INTRODUCTION
Developing and augmenting paleoclimate proxies and increasing the number of paleoclimate studies are necessary and urgent tasks to improve our understanding of climate change and variability at various spatiotemporal scales. In particular, proxies from the continental realm that offer hemispheric to near-global coverage are rare. For example, tree-ring records are generally limited to middle latitudes of the Northern Hemisphere, speleothems develop in cave settings only, and ice-core records are only accessible at the highest latitudes or very high elevations. In this context, land snail shells are an exceptionally valuable proxy because they exhibit a near-global spatial distribution within the continental realm (except Antarctica), ranging from the tropics to the high Arctic, and are present in almost every environment and biome, ranging from deserts to wetlands.
Within North America, more than 1200 species of land snails have been identified spanning nearly the entire continent and all terrestrial biomes except for hyperarid deserts and the high Arctic (Nekola, Reference Nekola2014). Approximately 40% of North American land snails have shells that are smaller than 10 mm in maximum shell length, with 10 families accounting for more than 75% of the small species, including Vertiginidae (106),Footnote 1 Polygyridae (80), Pristilomatidae (72), Oxychilidae (32), Gastrodontidae (30), Helicodiscidae (27), Succineidae (19), Pupillidae (16), Discidae (14), and Valloniidae (13) (Nekola, Reference Nekola2014). The same families are also common in the Quaternary fossil record of North America, as they are preserved in paleosols, loess sequences, and wetland, fluvial, alluvial, colluvial, and lake deposits. All class sizes of shells, from minute (<5 mm) to large (>20 mm), are frequently preserved in a variety of Quaternary sedimentary settings. Fossil shells of these and other taxa are also found at paleontological (Goodfriend, Reference Goodfriend1992, Reference Goodfriend1999) and archaeological (Evans, Reference Evans1972; Lubell, Reference Lubell2004) sites worldwide. Because of their ubiquity and excellent preservation potential, land snail shells offer unique opportunities to investigate climate and ecosystems of the past at broad spatiotemporal scales within the continental realm.
Increasingly, researchers are using the oxygen isotope composition (δ18O) of land snail shells to reconstruct aspects of Quaternary climate (Fig. 1). Land snails form their shells in isotopic equilibrium with snail body fluids, which, in turn, are strongly correlated with meteoric precipitation δ18O values (Prendergast et al., Reference Prendergast, Stevens, Barker and O’Connell2015). Therefore, the δ18O of snail shells can be used as a paleoenvironmental proxy if the dominant environmental factors controlling the oxygen isotopic system are reasonably well characterized (Balakrishnan and Yapp, Reference Balakrishnan and Yapp2004). Although several studies have tested and assessed the utility of land snails in paleoenvironmental studies, the majority of published articles have been limited in scope, targeting one or a few species and often exploring just a single biome or region.
Figure 1 Summary of published studies (1979–2018) of oxygen isotopes of modern and fossil land snails worldwide. (A) Cumulative number of published articles by year of publication. (B) Proportion of studies published by major geographic region.
In North America, for example, only six published studies have attempted to calibrate the δ18O systematics of modern land snails. Yapp (Reference Yapp1979) analyzed 30 aliquots of shells of ecologically disparate species, ranging from the Yucatan Peninsula (20°N) to Wisconsin (43°N), and observed that the difference (Δ18O) between the calculated δ18O of the snail body fluid and the δ18O of the average annual precipitation correlated with the reciprocal of the average annual relative humidity (RH) at those locales. Goodfriend and Ellis (Reference Goodfriend and Ellis2002) analyzed snails of the genus Rabdotus living in Texas and concluded that shell δ18O values are primarily controlled by precipitation δ18O values, rather than precipitation amount or evaporation rates. Balakrishnan et al. (Reference Balakrishnan, Yapp, Theler, Carter and Wyckoff2005b) studied land snails living in the southern Great Plains along an east-to-west transect between Oklahoma and New Mexico and concluded that RH, precipitation δ18O, and temperature all seemed to play roles in shell δ18O values. Yanes (Reference Yanes2015) found that the differences between the δ18O values of the shells of modern snails of the genus Succinea retrieved from tropical (San Salvador, the Bahamas) and high (Fairbanks, Alaska) latitudes primarily record the differences in the δ18O values of the meteoric waters at the two locales. Yanes et al. (Reference Yanes, Nekola, Rech and Pigati2017) observed that small land snails from northwest Minnesota mainly tracked precipitation δ18O values; however, coexisting species showed significantly different isotopic values. Finally, Yanes et al. (Reference Yanes, Graves and Romanke2018) observed that large body-size snails of the genus Neohelix from the Appalachian Mountains primarily record changes in precipitation δ18O values and RH along an elevation transect between ~700 and ~1600 meters above sea level (m asl).
Despite the relative paucity of calibration studies, researchers are applying fossil land snail shell δ18O values for paleoclimatic reconstructions at a rapid rate (Fig. 1A). It is therefore essential that systematic and comprehensive calibration studies be conducted to resolve how the fossil shell δ18O data should be interpreted. Such calibration studies should have large spatial and ecological coverage so that future applications do not extrapolate beyond the limits of the data, and they should focus on taxa that are most common in the fossil record.
In this article, we present oxygen isotope data for several species of modern small land snails in North America collected across an unprecedented spatial scale, ranging from northern Florida (30°N) to northern Manitoba (58°N). We then augment this data with published oxygen isotope results on land snails from North America and the Caribbean, which expand our latitudinal coverage from Jamaica (18°N) to Fairbanks, Alaska (64°N). We also evaluate shell δ18O values of modern land snails using instrumental climate data and discuss the usefulness and limitations of land snails as paleoclimatic proxies in North America. We then review possible mechanisms explaining the observed patterns and identify several avenues of potential research that we think will advance the field of land snail isotope research significantly.
Overview of published land snail isotopic studies worldwide
Since Yapp’s seminal article on the δ18O of land snail shells was published nearly four decades ago (Yapp, Reference Yapp1979), approximately 70 articles have been published on the topic of stable isotopes in land snail shells (Supplementary Table 1). The bulk of these studies have attempted to use snail δ18O values to reconstruct local precipitation δ18O values for various periods of geologic time, primarily the Holocene and late Pleistocene (Supplementary Table 1). However, inferences on precipitation δ18O values from snail shells remain semiquantitative, as other climatic parameters such as RH, air temperature, water vapor, species ecology, and perhaps vital effects also appear to play a role in the snail oxygen isotope budget (Balakrishnan and Yapp, Reference Balakrishnan and Yapp2004).
Worldwide, ~37 studies have focused on modern calibration of the oxygen isotope composition of snail shells for a target region and taxa (Supplementary Table 1). However, nearly all published studies focusing on fossil shells generally include modern shell analyses for comparative purposes, an indispensable step if the species and locale have never been explored before. Some of these studies report site-specific and/or species-specific regression equations (Lécolle, Reference Lécolle1985; Goodfriend and Ellis, Reference Goodfriend and Ellis2002; Balakrishnan et al., Reference Balakrishnan, Yapp, Theler, Carter and Wyckoff2005b; Zanchetta et al., Reference Zanchetta, Leone, Fallick and Bonadonna2005; Yanes et al., Reference Yanes, Delgado, Castillo, Alonso, Ibáñez, De la Nuez and Kowalewski2008, Reference Yanes, Romanek, Delgado, Brant, Noakes, Alonso and Ibáñez2009; Prendergast et al., Reference Prendergast, Stevens, Barker and O’Connell2015), which is helpful, although we note that such equations are only valid for the taxa and location investigated. Balakrishnan and Yapp (Reference Balakrishnan and Yapp2004) proposed a universal mathematical model that combines empirical and theoretical data that can be used to evaluate the impact of multiple atmospheric variables (meteoric precipitation δ18O, water vapor δ18O, RH, and temperature) on shell δ18O values. The model is robust and can be applied to land snails in all parts of the world, but many studies conducted after their article was published in 2004 have not employed it, partly because of the large number of unknown variables and assumptions required.
Geographically, one third of the ~71 published articles on stable isotopes of land snail shells have been conducted in Europe (Supplementary Table 1; Fig. 1B), with fewer in North America (Yapp, Reference Yapp1979; Sharpe et al., Reference Sharpe, Forester, Whelan and McConnaughey1994; Goodfriend and Ellis, Reference Goodfriend and Ellis2000, Reference Goodfriend and Ellis2002; Balakrishnan and Yapp, Reference Balakrishnan and Yapp2004; Balakrishnan et al., Reference Balakrishnan, Yapp, Meltzer and Theler2005a, Reference Balakrishnan, Yapp, Theler, Carter and Wyckoff2005b; Zaarur et al., Reference Zaarur, Olack and Affek2011; Stevens et al., Reference Stevens, Metcalfe, Leng, Lamb, Sloane, Naranjo and González2012; Paul and Mauldin, Reference Paul and Mauldin2013; Yanes, Reference Yanes2015; Yanes et al., Reference Yanes, Nekola, Rech and Pigati2017, Reference Yanes, Graves and Romanke2018; Nash et al., Reference Nash, Conroy, Grimley, Guenthner and Curry2018) and the Caribbean (Baldini et al., Reference Baldini, Walker, Railsback, Baldini and Crowe2007; Yanes and Romanek, Reference Yanes and Romanek2013). As stated previously, of the North American studies, only six conducted modern calibration assessments for shell δ18O values (Yapp, Reference Yapp1979; Goodfriend and Ellis, Reference Goodfriend and Ellis2002; Balakrishnan et al., Reference Balakrishnan, Yapp, Theler, Carter and Wyckoff2005b; Yanes, Reference Yanes2015; Yanes et al., Reference Yanes, Nekola, Rech and Pigati2017, Reference Yanes, Graves and Romanke2018), and all of them were limited in scope.
The vast majority of snail isotope studies worldwide have analyzed δ18O values of entire shells rather than employing a time-series approach (Supplementary Table 1), although eight published articles have explored high-resolution oxygen isotope time series along shell growth direction (Goodfriend, Reference Goodfriend1992; Leng et al., Reference Leng, Heaton, Lamb and Naggs1998; Baldini et al., Reference Baldini, Walker, Railsback, Baldini and Crowe2007; Yanes et al., Reference Yanes, Yapp, Ibáñez, Alonso, De-la-Nuez, Quesada, Castillo and Delgado2011b, Reference Yanes, Gutiérrez-Zugasti and Delgado2012, Reference Yanes, Izeta, Cattáneo, Costa and Gordillo2014; Rangarajan et al., Reference Rangarajan, Ghosh and Naggs2013; Ghosh et al., Reference Ghosh, Rangarajan, Thirumalai and Naggs2017). In addition, one study focused on shell margin or last growth episode oxygen isotope values (Yanes and Fernandez-Lopez-de-Pablo, Reference Yanes and Fernandez-Lopez-de-Pablo2017), and two have analyzed snail body fluid δ18O values in conjunction with shell δ18O (Goodfriend et al., Reference Goodfriend, Magaritz and Gat1989; Prendergast et al., Reference Prendergast, Stevens, Barker and O’Connell2015).
Four studies used laboratory-controlled experiments to examine stable carbon isotope systematics in snail diet, shell, and body tissue (Stott, Reference Stott2002; Metref et al., Reference Metref, Rousseau, Bentaleb, Labonne and Vianey-Liaud2003; Liu et al., Reference Liu, Gu, Wu and Xu2007; Zhang et al., Reference Zhang, Yamada, Suzuki and Yoshida2014), although none have investigated the oxygen isotope composition of land snails under laboratory conditions. The absence of laboratory-based oxygen isotope studies is likely attributable to the difficulty in quantitatively constraining the relatively high number of environmental variables affecting the snail oxygen isotope composition (Balakrishnan and Yapp, Reference Balakrishnan and Yapp2004). Such experiments are extremely challenging, as they require establishing, stabilizing, and monitoring multiple climatic variables simultaneously and sustaining the snail communities healthy and alive, which can be difficult for long periods of time (>4 months).
Finally, three studies have measured clumped isotopes in land snail shells (Zaarur et al., Reference Zaarur, Olack and Affek2011; Eagle et al., Reference Eagle, Risi, Mitchell, Eiler, Seibt, Neelin, Li and Tripati2013; Wang et al., Reference Wang, Cui, Zhai and Ding2016). Some clumped isotope results showed that land snail calcification temperatures were significantly higher than ambient average annual temperatures or snail active period temperatures, especially at mid- to high latitudes (Zaarur et al., Reference Zaarur, Olack and Affek2011). This suggests that snails preferentially add shell material when environmental conditions are at optimum (warmer) levels, which may vary between species (Wang et al., Reference Wang, Cui, Zhai and Ding2016). The results of these pioneering studies are intriguing and suggest that further investigations of clumped isotope paleothermometry in modern land snail shells from contrasting environments and ecologies are warranted.
Land snail physiology and ethology
Although many terrestrial gastropods are detritivores that live among leaf litter or under rocks at the soil-air interface, as a whole they exhibit a range of trophic states (including numerous herbivores and even some carnivores) and microhabitat preferences (from subterranean to arboreal) (Wilbur and Yonge, Reference Wilbur and Yonge1964; Pearce and Örstan, Reference Pearce and Örstan2006; Meyer and Yeung, Reference Meyer and Yeung2011). Land snails form their shells under a relatively limited range of environmental conditions, generally when temperatures are between ~10 and ~27°C and RH is greater than ~75 (Herreid, Reference Herreid1977; Balakrishnan and Yapp, Reference Balakrishnan and Yapp2004; Pearce and Örstan, Reference Pearce and Örstan2006). Accordingly, snail activity and life cycles may vary significantly between species, latitudes, or contrasting habitats. The extreme conditions at high latitudes characterized by colder temperatures, marked seasonality, and shorter growing seasons, likely result in a decrease in snail metabolic rates and a reduction in annual growth rates (Gaitán-Espitia and Nespolo, Reference Gaitán-Espitia and Nespolo2014). For example, large body-size snails appear to tolerate freezing temperatures (Ansart and Vernon, Reference Ansart and Vernon2004) and drier conditions (Nevo et al., Reference Nevo, Bar-El and Bar1983; Yanes et al., Reference Yanes, Gutiérrez-Zugasti and Delgado2012) better than their smaller counterparts, which may result in differing active periods (Yanes et al., Reference Yanes, Nekola, Rech and Pigati2017). Thus, specific taxa may exhibit different tolerance ranges, activity, and behavior at various latitudes, and possibly within and between microhabitats (Yanes et al., Reference Yanes, Nekola, Rech and Pigati2017).
In low-latitude and maritime climates, where temperature and precipitation are relatively constant and RH remains high (>75%, on average) year-round, land snails may be active throughout all seasons (Yanes et al., Reference Yanes, Delgado, Castillo, Alonso, Ibáñez, De la Nuez and Kowalewski2008, Reference Yanes, Romanek, Delgado, Brant, Noakes, Alonso and Ibáñez2009, 2011, Reference Yanes, García-Alix, Asta, Ibáñez, Alonso and Delgado2013; Yanes and Romanek, Reference Yanes and Romanek2013). In contrast, land snails in temperate and cold regions are expected to be most active during the warmer months (Yanes, Reference Yanes2015), whereas in hot or arid environments they are mainly active during wet periods or at night when RH is high (Cook, Reference Cook2001). This complex behavior across contrasting environments has significant implications for interpreting shell δ18O records because reconstructed climate parameters may only reflect the season, or time of day, that snails are active, rather than mean annual conditions, and may be biased toward microclimatic conditions. The main challenge in this field of study is that specific details regarding snail active periods and behavior are virtually unknown for most (if not all) land snail species in North America. Therefore, isotope geochemists must make assumptions about expected snail active periods when relating the snail isotopic data to climate data, assigning them to specific seasons, or using a range of temperatures and humidity levels to discuss active periods in general.
Land snail shell formation
Land snail shell calcification is performed by the mantle, an organ composed of two epithelia separated by connective tissue and located at the inner surface of the shell (Wilbur and Yonge, Reference Wilbur and Yonge1964). Accretionary shell growth takes place at the shell margin, and the mantle edge is the most active calcification zone (Wilbur and Yonge, Reference Wilbur and Yonge1964). The epithelial cells of the mantle obtain the calcium and bicarbonate ions necessary to form shell from the hemolymph or snail’s internal body fluid. These ions are primarily obtained from the diet (e.g., carbon from consumed plant matter and, to a lesser extent, carbonate rocks) and the environment (e.g., oxygen from imbibed environmental water) (Goodfriend and Hood, Reference Goodfriend and Hood1983; Goodfriend et al., Reference Goodfriend, Magaritz and Gat1989).
Unlike marine mollusk shells, which are often arranged in several layers of two different calcium carbonate polymorphs, most species of land snails build a shell composed of a single layer of aragonite, which is a metastable polymorph of calcium carbonate. The mineral portion of the snail shell can account for >90% of the total mass for many species and is covered by a thin outer layer known as the periostracum, a protein-rich film that decays quickly after snail death. Fortunately, aragonite shells are fairly resistant to decay and can be preserved in the geologic record for thousands, and sometimes millions, of years (Pearce, Reference Pearce2008; Rech et al., Reference Rech, Pigati, Lehmann, McGimpsey, Grimley and Nekola2011). Moreover, potential diagenetic alteration can be examined through various techniques including X-ray diffraction, scanning electron microscopy, and Raman spectroscopy, among others (Rech et al., Reference Rech, Pigati, Lehmann, McGimpsey, Grimley and Nekola2011).
Environmental controls on land snail shell δ18O values
The δ18O values of aquatic mollusk shells (both marine and freshwater) are primarily determined by two environmental variables: (1) the δ18O values of the host water and (2) temperature during calcification (Epstein et al., Reference Epstein, Buchsbaum, Lowenstam and Urey1951; Grossman and Ku, Reference Grossman and Ku1986). The systematics of oxygen isotopes in land snail shells are more complicated, with at least four environmental variables contributing to shell δ18O values: (1) δ18O values of local precipitation, (2) RH, (3) temperature, and (4) δ18O values of ambient water vapor (Balakrishnan and Yapp, Reference Balakrishnan and Yapp2004). Shell δ18O values thus depend not only on local and regional climate conditions, but also on the interaction of species ecology and behavior with meso- and microclimatic conditions that can vary both between and within sites (Yanes et al., Reference Yanes, Nekola, Rech and Pigati2017). To accurately interpret shell δ18O values, we must understand the impact that each parameter has on the shell δ18O value under a variety of environmental and climatic conditions at different spatial scales and for the taxa that are most abundant in the fossil record.
MATERIAL AND METHODS
Field sampling strategy
Modern land snails were collected from seven localities in the eastern continental United States (Fig. 2), including the following: (1) Fort George Island, Florida (30.410°N); (2) Norfolk Bluff, Arkansas (36.224°N); (3) Cedar Bog Preserve, Ohio (40.060°N); (4) Heritage Farm, Iowa (43.382°N); (5) Randeen Ridge, Minnesota (48.479°N); (6) Buffalo Lake, Manitoba (53.410°N); and (7) Goose Greek Road, Churchill, Manitoba (58.709°N) (Table 1; Supplementary Table 2). This transect spans a wide range in average annual temperatures (−6.5 to +21.2°C), precipitation amount (276 mm to 1271 mm), and weighted average annual precipitation δ18O values (−17.5 to −3.0‰), while keeping average annual RH values relatively high (~69–76%). At each sampling site, snails were collected by hand for larger shells and litter sampling for smaller taxa from representative 100–1000 m2 areas (Nekola, Reference Nekola2003). Soil litter sampling was primarily used as it provides the most complete assessment of site faunas (Oggier et al., Reference Oggier, Zschokke and Baur1998). As suggested by Emberton et al. (Reference Emberton, Pearce and Randalana1996), collections were made at places of high micromollusk density, with a constant volume of soil litter of ~4 L being gathered at each site (Nekola, Reference Nekola2014).
Figure 2 (A) Geographic locations of field-collected modern land snails from North America. Black dots depict new data presented in this study, and red dots represent data of modern snails from previously published work in North America. (B) General view of small land snails from North America on a penny as scale. (C) Detailed view of several species of Gastrocopta, a highly abundant genus found in modern settings and the fossil record across North America. Scale bar=1 mm. Note that the majority of land snails used in this study are categorized as small taxa, with a shell length <10 mm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Table 1 Site-average oxygen stable isotope values of land snails from North America and relevant instrument climate data. Geographic locations of these sites are indicated as black dots in Figure 2. m asl, meters above sea level; PDB, Pee Dee belemnite; RH, relative humidity; SMOW, standard mean ocean water.
a Data from U.S. Climate Data site (www.usclimatedata.com) and Canadian Climate Normals site (http://climate.weather.gc.ca/climate_normals/index_e.html).
b Data calculated from the Online Isotopes in Precipitation Calculator (http://wateriso.utah.edu/waterisotopes/pages/data_access/oipc.html).
c Snail active period values were calculated for months with mean air temperatures between 10°C and 27°C.
Relevant climatic variables including mean annual temperature (MAT), precipitation amount (in millimeters), and average annual RH were gathered from data reported at the U.S. Climate Data website (www.usclimatedata.com) and the Canadian Climate Normals website (http://climate.weather.gc.ca/climate_normals/index_e.html). Oxygen isotope values of mean annual and monthly precipitation were obtained from the Online Isotopes in Precipitation Calculator of the Isoscapes website (http://wateriso.utah.edu/waterisotopes/pages/data_access/oipc.html) (Bowen, Reference Bowen2018). The calculated precipitation δ18O values are derived from global data sets and equations presented and discussed in Bowen and Wilkinson (Reference Bowen and Wilkinson2002), Bowen and Revenaugh (Reference Bowen and Revenaugh2003), and Bowen et al. (Reference Bowen, Wassenaar and Hobson2005).
Land snail species
Ideally, the same taxon would be collected at each site to minimize isotopic variability derived from variations in species-specific behavior, diet, resistance to dryness, mobility, and perhaps vital effects (Yanes et al., Reference Yanes, Nekola, Rech and Pigati2017). However, in studies with a large spatial coverage, like the one presented here, this is not possible because the taxonomic composition of snail assemblages varies tremendously between the tropics and the Arctic. Thus, we collected and analyzed a variety of species encountered during our field sample collection, and focused on analyzing the smaller taxa of the assemblages collected. Twenty-two taxa were collected for isotopic analysis along the latitudinal transect, including Anguispira alternata, Cochlicopa lubricella, Discus catskillensis, Euconulus fulvus, Gastrocopta contracta, Glyphyalinia umbilicata, Hawaiia minuscula, Helicina orbiculata tropica, Helicodiscus parallelus, Hendersonia occulta, Nesovitrea binneyana, Nesovitrea dalliana, Oxyloma verrilli, Polygyra lithica, Polygyra pustula, Pupilla muscorum, Pupoides albilabris, Rabdotus dealbatus, Succinea strigata, Succinea sp., Vallonia gracilicosta, and Vertigo modesta modesta (Supplementary Table 1). Although these taxa exhibit significant differences in behavior, mobility, body size, and dietary habits and, consequently, will probably differ in oxygen isotope values (Yanes et al., Reference Yanes, Nekola, Rech and Pigati2017), we hypothesize that over broad spatial (latitudinal) scales, environmental variations will be large enough to overwhelm species-specific variations that can be significant at the microhabitat and habitat scales. Additionally, even though snails from low latitudes will likely exhibit longer active periods throughout the year than individuals from higher (colder) latitudes (Gaitán-Espitia and Nespolo, Reference Gaitán-Espitia and Nespolo2014), we further hypothesize that oxygen isotope shell data will reflect similar aspects of local environmental conditions.
Laboratory analyses
We analyzed a total of 112 entire shells with an average shell size of about 3 mm. Snail bodies were removed from the shells, which were then rinsed in deionized water and subjected to ultrasonication to remove organic and detrital contaminants. This procedure was repeated multiple times until all detritus adhering to the shell had been removed. Shell samples were also treated with 6% sodium hypochlorite (NaOCl) for 48 hours at room temperature to remove all remnants of adhering organic materials. Cleaned and treated shells were then crushed in an agate mortar and pestle and carefully homogenized. An aliquot of each pulverized shell (~150 μg) was then loaded into 12 mL Exetainer vials and analyzed at the stable isotope laboratory at the University of Northern Illinois where vials were flushed with helium, reacted with 100‰ phosphoric acid (H3PO4), and equilibrated at room temperature for 24 hours. They were then analyzed with a GasBench II and a MAT 253 stable isotope ratio mass spectrometer. Isotope measurements were calibrated using National Bureau of Standards (NBS)-19 and NBS-18 international standards. Results are described using δ notation relative to the international standard Pee Dee Belemnite. Analytical precision was ±0.1‰ for δ18O and ±0.08‰ for δ13C (1-sigma) based on the repeated measurements of NBS-19 and NBS-18 standards throughout a sequence.
Statistical analyses
All statistical analyses were performed using PAST 3.12 software (Hammer et al., Reference Hammer, Harper and Ryan2001) considering statistical significance at α=0.05. Pearson correlation was conducted to assess the significance of monotonic relationships between two variables. Regression equations were computed to determine the slope and, therefore, the lapse rate of change of the dependent variable (shell δ18O) with respect to changes in relevant independent variables (i.e., latitude, longitude, altitude, meteoric precipitation δ18O, precipitation amount, air temperature, and RH; Tables 1 and 2). It should be noted that many of the examined variables are correlated with each other across the studied sampling sites. However, each environmental variable was compared with snail shell δ18O separately and discussed in the text.
Table 2 Site-average of published (1979–2018) oxygen stable isotope values of land snails from North America and relevant instrument climate data. Geographic locations of these sites are indicated as red dots in Figure 2. m asl, meters above sea level; PDB, Pee Dee belemnite; RH, relative humidity; SMOW, standard mean ocean water.
a Data calculated from the Online Isotopes in Precipitation Calculator (http://wateriso.utah.edu/waterisotopes/pages/data_access/oipc.html).
RESULTS
Transect from Florida to Manitoba (this study)
The oxygen isotope values of modern land snails from the sampled transect ranged from −9.5‰, for a Vallonia gracilicosta shell collected in Buffalo Lake, Manitoba, to +1.1‰, for a Polygyra pustula shell from Fort George Island, Florida, for the 112 individual shells analyzed (Supplementary Table 1). Average values for the seven sampling sites ranged from −1.3 ± 0.8‰ (n=16) for the Fort George Island site in Florida (30.4°N) to −8.7 ± 0.5‰ (n=13) for the Buffalo Lake site (53.4°N) in Manitoba (Table 1).
Average δ18O values for each site decrease 0.3‰ for every 1° increase in latitude (Fig. 3A), which reflects the influence of both a decrease in precipitation δ18O values and cooler air temperatures with increasing latitude. Shell δ18O values did not show a relationship with altitude (Fig. 3B) likely because all samples were collected at sites below 400 m asl.
Figure 3 (A–J) Bivariate relationships between site-average shell δ18O values and relevant instrument climate parameters from this study using all land snail taxa combined. m asl, meters above sea level; MAT, mean annual temperature; PDB, Pee Dee belemnite; RH, relative humidity; SMOW, standard mean ocean water.
Latitude, precipitation δ18O, precipitation amount, and air temperature were all highly correlated with each other (P<<0.01), whereas RH, longitude, and elevation did not show correlations with each other or with other climate parameters (P>0.05).
Shell δ18O values increased at the rate of 0.5‰ for every 1‰ increase in average annual precipitation δ18O values (Fig. 3C). This marked positive correlation (R 2=0.89; P=0.001) suggests that 89% of the shell δ18O variability can be explained by variations in precipitation δ18O values alone. When considering climate data during snail active periods (i.e., activity during months with average air temperatures between 10°C and 27°C), shell δ18O values increase at the rate of 0.7‰ for every 1‰ increase in snail “active period” precipitation δ18O values (Fig. 3D). The relationship between both variables remained significant (R 2=0.84; P=0.004). Shell δ18O values increase by 0.3‰ for every 1°C increase in MAT (Fig. 3E). Both variables are strongly correlated (R 2=0.83; P=0.005). Snail shell δ18O values are also significantly correlated with snail active period temperature (R 2=0.77; P=0.009), increasing by 0.7‰ for every 1°C increase in snail active period temperature (Fig. 3F). Site-averaged land snail δ18O values increase by 1‰ for every 100 mm increase in precipitation amount (Fig. 3G and H). In contrast to a previous study conducted in North America (Yapp, Reference Yapp1979), no relationship was observed between the oxygen isotope offset (Δ 18O) of the calculated δ18O snail body fluid and average annual precipitation δ18O in isotopic equilibrium with the reciprocal of average annual RH (Fig. 3I and J), possibly because the high average annual RH values from all sites sampled in this study (all above 70%). All of these bivariate relationships remained statistically significant when only minute (<5 mm) taxa were considered.
Transect from Jamaica to Alaska (combined published work)
The monotonic relationships identified previously continue and are significant when combining our shell δ18O values, which ranged from Florida to Manitoba, with previously published data from North America and the Caribbean, which span from Jamaica to Fairbanks, Alaska (Table 2, Fig. 4). Combining these data sets allows us to incorporate a much larger and contrasting range of habitats (from semiarid to wet woodlands), altitudes (from 1 to 2800 m asl), and snail species (from minute to very large species with variable ecologies and behaviors). In this larger data set, correlations between variables remain in many instances significant but display greater scatter (Fig. 4).
Figure 4 (A–J) Bivariate relationship between site-average land snail shell δ18O and relevant instrument climate parameters from all published snail work in North America, including the new data presented in Figure 3. Left column includes all snail data, whereas right column shows data from below 400 m asl only. m asl, meters above sea level; MAT, mean annual temperature; PDB, Pee Dee belemnite; RH, relative humidity; SMOW, standard mean ocean water.
When all shell oxygen isotopic data are considered, shell δ18O values decline 0.2‰ for every 1° increase in latitude (Fig. 4A and B), increase 0.5–0.6‰ for every 1‰ increase in average annual precipitation δ18O value (Fig. 4C and D), and increase 0.2–0.3‰ for every 1°C increase in MAT (Fig. 4E and F). These bivariate relationships are all significant (see the R 2 and P values reported in Fig. 4). The only correlation that is no longer significant is the one between shell δ18O values and annual precipitation amount (Fig. 4G and H). Finally, as before, no significant relationship was documented between the oxygen isotope offset (Δ 18O) of the calculated snail body fluid δ18O and the average annual precipitation δ18O with respect to the reciprocal of average annual RH (Fig. 4I and J).
DISCUSSION
Relationship between shell δ18O values and instrument climate data
Land snail shell δ18O values yielded significant and comparable best-fit linear regression equations when compared against both average annual climate data and expected snail active period climate data (Figs. 3 and 4)—that is, when air temperature is between 10 and 27°C, and RH is >70%. For simplicity, and to facilitate direct comparison with similar snail calibration studies elsewhere, the subsequent discussion focuses on average annual climate data.
Data from our new latitudinal transect (between 30°N and 58°N) suggest that shell δ18O values strongly correlate with average annual precipitation δ18O (Fig. 3C). We obtained the following linear regression equation for our sites from Florida to Manitoba (between 30°N and 58°N) (Fig. 3C):
$$\eqalignno{ & \delta ^{{18}} {\rm O}_{{{\rm shell}}} {\equals}0.5\,(\,\pm\,0.08){\times}\delta ^{{18}} {\rm O}_{{{\rm precip}{\rm .}}} -0.1\,(\,\pm\,0.9).$$
This equation is comparable to that obtained from the limited data reported by the only other large coarse-scale study in North America by Yapp (Reference Yapp1979) for sites with RH >70%:
$$\eqalignno{ & \delta ^{{18}} {\rm O}_{{{\rm shell}}} {\equals}0.43\,(\,\pm\,0.12){\times}\delta ^{{18}} {\rm O}_{{{\rm precip}{\rm .}}} -0.4\,(\,\pm\,0.8).$$
The relationship was similar when all published data from North America and the Caribbean (between 18°N and 64°N) are combined (Fig. 4C):
$$\eqalignno{ & \delta ^{{18}} {\rm O}_{{{\rm shell}}} {\equals}0.5\,(\,\pm\,0.05){\times}\delta ^{{18}} {\rm O}_{{{\rm precip}{\rm .}}} -1.1\,(\,\pm\,0.5).$$
The slopes of these regression equations for North America are comparable to those obtained from modern land snails collected in other regions of the world, including central Europe (Lécolle, Reference Lécolle1985), the Italian Peninsula (Zanchetta et al., Reference Zanchetta, Leone, Fallick and Bonadonna2005), and Libya (Prendergast et al., Reference Prendergast, Stevens, Barker and O’Connell2015), which range between 0.5 and 0.8.
From these equations, we can infer that at broad spatial scales (1) land snail shell δ18O values primarily track variations in meteoric precipitation δ18O and (2) shell δ18O values increase at a lapse rate of 0.5‰ for every 1‰ increase in average annual precipitation δ18O values (Fig. 3C and D; Fig. 4C and D). The slight differences with respect to other calibration studies from different regions of the world suggest that the relationship between the δ18O values of shell and average annual precipitation may be region specific, as previously proposed (Prendergast et al., Reference Prendergast, Stevens, Barker and O’Connell2015), and also may vary with the spatial scale and range of environments considered. In future studies in North America, the scientific community will be able to collect more data from underinvestigated regions and test the applicability of these equations (Eqs. 1 and 2). New data from different locales will reveal the presence or absence of correlations between shell δ18O and environmental variables noted herein. It is likely that additional data from high altitude and extreme environments are needed to generate equations that can be applied at the global scale.
On average, the calculated body fluid δ18O values in isotopic equilibrium with water δ18O values using the equation by Grossman and Ku (Reference Grossman and Ku1986) are 0.6–3.8‰ higher than observed average annual precipitation δ18O values per site (Fig. 3I and J). This offset is consistent with the majority of previous studies and suggests that snail shell aragonite is likely formed from snail body fluid that has undergone some evaporation (Yapp, Reference Yapp1979; Goodfriend et al., Reference Goodfriend, Magaritz and Gat1989; Balakrishnan and Yapp, Reference Balakrishnan and Yapp2004; Prendergast et al., Reference Prendergast, Stevens, Barker and O’Connell2015).
Our data also show that land snail shell δ18O values also increase by 0.3‰ for every 1°C increase in average annual temperature (Fig. 3E). This lapse rate matches well with the known relationships between temperature and precipitation δ18O, and the effect of temperature on oxygen isotope fractionation between aragonite and water. For example, precipitation δ18O values decrease, on average, 0.58‰ for every 1°C decrease of temperature (Rozanski et al., Reference Rozanski, Araguás-Araguás and Gonfiantini1993). In contrast, the equilibrium δ18O values of aragonite increase ~0.23‰ for every 1°C decrease in temperature (Grossman and Ku, Reference Grossman and Ku1986). Together, therefore, this should result in a 0.35‰ increase in the land snail shell δ18O values for every 1°C increase in temperature (Balakrishnan et al., Reference Balakrishnan, Yapp, Theler, Carter and Wyckoff2005b), a value that is similar to that observed in our study (see Fig. 3E).
We found that the relationship between shell δ18O values and average annual temperature was similar when considering only small taxa as well as when we included all taxa regardless of body size. However, the rate increases to 0.7‰ for every 1°C increase in temperature when considering snail active period temperatures rather than average annual temperatures (Fig. 3F). These matching patterns illustrate that despite the complexities inherent to interpreting snail shell oxygen isotope data sets, it is clear that shell δ18O values increase with increasing precipitation δ18O values and increasing air temperatures. Thus, at broad spatial scales, higher shell δ18O values could be interpreted as representing higher precipitation δ18O values and warmer temperatures, whereas lower shell δ18O values should generally depict lower precipitation δ18O values and cooler conditions, although species-specific variations should probably be considered at the habitat and microhabitat scales (Yanes et al., Reference Yanes, Nekola, Rech and Pigati2017).
The evaporative steady-state flux balance model by Balakrishnan and Yapp (Reference Balakrishnan and Yapp2004) suggests that shell δ18O values should decrease by 0.4‰ for every 1% increase in RH if all other climatic parameters stay constant. Our data (Fig. 3I and J), as well as combined published data (Fig. 4I and J), did not show a conclusive relationship between the isotopic offset of snail body fluid δ18O and average annual precipitation δ18O against the reciprocal of average annual RH. Further research is still needed to effectively assess the effects of RH variations on land snail shell δ18O values across species, biomes, and spatial scales. We hypothesize that at high elevations characterized by extremely cold and dry conditions, where absolute humidity is low, as well as in desert environments, RH will play an important role in determining shell δ18O values (see Fig. 5) (see also Yanes et al., Reference Yanes, Romanek, Delgado, Brant, Noakes, Alonso and Ibáñez2009).
Figure 5 (color online) Theoretical relationship between δ18O values of land snail shell carbonate and major continental biomes. The relation between shell δ18O values and latitude is predicted to be nonlinear in cold (average T<0°C) and relatively dry (average RH<50%) regions. PDB, Pee Dee belemnite; RH, relative humidity; SMOW, standard mean ocean water.
Figure 5 summarizes the theoretical relationship between average annual precipitation δ18O and land snail shell δ18O along the total range of continental environments. Our predictions suggest that at coarse scale, multiple species of land snails growing shells in tundra, taiga, temperate forest, and grassland biomes should primarily record 18O-enriched body fluid δ18O values in isotopic equilibrium with average annual precipitation δ18O values (Fig. 5). However, in environments with extreme dryness (e.g., deserts, high elevations) or that are extremely cold (e.g., polar latitudes, high elevations), we expect that shell δ18O values will be higher than predicted relative to the relationship of shell and precipitation δ18O values at other biomes with milder environmental conditions (Fig. 5). These hypotheses should be tested in future snail research investigations.
Processes driving snail shell δ18O spatial patterns
Bowen and Wilkinson (Reference Bowen and Wilkinson2002) evaluated global patterns of meteoric precipitation δ18O values using empirical and theoretical data and found that variations in global precipitation δ18O values versus latitude are best explained by variations of air temperature following a Rayleigh fractionation process between liquid water and water vapor (Dansgaard, Reference Dansgaard1964). Thus, the relationship between global precipitation δ18O values and latitude for low-altitude stations (<200 m asl) was best described by a second-order polynomial equation (Eq. 4) depicting the nonlinear relationship between latitude and temperature, which, in turn, is amplified by the negative relationship between annual precipitation amount and precipitation δ18O values in the tropics (Rozanski et al., Reference Rozanski, Araguás-Araguás and Gonfiantini1993; Bowen and Wilkinson, Reference Bowen and Wilkinson2002).
$$\eqalignno{ \delta ^{{18}} {\rm O}_{{{\rm global}\,{\rm precip}{\rm .}}} {\equals}&{\hskip1pt -0.0051\,\left( {{\rm Latitude}} \right)^{2} } \cr &{\plus}0.1805\left( {{\rm Latitude}} \right)-5.247$$
When this global meteoric precipitation δ18O equation (Eq. 4) is compared with the equation obtained from all low-altitude published snail shell δ18O values in North America along latitude (Eq. 5), we can observe a strong match between both data sets at the considered spatial scale (Fig. 6).
$$\eqalignno{ \delta ^{{18}} {\rm O}_{{{\rm shell}}} {\equals}&-0.0056\left( {{\rm Latitude}} \right)^{2} \cr &{\plus}0.2178\left( {{\rm Latitude}} \right)-2.737$$
Figure 6 Comparison between δ18O values of global meteoric precipitation by Bowen and Wilkinson (Reference Bowen and Wilkinson2002) (in SMOW scale) and published multitaxon land snail shells from North America from low-altitude (<400 m) settings along latitude (in PDB scale). Note that both proxies show a remarkably similar distribution of δ18O values as a function of latitude, reinforcing that low-altitude land snail shell δ18O values mimic variations in meteoric precipitation δ18O values. PDB, Pee Dee belemnite; SMOW, standard mean ocean water.
The striking match between global meteoric precipitation and snail shell δ18O values along latitude further reinforces that multitaxa land snail shells can be used as paleoprecipitation δ18O proxies at large spatial scale, at least for low altitude (<400 m asl) settings. Additional research along elevation gradients, desert and polar regions (Fig. 5), and areas with strong water vapor masses transport need additional research to further examine factors and mechanisms controlling snail shell δ18O values (see also Bowen and Wilkinson, Reference Bowen and Wilkinson2002).
Interpreting glacial land snail shell δ18O values
Oxygen isotope values of land snails have been used to infer climate conditions during the last glacial maximum (Kehrwald et al., Reference Kehrwald, McCoy, Thibeault, Burns and Oches2010; Yanes et al., Reference Yanes, Yapp, Ibáñez, Alonso, De-la-Nuez, Quesada, Castillo and Delgado2011b, Reference Yanes, García-Alix, Asta, Ibáñez, Alonso and Delgado2013; Nash et al., Reference Nash, Conroy, Grimley, Guenthner and Curry2018) and Younger Dryas (Yanes et al., Reference Yanes, Gutiérrez-Zugasti and Delgado2012). Many of these studies have documented significantly higher shell δ18O values than expected from present-day regression equations between mean annual meteoric precipitation δ18O and shell δ18O values. The effects of considerably low humidity during glacial times (as shown in Fig. 5) on shell δ18O values and/or the impact of higher glacial seawater δ18O values on meteoric precipitation δ18O values have been proposed to explain these relationships (e.g., Yanes et al., Reference Yanes, Yapp, Ibáñez, Alonso, De-la-Nuez, Quesada, Castillo and Delgado2011b). Accordingly, additional research on modern settings characterized by extremely cold and/or arid environments (e.g., high-elevation sites, deserts, high-latitude areas, etc.) may help to elucidate environmental controls on glacial or stadial snail shell δ18O values in future paleoclimatic studies in North America.
FUTURE LAND SNAIL RESEARCH DIRECTIONS IN NORTH AMERICA
The review presented here has allowed us to identify some areas of investigation in land snail oxygen isotope research throughout North America and to propose some future directions in land snail studies.
Spatial sampling deficiencies: Overall, there have been a reasonably large number of snail oxygen isotope studies conducted in eastern North America, covering middle latitudes (~30–58°N) and relatively humid environments (average RH >69%) (see Fig. 2A). In contrast, fewer studies have been conducted in the drier environments of the central and western United States (Fig. 2A). Tropical (<30°N) and high (>65°N) latitudes are still minimally surveyed, and future studies are necessary to assess the main climatic controls on snail shell δ18O values for individuals living under more marginal or extreme environmental conditions (see hypotheses in Fig. 5). Sampling at high latitudes is especially important if fossil shells are to be used to infer past glacial or stadial environmental conditions, as these fauna are mostly restricted to high latitudes today.
Altitudinal gradient studies: The majority of published snail oxygen isotope data from North America have focused on areas of low elevation, generally below 400 m asl (see Table 2). Altitudinal gradients remain to be further investigated and can help to test the effects of decreasing air temperatures with increasing elevation on snail shell δ18O values. This could be particularly useful to better interpret the oxygen isotope values of Pleistocene snail shells recovered from sites across North America, as the highest elevation sites may be able to act as analogs to glacial conditions in middle latitudes.
Time-series analysis: The bulk of the published work has focused on entire shell isotope analysis regardless of snail shell size or longevity, whereas time-series analysis along shell growth direction has been explored in just a few published studies so far (Supplementary Table 1). Entire shell analyses are probably the best approach for small snail species (<10 mm in shell length), as they are short lived and their shells are too small and thin to efficiently conduct high-resolution intrashell analyses. However, medium to large snail shells (>10 mm) from species that have lived longer lives (1 or more years), could also be further tested and investigated to measure seasonal variations along snail life span. Furthermore, intrashell isotope values of snail shells have the potential to offer relevant information about the season of land snail harvest when studying archaeological land snail shells (Yanes and Fernandez-Lopez-de-Pablo, Reference Yanes and Fernandez-Lopez-de-Pablo2017), much like marine shell midden research.
Laboratory-controlled experiments: growing snails under controlled conditions to monitor shell δ18O values remain to be undertaken. Although four published studies have grown snails in the laboratory, these studies focused on carbon isotope values and snail diet. Future controlled experiments should attempt to monitor and quantify the effects of variations of water δ18O values, RH, and ambient temperature on land snail shell δ18O values.
Clumped isotopes: Thus far, three studies have measured clumped isotopes in land snail shells (Supplementary Table 1). These studies, although informative, suggest that additional research should be pursued because the combination of clumped isotopes and oxygen isotopes in land snail shells could provide more informed details about the environmental conditions at which snails grew their shells. Published clumped-isotope studies suggest that some snails appear to grow at warmer temperatures than environmental air temperatures, which also challenges the idea of land snails being fully ectothermic. This kind of research will be useful not only for paleoenvironmental investigations, but also to better understand land snail ecology, physiology, and behavior.
Multitaxon data sets: We have found that published studies sometimes focus on a single taxon, whereas others combine data from multiple species, especially in broader spatiotemporal-scale studies. A recent study by our research group (Yanes et al., Reference Yanes, Nekola, Rech and Pigati2017) showed that different taxa of small land snails living at the same sites in northwestern Minnesota (USA) exhibited significantly different shell δ18O values. This likely reflects some combination of differing snail species ecologies, mobility habits, and/or microhabitat preferences. These results suggest that similar studies of modern snails should be undertaken for different regions and species of the world to assess the utility and potential limitations of multitaxa data sets. The data presented here, however, suggest that large spatial-scale studies may combine multiple species for paleoenvironmental reconstructions, but only if environmental variations are large enough to overwhelm species-specific variations that are significant at local or regional scales.
Quaternary continental paleoclimate studies: In mainland North America, isotopic studies involving ancient land snail shells have been conducted at archaeological sites (Yapp, Reference Yapp1979; Goodfriend and Ellis, Reference Goodfriend and Ellis2000; Balakrishnan et al., Reference Balakrishnan, Yapp, Meltzer and Theler2005a; Paul and Mauldin, Reference Paul and Mauldin2013), whereas the Quaternary fossil record of land snails has not been generally used for paleoclimatic reconstructions (but see Nash et al., Reference Nash, Conroy, Grimley, Guenthner and Curry2018). This is surprising considering the vast abundance and accessibility of Quaternary shells preserved in loess, wetland, alluvial, colluvial, and fluvial deposits throughout North America (e.g., Pigati et al., Reference Pigati, Quade, Shahanan and Haynes2004, Reference Pigati, Rech and Nekola2010, Reference Pigati, McGeehin, Muhs and Bettis2013; Rech et al., Reference Rech, Nekola and Pigati2012; Nash et al., Reference Nash, Conroy, Grimley, Guenthner and Curry2018). Our calibration study presented here suggests that land snails from North America primarily track the oxygen isotope signature of the average annual precipitation in a predictive near-linear fashion, although we acknowledge that there is a significant scatter from snail data sets because of the influence of other parameters such as temperature, RH, water vapor, and variations in snail mobility habits and behavior. Nevertheless, fossil shells across North America may be an invaluable continental proxy to make semiquantitative inferences of precipitation δ18O values during the Quaternary.
CONCLUSION
Land snail shells collected from a large range of middle latitude (30–58°N) sites across eastern North America exhibit δ18O values that show a strong positive correlation with both meteoric precipitation δ18O values and air temperature. Shell δ18O increases at a lapse rate of 0.5–0.7‰ for every 1‰ increase in precipitation δ18O, and 0.3–0.7‰ per 1°C increase in temperature (for both average annual and active period values). These relations persist when solely minute species (<5 mm) are included, as well as when all snail taxa are considered, regardless of their body size, ecology, and ethology. Other environmental factors, such as RH or precipitation amount, seem to play only minor roles at coarse spatial scales. Accordingly, middle-latitude shell δ18O values of small to minute species should be reliable semiquantitative proxies for precipitation δ18O in mainland North America. Future investigations of land snail shells should include additional modern calibration studies in low and high latitudes, desert and cold environments, and along altitudinal gradients to gain a better understanding of the isotope systematics for snails living in extreme environments, which could reasonably resemble glacial or stadial scenarios and can help us to better interpret glacial shell δ18O values extracted from the Quaternary fossil record of North America.
ACKNOWLEDGMENTS
This research was funded by the National Science Foundation grant EAR–1529133, a student research grant from the Geological Society of America to Nasser M. Al-Qattan, and the U.S. Geological Survey’s Climate and Land Use Change Research and Development Program. Special thanks go to Josh Miller (University of Cincinnati) for discussions on data analyses. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. government. The detailed and constructive revisions by Kathleen R. Johnson, Crayton J. Yapp, Adam Hudson, and an anonymous reviewer have greatly improved the clarity and quality of this manuscript.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/qua.2018.79
