaska. Scales are in millimeters.

Table 1. Carbon and oxygen isotope compositions of modern and fossil Succinea shells from central Alaska with median calibrated ages (BP).^a

^a Modern snails were live collected in 2015/2022.

The Cold Regions Research and Engineering Laboratory (CRREL) permafrost tunnel is a human-made tunnel that penetrates sediments of the Goldstream Creek valley near Fairbanks, Alaska (Figs. 1 and 2). The tunnel exposes fossiliferous reworked loess and alluvial sediments. Much has been published on the permafrost stratigraphy, ice formations, and preserved fossils at this site (e.g., Hamilton et al., 1988; Shur et al., 2004; Kanevskiy et al., 2022). However, analyses have yet to be done on the snails in these deposits. Ages for loess within the tunnel range from ~34 to ~9.5 ka (Hamilton et al., 1988). Snails from the tunnel were collected via handpicking or sieving of bulk sediment. Snails were found in two areas within the tunnel, referred to as facies A and facies B. Facies B is an organic-rich area on the ceiling of the tunnel located at the top of the winze (for more details, see Hamilton et al., 1988). Facies A is poorly sorted loess to large cobble–sized sediment located on the floor of the tunnel 12–13 m from the tunnel entrance. The tunnel was constrained chronologically for each facies by several radiocarbon ages from shells and plant remains (Table 2). The isotopic values of 13 Succinea shells were determined from these deposits, dating to ~14 ka.

Table 2. Sample identifications, radiocarbon ages, and calibrated ages for Succinea shells from the Cold Regions Research and Engineering Laboratory (CRREL) permafrost tunnel, the Bachner, and North Gerstle Point sites, and organic matter in the CRREL tunnel.

Snails were also collected from last glacial loess road outcrops along the Dalton Highway. Snails were found in two locations sampled by Muhs et al. (2018), at milepost 45.9 (AK-650A) and milepost 54.7 (AK-855). While Quaternary loess is common throughout Alaska, loess dating to the last glacial period in Alaska is rare (Muhs et al., 2018). The Dalton Highway sites present unusual occurrences of last glacial loess deposits dating from 19 to 12 ka in the Yukon-Tanana Upland (Figs. 1 and 2). At these sites, Succinea shells were found in the upper parts of loess sections. Calibrated radiocarbon ages of snails span an interval between ~13.4 and ~12.7 ka (Muhs et al., 2018). At these sites, 36 Succinea shells were selected for isotopic analysis.

To complement sampling at paleontological sites, we also retrieved land snail shells from four Alaskan archaeological sites: Mead, Upward Sun River (USRS), North Gerstle, and Bachner (Figs. 1 and 2). Each site has loess deposits that are greater than 1 m in thickness. The Mead archaeological site is located along Shaw Creek Bluff near Delta Junction (Dilley, 1998) and is interpreted as a human base camp (Potter et al., 2013). Succinea shells were found in a trench away from the main archaeological excavation in stratigraphic zones dated to ~12.6 to 11.8 ka (Kielhofer et al., 2020). Here, 10 Succinea shells were analyzed for isotopes. The Upward Sun River site is located on a loess-covered dune near the Tanana River valley (Reuther, 2013). Cultural remains come from repeated occupations of the dune dating back to 13.3 ka and include a late-glacial residential structure and the cremated and buried remains of a child (Potter et al., 2011, 2014). 20 Succinea shells were analyzed for isotopes from horizons dated to ~13.1–12.6 ka. The North Gerstle site is also in the Tanana River valley on a bluff overlooking the Tanana River. Succinea shells were collected from horizons dated to 12.9–12.8 ka (Vanderhoek et al., 1997). Here, 18 Succinea shells were selected for analyses. The Bachner site overlooks Quartz Lake and is adjacent to Shaw Creek Flats, also within the Tanana River valley, within 10 km of the Mead and North Gerstle sites (Reuther, 2013). Succinea shells were found at layers dated to ~12.8–11.9 ka, and 25 Succinea shells were analyzed for isotopes.

Radiocarbon dating

For this study, fossil Succinea shells were either radiocarbon dated directly or collected from a horizon where the chronology is well constrained by radiocarbon dating of other shells, plant macrofossils, wood, or charcoal. Three snail samples collected from the CRREL permafrost tunnel were selected for paired graphite-target and carbonate-target radiocarbon dating analysis (Bright et al., 2021; Fig. 4). Historically, land snail shells have been avoided for radiocarbon dating because of the potential incorporation of ¹⁴C from limestone during shell formation (Pigati et al., 2004). However, several genera of small gastropods, including Succineidae, have been shown to provide reliable radiocarbon ages (Pigati et al., 2004). To further verify the age of snail shells from the permafrost tunnel, two plant remains samples taken from the same horizon as the snail assemblage were radiocarbon dated and yielded results similar to those of the shells (Table 2).

Figure 4. (A) Scatter plot comparing carbonate-target vs. graphite-target radiocarbon-dated Succinea shells collected from the Cold Regions Research and Engineering Laboratory (CRREL) permafrost tunnel with error bars for age uncertainty. (B) Data from this study (filled red dots) plotted on figure from Bright et al. (2021). Reduced major axis regression of paired direct carbonate and graphite pMC (percent modern carbon) determinations.

Late-glacial Succinea shells from loess deposits and archaeological sites have been radiocarbon dated in prior work (Potter et al., 2011, 2013; Reuther, 2013; Muhs et al., 2018; Kielhofer et al., 2020). Sections at the Bachner and North Gerstle site were previously dated using charcoal and collagen from animal bones (Reuther, 2013). For the present study, shells from these deposits were radiocarbon dated (Table 2) at the Center for Applied Isotope Studies at the University of Georgia, following procedures outlined in their laboratory protocols (https://cais.uga.edu/service/radiocarbon-dating-by-ams). Shell and plant remains from the CRREL permafrost tunnel were radiocarbon dated at the W.M. Keck Carbon Cycle Accelerator Mass Spectrometry Laboratory at the University of California–Irvine following the procedures outlined in their laboratory protocols (https://www.ess.uci.edu/group/ams/protocols). Shells were dated using the carbonate-target and graphite-target radiocarbon methods, and detailed methodology for both dating methods can be found in Bright et al. (2021). The resultant radiocarbon ages were all calibrated to calendar years using the IntCal20 data set and CALIB 8.20 (Reimer et al., 2020).

Stable isotope analyses

In total, 174 individual shells were selected for isotope analysis by choosing the cleanest and best-preserved shells. Shells were mechanically cleaned with deionized water to remove any remaining debris and air-dried before laboratory analysis. Land snail shells were analyzed at the University of Florida using a Finnigan-MAT 252 Isotope Ratio Mass Spectrometer coupled with a Kiel III automated carbonate preparation device. Additional details of the methodology can be found in Nield et al. (2022). Analytical precision was better than ±0.1‰, as shown by repeated measurements of samples and standards. Results were calibrated against the international standards NBS-18 and NBS-19.

Statistical analyses

All statistical analyses were done using R v. 2.5.1 software (R Core Team, 2023). All data were treated as nonnormally distributed. The Kruskal-Wallis test was used to determine whether the multiple groups exhibited statistically equivalent median values. The Mann-Whitney test was used when comparing the median value between two groups only. Statistical significance was accepted when the P value was less than 0.05. Pearson correlation analysis was conducted to assess the significance of monotonic relationships between two variables. The coefficient of determination (R ²) was used to identify the strength of the model and relationship.

RESULTS

Radiocarbon results

Results from the paired graphite-target and carbonate-target radiocarbon dating analysis (Bright et al., 2021) of shells from the CRREL permafrost tunnel show the two dating methods are not statistically different (R ² = 0.98, P = 0.7; Fig. 4, Table 2). Both methods yield equivalent ages, although the graphite-target method (±35 yr) is more precise than the carbonate-target method (±140 yr). The carbonate-target method is less expensive and uses a far smaller sample, <1 mg compared with 20+ mg, which makes it an attractive option for land snail shell radiocarbon dating. All analyzed snails from the permafrost tunnel date to ~14 ka, in agreement with plant remains samples that were taken from the same horizon (Table 2). Ages of snail shells from the tunnel range from 14.0 to 13.9 ka, and plant remains ages range from 13.9 to 13.6 ka. Other snails sampled from the same assemblage of the CRREL tunnel are provisionally assumed to be the same age. Additional traditional graphite carbon radiocarbon analysis was performed on shells from the Bachner and North Gerstle sites (Table 2). Snails from Bachner were dated to 12.6 and 11.9 ka, and snails from North Gerstle were dated to 12.9 and 12.8 ka.

Carbon isotopic composition

Carbon isotope analyses were performed on 121 fossil and 53 modern Succinea shells, and the results are shown as box plots in 250 yr time bins (Fig. 5). Late-glacial shell δ¹³C values average −7.7‰ and range between −9.9‰ and −5.9‰ across the time intervals (Fig. 5, Table 1). Median δ¹³C values of shells from the early Bølling–Allerød (~14 ka), early Younger Dryas (12.9–12.3 ka), and late Younger Dryas time (12.3–11.7 ka) are −7.5‰, −7.8‰, and −8.2‰, respectively (χ² = 9.9792, df = 4, P = 0.0408). Median δ¹³C values in late-glacial time bins vary by only 0.7‰. Modern shell δ¹³C values have a median value of −10.9‰, ranging from −12.7‰ to −7.9‰ (Fig. 5). All late-glacial time bins differ significantly from values in modern shells, being on average 2.8‰ higher than modern shells (W = 6048, P < 0.001).

Figure 5. Succinea shell δ¹³C values (in PBD, Pee Dee Belemnite) binned into 250 yr time intervals. Box extremes represent lower and upper quartiles. Whiskers depict the range of values. The solid line inside the plot depicts the median δ¹³C value. Number in parentheses is the number of samples. The dots represent outliers that were outside the interquartile range. The three shaded regions mark the Bølling–Allerød (BA) (14.7–12.9 ka), the early Younger Dryas (YD) (12.9–12.3 ka), and the late Younger Dryas (12.3–11.7 ka).

Oxygen isotopic composition

Oxygen isotope analyses were performed on the same 121 fossil and 53 modern Succinea shells as the δ¹³C analysis. The results are illustrated again as box plots and separated into 250 yr bins (Fig. 6). Late-glacial (14–11.5 ka) shell δ¹⁸O values have a median value of −11.9‰ and range from −13.7‰ to −8.3‰ (Fig. 6B, Table 1). Within this time period, early Bølling–Allerød (~14 ka), early Younger Dryas (12.9–12.3 ka), and late Younger Dryas (12.3–11.7 ka) have median shell δ¹⁸O values of −10.6‰, −12.1‰, and −11.5‰, respectively, and differ significantly from each other (χ² = 29.636, df = 4, P < 0.001). Modern shell δ¹⁸O has a median value of −10.8‰ and ranges from −12.8‰ to −8.9‰ (Fig. 6). Late-glacial shell values differ significantly from values in modern shells (W = 778.5, P < 0.001).

Figure 6. δ¹⁸O values binned into 250 yr time intervals. The three shaded regions mark the Bølling–Allerød (BA) (14.7–12.9 ka), the early Younger Dryas (YD) (12.9–12.3 ka), and the late Younger Dryas (12.3–11.7 ka). (A) Seawater δ¹⁸O values (in SMOW) calculated from benthic foraminifera off the coast of Alaska (Grebmeier et al. 1990; Praetorius et al., 2015, 2016); (B) snail shell δ¹⁸O values (in PDB) of Succinea shells from central Alaska; and (C) North Greenland Ice Core Project (NGRIP Members, 2004) and North Greenland Eemian Ice Drilling (NEEM; Masson-Delmotte et al., 2015) Greenland ice core δ¹⁸O values in SMOW. Box extremes represent lower and upper quartiles. Whiskers depict the range of values. The solid line inside the plot depicts the median δ¹⁸O value. The dots represent outliers that were outside the interquartile range. The number in the parentheses is the number of samples.

DISCUSSION

Paleo-vegetation inferences

Median modern shell δ¹³C values are 2.8‰ lower than median late-glacial δ¹³C values (Fig. 5). This offset between modern and late-glacial values is likely best explained by shifts in the carbon isotope composition of one or more of the three main components affecting the carbon isotope values of land snail shells: atmospheric carbon, diet, and ingested carbonate (Goodfriend and Ellis, 2002). Of these, changes in atmospheric carbon and diet may be most likely to have produced differences in shell δ¹³C. While land snails may consume limestone as a source of calcium to build their shells, and limestone ingestion may be capable of positively shifting (by several per mil) shell δ¹³C values, recent studies have found that small-bodied Succineidae gastropods may consume only minimal amounts of carbonate sediments (Pigati et al., 2004, 2013).

Since the Industrial Revolution, increased burning of fossil fuels has not only increased CO₂ concentrations but also affected carbon isotopic composition of atmospheric CO₂. Overall, δ¹³C values of atmospheric CO₂ have become more negative following the Industrial Revolution (Friedli et al., 1986). This decline in atmospheric CO₂ δ¹³C has caused a decline in plant (Dawson et al., 2002), and therefore snail shell, δ¹³C values. This phenomenon is known as the Suess effect (Keeling, 1979). However, the Suess effect can only account for a 1–1.5‰ negative shift in δ¹³C values (Wang et al., 2011). If we apply the correction to the snail shells analyzed here, there would still be an ~1–2‰ decrease in shell δ¹³C composition from the late glacial to today.

The remaining offset between modern and late-glacial shell δ¹³C values is most likely due to differences in the δ¹³C values of the snail diet. Generally, modern tundra, boreal forest, and coastal forest biomes in Alaska are all dominated by C₃ plants (average δ¹³C value of −26‰), which include trees and cool-season grasses and are depleted in ¹³C compared with C₄ plants (average δ¹³C value of −12‰) (O'Leary, 1988; Muhs et al., 1999, 2000). C₃ and C₄ plants fractionate carbon isotopes in different ways. Soil organic matter in most Alaskan biomes shows δ¹³C values of −25‰ (Muhs et al., 2000). Additionally, none of the δ¹³C analyses from samples taken from the CRREL permafrost tunnel or from nearby Quartz and Goldbottom Creeks are in the range of δ¹³C values of C₄ grasses (Wooller et al., 2018). Laboratory studies have found aragonite carbon of land snails shells to be positively fractionated with respect to the bicarbonate carbon from which it precipitated (Rubinson and Clayton, 1969). This can in part explain why shell δ¹³C values are more positive than consumed plant δ¹³C.

C₄ plants are supported by lower atmospheric CO₂ concentrations (Wang et al., 2020). In theory, this could suggest higher concentrations of C₄ plants in Alaska during the last glacial period, when atmospheric CO₂ concentrations were much lower than during interglacial times (Sigman and Boyle, 2000). However, it is likely that regional temperatures override this factor, with lower last glacial temperatures being more important than CO₂ in determining the relative abundance of C₃ and C₄ plants, as seen in Muhs et al. (1999).

C₄ plant occurrences are mainly found in mid-latitudes and tropical grasslands. Hence, a negative shift in shell δ¹³C values is most likely explained by lower δ¹³C values in the vegetation composition of C₃ plants today, compared with the late glacial. This may be caused by changes in the water-use efficiency in C₃ plants in slightly drier conditions during the late glacial compared with today (Policy et al., 1993). Water-use efficiency of plants refers to the amount of CO₂ uptake through photosynthesis versus water loss to transpiration. Carbon isotope discrimination occurs during photosynthesis, when plants preferentially take up the lighter ¹²C isotope (Hubrick et al., 1986). When water is more limited, a plant's ability to discriminate between carbon isotopes is reduced, leaving plants with higher ¹³C concentrations and more positive δ¹³C values at drier conditions (Hubrick et al., 1986).

Paleo-temperature and precipitation inferences

Shell δ¹⁸O values were compared with the oxygen isotope values of two well-studied paleoclimate proxies: (1) Recent and Pleistocene seawater δ¹⁸O and (2) Greenland ice core δ¹⁸O values (Fig. 6). For seawater, we can evaluate time series of δ¹⁸O values of benthic foraminifera collected from the southwest coast of Alaska (Grebmeier et al., 1990; Fig. 1). Late-glacial seawater δ¹⁸O averaged 1‰, with low variability (<1.5‰). Modern seawater δ¹⁸O off the coast of Alaska averages −1.1‰, which is ~2‰ lower than late-glacial seawater δ¹⁸O (Fig. 6A). Late-glacial δ¹⁸O values were reported from the North Greenland Ice Core Project ice core (NGRIP Members, 2004) and modern δ¹⁸O values were reported from the North Greenland Eemian Ice Drilling (NEEM) ice core (Masson-Delmotte et al., 2015). Greenland ice core δ¹⁸O compositions vary through the late glacial, averaging −39‰ during the early Bølling–Allerød (~14 ka) and shifting to −40.4‰ during both the early Younger Dryas (12.9–12.3 ka) and the late Younger Dryas (12.3–11.7 ka). Modern Greenland ice core δ¹⁸O values are the highest, averaging −32.4‰ (Fig. 6C).

When comparing snail shell δ¹⁸O values with those of seawater and Greenland ice core records, it is important to recognize that they are recording different periods of time: seawater and ice core records generally average inputs across annual timescales (Grebmeier et al., 1990; Masson-Delmotte et al., 2015), while snail values record signatures of their growing seasons (Yanes et al., 2019). Despite the differences in temporal resolutions, comparing snail shell δ¹⁸O values to these well-established paleoclimatic proxies provides context that aids in the interpretation of the respective roles of precipitation and temperature in determining fossil snail shell δ¹⁸O values in interior Alaska. Fluctuations in snail shell δ¹⁸O values do not mimic either Alaska seawater δ¹⁸O or Greenland ice core δ¹⁸O trends, as indicated by their nonsignificant Pearson correlation. Changes in shell δ¹⁸O values through the late glacial likely tracked a combination of several atmospheric conditions, including precipitation and water vapor δ¹⁸O composition, air temperature, and local RH (Balakrishnan and Yapp, 2004). Similar to this study, other temperature proxies from interior Alaska show results inconsistent with Greenland ice core records. The brGDGT temperature records from interior Alaska do not indicate warmer temperatures during the Bølling–Allerød or consistently cooler temperatures during the Younger Dryas, in contrast to Greenland ice core records (Kielhofer et al., 2023).

Flux balance model

Measured shell δ¹⁸O values from this study, along with local temperature and precipitation δ¹⁸O estimates gathered from previously published data (Rozanski et al., 1993; Kielhofer et al., 2023), were used to apply the Balakrishnan and Yapp (2004) evaporative steady-state flux balance mixing model to estimate changes in late-glacial RH in interior Alaska. The model assumes (1) snail body water is lost through evaporation and (2) precipitation δ¹⁸O and water vapor δ¹⁸O values are in isotopic equilibrium. Despite the model's reliance on various unknown variables and assumptions, it can still provide a valuable nominal estimate of the RH at which snails grow. The flux balance model was used in this study to assess how RH changes from the early Bølling–Allerød to the early and late Younger Dryas with respect to the modern. Each period was analyzed under two temperature scenarios.

The model was applied to three late-glacial time periods: the early Bølling–Allerød (~14 ka), with an average shell δ¹⁸O value of −10.3‰; the early Younger Dryas (12.9–12.3 ka), with an average shell δ¹⁸O value of −12.1‰; and the late Younger Dryas (12.3–11.7 ka), with an average shell δ¹⁸O value of −11.2‰. Scenario 1 assumes snails were active only above 10°C (Cowie, 1984; Thompson, 1996). Precipitation δ¹⁸O was calculated from the Rozanski et al. (1993) isotope–temperature coefficient at high latitudes. These investigations reported that precipitation δ¹⁸O decreases by 0.58‰ for every 1°C of temperature decrease. Therefore, if temperatures decreased by ~2°C during the last glacial period (Kielhofer et al., 2023) and modern rainwater δ¹⁸O is −17.1‰, paleontological rainwater δ¹⁸O would have been −18.2‰ (Standard Mean Ocean Water [SMOW]) at this site. If this scenario is valid, then interior Alaska snails dating to the early Bølling–Allerød (~14 ka), with an average shell δ¹⁸O value of −10.3‰, grew at times when snails were active at 10°C, precipitation δ¹⁸O was −18.2‰, and RH was 85% (Fig. 7B). Shells dating to the early Younger Dryas (12.9–12.3 ka PB), with an average shell δ¹⁸O value of −12.1‰, grew at times when snails were active at 10°C, precipitation δ¹⁸O was −18.2‰, and RH was 90% (Fig. 7B). Shells dating to the late Younger Dryas (12.3–11.7 ka), with an average shell δ¹⁸O value of −11.2‰, grew at times when snails were active at 10°C, precipitation δ¹⁸O was −18.2‰, and RH was 87% (Fig. 7B).

Figure 7. Calculated Succinea shell δ¹⁸O values as a function of relative humidity (RH) using the evaporative steady-state flux balance mixing model by Balakrishnan and Yapp (2004). (A) Calculations for modern (live-collected) land snail shells from central Alaska assuming snails were active for an extended growing season, months of April to October (Nield et al., 2022. With rain δ¹⁸O value of −17.1‰ SMOW, two possible temperature scenarios, 8°C (dashed line) and 10°C (solid line) (see text). Filled dots represents the average shell δ¹⁸O value for modern shells. (B) Paleoclimatic scenario 1. Calculations were performed assuming snails were active only above 10°C (Cowie, 1984; Thompson, 1996) and rain δ¹⁸O values (−18.2‰ SMOW) were estimated using the isotope–temperature coefficient at high latitudes (see text) (Rozanski et al., 1993). Filled dots represent the average shell δ¹⁸O value for the early Bølling–Allerød (BA) (~14 ka), early Younger Dryas (YD), late YD, and modern. (C) Paleoclimate scenario 2. Calculations were performed using glacial temperature (6°C) estimated from a branched glycerol dialkyl glycerol tetraethers (brGDGT) record from a loess–paleosol sequence in interior Alaska (Kielhofer et al., 2023) and rain δ¹⁸O value of −18.2‰ SMOW (Rozanski et al., 1993) (see text). Filled dots represent the average shell δ¹⁸O value for the early BA (~14 ka) (−10.3‰), early YD (−12.1‰), late YD (−11.2‰), and modern (−10.7‰).

Because snail activity has not been studied in detail in high-latitude regions, and some previous work has documented that snails track extended summer (April–October) atmospheric conditions (Nield et al., 2022), it is possible that Succinea snails could be active at temperatures below 10°C. Hence, scenario 2 assumes modern snails are active at 8°C. This scenario further assumes that late-glacial snails were active at 6°C, as inferred from a nearby brGDGT record, which found growing season temperatures during the Younger Dryas were approximately ~2°C colder than modern temperatures in interior Alaska, on average 8°C during the snail active period (Kielhofer et al., 2023). Precipitation δ¹⁸O was estimated at −18.2‰ (SMOW), based on the Rozanski et al. (1993) equation, as calculated earlier. If this scenario is valid, then interior Alaska snails dating to the early Bølling–Allerød (~14 ka), with an average shell δ¹⁸O value of −10.3‰, grew at times when temperatures were 6°C, precipitation δ¹⁸O was −18.2‰, and RH was 88% (Fig. 7C). Shells dating to the early Younger Dryas (12.9–12.3 ka PB), with an average shell δ¹⁸O value of −12.1‰, grew at times when temperatures were 6°C, precipitation δ¹⁸O was −18.2‰, and RH was 92% (Fig. 7C). Shells dating to the late Younger Dryas (12.3–11.7 ka), with an average shell δ¹⁸O value of −11.2‰, grew at times when temperatures were 6°C, precipitation δ¹⁸O was −18.2‰, and RH was 90% (Fig. 7C).

Calculations using the flux balance model for modern (live-collected) land snails from interior Alaska were made assuming snails were active for an extended growing season, April– October, as suggested by prior work (Nield et al., 2022). A precipitation δ¹⁸O value of −17.1‰ (SMOW) was assumed based on modeled results from IAEA/WMO (2015) by Bowen et al. (2005) (Fig. 7A). Model calculations were again performed at two possible temperature scenarios, 8°C, average extended growing season temperature in interior Alaska, and 10°C (Fig. 7A). If the assumptions behind these two scenarios are valid, then modern snails with an average shell δ¹⁸O value of −10.7‰, grew at times when RH was ~89–90% (Fig. 7A).

In both plausible situations, modern RH is predicted to be higher than in the early Bølling–Allerød, lower than in the early Younger Dryas, and similar to or slightly higher than in the late Younger Dryas (Fig. 7). Both scenarios predict high RH values (>85%) at all time intervals studied. This interpretation is plausible, as snails are mainly active during or directly after rain events, when RH is the highest (Ward and Slotow, 1992).

Paleo-humidity fluctuations during the Younger Dryas

As inferred from the Balakrishnan and Yapp (2004) flux balance model, RH varied through the late-glacial period. In this study (14–11.5 ka), the lowest RH was found during the early Bølling–Allerød (~14 ka). Similar to previously discussed paleoclimate proxies (e.g., Bigelow and Edwards, 2001; Kielhofer et al., 2023), the data presented here do not show a dramatic climate shift at the start of the Younger Dryas. Shell δ¹⁸O values from 13.5 to 12.8 ka have values similar to those of the early Younger Dryas, suggesting similar RH levels (Fig. 6A). Some paleoclimate studies in central and northern Alaska reported a prolonged dry period leading up to the Younger Dryas (Abbott et al., 2000; Edwards et al., 2001; Rabanus-Wallace et al., 2017), which we observed in the early Bølling–Allerød (~14 ka). This is likely due to a combination of factors, including the presence of the continental ice sheets, increased summer insolation, cool sea-surface temperatures, and lower sea levels during the late-glacial (Abbott et al., 2000). During the last glacial period, the Cordilleran and Laurentide Ice Sheets have been modeled as deflecting the westerlies south of their modern track, causing lower moisture in interior Alaska compared with today (COHMAP Members, 1988). Additionally, reduced North Atlantic Ocean sea-surface temperatures could have affected westerly flow and the strength of the Aleutian low air-pressure system (Kokorowski et al., 2008). Sea level was ~100 to 120 m lower than today, leaving portions of the Bering shelf exposed, increasing the distance for moisture transport to interior Alaska (Clark et al., 2014).

Our data suggest an increase in RH after the drier early Bølling–Allerød. RH through the Younger Dryas does not seem to be constant, as our data imply a two-stage climate pattern. The offset in snail shell oxygen isotope values between the early Younger Dryas (12.9–12.3 ka) and the late Younger Dryas (12.3–11.7 ka) is ~1‰. According to results from the flux balance model (Balakrishnan and Yapp 2004), this implies a wetter early Younger Dryas and subsequent drier late Younger Dryas. This two-step Younger Dryas phenomenon has been observed in other locations in Alaska (Kaufman et al., 2010) and globally (Brauer et al., 1999; Bakke et al., 2009; Schlolaut et al., 2017; Pigati and Springer, 2022).

The Younger Dryas is well documented in southern and coastal Alaska (Kokorowski et al., 2008). In these regions, the results contradict the findings reported in this study, pointing to a mild early Younger Dryas and subsequently wetter late Younger Dryas. Jones et al. (2009) observed an increase in Polypodiaceae (fern) spores from a peat core in Swanson Fen (southwestern Alaska), indicating an increase in moisture around 12.2 ka. This agrees with paleoclimatic inferences from diatom assemblages at Arolik Lake (Hu et al., 2006), pollen records in nearby Nimgun Lake (Hu et al., 2002), and pollen and microfossil assemblages from Discovery Pond (Kaufman et al., 2010). However, all these sites are closer to coastal moisture sources compared with interior Alaska (this study).

Late-glacial paleoclimate records in interior Alaska, although not as abundant as those in southern Alaska, can be compared with our results. Abbott et al. (2000) studied lake-level history from seismic profiles and sediment cores of Birch Lake, located ~50 km from three of the archaeological sites analyzed in this study: Mead, Bachner, and North Gerstle. Lake systems in interior Alaska are sensitive to climate change and respond rapidly to moisture changes. In agreement with snail shell δ¹⁸O values, the lake levels at Birch Lake suggest there was an increase in humidity at the beginning of the Younger Dryas, with lake levels rising dramatically (by 18 m) between 12.7 and 12.2 ka and then stabilizing during the late Younger Dryas.

Today, the moisture in interior Alaska is mainly obtained from a westerly flow from the North Pacific Ocean and is related to large-scale atmospheric circulation patterns (Abbott et al., 2000; Kokorowski et al., 2008; Chipman et al., 2012). At the start of the Younger Dryas, interior Alaskan paleoclimate proxies suggest an increase in moisture. This change in RH could be caused by the change in sea level, which rose 30 m from 12.5 to 11.6 ka (Clark et al., 2014), or the decrease in size of the Cordilleran and Laurentide Ice Sheets, which affected circulation. As the ice sheets decreased, their impact on circulation would have diminished, making circulation patterns more like those of today (Bartlein et al., 1991).

CONCLUSIONS

The δ¹³C values of Succinea shells in interior Alaska suggest that δ¹³C values of late-glacial vegetation were consistently more positive than those of modern shells. Part of this difference is probably due to changes in the δ¹³C value of CO₂ because of the increased burning of fossil fuels (Suess effect), which affects the δ¹³C values of consumed plants. The rest of the offset is likely explained by changes in the δ¹³C composition of C₃ plants consumed by snails. This is possibly explained by changes in the water-use efficiency of C₃ plants during the slightly drier late-glacial period compared with today, which would leave plants, and therefore snail shells, more enriched in ¹³C due to changes in isotope discrimination during slightly drier conditions.

The δ¹⁸O values of late-glacial Succinea shells in interior Alaska appear to record changes in RH over the late glacial. RH was modeled using the Balakrishnan and Yapp (2004) flux balance mixing model. The results show RH increased from ~14 ka to the early Younger Dryas (~12.9–12.3 ka), then decreased in the latter half of the Younger Dryas (~12.3–11.7 ka), with modern conditions being similar to those in the late Younger Dryas. Shell δ¹⁸O values from interior Alaska highlight a two-stage humidity pattern during the Younger Dryas, which other studies have found evidence of at other sites in interior Alaska and elsewhere in North America. Published climate and humidity records in Alaska during the Younger Dryas are somewhat complex; however, lake proxies from interior Alaska agree with our findings and show evidence for a wetter early Younger Dryas compared with other periods of the late glacial.

This study illustrates that ancient land snails may be useful for clarifying the complex climatic history of the late glacial in interior Alaska. Isotope records dating to the late glacial in interior Alaska are rare. Additional isotope records dating to the Younger Dryas may help solidify emerging patterns about late-glacial humidity in interior Alaska.