Factors controlling shell carbon isotopic composition of land snail Acusta despecta sieboldiana estimated from laboratory culturing experiment

The carbon isotopic composition ( δ13C) of land snail shell carbonate derives from three potential sources: diet, atmospheric CO 2, and ingested carbonate (limestone). However, their relative contributions remain unclear. Under various environmental conditions, we cultured one land snail subspecies,Acusta despecta sieboldiana , collected from Yokohama, Japan, and confirmed that all of these sources affect shell carbonate δ13C values. Herein, we consider the influences of metabolic rates and temperature on the carbon isotopic composition of the shell carbonate. Based on results obtained from previous works and this study, a simple but credible framework is presented to illustrate how each source and environmental parameter affects shell carbonate δ13C values. According to this framework and some reasonable assumptions, we estimated the contributions of different carbon sources for each snail individual: for cabbage-fed (C 3 plant) groups, the contributions of diet, atmospheric CO 2, and ingested limestone vary in the ranges of 66–80, 16–24, and 0– 13 %, respectively. For corn-fed (C 4 plant) groups, because of the possible food stress (less ability to consume C 4 plants), the values vary in the ranges of 56–64, 18–20, and 16–26 %, respectively. Moreover, according to the literature and our observations, the subspecies we cultured in this study show preferences towards different plant species for food. Therefore, we suggest that the potential food preference should be considered adequately for some species in paleoenvironment studies. Finally, we inferred that only the isotopic exchange of the calcite-HCO−3 -aragonite equilibrium during egg laying and hatching of our cultured snails controls carbon isotope fractionation.

Land snail shell carbon has three sources: food, atmospheric CO 2 , and ingested carbonate (limestone) (Goodfriend and Hood, 1983).Food sources such as plants, fungi, and other organic matter can be transformed into metabolic CO 2 in the body of a land snail by two pathways: direct digestion and breakdown of urea (Stott, 2002).The resulting CO 2 will dissolve into the bicarbonate pool in the hemolymph and then precipitate as shell carbonate.Atmospheric CO 2 can be introduced into the bicarbonate pool via Published by Copernicus Publications on behalf of the European Geosciences Union.respiration.Similarly, ingested carbonate can first dissolve in stomach acid to produce gaseous CO 2 .Then it can be introduced into the bicarbonate pool too.Some previous works have been undertaken to clarify their relative contributions both from observations and model calculations.Nevertheless, this topic remains poorly understood (Goodfriend and Hood, 1983;Goodfriend and Stipp, 1983;Stott, 2002;Metref et al., 2003;Balakrishnan and Yapp, 2004;Romaniello et al., 2008).The following are estimations of each carbon source based on published works: food source ratios vary in the range of 25-40 % (Goodfriend and Hood, 1983), 36-73 % (Romaniello et al., 2008), and 100 % (Stott, 2002;Metref et al., 2003); atmospheric CO 2 has been estimated as negligible (Stott, 2002;Metref, 2003), 16-48 % (Romaniello et al., 2008), and 30-60% (Goodfriend and Hood, 1983); ingested carbonate has been estimated as up to 30% (Goodfriend and Stipp, 1983), often negligible for small terrestrial gastropods of less than 10 mm, and as always much less than 20-30 % for larger species (Pigati et al., 2004, 2010), 3-23 % (Romaniello et al., 2008), and ∼ 20 up to ∼ 40 % (Yanes et al., 2008a).
For solving problems of this kind, additional studies including laboratory culturing experiments must be done.Stott (2002) and Meterf et al. (2003) reported two independent works related to land snail culture experiments.Both show a marked but discrepant correlation between land snail shell carbonate δ 13 C and diet δ 13 C, with slopes less than one.Additionally, Stott pointed out that ingested carbonate does not contribute to shell carbonate δ 13 C values based on results of snails fed with and without an added CaCO 3 source.Finally, according to their calculations and discussions, both papers reported that atmospheric CO 2 does not contribute to shell carbonate δ 13 C values.Those reports notwithstanding, inconsistent observations and discussions emerged soon thereafter.Yanes et al. (2012) reported that higher δ 13 C values were observed during the younger growth stages of both living and fossil snails, which showed signs of a higher contribution of ingested limestone.Results of this research suggest strongly that environmental carbonate is incorporated as an important source for precipitating land snail shells (at least some snail species) and suggest that it can affect their shell δ 13 C values.Regarding the atmospheric CO 2 , after summarizing the previous studies of the relationship between land snail shell carbonate δ 13 C and shell organics/body tissue organics/diet (Stott, 2002;Goodfriend and Ellis, 2002;Balakrishnan and Yapp, 2004), McConnaughey and Gillikin (2008) pointed out that the offsets between δ 13 C of shell carbonate and their diet (or body organics) are greater when the δ 13 C of the latter diverge more from atmospheric CO 2 , suggesting that the atmospheric CO 2 does contribute to shell carbonate.
These great discrepancies from reports of the literature are probably attributable to (1) the variation of land snail species studied with different ecological requirements, ethology, and other species-dependent behavior; (2) the variation of environmental conditions where these snails were living (e.g., CaCO 3 -rich areas vs. CaCO 3 -poor areas, wet areas vs. dry areas, hot/warm areas vs. cold/cool areas); (3) the limitations of calculations.Therefore, a better understanding of the contribution of each carbon source and related environmental controlling factors can promote this isotopic tool in the field of paleoenvironment reconstruction.For this study, we cultured one land snail subspecies (Acusta despecta sieboldiana) under different controlled conditions.Based on previous works and results of this study, a simple but credible framework was raised to discuss the mechanism of how each possible source and environmental parameter can affect shell carbonate δ 13 C values.According to this framework and some reasonable assumptions, we estimated the contributions of different carbon sources for each snail individual.This report is the first to describe an attempt at estimating the contributing proportion of limestone ingested by snail individuals using stable isotope values compared with a previous method using calculations including radiocarbon dating (Goodfriend and Stipp, 1983;Romaniello et al., 2008;Pigati et al., 2010).
In this study, eight adult snails of Acusta despecta sieboldiana (a subspecies of Acusta despecta) were collected in Suzukakedai, Yokohama, Japan and were cultured at room temperature (ca. 25 • C) beginning in January 2012.These snails began to lay eggs in March 2012.The eggs were transferred into a stainless steel container and were covered with moist cloth in a 25 • C incubator.Most eggs hatched at around 3-4 weeks.
Larvae were distributed randomly into around 30 small transparent plastic boxes that had been perforated to allow air and vapor exchange: each box contained 3-5 snails.Some later-hatched larvae were added in May and June 2012.Then four small boxes were put into semi-sealed big plastic boxes: two parallel groups were fed cabbage (Brassica oleracea var.capitata, green cabbage, C 3 plant, δ 13 C = −28.4± 1.2 ‰, n =12) that had been sprinkled with fine calcium carbonate powder (CaCO 3 , δ 13 C = 4.0 ‰).Another two were fed cabbage sprinkled with calcium phosphate (Ca 3 (PO 4 ) 2 ).A NaCl saturated solution was used to produce high humidity conditions in each large box.The snails grew under natural light/dark cycles.The air and food were changed every 2 days.Then the big plastic boxes were put into three incubators with respective temperatures of 20, 25, and 30 • C, until January 2013; some snails from 20 • C group were cultured until May 2013.At the end of culturing period, we recorded the length and height for each snail (see Table S2 in the Supplement).Additionally, we checked the growth stage of one individual cultured at 25 • C (S43, Table S2).Its length was 10.1 mm.The results suggest that this snail had already become an adult.Consequently, the lengths for most other snails were larger than 10.0 mm, suggesting that they had probably reached adult stage.The growth phases of snails are presented in Fig. 1.
We also cultured some snails using corn (C 4 plant, δ 13 C = −12.0± 0.7 ‰, n = 4; first we used Miscanthus sinensis, but the snails did not eat it; then we changed the food to corn) sprinkled with fine calcium carbonate powder in a 20 • C incubator.
Ultimately, the snails were collected into labeled sampling bottles and were preserved at −40 • C for additional treatment.

Isotopic samples preparation and analysis
Frozen snails were dried using a cryogenic vacuum line.They were then washed with distilled water and were kept in the water for 10 min.The soft body tissues of respective snails were separated from shells using a nipper causing no damage.Then they were immersed into 3N HCl for 2-4 h, rinsed with distilled water and then lyophilized.All dry tissues were ground to powder and were wrapped in tinfoil (Sn) capsules.Samples were introduced into a combustion tube from the auto-sampler one by one and were converted into gaseous CO 2 at 980 • C. Then the resultant CO 2 was injected into a cavity ring-down spectroscope (CRDS, L1121-I; Picarro, Inc., Santa Clara, CA, USA) for δ 13 C measurement.More details related to the CRDS method and system were presented by Wahl et al. (2006) and Balslev-Clausen et al. (2013).All measured δ 13 C values were normalized against two simultaneously measured standards, acetanilide (δ 13 C = −33.62‰; Costech Analytical Technologies, Inc., Valencia, CA, USA) and sucrose (δ 13 C = −13.55‰; Kanto Chemical Co., Inc., Tokyo, Japan) using a two-point calibration method (Coplen et al., 2006).The analytical precision was better than 0.4 ‰ (n = 3).
Shells were washed successively with distilled water, then with acetone under an ultrasonic bath for 60 min to eliminate the organic residues.After washing again using distilled water, they were put into labeled sampling bottles containing diluted hydrogen peroxide (10 %) to remove the remaining organic matter.These bottles were kept overnight in a shaking machine with a speed around 100 rpm.Finally, all the shells were dried using lyophilization and were crushed into a homogeneous powder using an agate mortar for isotopic analyses.The first two internal spirals were removed, which were inherited from their parents and which accounted for less than 5 % of the total shell mass.Regarding the collected adults and cultured snails, their shell powder (7-10 mg) was reacted with 103 % phosphoric acid for more than 2 h in a 25 • C water bath.Then the purified CO 2 was analyzed using isotope ratio mass spectrometry (MAT 253; Thermo Fisher Scientific Inc., Waltham, MA, USA) and calibrated against an synthetic calcium carbonate standard (δ 13 C = −9.13‰; Wako Pure Chemical Industries Ltd., Osaka, Japan).The precision was better than 0.1 ‰ (n = 3).
Regarding the larvae and eggshell, the powder (400 ± 50 µg) was reacted with 103 % phosphoric acid for more than 1 h at 72 • C. The resultant CO 2 was analyzed using an isotope ratio mass spectrometer (Delta XL; Thermo Fisher Scientific Inc., Waltham, MA, USA) coupled with a PAL auto-sampler (GCPAL; CTC Analytics AG, Zwingen, Switzerland), and a GasBench II preparation device (Thermo Fisher Scientific Inc., Waltham, MA, USA).were normalized against two simultaneously measured synthetic calcium carbonate standards, Wako (δ 13 C = −9.13‰; Wako Pure Chemical Industries Ltd., Osaka, Japan) and Kanto (δ 13 C = −20.62‰; Kanto Chemical Co., Inc., Tokyo, Japan) using two-point calibration.The analytical precision was around 0.1 ‰ (n = 3).All of our in-house working standards described here are calibrated with respect to IAEA standards such as NBS-19 and the δ 13 C values are reported relative to Vienna Peedee Belemnite (VPDB).
In addition, the crystal structure of shell carbonate powder was checked using x ray diffractometry (XRD, MXP3TA; Mac Science Ltd.).The results revealed that our snail shell carbonate is aragonite, although the eggshell carbonate is calcite.

Summary of culturing experiment
Figure 2a shows that snails that grew at low temperature (20 • C) have a higher survival rate (72 %) than those at 30 • C (16 %), which is consistent to the observations by Kohno (1976).The reason is probably that, in hightemperature groups, the humidity is slightly higher so that the leaves are more easily putrefied.Because the cultured environment is a sealed system with an air exchange every 2 days, the perishable leaves consumed oxygen and perhaps produced noxious gases, which might have caused all the snails in one small box to have died.Therefore, these snails show higher mortality.Egonmwan (2008) reported a positive correlation between calcium provision, snail body weight, and shell length.High mortality was observed for snails that were deprived of a calcium source.Similarly, in our study, snails fed Ca 3 (PO 4 ) 2 exhibited lower weight and shorter length.They were more translucent, more fragile, and had thinner shells than snails fed CaCO 3 during our cultivation.Moreover, Fig. 2b shows a lower survival rate (30 %) of snails fed Ca 3 (PO 4 ) 2 than those fed CaCO 3 (50 %), which is consistent with results of the previous study, suggesting that calcium from the powder of Ca 3 (PO 4 ) 2 was ingested only to a slight degree by land snails.

Carbon isotope results
Body tissues of cultured snails yield a similar δ 13 C value with the same food source (Table 1), which are −27.9 ± 0.6 ‰ (fed cabbage, n = 38) and −10.8 ± 0.7 ‰ (fed corn, n = 3).No marked differences were found among temperatures or between those fed with and without CaCO 3 .

Land snail shell carbonate precipitation and carbon source estimation
The solid line of Fig. 3 shows a model of diet-controlled snail shell carbonate δ 13 C (Francey, 1983;Stott, 2002), whereas the dashed line shows a flux balance model based on CO 2 diffusion when the input flux of CO 2 is equal to respired flux of CO 2 in the body fluid of land snails (Balakrishnan and Yapp, 2004).However, additional enrichments were observed in the published results and in this study (e.g., snails cultured at 20 and 25 • C with carbonate), suggesting a contribution ) 2 , which is probably hard and rarely ingested by snails; b snails regarded as a control group, cultured under an air-free exchange system with a small amount of soil; c mixture of four individuals cultured in the same condition (sizes are similar); d average value of snails of S34-S41; * x, calculated proportion of metabolic CO 2 ; y, proportion of atmospheric CO 2 ; z, proportion of CO 2 produced from ingested carbonate from ingested carbonate.In addition, because a common model seems unsuitable for all snails because of the different metabolic rates among snail individuals growing up at different conditions (Barnhart and McMahon, 1987), the estimation of shell carbon sources for each snail is expected to be useful and necessary.
The main sources of shell carbon are diet, atmospheric CO 2 , and ingested carbonate (Goodfriend and Hood, 1983).A simple framework of the possible mechanisms to precipitate shell carbonate is presented in Fig. 4 based on re-sults of a previous study as well as results from the present study.Almost all of these three carbon sources are expected to transform into gaseous CO 2 and then dissolve into the hemolymph of a snail (so-called bicarbonate pool) for reaching isotopic equilibrium with bicarbonate (HCO − 3 ).The fractionation between gaseous CO 2 and HCO − 3 under equilibrium is controlled by temperature, shown as 13 C HCO 3 −g = − (0.114 ± 0.003) T ( • C) + (10.78 ± 0.05) ‰ (Mook et al., 1974;Zhang et al., 1995;Szaran, 1997) Stott, 2002;Metref et al., 2003;Balakrishnan and Yapp, 2004).
carbon isotope fractionations under our culturing temperatures might be 8.5, 7.9, and 7.4 ‰, respectively, at 20, 25, and 30 • C. Finally, the shell aragonite will precipitate from bicarbonate pool in isotopic equilibrium (also designated as carbon isotope steady state, Balakrishnan and Yapp, 2004) and the carbon isotope ratio might be enriched to around 2.7 ‰ (Rubinson and Clayton, 1969;Romanek et al., 1992).Therefore, the total fractionation between each carbon source (diet, atmospheric CO 2 , and ingested carbonate) and shell aragonite is expected to be an enrichment of 11.2, 10.6, and 10.1 ‰, respectively, at 20, 25, and 30 • C. In addition, trace amounts of dissolved inorganic carbonate (DIC) dissolved in leaf water and drinking water might also introduce isotopic fractionations of shell carbonate.For instance, Pigati et al. (2004), based on radiocarbon dating technique, reported that aqueous carbon sources account for approximately 10 % of the shell carbon for one species of Catinella, which is a semi-aquatic gastropod.However, no report to date describes a study of this source related to terrestrial gastropods.
For simplification of the model and calculations, we made two assumptions.First, no carbon isotopic fractionation takes place when the ingested carbonate dissolves into the water and reacts with stomach acid.Second, because of the variable sources and probably low contribution of DIC from directly ingested water, which are similar to all earlier published discussions, we will not include such sources in our calculation.These assumptions can be evaluated for their propriety in future research.Results of those studies can be expected to improve the accuracy of this estimation method.
According to this framework, we will explain briefly how to calculate the contribution of each source to shell carbon based on our observed results.First, from the mass balance model, we obtain the following:   2010) is approximately 0.02 ‰ yr −1 .Therefore, we use −9.5 ‰ as the δ 13 C a value); δ 13 C c is the isotopic composition of ingested carbonate (4.0 ‰); x is the proportion of metabolic CO 2 ; y is the proportion of atmospheric CO 2 ; z is the proportion of CO 2 from ingested carbonate; b is the isotope fractionation value from gaseous CO 2 to aragonite at different temperatures (11.2, 10.6 and 10.1 ‰, at 20, 25 and 30 • C, respectively).
Here we chose 20 • C group snails to demonstrate how we calculated the contribution of each source.When there is no CaCO 3 added, z = 0.
From inputting the measured data, we obtain the x and y values shown in Table 1.For simplification of the calculation, for the snails fed CaCO 3 , we assumed a similar x/y ratio to those of all the snails, although individual differences might happen among snails attributable to different growth rates.Therefore we calculated the average x/y ratio in noncarbonate groups and obtained x/y = 3.2 (6) By combining Eqs. ( 2), (6), and (7), we calculated x, y, z values separately.Similar calculations were done at 25 • C and 30 • C. Finally, for the C 4 plant groups, we assumed the same x/y ratio for the C 3 plant groups, which might not be accurate because the food consumption preference differs and C 3 plant groups have a higher growth rate than C 4 plant groups.
Then we calculated x, y, and z using the same method.All calculated results are presented in Table 1.
The calculated x, y, z values for snails fed C 3 plants reveal that the contributions of diet, atmospheric CO 2 , and ingested limestone varied, respectively, in the ranges 66-80, 16-24, and 0-13 %.Furthermore, for those fed C 4 plants, because of the potential food stress (less ability to consume C 4 plants), they vary in the ranges of 56-64, 18-20, and 16-26 %, respectively.We observed a higher shell weight proportion of snails that had been fed corn compared with those that had been fed cabbage, suggesting a higher ingested limestone contribution, which is coincident with our calculations.
Figure 5a presents a positive correlation between the calculated contribution of ingested carbonate and the shell weight proportion.Calcium carbonate can be transformed into Ca 2+ in the stomach of land snails and can then be ingested into the hemolymph, and can finally be precipitated into the shell carbonate.Therefore, high calcium carbonate consumption is always correlated with a high shell weight proportion.The δ 13 C of fed carbonate is more positive than δ 13 C of food and atmospheric CO 2 .For that reason, if snails consume more carbonate, then their shell δ 13 C values are expected to be more positive, which is consistent with the relationship presented in Fig. 5b.
As described earlier, the x/y ratio can vary among snail individuals.Therefore, the robustness of our estimation method was tested using the maximum and minimum x/y values observed from snails fed without CaCO 3 at different temperatures.The results are presented in Table 2.For snails fed with cabbage and CaCO 3 at 20 • C, the x/y values vary in the range of 2.9-3.3.Maximum discrepancies of estimations by the maximum and minimum x/y value could be 1.3, −2.9, and 1.6 % for x, y, and z values, respectively.For those fed with cabbage and CaCO 3 at 25 • C, x/y values vary in the range of 3.4-5.2.Maximum discrepancies could be 3.0, −6.9, and 4.1 % for x, y, and z values, respectively.Considering individual differences among snails (calculated standard deviations of x, y, and z for the snails fed with cabbage and CaCO 3 are 3.9, 1.6, and 3.9 %, respectively), these discrepancies are acceptable, especially for the estimated contribution of diet and ingested carbonate.Moreover, almost all of these re-estimated x, y, and z values are within in the range we have calculated based on an average x/y value, showing a satisfactory robustness of this estimation method.
We have become aware of another possibility for the carbon isotope fractionation related to metabolic CO 2 , which probably has already dissolved into the snail body water when produced with no isotope fractionation generated from gaseous state to aquatic state.In this case, the total carbon isotope fractionation values from metabolic CO 2 to shell aragonite would be 12.4, 11.8, and 11.2 ‰, respectively, at 20, 25 and 30 • C. We calculated the contributions of each carbon source using this assumption and present the results in Table S1 in the Supplement.In such circumstances, the contributions of diet, atmospheric CO 2 , and ingested limestone for snails fed C 3 plants varied, respectively, in the ranges 70-85, 13-19, and 0-13 %, whereas those fed C 4 plants varied, respectively, in the ranges 64-73, 15-17, and 11-22 %.This estimation shows a lower contribution of atmospheric CO 2 and higher contribution of diet than the previous one.Although we cannot discount either of these possibilities with present knowledge, both imply similar discussions leading to similar conclusions.Consequently, in the following discussion, we will consider only the data presented in  x min , y min , and z min are calculated using the minimum x/y values; x max , y max , and z max are calculated using the maximum x/y values; x, y, and z show the discrepancies of estimations by maximum and minimum x/y values for each snail individual.

Contribution of each carbon source to shell δ 13 C
4.2.1 Ingested carbonate (limestone in nature) Stott (2002) reported no apparent differences in shell δ 13 C of land snail Helix aspersa fed with and without CaCO 3 .However, the deviation of 14 C age estimation for some snail species in nature shows an incorporation of limestone (Goodfriend and Stipp, 1983;Pigati et al., 2010).Consequently, ingested limestone is expected to play an important role in controlling shell carbonate δ 13 C, at least for some species.Higher δ 13 C values were observed in our cultivated land snail Acusta despecta sieboldiana fed CaCO 3 compared with those fed Ca 3 (PO 4 ) 2 , especially at 20 and 25 • C (Fig. 6a).These values are even higher than the values predicted from a flux balance model (Balakrishnan and Yapp, 2004), which considers both metabolic CO 2 and atmospheric CO 2 (Fig. 3), strongly suggesting the involvement of ingested CaCO 3 powder (Fig. 6b).The estimated contribution for C 3 plant group is 0-13 %; and for C 4 plant, it varies: 16-26 %.The different estimated proportions reveal different growth rates (metabolic rates) among snails.Apparently, snails fed C 4 plants might prefer ingesting higher proportions of limestone, which is probably caused by diet stress.Figure 7a (see Table 3) portrays a decreasing trend of shell δ 13 C along with the snail growth direction, which is consistent with the measurements of living land snail individuals reported by Yanes et al. (2012), who observed higher δ 13 C values during the younger growth stages of land snails.This fact is explainable by the larger proportion of limestone ingested during the early period of land snails to enhance their growth rates.Egonmwan (2008) reported that the amount of food ingestion increases gradually, although calcium ingestion first increases (1-3 months, but the increase rate is lower than diet consumption) and then decreases (4-6 months) during the first 6 months, suggesting that the contribution of ingested carbonate can be expected to decrease along with the snail growth.Our estimated contributions of ingested carbonate (Fig. 7b) support this presumption.To elucidate this phenomenon, similar studies should be conducted to observe snails fed with C 4 plants and those fed without CaCO 3 .

Diet and food selectivity
DeNiro and Epstein (1978) reported a slight enrichment (ca. 1 ‰) of the snail body tissue δ 13 C relative to their diet: romaine lettuce.Stott (2002) shown no significant isotopic offset between snail body tissue δ 13 C fed lettuce and their diet, whereas about 2-3 ‰ depletion was observed for snails fed corn (Fig. 8a), which is regarded as attributable to the contribution of pre-culture carbon because they cultured snails from juveniles but not from eggs.Our observations show slight enrichment among all snails relative to their diet: approximately 0.5 ‰ for cabbage and approximately 1.2 ‰ for corn.However, this enrichment is negligible when considering measurement precision among both snail individuals and vegetable samples.We suspect that the small discrepancies of the δ 13 C values between snail body tissue and their diet observed in the literature and this study may depend on the analytical precision and/or limitation of samples (e.g., sample size, growth condition).Therefore, we infer that the δ 13 C values of snail body tissue are similar and that they directly reflect the δ 13 C values of their food.
Figure 8b shows the observed relations between shell carbonate δ 13 C of snails and their diet (This study; Stott, 2002;Metref et al., 2003).Evidently, all of these slopes are less than the expected slope of 1, which is considering diet as a single source controlling shell carbon (Francey, 1993;Stott, 2002).Offsets suggest the influence of other sources such as limestone and atmospheric CO 2 cause these more positive values.The relative contributions of C 3 -and C 4 -type food might be different.Our estimation confirms this inference: the proportions of shell carbon from cabbage are 66-80 % and from corn are 56-64 %, which might reflect food stress among different plants.Metref et al. (2003) reported a C 3 vs.C 4 mixed experiment.Most snails show a preferential use of C 3 food.This study also found a preference of this kind in consuming C 3 but not C 4 plants, e.g., snails grew faster when eating C 3 plants such as lettuce and cabbage than snails fed a C 4 plant such as corn, which agrees with the observations reported by Metref et al. (2003).Moreover, almost all the snails fed Miscanthus sinensis (C 4 plant) died after 20-30 days, except one or two large individuals, and their shell δ 13 C values (−11.1 ± 3.4 ‰) show no marked differences from newly hatched snail larvae (−11.0±0.1 ‰), suggesting that the snails were unable to consume Miscanthus sinensis at all.However, Stott (2002) reported a lower growth rate of snails cultured with dried sour orange tree leaves (C 3 plant) compared to those cultured by lettuce and corn, suggesting food quality such as water content or physical structure as the reason for food preference.Regarding the land snail Acusta Relative culturing conditions are shown in Table 1.Meanings of symbols x, y, and z are the same as those shown in Table 1.
despecta, food preference has been reported by Suzuki and Yamashita (1967) and Takeuchi and Tamura (1995).For instance, this species were unable to eat Oxalis corniculata (C 3 plant), Commelina communis (C 3 plant), or Yoshinagella japonica (Fungi) at all.Similar phenomena have also been observed in nature.Hatziioannou et al. (1994) and Iglesias and Castillejo (1999) observed that land snails do not eat plant species arbitrarily.Baldini et al. (2007) reported that the feces of land snail Cerion collected from Sporobolus domingensis (C 4 ) plant exhibited δ 13 C values more typical of a C 3 plant, suggesting a preference of C 3 plant as food of these snails.However, some other studies, such as those of Yanes et al. (2013b), indicate an opposite conclusion.Their results show that individuals from the same land snail species consume different plants in relation to their relative abundance in nature.To summarize, we infer that the food selectivity of land snails might depend on species, which would increase the difficulty of their application in the paleoenvironment reconstruction, especially for the accurate study of C 3 /C 4 vegetation distribution.
In addition, although some previous works reported that the fresh plants are the main diet for some land snail species (e.g., Suzuki and Yamashita, 1967;Colonese et al., 2014), it may not be common for all the species.According to the literature, land snails can eat decayed plant matter (Richardson, 1975), fungi, animal tissue (Mason, 1970), lichens (Baur et al., 1994) and other organic matter in nature.The consumed proportions of these food sources vary among land snail species (Mason, 1970).
Consequently, we suggest that pre-investigation of food preference on living snails is important before applying shell carbon isotopic values in the paleoenvironment reconstruction of a certain species.In addition, information from future studies of food quality's influence on snail shell δ 13 C values might be helpful too.

Atmospheric CO 2
Although Stott (2002) and Metref et al. (2003) reported that atmospheric CO 2 does not contribute to shell carbonate δ 13 C values based on culturing land snail Helix aspersa, some works clarify it as an important source of shell carbon.For instance, according to the combination of δ 13 C and 14 C values observed from modern land snail shells, Romaniello (2008) calculated the contribution of atmospheric CO 2 varying from 16 to 48 %.McConnaughey and Gillikin (2008) summarized the previous studies of the relationship between land snail shell carbonate δ 13 C and shell organics (or body tissue organics / diet) and pointed out that the offsets are greater when the δ 13 C of organics diverge more from atmospheric CO 2 , i.e., when the slope of the shell vs. organic regression line is less than unity, suggesting that the latter does contribute to shell carbonate.
In this study, our calculation results revealed that atmospheric CO 2 can affect shell carbonate isotopic values, and our estimated contributions are, respectively, 16-24 % (fed cabbage) and 18-20 % (fed corn).More clear evidence might be that, for snails developing without CaCO 3 (Fig. 3), shell δ 13 C values became higher than the expected values controlled by one end member (diet).These estimated values can be attributed to two pathways: (1) atmospheric CO 2 directly being introduced into the bicarbonate pool via respiration; (2) using the imbibed water as dissolved atmospheric CO 2 .The carbon isotope fractionation exhibits no difference between these two pathways.
However, because the CO 2 concentration and its in situ carbon isotopic value were not monitored in the semi-sealed system, the CO 2 accumulated gradually from the respiration of snails and plant tissues might affect the accuracy of our estimations of atmospheric CO 2 .Although Balakrishnan and Yapp (2004) inferred that the accumulated CO 2 (ca.240 ppm, vs. ambient background) produced by the respiration under forest canopy would contribute an insignificant amount of variation of shell δ 13 C values (ca.0.1 ‰), we suggest that additional studies based on experimentation are needed.Furthermore, the limited accuracy of our assumption of similar diet/atmospheric CO 2 ratio at the same temperature must be acknowledged because the snails have individual differences of metabolic rates during their growth (Barnhart and McMahon, 1987).Therefore, to learn more accurate contributions of atmospheric CO 2 , further incubation experiments are necessary, which are expected to include several parallel groups with 14 C-labeled or δ 13 C-different CO 2 compositions, and also to record the concentration and isotopic composition variations of in situ atmospheric CO 2 during cultivation.

Metabolic rate
Snail individuals have different metabolic rates in different environment conditions, or even in different growth phases of a single individual.Different metabolic rates reflect different partial pressures of CO 2 produced from the diet, which consequently cause different flux ratios among CO 2 produced from metabolism, digestion of limestone, and atmosphere to approach a carbon isotope steady state in snail body fluids, and which eventually produce different δ 13 C values.Much of the discussion we have presented above shows influences of this kind, such as (1) snails cultured at 30 • C groups with CaCO 3 show more-depleted δ 13 C values than those at 20 and 25 • C groups (Fig. 6a).Correspondingly, they have a lower survival rate and lighter mass, suggesting stress from poor growth conditions (e.g., perishable leaves, lack of oxygen).( 2) For one snail individual, the trend of a decrease in both δ 13 C values of shell carbonate (Fig. 7a) and the estimated contribution of ingested carbonate (Fig. 7b) reflect a decreasing metabolic rate along with snail development.(3) Also, the different slopes presented in Fig. 8b reveal different metabolic rates among different snail species or different cultivation conditions.
Therefore, the variations in metabolic rates attributable to the shift of environmental conditions, which can produce discrepancies of shell carbonate δ 13 C values, should be considered in paleoenvironment studies.For example, Yanes et al. (2011) reported 3 ‰ higher moving average shell δ 13 C values during the glacial interval (ca.50-15 ka BP) than today and inferred a larger proportion of C 4 plant during that period.However, this study revealed that shell carbonate δ 13 C values of land snails, for those fed same diet (cabbage) and carbonate but maturing at different temperatures, can also vary as greatly as 3.5 ‰ (Fig. 6a).
Consequently, the carbon isotopic composition controlled by metabolic rate in land snail fossils can be regarded as an auxiliary tool for ascertaining changes of paleoenvironment conditions, such as suddenly changed environment conditions during the Younger Dryas event.

Temperature
Theoretically, temperature can affect the fractionation factor between gaseous CO 2 and HCO − 3 in land snail body fluid from a relationship presented as 13 C HCO 3 −g = − (0.114 ± 0.003) T ( • C) + (10.78 ± 0.05) ‰ (Mook et al., 1974;Zhang et al., 1995;Szaran, 1997).Therefore, for shells precipitated at two temperatures, the discrepancy between δ 13 C values ( 13 C, ‰) is −0.11 • C −1 ×T (Zhang et al., 1995; see also in Romanek et al., 1992, −0.13 • C −1 and Szaran, 1997, −0.10 • C −1 ).We observed a relationship of −0.17 ± 0.04 • C −1 among snails fed without CaCO 3 (Fig. 9), which is not significantly different from the theoretically expected relation.The small sharper trend might somehow reflect metabolic differences among the snails at different temperature conditions.The δ 13 C discrepancy of 0.17 ‰ • C −1 is small compared with the contributions of different carbon sources.For that reason, no special consideration of environmental parameters of this kind is necessary on most occasions.

Carbon isotopic fractionations during egg laying
and hatching Metref et al. (2003) found that hatched and one-day-old snails showed δ 13 C values that were depleted by 2.5 ‰ compared to those of their parents.They inferred that shells of the hatching juveniles were built not only from eggshell calcite but also from isotopic-depleted albumen.We have observed these δ 13 C results for Acusta despecta sieboldiana: adult snails (parents), −11.5±1.2 ‰ (n = 6); hatched snails: −11.0±0.1 ‰ (n > 10, mixed samples); eggs: −13.1±0.9 ‰ (two groups of eggs, for each group, n > 10).No marked δ 13 C depletion was found between hatched snails and adults.Romanek et al. (1992) pointed out that fractionation factors among calcite, HCO − 3 , and aragonite are temperatureindependent when they are in equilibrium conditions.Their values were reported as 13  Rubinson and Clayton, 1969; see also Romanek et al., 1992, 13 C Calcite−HCO − 3 = 1.0 ± 0.2 ‰; 13 C Aragonite−HCO − 3 = 2.7 ± 0.6 ‰; 13 C Aragonite−Calcite = 1.7 ± 0.4 ‰).According to the XRD results, all of our cultured snail shells are made of aragonite, whereas eggshells are made of calcite.We are striving to ascertain the isotopic fractionations based on calcite-HCO − 3 -aragonite equilibrium.At this isotopic equilibrium condition, (a) for the laying process: when the snail precipitates shell (aragonite) from bicarbonate pool, δ 13 C will enrich 2.7 ‰ and when it precipitates eggshell (calcite), δ 13 C will enrich 0.9-1.0‰.Therefore, the difference between shell aragonite and eggshell calcite is expected to be 1.7-1.8‰; we obtain a similar value of 1.6 ‰, suggesting that snail eggshells are also precipitated from the bicarbonate pool and that they follow the equilibrium fractionation.(b) For the hatching process, when the larvae are hatching, we assume that the shell calcite will dissolve gradually into the egg water to form a bicarbonate pool to precipitate egg aragonite.Then the fractionation between egg calcite and hatched snail shell aragonite is expected to be around −1.8 to −1.7 ‰ .Here we obtained a value of −2.1 ‰, which is also consistent with the assumed value if we consider the measurement error, suggesting that the shell aragonite is transferred from eggshell calcite at an isotopic equilibrium condition.

Conclusions
From culturing the land snail Acusta despecta sieboldiana under controlled environments, we confirmed that diet, atmospheric CO 2 , and ingested limestone are important sources controlling shell carbon isotopic composition.We have also discussed the influences of metabolic rates.Furthermore, temperature can affect shell carbonate δ 13 C values by controlling the carbon isotopic fractionation of gaseous CO 2bicarbonate equilibrium.We presented a simple but credible framework to assess the mechanisms of how each source and environmental parameter might affect shell carbonate δ 13 C values based on this and earlier studies.According to this framework and some reasonable assumptions, we estimated the contribution of different carbon sources for each snail individual: for cabbage-fed (C 3 plant) snails, the contributions of diet, atmospheric CO 2 , and ingested limestone vary in the ranges of 66-80, 16-24, and 0-13 %, respectively.Furthermore, for corn-fed (C 4 plant) snails, because of the potential food stress (less ability to consume C 4 plants), the contributions vary in the ranges of 56-64, 18-20, and 16-26 %, respectively.Secondly, results show that Acusta despecta sieboldiana discriminate in their choices of different plant species as food.For instance, they can grow faster when eating C 3 plants such as cabbage compared with C 4 plants such as corn, and they cannot eat certain C 4 plants such as Miscanthus sinensis (no growth).This kind of food selectivity of land snails increases the difficulty of their application in the paleoenvironment reconstruction, especially for the accurate study of C 3 /C 4 vegetation distribution.Finally, we inferred that during the egg laying and hatching of our cultured snails, carbon isotope fractionation is controlled only by the isotopic exchange of the calcite-HCO − 3 -aragonite equilibrium.To initiate the application of carbon isotopes for paleoenvironment reconstruction, additional work should be undertaken in future culture experiments, especially in the following aspects: (1) intra-shell measurements of snails fed with C 4 plants and those fed without CaCO 3 should be taken in future studies to verify the phenomenon of decreasing trend of shell δ 13 C along with the snail growth direction; (2) influences of food quality such as water contents or physical structure on snail shell δ 13 C values should be investigated; (3) snails should be cultured in air with 14 C-labeled or δ 13 Cdifferent CO 2 compositions to ascertain the contribution of atmospheric CO 2 more accurately; (4) snails fed with C 4 plants under a non-carbonate condition should be evaluated; (5) more land snail species preferring varied environmental conditions should be studied to ascertain which inferences are common to all species and which are particular for a given species.
The Supplement related to this article is available online at doi:10.5194/bg-11-5335-2014-supplement.

Figure 4 .
Figure 4. Simple framework of shell carbon sources and their relative isotope fractionation processes based on this and earlier studies.

δ
13 C S = xδ 13 C t + yδ 13 C a + zδ 13 C c + b (1)x + y + z = 1 (2)Therein, δ 13 C S is the isotopic composition of shell aragonite; δ 13 C t is the isotopic composition of snail tissue.Based on our observations and discussions (see Sect. 4.2.2),we use the measurements for each snail, not an average value, because of the individual differences among snails.δ 13 C a is the isotopic composition of atmospheric air (δ 13 C of atmospheric CO 2 was observed from June 2010 to July 2011 in Suzukakedai, Yokohama, Japan.The average value is −9.5 ‰ (n = 42; Zhang et al., unpublished data); the annual rate of decrease reported by Keeling et al. (

Figure 5 .
Figure 5. Relationship between (a) estimated contribution of ingested carbonate and shell weight percentage and (b) δ 13 C of snail shell aragonite and shell weight percentage.

Figure 6 .
Figure 6.Bulk δ 13 C of snail shell aragonite and estimated contribution of ingested carbonate at different culturing conditions.

Figure 7 .
Figure 7. δ 13 C of snail shell sections and their estimated contribution of ingested carbonate at different growth phases.

Figure 9 .
Figure 9. Relationship between δ 13 C of snail shell aragonite (fed without carbonate) and different culture temperatures.

Table 1 .
Stable isotope results of snails cultured under different conditions and the estimated contributions of their shell carbon sources.

Table 2 .
Estimated x, y, and z under observed maximum and minimum x/y values at 20 • C and 25 • C.

Table 3 .
Stable isotope results of snail individual fractions and the estimated contributions of their shell carbon sources.