Introduction
The transition to terrestrial life represents a pivotal event in the evolutionary history of animals, and how and why it occurred are topics of interest in evolutionary biology (Sayer and Davenport, Reference Sayer and Davenport1991; Rosenberg, Reference Rosenberg1996; Greenaway, Reference Greenaway2003; Weihrauch et al., Reference Weihrauch, Morris and Towle2004; Vermeij and Watson-zink, Reference Vermeij and Watson-zink2022). Although various potential causal factors driving terrestrial emergence, such as poor food quantity or quality, inter- and intraspecific competition, and predation risk, have been previously discussed and hypothesised, the evolutionary mechanisms underlying this transition remain unclear (Sayer and Davenport, Reference Sayer and Davenport1991). In particular, the selective pressures and adaptations required to overcome the challenges of terrestrial environments, including fluctuations in physicochemical parameters and their internal regulation, are still poorly understood in many animal taxa, including hermit crabs (Sayer and Davenport, Reference Sayer and Davenport1991; Greenaway, Reference Greenaway2003). Accumulating knowledge on the ecological significance of the utilisation of terrestrial habitats by marine animals is essential to elucidate the evolutionary processes underlying animal water-to-land transitions.
During low tides, some species of diogenid hermit crabs are known to actively climb from tidal pools onto rocky surfaces, exposing themselves to air. This behaviour has been scientifically documented in three Calcinus species, namely, Calcinus laevimanus (Randall, 1840), Calcinus obscurus (Stimpson, 1859), and Calcinus elegans (H. Milne Edwards, 1836) (Reese, Reference Reese1969; Perez-Miguel et al., Reference Perez-Miguel, Wehrtmann, Drake and Cuesta2020; Yoshikawa and Kosemura, Reference Yoshikawa and Kosemura2023), as well as in five species of Clibanarius (Dana, 1852), including Cl. corallines (H. Milne Edwards, 1848), Cl. erythropus (Hilgendorf, 1879), Cl. taeniatus (H. Milne Edwards, 1848), Cl. virescens (Krauss, 1843), and Cl. vittatus (Bosc, 1802) (Rittschof et al., Reference Rittschof, Sarrica and Rubenstein1995; Dunbar, Reference Dunbar2001; Perez-Miguel et al., Reference Perez-Miguel, Drake and Cuesta2015; Yoshikawa et al., Reference Yoshikawa, Goto, Yasuda and Asakura2020). When exhibiting this behaviour, hermit crabs adopt a specific posture, wherein the shell aperture points upwards (Reese, Reference Reese1969; Dunbar, Reference Dunbar2001; Perez-Miguel et al., Reference Perez-Miguel, Drake and Cuesta2015). Moreover, the molecular phylogenetic tree constructed by Tsang et al. (Reference Tsang, Chan, Ahyong and Chu2011) indicates that Calcinus and Clibanarius are closely related to the land hermit crab family Coenobitidae Dana, 1851. Therefore, species in these genera represent valuable models for studying the evolutionary forces and selective pressures that drive terrestrial adaptation in marine animals. Obtaining comprehensive data on species exhibiting this behaviour is essential to understand the adaptive significance and evolutionary relationships and conduct future studies on the role of natural selection in the terrestrial adaptation of hermit crabs.
Despite these insights, the ecological significance of air exposure (AE) in Calcinus spp. has received little attention. To date, the adaptive significance of AE behaviour has primarily been investigated in Clibanarius spp. (Rittschof et al., Reference Rittschof, Sarrica and Rubenstein1995; Dunbar, Reference Dunbar2001; Perez-Miguel et al., Reference Perez-Miguel, Drake and Cuesta2015; Yoshikawa et al., Reference Yoshikawa, Goto, Yasuda and Asakura2020). For example, Dunbar (Reference Dunbar2001) compared the shell-usage patterns of Cl. taeniatus and Cl. virescens between AE and non-AE individuals in Keppel Bay, Queensland, Australia. They found that the shells of three gastropods, namely, Planaxis sulcatus (Born, 1778), Clypeomorus petrosa (W. Wood, 1828), and Cronia sp., were more frequently used by AE hermit crabs than by others. Yoshikawa et al. (Reference Yoshikawa, Goto, Yasuda and Asakura2020) observed that female Cl. virescens exhibited a stronger tendency for AE behaviour compared with males, with this behaviour peaking at shield sizes of approximately 3.0–4.0 mm (which is relatively larger in females and mid-sized in males) before decreasing beyond this size range. Additionally, Rittschof et al. (Reference Rittschof, Sarrica and Rubenstein1995) suggested that the presence of Cl. vittatus on mud-sand and oyster reef substrates during low tide represents an adaptation by larger individuals to avoid shell competition when carrying suitable shells. These studies suggest that competition for shells, both within and between species, may be particularly intense among hermit crabs of specific size ranges, prompting individuals carrying suitable shells to avoid competition by leaving tide pools during low tides.
AE behaviour was identified according to previously published descriptions. This behaviour has been referred to by various terms in prior studies, including ‘air-exposure behaviour’, ‘posing behaviour’, and ‘sunbathing behaviour’ (Reese, Reference Reese1969; Dunbar, Reference Dunbar2001; Perez-Miguel et al., Reference Perez-Miguel, Drake and Cuesta2015; Yoshikawa et al., Reference Yoshikawa, Goto, Yasuda and Asakura2020; Yoshikawa and Kosemura, Reference Yoshikawa and Kosemura2023). For consistency, in this study, we adopt the term ‘air-exposure (AE) behaviour’, as described by Yoshikawa et al. (Reference Yoshikawa, Goto, Yasuda and Asakura2020) and Yoshikawa and Kosemura (Reference Yoshikawa and Kosemura2023).
To address this gap in knowledge between Clibanarius and Calcinus, the AE behaviour of Ca. laevimanus, namely, the dominant species on subtropical and tropical rocky shores, was investigated on the rocky shores of Naze, Amami, Kagoshima, Japan to understand the biological significance of this behaviour. We tested the hypothesis that Calcinus species avoid shell competition when inhabiting suitable shells by engaging in AE behaviour.
Materials and methods
Field survey and sample measurements
To investigate the significance of AE behaviour, hermit crabs exhibiting both AE and non-AE behaviours were collected from the field for comparison. Field surveys were conducted on rocky shores along the coast of Naze, Amami, Kagoshima, Japan (28°24′42.2″ N, 129°27′23.8″ E) (Figure 1A, B). Quantitative sampling was conducted during the lowest tidal periods on June 3–5, July 4–6, and August 3, 2023 using a 50 cm × 50 cm quadrat.
Map of the sampling site, photograph of the tidal pool, and air-exposure (AE) individuals along the side of the tide pool. (A) Location of the sampling site on the rocky shores of Naze, Amami, Kagoshima, Japan (indicated by a closed arrowhead). (B) AE individual of Calcinus laevimanus (Randall, 1840). (C) Schematic diagram of sampling rules 1 and 2 of this study; rules 1 and 2 are for individuals on the isolated rock and bedrock, respectively. In both rules, a quadrat was set on the nearest seawater area from each AE-exhibiting individual to collect the non-AE individuals. (D) Tide pool and representative substrate where AE behaviour was observed, and the above rules were applied.
First, individuals displaying AE behaviours were identified and collected. A quadrat was subsequently placed on an area of the sea bottom closest to the terrestrial area where AE behaviour was observed, and all non-AE individuals within the quadrat were collected following sampling rules 1 and 2 of this study. Rules 1 and 2 are for individuals on the isolated rock and bedrock, respectively. In both rules, a quadrat was set on the nearest seawater area from each AE individual to collect the non-AE individuals (Figure 1C, D). Finally, quadrat sampling was repeated 24 times.
In the laboratory, hermit crab species, shield length, sex, gastropod shell species, shell length, and shell width were recorded for both AE and non-AE individuals of Ca. laevimanus. For female individuals, the presence or absence of eggs was also recorded. The carapace shield length of the crabs, namely, an index of body size, was measured using callipers with a precision of 0.01 mm. Following the focus of previous studies (e.g., Yoshikawa et al., Reference Yoshikawa, Goto, Yasuda and Asakura2020), the presence of parasitic bopyrid isopods was also examined; however, no infestations were detected.
Statistical analysis
To determine the factors associated with AE behaviour, a generalised linear model (GLM) with a binomial error distribution was constructed based on the dataset of Ca. laevimanus. The response variable was binary data indicating the presence or absence of AE behaviour (non-AE = 0; AE = 1), and continuous variables of shield length, shell length, and shell width and categorical variables of sex (non-ovigerous female, ovigerous female, male) were used as explanatory variables. In categorical explanatory variables of sex, non-ovigerous female was used as the baseline and ovigerous female and male were used as the dummy variables. The significance of each explanatory variable was assessed using likelihood ratio tests. Post hoc comparisons among factor levels of sex were conducted using estimated marginal means (emmeans) based on the fitted GLM, with Tukey-adjusted P-values to control for multiple comparisons. To examine multicollinearity, the adjusted generalised variance inflation factor (GVIF) was also calculated (GVIF(1/(2Df))). Adjusted GVIF values ranged from 1.07 to 2.65, indicating that multicollinearity was not a serious concern in this analysis (Supplementary Table S1).
To evaluate the robustness of the results, a sensitivity analysis with 1,000 resampling iterations was conducted to equalise the sample sizes between AE and non-AE individuals. As the number of AE individuals was smaller than that of non-AE individuals, AE individuals were fixed and non-AE individuals were randomly sampled from the complete dataset in the same number as AE individuals during each iteration. For each resampled dataset, GLMs were constructed with AE behaviour as the response variable and shield length, shell length, shell width, and sex as the explanatory variables. The robustness of the results was assessed by examining the consistency in the sign (positive or negative) of the estimated coefficients across models.
Additionally, the difference in shell length and shield length between AE and non-AE individuals were also tested using GLMs with a gamma distribution and log link function. Analyses were conducted separately for all individuals combined and for each sex category (male, female, ovigerous female, and non-ovigerous female). In these models, shell length was used as the response variable and AE behaviour (non-AE = 0; AE = 1) was used as the explanatory variable. The differences in shell length between AE individuals of Ca. laevimanus and non-AE individuals of all collected species were also examined using GLMs with a gamma distribution and log link function. Additionally, one AE-exhibiting non-ovigerous female of Ca. latens and three AE-exhibiting individuals of Cl. corallinus (one male and two non-ovigerous females) were also collected during field sampling (Supplementary Table S2). These data were removed from this analysis to focus on the trigger and ecological significance of AE behaviour of Ca. laevimanus. Moreover, data of the juvenile-immature individuals of Ca. latens and Pagurus sp., whose sex was not detected, were also removed from the analysis. All data used in this analysis are summarised in Supplementary Table S2.
Sensitivity tests with 1,000 resampling iterations were similarly conducted for the GLMs comparing the parameters with significance. As specified above, AE individuals were fixed, and non-AE individuals were randomly sampled from each dataset in the same number as AE individuals. For each resampled dataset, GLMs were constructed with shell length as the response variable and AE behaviour as the explanatory variable. Robustness was evaluated by examining the consistency in the sign of the estimated coefficients across models.
These GLMs were implemented using the ‘glm’ function in R version 4.2.3 (R Core Team, 2023). Multicollinearity was tested using the ‘vif’ function in the ‘car’ package (Fox and Weisberg, Reference Fox and Weisberg2019), and post hoc comparison was tested by the ‘emmeans’ function in the ‘emmeans’ package (Lenth and Piaskowski, Reference Lenth and Piaskowski2025).
Results
A total of 398 individuals of Ca. laevimanus were collected, including 25 AE individuals (12 non-ovigerous females, 1 ovigerous female, and 12 males) and 373 non-AE individuals (56 non-ovigerous females, 194 ovigerous females, and 123 males) (Supplementary Table S2). The average body sizes of AE individuals were 3.99 ± 0.66 (SE) and 4.75 ± 1.03 (SE) mm for females and males, whereas the average body sizes of non-AE individuals were 4.06 ± 0.68 (SE) and 4.48 mm ± 1.31 (SE) mm for females and males, respectively.
The GLM for determining the factors associated with the AE behaviour of Ca. laevimanus showed a significantly higher frequency of AE behaviour in individuals with a larger host shell length (Table 1; Figure 2). Additionally, the model indicated a significant effect of sex, and post hoc comparisons indicated a significantly lower frequency of AE behaviour in ovigerous females than in non-ovigerous females (estimate = −3.76, SE = 1.07, P-value = 0.0012) and males (estimate = −2.57, SE = 1.07, P-value = 0.042) (Table 1; Supplementary Table S3). The shield length and shell width did not show a significant influence on AE behaviour, and the difference in the frequency of AE behaviour between non-ovigerous females and males was not significant (estimate = −1.19, SE = 0.53, P-value = 0.065) (Supplementary Table S3). Moreover, the sensitivity test with 1,000 resamples indicated consistency in the influence of shell length (positive: 99.8%) and sex of ovigerous female (negative: 100%) (Supplementary Table S4).
Predicted probability of air-exposure (AE) behaviour in relation to the shell length predicted by the generalised linear model (GLM). The histograms indicate the distribution of shell length in AE (upper) or non-AE (lower) individuals. All the (A) non-ovigerous females, (B) ovigerous females, and (C) males (non-AE = 0; AE = 1). Shaded areas represent 95% confidence intervals around the GLM estimates.
Result of the generalised linear model (GLM) on the dataset of Calcinus laevimanus for predicting the probability of air-exposure (AE) behaviour based on the parameters measured in this study
P-values were obtained from likelihood ratio tests. Significant effects are indicated in bold font. The categorical variable containing non-ovigerous female, ovigerous female, male, and non-ovigerous female was used as the baseline for the explanatory variable. *P-value < .05, **P-value < .01, P-value < .001
The GLM for comparing the shell length between AE and non-AE individuals indicated significantly longer shells in AE individuals in all sexes and status (Supplementary Table S5; Figure 3). Moreover, the shells of AE individuals of Ca. laevimanus were significantly longer than those of non-AE individuals of all collected species (Figure S1). The sensitivity tests with 1,000 resamples indicated consistency in the influence of AE behaviour on shell length (all positive: all sexes = 100%; male = 99.3%; female = 100%; ovigerous female = 100%; non-ovigerous female = 100%; Ca. laevimanus vs. all species = 100%). On the other hand, the GLMs for comparing the shield length indicated non-significant difference between AE and non-AE individuals in all sexes and status (Figure S2; Table S6).
Violin and strip plots showing the size comparison of the shell length between air-exposure (AE) and non-AE individuals of Calcinus laevimanus. All the (A) females, (B) males, (C) non-ovigerous females, and (D) ovigerous females. This analysis showed that individuals exhibiting AE behaviour used significantly taller shells than non-AE individuals (*significant differences, see Supplementary Table S5).
Discussion
In the intra- and interspecific comparisons, our results indicate that individuals inhabiting larger gastropod shells, which provide the potential for greater growth, more frequently engage in AE behaviour during low tides. As the growth and reproductive success of hermit crabs primarily depend on the availability and quality of shells (Reese, Reference Reese1969; Wilber, Reference Wilber1989; Asakura, Reference Asakura1991), this behaviour is likely to have an adaptive significance in preventing preferable or high-quality shells from being taken up by potential competitors among Ca. laevimanus individuals. However, ovigerous females exhibited this behaviour at significantly lower frequencies. This suggests that engaging in AE behaviour may pose risks to eggs, such as desiccation, exposure to high temperatures, and/or hypoxic conditions. Although the reproductive season of this species in this study area has not been documented in detail, Ca. laevimanus is generally considered to have an extended reproductive period. Reese (Reference Reese1968) reported that females are in the breeding cycles and carry eggs from February to early October in the tropical zone. Moreover, a few ovigerous females were found throughout the year (Reese, Reference Reese1968). As ovigerous females were included in our analysis conducted on the subtropical zone, the study period likely overlapped at least partly with the reproductive season of Ca. laevimanus along the coast of Naze, Amami, Kagoshima, Japan. Accordingly, although the exact reproductive timing of this local population remains undocumented, these results may be relevant to understand the adaptive significance of this behaviour during most of the year in the study area. Consequently, our results partially support the previously proposed hypothesis that ‘individuals carrying suitable shells may avoid competition by actively escaping from tide pools during low tide’ (Yoshikawa et al., Reference Yoshikawa, Goto, Yasuda and Asakura2020).
In contrast to Cl. virescens, wherein mature females frequently engage in AE behaviour to retain high-quality shells, regardless of the season, no clear size bias was detected for Ca. laevimanus. Yoshikawa et al. (Reference Yoshikawa, Goto, Yasuda and Asakura2020) also suggested that intra- and interspecific competition for shells was particularly intense among hermit crabs with overlapping size ranges. Individuals with suitable shells may avoid such competition by actively escaping from tidal pools during low tides. Other hypotheses have been proposed to explain the selective advantages of AE behaviour, such as conserving energy for growth by reducing movement, avoiding shell fights (Rittschof et al., Reference Rittschof, Sarrica and Rubenstein1995), and maintaining or cleaning shells (Perez-Miguel et al., Reference Perez-Miguel, Drake and Cuesta2015). These findings suggest that AE behaviour in Calcinus species is triggered by external factors (i.e. shell size), which may appear to have a more complicated mechanism in that of Clibanarius species, as opposed to internal factors (i.e. body size or sex) in the intra- and interspecific comparison. Further controlled experiments under various conditions, such as the variation in shell quality within individual hermit crabs or fluctuations in the density of conspecific and heterospecific hermit crabs, are required to rigorously evaluate this hypothesis. These evaluations must not only focus on the ecological significance of this behaviour but also on its differential functional roles between species of Clibanarius and Calcinus.
The AE behaviour of Calcinus species has been previously documented in Ca. laevimanus, Ca. obscurus, and Ca. elegans (Reese, Reference Reese1969; Perez-Miguel et al., Reference Perez-Miguel, Wehrtmann, Drake and Cuesta2020; Yoshikawa and Kosemura, Reference Yoshikawa and Kosemura2023). According to the molecular phylogenetic tree constructed using three genes (16S rDNA, COI, and Histone 3), Ca. laevimanus is most closely related to Ca. seurati (Malay and Paulay, Reference Malay and Paulay2010). Additionally, Ca. obscurus is included in a different clade from Ca. laevimanus, but the most closely related clade comprises Ca. californiensis, Ca. maclaughlinae, Ca. tibicen, Ca. talismani, and Ca. explorator. Moreover, Ca. elegans is also grouped with phylogenetically different species, such as Ca. gaimardii, Ca. imperialis, Ca. isabellae, Ca. mogami, Ca. orchidae, and Ca. vaninii. Observations by Yoshikawa and Kosemura (Reference Yoshikawa and Kosemura2023) also suggested that a few individuals of Ca. latens, Ca. mogami, and Ca. vahoni may engage in AE behaviour on rocky tidal pool surfaces. Moreover, although they were not used in the statistical analysis, one Ca. latens and three Cl. corallinus individuals were collected as AE individuals during this study. As AE behaviour appears to have evolved in a broader range of Calcinus and Clibanarius species or multiple times independently, continuous comparative investigations on phylogenetically different species are necessary to understand the general adaptive significance of this behaviour and explore how water-to-land transitions occur within hermit crab lineages.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0025315426101283.
Acknowledgements
We are grateful to the following people for their invaluable support with sampling, field surveys, and data collection: Dr. Kei Kawai, Ms. Minori Takarabe, Ms. Mikako Wakashi (International Center for Island Studies, Amami Station, Kagoshima University), all the staff of the International Center for Island Studies, Kagoshima University, all the staff members of Amami Ocean Exhibition Hall, Mr. Kosuke Uchiwa, Mr. Hiroyoshi Kotobuki, and Mr. Misao Higashi (Amami, Kagoshima, Japan). We are also grateful to Mr. Ryo Nakayama (Fisheries Research Institute, Aomori Prefectural Industrial Technology Research Center) for identifying the gastropod species whose shells were inhabited by the hermit crabs. We thank Editage for reviewing the English language in this manuscript.
Author contributions
AY and TM designed the research. AY, TM, MW, TK, TK, and KT conducted field surveys. TM analysed data. AY, TM, MW, and TK wrote the paper. All authors have accepted responsibility for the entire content of this manuscript and approved its submission.
Funding
This study was financially supported by the Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research (KAKENHI; Grant No. JP 21K20591), and an Innovative Model Fusing Conservation of Biological and Cultural Diversity with Regional Revitalisation in the Amami Archipelago, Kagoshima University.
Competing interests
We declare we have no competing interests.
Ethical standards
This work did not require ethical approval from a human subject or animal welfare committee.
