Polar gigantism and the oxygen–temperature hypothesis: a test of upper thermal limits to body size in Antarctic pycnogonids

Abstract

The extreme and constant cold of the Southern Ocean has led to many unusual features of the Antarctic fauna. One of these, polar gigantism, is thought to have arisen from a combination of cold-driven low metabolic rates and high oxygen availability in the polar oceans (the ‘oxygen–temperature hypothesis'). If the oxygen–temperature hypothesis indeed underlies polar gigantism, then polar giants may be particularly susceptible to warming temperatures. We tested the effects of temperature on performance using two genera of giant Antarctic sea spiders (Pycnogonida), Colossendeis and Ammothea, across a range of body sizes. We tested performance at four temperatures spanning ambient (−1.8°C) to 9°C. Individuals from both genera were highly sensitive to elevated temperature, but we found no evidence that large-bodied pycnogonids were more affected by elevated temperatures than small individuals; thus, these results do not support the predictions of the oxygen–temperature hypothesis. When we compared two species, Colossendeis megalonyx and Ammothea glacialis, C. megalonyx maintained performance at considerably higher temperatures. Analysis of the cuticle showed that as body size increases, porosity increases as well, especially in C. megalonyx, which may compensate for the increasing metabolic demand and longer diffusion distances of larger animals by facilitating diffusive oxygen supply.

1. Introduction

Since the formation of the Antarctic Circumpolar Current during the Cenozoic Era, Antarctic marine species have evolved in one of the coldest and most temperature-stable marine environments on Earth [1,2]. Temperatures in the high Antarctic vary by less than 1.5°C above the freezing point of water (−1.9°C) all year round [3]. This extremely cold, stable environment is thought to have driven the evolution of unusual polar adaptations such as the formation of glycoprotein antifreeze in ice fish [4]. Many Antarctic marine organisms are also highly stenothermal and can only survive and function within a narrow temperature range [5–10]; loss of function occurs at temperatures only 1–2°C higher than ambient [5,9,11] and even brief excursions to 5–10°C can be fatal [9].

Another striking feature of the Antarctic fauna is the evolution of unusually large body sizes, a phenomenon known as polar gigantism [12–15]. There are many hypotheses about the ecological and evolutionary factors that underlie polar gigantism [16,17]; one that has received considerable attention is the oxygen–temperature hypothesis which postulates that polar gigantism arose from a combination of cold-driven low metabolic oxygen demand and high oxygen availability [18,19] owing to upwelling of deep waters around Antarctica and low biological demand [20]. That body size is limited by temperature–oxygen interactions is an old idea. In 1960, von Bertalanffy [21] first pointed out that for ectotherms, as temperatures increase, metabolic oxygen demand increases more steeply than oxygen supply. This creates a threshold body size above which diffusive supply cannot meet demand, meaning that at higher temperatures large animals face greater challenges to matching oxygen uptake with whole-body oxygen requirements [17,21–23]. The high ratio of oxygen supply to demand in the Southern Ocean is thought to allow ectotherms to reach larger body sizes without experiencing increased oxygen deprivation [16,18,24,25]. However, if the large body sizes of polar giants evolved in response to cold temperatures, then polar giants may be among the most vulnerable taxa in the face of ocean warming [18].

The pycnogonids, or sea spiders (Chelicerata, Pycnogonida), contain some of the most striking examples of polar gigantism [14,26–28]. Pycnogonids inhabit most marine environments [29,30] and there are approximately 1400 known species in 80 genera [29,31,32], with about 190 species occurring in Antarctic waters [28,33]. Worldwide, most species are small as adults, with maximum leg spans of a few centimetres; Antarctic and abyssal species, however, can have leg spans of over 70 cm [29]. On the Antarctic benthos, pycnogonids are abundant, diverse and ecologically important as predators and scavengers [28,34,35]. Pycnogonids do not have specialized respiratory organs or pigments [36,37] and obtain oxygen through diffusion across the cuticle [38–40]. Oxygen is transported throughout the body both by a dorsal heart located in the trunk and by peristaltic contractions of the gut [41], which extends into the legs.

Woods et al. [42] explored interactions between body size and oxygen availability in Antarctic sea spiders by measuring the righting ability of animals—a metric of physiological performance, for individuals across 12 species in five families under a range of oxygen saturations. While hypoxia strongly reduced performance, large-bodied animals were not more susceptible to oxygen limitation than small ones. Woods et al. [42] listed three possible explanations for this surprising result. First, righting experiments were performed at near-ambient temperatures (−1.13 ± 0.02°C) and cold-induced depression of metabolic rate may have ameliorated the potential detrimental effects of large body size on oxygen supply. This interpretation is supported by results from Peck et al. [9], who found interactive effects of oxygen and body size in the Antarctic clam, Laternula elliptica, but only at temperatures greater than 0°C. Second, large-bodied animals may have undergone evolutionary increases to their capacity for oxygen uptake and delivery to maintain performance as body size increased (i.e. symmorphosis; Weibel et al. [43]) [42]. Third, some species were much more sensitive to changes in dissolved oxygen concentration than others; phylogenetic effects could have potentially masked interactions between body size, performance and oxygen availability as their dataset combined multiple species from several different families of pycnogonids.

We directly tested the interactions between body size, temperature and performance of Antarctic pycnogonids using animals of a range of sizes within congeneric and conspecific groups. We hypothesized that larger-bodied pycnogonids would show disproportionately poor performance compared to small-bodied pycnogonids at elevated temperatures, consistent with the predictions of the oxygen–temperature hypothesis. We also predicted that like most other Antarctic ectotherms [7–9], Antarctic pycnogonids would be highly stenothermal and that small amounts of warming would have strong effects on performance.

Second, we explored the relationship between animal size and cuticular structure in the same taxa. Across different species and families, pycnogonids compensate for the greater diffusion distances and higher metabolic demand of larger bodies by increasing oxygen flux across the cuticle [42], at least in part by increasing the porousness of the cuticle [40]. We examined the relationship between body size and cuticle structure (thickness and porosity) within two different species of giant Antarctic pycnogonids to test if cuticle morphology also changes to accommodate larger body size as animals grow.

2. Material and methods

(a). Collection and maintenance

Pycnogonids were collected on SCUBA near McMurdo Station, Antarctica (77°51′ S, 166°40′ E) between 10 and 40 m depth, in October and November 2015 and 2016. Pycnogonids were transported to McMurdo Station in coolers filled with chilled seawater (−1.8°C) and were kept in flow-through seawater tables maintained at 1–2°C above ambient seawater temperatures (−1.8°C) until used for experiments.

We collected animals from two genera, Ammothea (Ammotheidae) and Colossendeis (Colossendeidae). In McMurdo Sound, these are the two genera of sea spiders that attain very large body sizes as adults (leg spans greater than 10 cm) (figure 1) and can be readily found in abundance and across a broad range of body sizes. Individuals were identified to species using the keys of Child [44]. When animals could not be identified to species unambiguously, we used genetic barcoding (cytochrome c oxidase subunit 1) to identify them (methods in Lane et al. [45]). In total, our dataset contained six species of Colossendeis (C. australis, n = 2, C. hoeki, n = 13, C. megalonyx, n = 20, C. robusta, n = 2, C. scotti, n = 2 and C. sp., n = 3) and one species of Ammothea (A. glacialis, n = 26).

Figure 1.

(a) Ammothea glacialis; photo: Timothy R. Dwyer, PolarTREC 2016/courtesy of ARCUS, (c) cuticle cross-section of A. glacialis, (b) Colossendeis megalonyx; photo: Timothy R. Dwyer, PolarTREC 2016/courtesy of ARCUS, and (d) cuticle cross-section of C. megalonyx. Arrows on bottom pictures indicate pores. (Online version in colour.)

(b). Temperature–body size experiments

To determine if pycnogonid performance was affected by body size, temperature and their interaction, we measured righting ability over a range of body sizes and temperatures. In 2015, we tested 30 individuals of Colossendeis at three different temperatures: −1.8, 4 and 9°C. Because the animals showed a sharp decline in performance between 4 and 9°C, in 2016, we added an intermediate temperature and tested an additional 12 individuals of Colossendeis at −1.8, 4, 7 and 9°C. Also, in 2016, we performed these tests on 26 individuals of A. glacialis at all four temperatures. Righting experiments were carried out in a 500 l aquarium submerged in a 1000 l tank with temperature regulation provided by a submersible ‘cold finger’/bath cooler (model PBC-2II, Neslab) and a heating element (immersion circulator, Fisher Scientific). This experiment was an acute temperature challenge rather than an emulation of ecologically relevant warming, so animals were moved directly from sea tables into water at the experimental temperature. The temperature of the water in the aquarium was always within 0.3°C of the target temperature. Air was constantly and gently bubbled into the aquaria to maintain 100% air saturation.

Each trial was performed with a single pycnogonid in the aquarium. Prior to each trial, each animal was allowed to adjust to its surroundings for 20 min (a length of time we had previously determined was sufficient for pycnogonids to return to their normal orientation and stance after being moved into reduced oxygen (Woods et al. [42]) or a new experimental temperature (this study). Then, the animal was gently flipped upside-down and allowed to right itself; this process was repeated until we had counted the number of times the animal righted itself in a 1 h period. Each animal was tested at all three (n = 30) or four (n = 38) experimental temperatures. The order of temperature exposure was determined randomly for each individual and no animal was tested more than once per day. Each animal was assessed daily for normal appearance, movement and behaviour over the course of the experiment.

(c). Cuticle morphometrics

After the performance of an animal had been assessed at all temperatures, the animal was blotted dry and weighed on an electronic balance (Model PE1600, Mettler Toledo) to obtain wet mass, then photographed for morphological measurements. A subset of individuals was preserved in 100% ethanol (EtOH) for later genetic and cuticular analysis (preliminary tests established that EtOH preservation did not alter cuticular thickness or morphology). Cuticular analysis was performed on a combination of animals from the righting experiments (n = 10 C. megalonyx and n = 19 A. glacialis) and 14 additional C. megalonyx and eight A. glacialis collected during 2015 and 2016 at the same sites (weighed, photographed and preserved as above). Measurements of cuticle morphometrics were not performed on other species of Colossendeis because of the comparatively small sample sizes. In total, we measured 24 C. megalonyx and 27 A. glacialis for four cuticular parameters: areal porosity (AP, %), cuticle thickness (CT, cm), surface area (SA, cm²) and pore volume (PV, cm³). All parameters were measured from digital images using ImageJ software (v. 1.51j8, NIH) [46] (detailed methods in the electronic supplementary material, S1).

(d). Statistical analysis

We analysed the performance data in three categories; all species of Colossendeis combined, C. megalonyx alone and A. glacialis alone. Righting data for all groupings were strongly zero-inflated at 7 and 9°C because some animals never righted themselves at these higher temperatures. Therefore, to determine if there was an interactive effect between mass and temperature on righting performance, we used a zero-inflated generalized linear mixed-effects model (ZIGLMM) fitted with the ‘glmmadmb’ function in the ‘glmmADMB’ package in R [47]. The data for the masses of both Colossendeis and A. glacialis were log-transformed to meet the assumptions of normality and to give reasonable distributions of the residuals. We used number of times an animal could right itself as the response variable (ZIGLMM was fitted with a Poisson distribution to account for count data), temperature treatments and size as explanatory variables and incorporated individual pycnogonids as random effects. Pycnogonids that did not right themselves at our control temperature (−1.8°C) were excluded from the analyses (C. hoeki n = 1, C. megalonyx n = 4). The current glmmADMBpackage in R did not support post hoc tests; thus temperature thresholds were assessed by comparing 95% confidence intervals around mean righting rates at each temperature.

Most statistical analyses were performed in JMP (v. 13, SAS Institute Inc., Cary, NC) except the ZIGLMM which was performed in Rstudio (v. 1.0.143) [48].

To test if exposure to our highest experimental temperature (9°C) had detrimental effects on subsequent performance, we used a Student's t-test (data met all assumptions) to compare the righting performance of pycnogonids at −1.8°C between those that received −1.8°C as their first exposure (Colossendeis combined, n = 12; C. megalonyx, n = 5; A. glacialis, n = 7) and those that received 9°C as their first exposure (Colossendeis combined, n = 11; C. megalonyx, n = 6; A. glacialis, n = 5). We also used an ANOVA (data met all assumptions) to determine if the order of exposure and subsequently the impact of multiple days of testing had detrimental effects on performance. We used the number of rightings as a response variable, order as an explanatory variable and incorporated individual pycnogonids as random effects (Colossendeis combined, n = 42; C. megalonyx, n = 20, A. glacialis, n = 26).

To determine how CT, PV and SA scaled with body mass (g), we log₁₀-transformed CT, PV and SA, and fitted the data to ordinary least-squares regressions. We then compared the slopes of our regressions against the scaling exponents expected from isometric geometric scaling; we considered our measured slopes to be significantly different from isometry if the 95% confidence intervals of the slope did not include the expected isometric scaling coefficient. To determine if scaling differed between species, we used ANCOVA to compare slopes between C. megalonyx and A. glacialis. We also compared the relationship between AP and body mass (g) for C. megalonyx and A. glacialis using linear regression.

3. Results

(a). Temperature and righting

For all species and genera combined, the temperature had a significant and negative effect on righting performance (ZIGLMM, p < 0.0001) (figures 2 and 3; electronic supplementary material, tables S2 and S3). Across all individuals, righting frequency ranged from 175 rights hr⁻¹ (RPH) at ambient temperature to 0–23 RPH at the highest temperature (electronic supplementary material, table S3). Average RPH decreased approximately 10-fold for C. megalonyx and all Colossendeis combined between ambient and 9°C, and more than 20-fold for A. glacialis. The righting ability of individuals of Colossendeis combined and C. megalonyx was not impacted until 9°C (figure 2; electronic supplementary material, table S4). RPH of individuals of A. glacialis dropped at each step increase in temperature starting at 4°C (figure 2; electronic supplementary material, table S4). No pycnogonids died and the behaviour and activity level of animals did not visibly change over the course of the experiment. t-tests within each genus also showed no significant difference between the number of rightings at −1.8°C when we compared pycnogonids that had received their first treatment at −1.8°C to those that had received 9°C as the first exposure: Colossendeis (p = 0.27), C. megalonyx (p = 0.51), A. glacialis (p = 0.58). The order of exposure to experimental temperatures and impact of multiple days of testing also did not have any effect on the performance: Colossendeis combined (F = 0.37; p = 0.77), C. megalonyx (F = 1.17; p = 0.33), A. glacialis (F = 0.069; p = 0.98).

Figure 2.

Rate of self-righting (mean ± s.e.) rightings hr⁻¹ for all species of (a) Colossendeis p < 0.0001, (b) Colossendeis megalonyx p < 0.0001 and (c) Ammothea glacialis p < 0.0001, at ambient (−1.8°C) and three elevated (4°, 7° and 9°C) temperature treatments. See the electronic supplementary material, table S4 for summary of 95% confidence intervals. (Online version in colour.)

Figure 3.

Righting performance (number of rightings hr⁻¹) compared to body mass (g) of (a) Colossendeis, (b) Colossendeis megalonyx and (c) Ammothea glacialis. See the electronic supplementary material, table S2 for results of ZIGLMM. (Online version in colour.)

Mass did not have a significant effect on RPH of A. glacialis (p = 0.11) or on the combined Colossendeis individuals (p = 0.44), and there was no significant interaction between temperature and body size in either group (A. glacialis, p = 0.18; all Colossendeis, p = 0.86) (figure 3; electronic supplementary material, table S2). By contrast, mass did significantly affect RPH of C. megalonyx (p = 0.002) and there was also a significant interaction between temperature and mass (p = 0.005) for this group.

(b). Cuticle morphometrics

Mass, SA and CT all ranged over close to an order of magnitude in C. megalonyx and more than an order of magnitude for A. glacialis. No cuticle measurements were made for Colossendeis from other species, but mass for these ranged over two orders of magnitude from 0.21 to 21.80 g.

The log-linear relationships between CT (cm) and mass (M, g) for the two genera were CT_C.megalonyx = −2.26 × M^0.51 and CT_A.glacialis = −2.21 × M^0.58. The slopes of the relationship between CT and M for both species were not significantly different from each other (ANCOVA, p = 0.1925). The 95% confidence intervals of the scaling exponents of the two groups (0.51 for C. megalonyx and 0.58 for A. glacialis, electronic supplementary material, table S4) overlapped substantially, but in neither case did the confidence intervals contain one-third (isometric scaling; electronic supplementary material, table S5). The relationship between SA (cm²) and M was SA_C.megalonyx = 1.45 × M^0.62 and SA_A.glacialis = −1.31 × M^0.78. There was a significant difference in the slopes in the relationship of SA to M between C. megalonyx and A. glacialis (ANCOVA, p < 0.0001*; figure 4). The 95% confidence intervals of the scaling exponents did not include the predicted scaling exponent of two-thirds for either of the two groups (0.62 for C. megalonyx and 0.78 for A. glacialis) (electronic supplementary material, table S5). The relationship between porosity (AP, %) and M was AP_C.megalonyx = 0.18 × M^0.11 and AP_A.glacialis = 0.10 × M^0.015. Because there is no expected scaling exponent for the relationship between AP and M, we could not make that comparison. However, the slope of this relationship was significantly higher for C. megalonyx than for A. glacialis (ANCOVA, p = 0.0126, figure 5). Finally, the relationship between (PV, cm³) and M was PV_C.megalonyx = −1.62 × M^1.50 and PV_A.glacialis = −1.93 × M^1.61. There was no significant difference in the slope of PV and mass between C. megalonyx and A. glacialis (ANCOVA, p = 0.5026; figure 4). The 95% confidence intervals of the scaling exponents for both species (1.50 for C. megalonyx and 1.61 for A. glacialis, electronic supplementary material, table S5) did not contain the expected scaling exponent of 1 (electronic supplementary material, table S5).

Figure 4.

Regression lines for scaling relationships between (a) cuticle thickness (cm), (b) surface area (cm²), and (c) pore volume (cm³). Individuals of Ammothea glacialis are represented by closed circles and Colossendeis megalonyx are represented by the open circles. See the electronic supplementary material, table S5 for summary of scaling coefficients.

Figure 5.

Regression lines for scaling relationships between areal porosity and mass (g). Individuals of Ammothea glacialis are represented by closed circles and Colossendeis megalonyx are represented by the open circles.

4. Discussion

Temperature strongly and negatively affected the righting performance of all Antarctic pycnogonids in our study. These data are consistent with patterns seen in Antarctic bivalves [10,11,49,50], brachiopods [51] and limpets [52], and with the overall paradigm of strong stenothermality of Antarctic marine ectotherms [53]. However, the two genera of sea spiders in our study differed in their sensitivity to temperature; for Ammothea, righting ability dropped at each temperature increment, and there was a two-fold drop between −1.8 and 4°C. For individuals of C. megalonyx alone, however, righting rates did not drop until our highest temperature treatment, 9°C. Antarctic ectotherms are typically very sensitive to small increases in temperature, so the ability of C. megalonyx to maintain performance up to 7°C and to survive excursions up to 9°C with no observed after-effects on survival or behaviour was intriguing (these temperatures are 7.5°C and 9.5°C, respectively, above the annual summer maximum of -0.5°C in McMurdo Sound [3]). However, these results are consistent with Peck et al. [54] who found that some Antarctic species such as the amphipod Cheirimedon femoratus survived temperatures up to 17.6°C when exposed to a rapid rate of warming (1°C day⁻¹). We used an acute thermal challenge, while previous studies [10,50] that tested the effects of temperature on Antarctic invertebrates used slower ramping rates and/or longer exposure periods from 24 h [10] to 20 days [50], both of which can increase thermal sensitivity [54]. It remains to be seen, therefore, whether other types of Antarctic organisms can also survive large but short-term increases in temperature, or if the unique morphology and physiology of pycnogonids, particularly with regard to gas exchange, make Antarctic species (particularly Colossendeis) unusually robust to short-term exposure to elevated temperatures.

In contrast to temperature, which had strong effects on performance, the effects of body size on performance varied. In two of three groupings (A. glacialis and all Colossendeis combined), we found no effects of body size, measured as mass, on righting performance. In C. megalonyx, however, righting performance increased with body size. Such an outcome could arise from several biomechanical or physiological mechanisms, e.g. a shift in the centre of gravity or a hypermetric increase in muscle mass with body size. However, when data from all species of Colossendeis were combined, the intraspecific pattern was obscured, probably driven by the fact that the other Colossendeis species were much larger than C. megalonyx but righted themselves less frequently.

The significance of the interaction between mass and temperature also varied among groupings. Neither A. glacialis nor combined Colossendeis showed this interaction, suggesting that temperature affected righting performance similarly across all body sizes. These results are consistent with those of Woods et al. [42] for a combined dataset including 12 species from five families, in which hypoxia strongly affected performance but these effects were not size-dependent. We did, however, find an interaction between body size and temperature for C. megalonyx; compared with smaller animals, larger animals were more strongly affected by warm temperatures. This pattern was driven by a change in the relationship between body size and righting rates at the highest temperature, 9°C. At the three lower temperatures, larger animals righted themselves more times per hour than small animals; at 9°C, by contrast, RPH was reduced to close to 0 for all body sizes. Two interpretations of these results are that (i) larger animals were in fact more strongly affected by the highest temperature, supporting the temperature–size hypothesis, or (ii) 9°C exceeded the performance threshold of most animals regardless of size. We think the latter is more likely, since more than half of the C. megalonyx did not right themselves at all at 9°C, and 9°C is higher than, or in some cases very close to, the thermal limits that have been observed in other Antarctic marine ectotherms [9,10,50]. In general, these results suggest that temperature–size interactions should be interpreted cautiously when measurements are made close to the physiological limits of performance.

Our experiments testing the temperature–size rule are consistent with previous work [42] in showing that these organisms do not meet the predictions of the oxygen–temperature hypothesis; limitations of oxygen diffusion on performance do not increase with body size, regardless of whether oxygen is manipulated in isolation (as in Woods et al. [42]) or when metabolic oxygen demand increases with rising temperature (present study). One way to get around the constraints that the limitation of oxygen diffusion sets on body size is to increase the SA-to-volume ratio as body size increases [17,18,23]. In the case of our two species, the scaling of SA to mass was very close to geometric isometry for C. megalonyx (0.62) and not much greater than isometry for A. glacialis (0.78), similar to the interspecific relationship between SA and mass reported in Woods et al. [42]. Because pycnogonids acquire oxygen across the cuticle [38–40], an alternative mechanism could be to increase the permeability of the cuticle to oxygen diffusion as body size increases. We found that CT, pore volume and AP all increased with body mass for both A. glacialis and C. megalonyx. These results match an interspecific comparison [40] demonstrating that larger-bodied species of sea spiders have proportionally greater pore volume than smaller species. Our results confirm that these patterns also occur across body sizes; as animals grow, the increasing metabolic oxygen demand of a larger body and the longer diffusion distances within the body are offset by increasing cuticular porosity. This increase in porosity probably allows pycnogonids to maintain their activity levels as they grow and may also in part explain why large-bodied pycnogonids are able to cope well with elevated temperatures (though other mechanisms may also be at play, such as ontogenetic changes to the circulatory system or a reduction in the impact of boundary layers with increased size [17,55,56]). The increase in porosity with body size is also a potential example of ontogenetic symmorphosis, in that it represents a change in body structure that meets the increasing functional demands associated with larger body size.

Although both species showed an increase in porosity with body size, we also found differences in cuticle porosity between A. glacialis and C. megalonyx. Overall, the cuticle of C. megalonyx was considerably more porous than A. glacialis, and this difference became more pronounced at larger body sizes. All else being equal, this suggests that C. megalonyx has a greater capacity for gas exchange, and therefore may be capable of higher levels of aerobic activity than A. glacialis. Indeed, in greater than 200 h of casual observation in the laboratory and field, we most often saw C. megalonyx walking across the substrate, while other species were more stationary. Likewise, C. megalonyx is capable of rapid and active capture of pelagic prey [35], and maximum escape speeds of C. megalonyx were almost twice as fast as those of A. glacialis (0.023 cm s⁻¹ versus 0.013 cm s⁻¹) [57]. Greater cuticle porosity may be one mechanism that allows C. megalonyx to evolve both large size and maintain high levels of activity.

Porosity also scaled more steeply with size in C. megalonyx than in A. glacialis. This suggests that larger C. megalonyx may have a greater capacity for maintaining aerobic activity than small ones, which is consistent with the overall increase we found in righting performance with body size, although other factors might have affected this as well. Ecologically, larger body size may confer a fitness benefit on C. megalonyx because larger, more mobile animals have increased probabilities of encountering and capturing food [35]. The disproportionate increase in porosity with size may also reflect an ontogenetic shift in ecological feeding modes from comparatively stationary, parasitic juveniles to more mobile and actively predatory adults. Mobile, active organisms are likely to have a higher aerobic scope than less active ones, which may confer greater thermal tolerance and a greater capacity to cope with warming sea temperatures [9]. Linkages between cuticle structure, functional performance and behaviour of C. megalonyx versus A. glacialis strongly suggest that the two species inhabit different niches in the Antarctic benthos and may have different vulnerabilities to warming oceans.

Like other Antarctic ectotherms, pycnogonids are sensitive to elevated temperatures; but we did not find evidence that larger-bodied pycnogonids were more strongly affected than small ones. We propose that Antarctic pycnogonids can attain giant sizes not only because their metabolic rates are limited by cold temperatures, but also because the porosity of the cuticle increases as they get larger. Future ocean warming will undoubtedly have profound effects on Antarctic marine organisms and ecosystems. However, even among these polar giants, whose large body size is thought to confer a particular vulnerability to climate change [18,42], predicting ‘winners’ or ‘losers’ requires a more nuanced, whole-organism approach that integrates across many levels of a species' ecology, life history and physiology.

Data accessibility

This article has no additional data.

Authors' contributions

C.M.S., H.A.W., A.L.M. designed the experiments. C.M.S. carried them out. C.M.S., M.A.T. and A.L.M. performed cuticle analysis. C.M.S., A.L.M., H.A.W., S.J.L. and B.W.T. wrote the manuscript. All authors gave final approval for the manuscript.

Competing interests

We have no competing interests.

Funding

Funding was provided by NSF grant nos. PLR-1341476 to A.L.M. and PLR-1341485 to H.A.W. and B.W.T.