Is there metabolic cold adaptation in terrestrial ectotherms? Exploring latitudinal compensation in the invasive snail Cornu aspersum

Environmental variation across geographical gradients is a key driver in the evolution of biological diversity, giving rise to genetic and phenotypic variation within species (Culumber et al., 2012), and affecting survival and reproduction of organisms (Gaston, 2009; Gunderson and Leal, 2012). Such variation has been well documented across latitudinal gradients, in which climatic conditions, season lengths, ambient temperatures and resource availability are paramount factors affecting the rate of most biological processes (Somero, 1995; Huey and Berrigan, 2001; Kingsolver and Huey, 2008; Hochachka and Somero, 2002; Lindgren and Laurila, 2009; Pörtner et al., 2005). For example, lower mean temperatures, extreme seasonality and shorter growing seasons at higher latitudes are expected to cause a decline in metabolic rates (Jose et al., 2009) and annual growth rates of ectotherms (Yamahira and Conover, 2002). Nevertheless, if a reduction in the rates of these biological processes involves a reduction in fitness, then organisms may evolve compensatory responses for the constraints imposed by high-latitude habitats (Yamahira and Conover, 2002; Levinton, 1983; Fangue et al., 2009).

In principle, two alternative models could explain the evolution of latitudinal compensation in ectotherms. One model is local thermal adaptation, which predicts that organisms from high latitudes should function better at low temperatures (i.e. shift of the thermal optimum) in relation to their counterparts from lower latitudes (Tattersall et al., 2012; Angilletta, 2009; Gaitán-Espitia et al., 2013a). An alternative model of latitudinal compensation focuses on latitudinal differences in the length of the growing season rather than the local mean temperature (Conover and Present, 1990). This model is based on the countergradient variation (CnGV), which occurs when genetic differences counteract environmental effects, reducing the phenotypic differentiation between populations along an environmental gradient (Conover and Schultz, 1995; Niewiarowski and Angilletta, 2008). The CnGV model predicts that organisms compensate for the climatic constraints at high latitudes by increasing performance, growth and fecundity relative to their lower latitude counterparts when they are reared and compared at the same temperature (Levins, 1969; Conover and Schultz, 1995; Schultz et al., 1996; Fangue et al., 2009; Yamahira and Conover, 2002). This compensation allows the completion of the juvenile stages during the short growing season and enhances overwinter survival under harsh conditions (Lindgren and Laurila, 2009).

These models of latitudinal compensation are not mutually exclusive. In fact, as temperature and length of growing season covary across latitudinal gradients, organisms may adopt a mixed strategy by adapting to differences in both temperature and seasonality (Conover and Present, 1990; Yamahira and Conover, 2002). Nevertheless, observations from eco-physiological studies have shown that adaptive responses to climatic constraints at high latitudes usually follow the CnGV model (Yamahira and Conover, 2002). Latitudinal compensation usually comprises a rise in mitochondrial ATP synthesis capacity (Sommer and Pörtner, 2002; Sommer and Pörtner, 2004), and in metabolic rates of organisms at high latitudes (Clarke, 1993; Pörtner et al., 2005; Conover and Schultz, 1995). Increased metabolic rates are predicted to be an adaptive response, allowing accelerated physiological processes in environments that feature shorter periods of optimal conditions (Addo-Bediako et al., 2002; Clarke, 1993; Clarke, 2001), and should effectively counteract the environmental effect of depressed temperatures on growth rates (Conover and Schultz, 1995; Chown and Gaston, 1999). This example of CnGV has been termed metabolic cold adaptation (MCA) (Clarke, 1993; Clarke, 2001; Schaefer and Walters, 2010; White et al., 2012).

In general, MCA has been the subject of great controversy because whereas some studies support the MCA hypothesis (Addo-Bediako et al., 2002; Gaston et al., 2009; White et al., 2012; Hodkinson, 2003), other studies have failed to find an increase in O₂ consumption at high latitudes (Lardies et al., 2004a; Jordan et al., 2001; Steffensen, 2002; Steffensen et al., 1994). Additionally, much of the controversy surrounding MCA has centred on methodological issues that could confound the outcome of these studies (Chown et al., 2003; Bozinovic et al., 2011). The most straightforward approach to test for a CnGV pattern such as MCA is the comparison of populations at a single test temperature (Schaefer and Walters, 2010) using a common-garden experiment. This approach permits elimination of the possibility that differences between groups are due to maternal effects (i.e. modifications of offspring phenotype caused by the environment provided by the mother) (Mousseau and Fox, 1998), other pre-fertilization environmental influences (Suarez-Alvarez et al., 2012) or acclimation (i.e. phenotypic flexibility) (Piersma and Drent, 2003).

Using a common-garden experiment, we examined the MCA hypothesis in ectotherms at the intra-specific level by testing for the existence of latitudinal compensation in the land snail Cornu aspersum (O. F. Müller 1774). This land snail is characterized by physiological and morphological adaptations (e.g. metabolic depression, supercooling ability, reduction of the rate of water loss, thicker shell and epiphragm) that enable it to inhabit a great variety of terrestrial environments (Vorhaben et al., 1984; Bishop and Brand, 2000; Machin, 1967; Gaitán-Espitia et al., 2013c). Here, we analysed the extent to which the level of energy turnover (measured as standard metabolic rate, SMR) and the energy budget (energy allocation to growth) are affected by climatic constraints in three populations distributed across a latitudinal gradient of 1300 km in Chile.

Populations of C. aspersum across the latitudinal gradient (La Serena 29°S; Constitución 35°S; Valdivia 29°S) exhibited similar parameters (Table 1) and patterns of growth curves (Fig. 1A). In the three populations, values of specific growth during the first month of age were the largest (Table 1). However, the parameters estimated for monthly growth rate with the Gompertz equation showed differences between populations (maximum likelihood test, P<0.05) characterized by an increasing pattern with age in the La Serena and Valdivia populations, but not in Constitución (Table 1). This difference is explained by the greater rate of growth exhibited by snails from the Constitución population at 4 months of age (Table 1 and Fig. 1A; one-way ANOVA, F_2,1451=13.01, P<0.001). However, the overall growth rate showed the greatest values in the Valdivia population (one-way ANOVA, F_2,1451=8.01, P<0.001; Fig. 1B), producing greater body mass (M_b) in pre-adult snails from Valdivia compared with the other two populations after 6 months under the same rearing conditions (one-way ANOVA, M_b F_2,1451=10.83, P<0.001). Moreover, analysis at early stages (i.e. eggs and hatchlings) revealed that the mass of the eggs (Table 2) differed among populations (one-way ANOVA, F_2,1451=9.0, P<0.001), with larger eggs in La Serena and Valdivia populations compared with the Constitución population (Table 2; Tukey HSD test, P<0.05). After eclosion, these differences were maintained in the mass of the hatchlings across the latitudinal gradient (one-way ANOVA, F_2,1451=8.13, P<0.001) only because of the greater values for snails of the Valdivia population (Table 1; Tukey HSD test, P<0.05).

Of the physiological variables, only SMR was highly and significantly correlated with M_b (r=0.71, P<0.05), whereas body dehydration was low but significantly correlated with apparent dry-matter digestibility (r=0.10, P<0.05) and SMR (r=0.08, P<0.05). Analysis of these variables among populations revealed a lack of differences in SMR (one-way ANCOVA; F_2,1450=0.872, P=0.42; Fig. 1C) and body dehydration (one-way ANOVA; F_2,1451=1.24, P=0.29). In contrast, dry-matter digestibility showed differences across the latitudinal gradient (one-way ANOVA; F_2,1451=13.72, P<0.05), which was explained by lower values in snails from Valdivia (i.e. high latitude) compared with the other populations (Tukey HSD test, P<0.05; Fig. 1D). Finally, after controlling for M_b, SMR and the other physiological variables were not correlated with the overall growth rate.

According to the hypothesis of MCA (Clarke, 1993; Wohlschlag, 1960), populations of ectotherms that inhabit colder environments should have increased metabolic rates to compensate for the negative effect of low temperatures and shorter growing season on growth rates (Conover and Schultz, 1995). This latitudinal compensation should allow faster growth in populations from high-latitude habitats relative to their counterparts from lower latitudes when compared at a common temperature (Levins, 1969; Conover and Schultz, 1995; Schultz et al., 1996; Álvarez et al., 2006; Yamahira and Conover, 2002). However, our results evidenced a lack of differences in the energetic costs of maintenance (measured as SMR) among populations of the land snail C. aspersum raised in a common-garden experiment after three generations. This could be the result of: (i) metabolic adjustments through phenotypic plasticity (Naya et al., 2011); (ii) the effect of common-garden conditions and acclimation in the laboratory (Gatten et al., 1988); or (iii) the minimization of climatic differences through a ‘microhabitat effect’, considered to be a behavioural adaptation to seek the optimum environment (e.g. orientation of the shell to facilitate heat flow or heat retention, climbing on to host plants to choose resting sites that are sheltered from harsh environmental conditions) (Hazel and Johnson, 1990). Nevertheless, the absence of differences in SMR could indicate that snails from the three populations required similar amounts of energy for maintenance during physical inactivity (Lindgren and Laurila, 2009) despite the differences in the climatic constraints of their original environments (Gaitán-Espitia et al., 2013a).

Despite the lack of evidence in support of the predictions of the MCA (i.e. differences in SMR among populations of C. aspersum), our results revealed a pattern of CnGV in growth rates. In fact, snails from the highest latitude exhibited a greater overall growth rate than their counterparts from lower latitudes. This kind of countergradient response in growth rates has been found in other ectotherms at both intra- and inter-specific levels (Schultz et al., 1996; Conover and Present, 1990; Álvarez et al., 2006; Niewiarowski and Angilletta, 2008; Fangue et al., 2009; Conover and Schultz, 1995; Yamahira and Conover, 2002). Such a pattern suggests that the evolutionary response to short growing seasons and long winters at high latitudes favours genotypes with the capacity for rapid summer growth (Yamahira and Conover, 2002). However, it is important to recognize that CnGV models such as the MCA, rely on an energy budget context in which the energy allocation to growth is subject to constraints (e.g. the energy intake rate, digestion and absorption efficiency) and competing demands (e.g. maintenance, activity and reproduction) (Clarke, 1993; Weiner, 1992; Piersma and van Gils, 2011). Therefore, it could be expected that higher growth rates in populations from high latitudes involve some costs in other contributors to the energy budget (Chown and Gaston, 1999; Lindgren and Laurila, 2009; Pörtner et al., 2005; Conover and Present, 1990). Nevertheless, this was not found in our study, at least regarding maintenance metabolism.

In principle, there are two ways to explain changes in growth rate in ectothermic organisms from different environments. Firstly, as noted above, growth can be maximized by minimizing energy expenditure for maintenance or activity (Pörtner et al., 2005). Alternatively, growth can be maximized when selection promotes increased energy turnover (i.e. maximizing energy extraction), which generates excess energy for growth (Pörtner et al., 2005; Piersma and van Gils, 2011). Unfortunately, neither of these explanations is compatible with our results. In fact, in our study, those animals from the southernmost population, which exhibited higher growth rates, also exhibited lower apparent dry-matter digestibility (i.e. lower digestive efficiency). In a recent study using C. aspersum populations in a similar geographical gradient in Chile, Naya et al. found that at 20°C those snails from the highest latitude exhibited lower size and mass of digestive organs compared with lower latitude populations (Naya et al., 2011). This evidence could be associated with the lower dry-matter digestibility that we found in our study, because a smaller digestive tract involves a greater rate of food passage and lower intestinal retention (Karasov and Martínez del Rio, 2007). However, in the present study, snails from the highest latitude population grew faster than those from lower latitude populations, which is completely counterintuitive as it is not clear how they could obtain the extra energy (Lindgren and Laurila, 2009).

A Darwinian demon is a hypothetical organism that is able to maximize all elements of its fitness simultaneously. It reproduces immediately after birth, produces the maximum number of young and lives indefinitely (Williams, 1966; Gadgil and Bossert, 1970). As no such organism could exist, this metaphor is frequently used as a null hypothesis to be rejected when trade-offs are detected, or to highlight the fact that our reasoning is incomplete (e.g. because there is some missing variable). In this study, we cannot reject the Darwinian demon hypothesis and we do not believe snails are such ‘magic’ organisms that maximize everything with no costs. Here, adjustments in energy allocation to the digestive machinery and to protein storage (Piersma and van Gils, 2011), as well as differential nutrient digestibility among populations (Bozinovic et al., 1997), are possible explanations for the puzzling result. Nevertheless, further research is warranted to search for this missing link.

Three coastal populations of the land snail C. aspersum were selected from a latitudinal gradient of ca. 1300 km from northern to southern Chile. We selected these localities based on their different climatic characteristics (Fig. 2) and genetic differentiation (Gaitán-Espitia et al., 2013c). La Serena (29°54′S, 71°15′W) is a mesomediterranean arid environment with a mean annual temperature of 13.43±2.17°C, whereas Constitución (35°20′S, 72°25′W) is a mesomediterranean subhumid environment with a mean annual temperature of 12.36±2.65°C, and Valdivia (39°38′S, 73°5′W) is a mesotemperate perhumid environment with a mean annual temperature of 10.79±3.05°C (Amigo and Ramírez, 1998; Luebert and Pliscoff, 2006; Naya et al., 2011). Climatic data were downloaded from http://www.meteochile.gob.cl.

To ensure that measured traits come from population differences and not phenotypic plasticity or maternal effects, we measured individuals from the third generation after capture, 2 years after field collection. Six-hundred adult snails of approximately the same M_b (mean ± s.d. 4.21±0.63 g, approximate age less than 1 year) (Gaitán-Espitia et al., 2013b), were collected by hand from gardens and parks in each of the three localities (total of 1800 snails), placed in plastic containers and transferred to the laboratory at the Universidad Austral de Chile in Valdivia. Snails were maintained at densities of 50 animals in plastic cages (60×60×13 cm), filled with 10 cm of humid soil, and fed ad libitum with a mix of corn–wheat flour and calcium carbonate (1:1:0.3). The temperature and photoperiod for rearing conditions were 20°C and 16 h:8 h light:dark, respectively. This temperature is within the thermal optimum range of these populations (Gaitán-Espitia et al., 2013a). Relative humidity was maintained at high levels by sprinkling the interior of the boxes with water every day. Visual inspections were made during the next 6 months to identify copulation and egg laying. Snails were isolated into laying boxes as soon as they were seen copulating. Eggs were kept at the same environmental conditions, recording the incubation time, size and mass at eclosion. After eclosion, snails were maintained in plastic boxes (15×8×3 cm) of 20 individuals during the growth period. Animals were reared until pre-adulthood (i.e. 6 months) in the third generation.

To avoid any noise or erroneous recordings generated by animal manipulation, we eliminated the first 10 min of the record (600 samples) (Gaitán-Espitia et al., 2012). From each individual record, we extracted the average of each transformed record () which is used here as a proxy of SMR. To achieve a post-absorptive state, metabolic rate was measured in individuals deprived of food for 18 h (Naya et al., 2011). All metabolic trials were performed during the day, when land snails are inactive, which corresponds to the rest phase in this species (J.D.G.-E. and R.N., personal observation). All individuals were weighed at the beginning and at the end of the test period (i.e. 45 min), recording the mean M_b. Finally, to determine the extent of evaporative water loss (i.e. body dehydration), the snails were weighed at the beginning of each trial, when the animals were fully hydrated, and at the end of the test period (i.e. 45 min), recording the mean M_b and also the differences between initial and final M_b as a proxy for body water loss (Gaitán-Espitia et al., 2012).

After the 1454 respirometric trials, snails were maintained for 5 days in laboratory conditions with a diet of 100% lettuce to eliminate any remnants of the previous diet (i.e. flour mix) in the intestine. Animals were then deprived of food for 2 days and starting on the third day they were provided with a known amount of wheat flour for 15 days. On day 16 they were exposed to the 100% lettuce diet for three more days. Every day, faeces were removed, dried and weighed at 72°C to a constant mass (±0.0001 g). At the end of the experiment apparent dry-matter digestibility was calculated as (I–E)/I, where I is food intake and E is egestion (both in g). Digestibility is apparent because faecal contributions of endogenous protein and gut microflora were not considered (Lardies et al., 2004b). Energy content of food and faeces were determined in a Parr 1261 computerized calorimeter (Parr Instrument Company, Moline, IL, USA). Because there was no difference in the faecal energy content among populations, we excluded this variable from the analysis.

To evaluate the correlation between M_b, growth rate, energy expenditure (i.e. SMR), the ability to maintain water balance (i.e. body dehydration) and the apparent dry-matter digestibility in C. aspersum among populations in the latitudinal gradient, we conducted Pearson's correlation test. Maximum likelihood estimates of Gompertz growth rate parameters for each population were obtained using TableCurve2D curve-fitting software (version 5.01; Systat Software Inc.) Growth rate curves were compared based on the confidence intervals of the parameters and their slopes were tested using the maximum likelihood test (Kimura, 1980). Comparisons of size and mass of eggs and hatchlings, as well as M_b, body size, growth rate, body dehydration and dry-matter digestibility in pre-adults among populations were done using ANOVA, whereas SMR was analysed by ANCOVA with mean M_b as a covariate. When differences in the means were significant at the P<0.05 level, an a posteriori Tukey (HSD) test was used to locate significant differences among pairs. We checked normality and homoscedasticity by the Lilliefors and Levene tests, respectively. All data were square-root or log₁₀ transformed before the analysis. These analyses were run in Statistica v7.0 software (StatSoft).

The authors express their thanks to Dr Derek Roff for his critical reading of the manuscript. Also, the authors want to thank Diana C. Guerra Forero, Avia Gonzalez, Andrea Bruning, Jose L. Bartheld and Fredy Mondaca for their help in lab experiments and data digitalization.