Influence of respiratory mode on the thermal tolerance of intertidal limpets

Predicting ecological responses to climate change requires an understanding of the mechanisms that influence species’ tolerances to temperature. Based on the idea that air and water breathing animals are differentially suited to life in either medium due to differences in their respiratory morphology, we examined the possibility that the thermal tolerances of co-existing intertidal pulmonate and patellogastropod limpets may differ in different breathing media. We tested this by determining each species’ median lethal temperature (LT50) and cardiac Arrhenius breakpoint temperature (ABT) as measures of upper thermal tolerance limits, in air and water. Although all these species can survive in air and water, we hypothesised that the pulmonate limpets, Siphonaria capensis and S. serrata, would have higher thermal limits than the patellogastropod limpets, Cellana capensis and Scutellastra granularis, in air and vice versa in water. The results did not support our hypotheses, since C. capensis had similar thermal tolerance limits to the pulmonate limpets in air and the pulmonate limpets had thermal tolerance limits similar to or higher than S. granularis in water. Thus, considering pulmonate and patellid limpets as groups, we found no differences in their collective upper thermal tolerance limits in either medium. We conclude that differences between these two limpet groups in their respiratory morphology do not influence thermal tolerance, but that tolerances are species-specific.


Introduction
Climate change is a statistically significant change in the long-term state of the global climate, caused by a combination of natural and external anthropogenic activity [1]. One of the most important consequences of climate change is the perceived change in environmental temperatures, which are likely to have numerous consequences for ecosystem level processes [2][3][4][5][6]. It has, therefore, become important to improve our understanding of the impact that climate change may have on individual organisms and thus overall ecosystems [7]. The rocky shore and its inhabitants have frequently been used to examine the ecological effects that global PLOS  low tide by quickly sliding a coarse scalpel between the muscular foot and the substratum. Any limpets not detached at the first attempt were left, to avoid using animals that may have been injured. Specimens were transported to the laboratory within 3 hours in small containers, moistened with sea water and kept inside an insulated box. In the laboratory, the specimens were housed in a 20L glass tank filled with 5L of aerated seawater at 22˚C for a minimum of 24h and a maximum of 48h before use. Before experimentation, limpets were submerged in 500mL containers filled with constantly aerated seawater for 1 hour to ensure they were fully hydrated.

Determining the median lethal temperature (LT 50 )
Thermal limits were determined first by finding the LT 50 values, using a protocol based on Clarke et al. [43]. LT 50 measurements were carried out on 150 individuals per species in each medium, over three trials (50 individuals/trial) to generate a mean LT 50 value. During the experiment, limpets were housed in 500mL containers (10 individuals/container) filled with natural aerated seawater to simulate aquatic conditions or dampened with seawater to simulate aerial exposure [44]. Temperatures within the containers were controlled by submerging them in a Grant programmable water bath (GP 200, Grant, Germany). A Fluke 54II thermometer (Fluke cooperation, USA) fitted with a T-type thermocouple (Fluke cooperation and Cromega) was used to measure temperature at the bottom of the containers, which were recorded with a PowerLab recording system. A wide range of heating rates (10˚C/min-1˚C/3.5 days) have been used to determine thermal limits in past studies [45]. Similar studies on various intertidal organisms, including limpets, have generally used heating rates between 0.1˚C.min -1 and 0.3˚C.min -1 [35,36,46,47]. In this study, temperature was programmed to increase at 0.2-0.4˚C.min -1 for the LT 50 measurements (S1 Table) using a water bath, following the ramping protocol described in S3 Fig. To avoid influencing the results through repeated thermal shock [48, 49] a new batch of specimens was used for each temperature interval. After each test run, limpet mortality was determined after a 24-hour recovery period in the holding tank (S3 Fig) and limpet shell lengths were measured to the nearest 0.02mm using Vernier callipers. Mortality was assessed by probing the limpets for tactile responsiveness using a blunt probe. Limpets were classified as dead if they showed no response to having the foot muscle or the edge of the mantle probed.

Heart rate measurements
Heart rate measurements were carried out on 20 individuals per species in each medium, housed in a total of 56 500mL containers with 3 individuals/container in 48 of the containers, and 2 individuals/container in the rest. To simulate aquatic conditions, the containers were filled with natural seawater aerated using air stones. To determine thermal limits in air, the containers were dampened with seawater to maintain relatively high levels of humidity during heat exposure [44].
Heart rate was recorded using non-invasive plethysmography [50] by attaching optoelectronic (infrared) sensors (Vishay semiconductors, V69 CNY70 732/735, Germany) to the shells of each limpet near the heart using Pattex super glue (Henkel (Pty) Ltd, South Africa). These sensors produced signals which were amplified by a custom-built preamplifier, after which Triangular-Bartlett smoothing was used to produce an additional smooth trace on a separate channel. The signal was then filtered before being recorded as beats per minute on a computerised recording system (PowerLab/4SP and 430, Chart version 5 and 7, ADInstruments, Australia). The amplitude ranged between 40 and 100mV at a sampling rate of 40Hz. The specimens were exposed to a temperature increase of 30˚C from 20-50˚C at a rate of 0.25˚C.min -1 , using the Grant programmable water bath heating system whilst simultaneously recording heart rate. Prior to heat exposure, the animals were allowed to settle at 20˚C in the water bath for 30 minutes. Limpet mortality and shell lengths were determined after each treatment, as described for the LT 50 measurements.

Data and statistical analyses
One caveat here is that individuals in the same container during pre-treatment could be considered to be pseudoreplicates. However, species were interspersed during pre-treatments and monitoring of the water bath temperature showed no significant or systematic variation. Similarly, there was no evidence that the presence of conspecifics influenced individual thermal physiology.
The LT 50 values for each trial were generated from limpet mortality at each temperature interval using probit analysis [51,52]. Thereafter, mean LT 50 values were compared between media and respiratory modes using a nested ANOVA with species nested in respiratory mode.
Cardiac thermal response curves were plotted and checked manually to control for artefacts and anomalous trends, for example due to movement of the animals. This meant that the final ABT and heart rate analyses were carried out on fewer than 40 individuals from each species (S2 Table). The temperature ranges used to determine ABT were 25-45˚C for all species in air (except Scutellastra granularis; 25-40˚C) and 25-40˚C for all species in water (except C. capensis; 25-45˚C). Outside of these temperature ranges the data points were distributed haphazardly.
Arrhenius plots were then generated from these thermal response curves using Eq 1 to analyse the effect of temperature on heart rate: In Eq 1, HR represents the heart rate (bpm), a is the normalization constant, E a is the activation energy (J.mol -1 ), R is the ideal gas constant (J.K -1 .mol -1 ) and T is the absolute temperature (K). Piecewise linear regression was used to calculate the breakpoints using transformed heart rate (L n (HR)) and temperature (1/T) data. These breakpoints were then converted back into degrees Celsius, to present them as ABTs (˚C) before conducting further analyses. The ABT values were then compared using a nested ANOVA as described for the LT 50 values above.
Once the ABT values were determined, the increase in heart rate as a function of temperature was compared among treatments, by considering only the data points below the ABTs. The slopes (E a /R) from the resultant, linear Arrhenius plots were compared using a nested ANOVA as described for the LT 50 and ABT analyses above.
Tukey HSD analyses were used post hoc to determine where significant differences lay for all the ANOVA models.All statistical analyses were performed with Statistica 13.  Table), indicated significant effects of Respiratory Mode and of Species nested in Respiratory Mode (p < 0.0001 in both cases), with no effect of Medium or its interaction with Respiratory Mode. In both media, Scutellastra granularis had significantly lower values than the other three species, whose LT 50 values were the same, and there were no significant differences between LT 50 values in air and water for any species. The significant effect of Respiratory Mode was not expected on the basis of the raw data and presumably reflects the low LT 50 of Scutellastra granularis, which decreases the mean value for the patellids. This effect is probably exacerbated by the small sample size (n = 3) used in the LT 50 analysis, particularly compared to the ABT (n~10) analysis.

Differences in thermal limits between species and/or media
ABT. The nested ANOVA of ABT data (S4 Table) indicated significant effects of Medium, Species nested in Respiratory Mode and the interaction between Medium and Respiratory Mode (p < 0.0001 in all cases). Other than C. capensis, the limpets had significantly higher ABT values in air than water (Fig 1). Although the effect of Respiratory mode was non-significant in the nested Analysis (p = 0.34), there was a significant effect of Species. Scutellastra granularis exhibited a significantly lower ABT in air than the other species, while C. capensis had a significantly higher ABT in water than the others (p < 0.05 in both cases).

Relationship between heart rate and temperature
The cardiac thermal response curves (Fig 2) and Arrhenius plots (Fig 3) displayed a high degree of inter-individual variability within each species in both media.
The nested ANOVA comparing slopes (S5 Table) revealed insignificant effects for all factors (Fig 4). Respiratory mode and thermal tolerance

Discussion
We hypothesised that differences in the respiratory morphology of the two limpet groups would be reflected in differences in their responses to increasing temperatures. While the nested analysis for LT 50 did show a significant effect of respiratory mode, the results from both the LT 50 and ABT analyses suggested species-specific effects, rather than an over-riding influence of respiratory morphology. A lower aerial temperature tolerance and aquatic LT 50 was measured for Scutellastra granularis compared to the other three species (which did not differ). Similarly, C. capensis had a higher ABT in water than the other three species whose aquatic ABTs were similar. In addition, there was no obvious effect of medium on species-specific LT 50 values, while ABT values were generally higher in air except in the case of C. capensis. There were no significant effects of either respiratory mode or medium on the slopes of the Arrhenius plots.

Aerial exposure
All experiments performed in air showed important differences between species. Scutellastra granularis had the lowest ABT and LT 50 . Conversely, the response of the other patellogastropod was not significantly different to those of the two pulmonates.
The lungs of the Siphonariid limpets were expected to give them a better breathing capability than patellids in air [17,22,53,54,55], making it easier for them to meet their mitochondrial O 2 demands. This in turn should allow for the efficient use of energy stores, a delay in the onset of anaerobic metabolism and relatively high upper thermal limits [25,26,56,57].
However, our results indicate that some patellogastropods, like C. capensis, can survive similar levels of thermal stress to pulmonate limpets during low tide. This may be explained by the fact that some high shore "gill-bearing" limpets can respire efficiently in air despite not having a lung [27,[58][59][60]. This helps reduce the accumulation of anaerobic by products and water loss, allowing for an increase in aerial thermal tolerance [25,61,62]. For example, unexpectedly high thermal limits in air have been measured for the high shore species Cellana toreuma (LT 50 = 41.29-43.36˚C) [63] and C. grata (ABT = 47˚C) [35].

Immersion
Increasing water temperature affected the limpet species differently. The two pulmonates had similar ABT and LT 50 values, while the two patellogastropods reacted differently. Scutellastra granularis had lower thermal limits compared to Cellana capensis, which had an exceptionally high ABT.
Pulmonate limpet accessory gills evolved secondarily after loss of the ctenidium, and are not primarily adapted to aquatic respiration, in contrast to the pallial gills of patellogastropod limpets [22,28]. Despite this, the pulmonate limpets had surprisingly high thermal limits in water, probably due to the efficient use of their accessory gills as shown previously by Koopman et al. [64]. These authors found that, when submerged the freshwater pulmonate limpets Physa fontinalis and P. acuta had higher upper thermal limits (CT max ) than the gill-bearing caenogastropods Bithynia tentaculata and Potamopyrgus antipodarum.

Air vs water
When comparing thermal performance in air and water, there were strong similarities between the two limpet groups. For both groups, there was no significant effect of medium when performance was measured as LT 50 . In the case of the pulmonates, both species showed significantly higher ABT values in air. Among the patellogastropods, the same was true for Scutellastra granularis, but not C. capensis. Regarding the slopes from the Arrhenius plots, the non-significant effects were probably related to the high inter-individual variability (Fig 4) in the species sensitivity to increasing temperature. The slope of the heart rate response represents thermal metabolic sensitivity [17,35,37,55,65], and this has previously been shown to be highly variable inter-individually [66].
Past studies have generally focused on the influence of geographic [37,67] or vertical distribution [35, 36, 68] on the thermal tolerances of intertidal organisms. Most such studies examined thermal tolerance in submersed animals, with only a few examining aerial thermal tolerances or comparing tolerances in different media [36,62,69]. Even fewer studies have compared the aerial thermal limits of air and water breathing gastropod molluscs [but see 26,70,71].
Aquatic animals adapted to aerial respiration (such as pulmonate limpets) should be able to respire more efficiently in air where oxygen concentrations are higher, but the emersion period is also characterized by extreme temperature and desiccation stress involving important energetic costs [25,[72][73][74][75]. Dye [76] found that Siphonaria capensis and S. concinna had higher respiration rates in air than water, indicating that it would be easier to avoid anaerobic metabolism, which should translate into higher aerial thermal limits [77][78][79].
Because patellogastropods respire primarily through their pallial gills, they were expected to have lower thermal limits in air. In the case of LT 50 , there was no difference between media for either Scutellastra granularis or Cellana capensis, while the ABT for S. granularis was unexpectedly higher in air. This has also been shown for the mid-to-high shore patellogastropod limpet Lottia digitalis, which had a higher thermal limit (final cardiac breakpoint temperature) in air [80], and in both cases, this presumably reflects the efficiency of O 2 uptake by the highly vascularised mantle cavity of many high shore patellogastropod limpets.

Conclusions
Although the results differed slightly between LT 50 and ABT, they provide no clear indication that respiratory morphology is important in determining either aerial or aquatic thermal limits in the study species, indicating that other factors play important roles. Within species, reproductive and nutritional state can influence susceptibility to high temperatures, and probably contributed to the high degree of individual variability we observed in the relationship between the limpet heart rates and temperature [81][82][83][84][85]. Recent thermal history, including the influence of microhabitat use, is also likely to influence thermal limits [16,30,74,[86][87][88]. In this regard, body temperature estimates generated from the range of microhabitats occupied by each species would have benefited this study. Nevertheless, our overall conclusion is that the data do not support the hypothesis that the respiratory morphology of these species has an overriding influence on the interaction of thermal tolerance and respiratory medium.