Climate-related environmental stress in intertidal grazers: scaling-up biochemical responses to assemblage-level processes

Study system

The study was carried out along the rocky coast south of Livorno (Italy, Western Mediterranean Sea; 43°28^′02N, 10°22^′19E) from March to May 2009.

Mediterranean rocky intertidal habitats are extremely variable environments due to the limited amplitude of tides. Barometric pressure is the primary force determining the position of the sea level, largely dictating the timing and duration of aerial exposure of organisms along the intertidal gradient. Changes in barometric pressure can therefore induce abrupt fluctuations in thermal conditions, particularly in the high-shore habitat, where temperature in Spring can vary over a range of more than 15°C and reach values up to 30°C (Benedetti-Cecchi et al., 2006).

Differences among algal assemblages at different heights on the shore are mainly due to changes in relative abundance of species. Low-shore assemblages are dominated by encrusting, filamentous and coarsely branched algae. In contrast, barnacles (mainly Chthamalus stellatus (Poli), but also C. montagui (Southward)) and cyanobacteria (Rivularia spp.) dominate the high-shore (Benedetti-Cecchi, 2000; Benedetti-Cecchi, 2001). Among most common grazers, Patella ulyssiponensis (Gmelin) is characterized by a relatively wide vertical range of distribution (from mid/high- to low-shore habitats). It is a generalist herbivore (Della Santina et al., 1993), feeding either on small-sized macroalgae, such as filamentous ones, or on the microscopic components of biofilm colonizing apparent bare rock at all heights (cyanobacteria, diatoms or macroalgal spores). A recent study indicated that Mediterranean limpets already exist on the edges of their thermal tolerance windows. In particular, P. ulyssiponensis, although characterized by a low metabolism, is very sensitive to exposure to elevated temperatures (Prusina et al., 2014), as those naturally experienced at the study site (Benedetti-Cecchi et al., 2006).

Experimental design and field sampling

We imposed increasing levels of desiccation stress by confining individuals of P. ulyssiponensis at increasing heights on the shore within their natural range of vertical distribution (from 0 (low-shore) to 15 (mid-shore) or 30 cm (high-shore) above MLWL), by means of 17 × 17 cm fenced plots. By precluding limpets to move down the shore for a period of about two months, we simulated a scenario of extreme thermal stress that can be expected under climate change. The in situ estimated difference in daily average temperature between low- and high-shore for the study period was 1.3°C (Benedetti-Cecchi et al., 2006), within the range of expected increases in extreme surface air temperature for the period 2016–2035 (Intergovernmental Panel on Climate Change, 2013). For each height on the shore, three replicate fenced plots were positioned within each of three areas (stretches of coast extending about 3 m alongshore and 10 s of meters apart). Limpets were taken from the low-shore with blunt metal sheets to avoid breaking the shell and were immediately transplanted to plots (n = 2 limpets per plot) by gently pressuring them against the rock for some minutes to facilitate their attachment. Each individual was marked with blue epoxy putty (Subcoat S, Venziani) either in the centre or at the margin of the shell, for subsequent identification. Length of the major axis of the shell was measured by means of a plastic calliper (size range: 13–23 mm, length of major axis of the shell). To control for possible artefacts associated with the manipulation of grazers, additional pairs of limpets were detached, marked and placed back to their original position from each of 3 open plots (17 × 17 cm) within each of 3 additional areas in the low-shore habitat. Occasionally missing individuals from fenced plots were replaced by new ones to maintain approximately constant limpet density during the experiment.

At the end of the experiment, marked limpets were collected, measured, and maintained at −80°C for subsequent analysis. Percentage cover of filamentous algae (their main macroscopic resource) within plots was visually estimated with a plastic frame of 15 × 15 cm, divided into 16 sub-quadrats. A score from 0 (absence) to 4 (full cover) was given in each sub-quadrat and a final estimate was obtained by summing individual scores over the 16 sub-quadrats (Dethier et al., 1993), for a maximum score of 64. Values were then expressed as percentages.

Biochemical analyses

Frozen limpets were shelled, weighed, cut in small pieces and homogenized with a hand-driven Potter Elvejhem glass homogenizer in 5 vol of 50 mM Tris HCl (pH 7.4). The homogenate was centrifuged in a Millifuge centrifuge (Millipore) for 30 s to remove debris. The supernatant (crude extract) was divided into two aliquots. An aliquot was used without further treatment for the assay of catalase and glutathione reductase; the other aliquot was immediately acidified to pH 2 by the addition of 4 N HCl in order to preserve glutathione in the extract, centrifuged again and used for the measurement of glutathione concentration.

The activity of catalase was measured at 25°C in accordance with Luck (1965), by following the decrease in absorbance at 240 nm (Δε = 0.0436 mM⁻¹ cm⁻¹). The reaction mixture contained, in 50 mM sodium phosphate buffer (pH 7.4), 12.5 mM hydrogen peroxide.

The activity of glutathione reductase was determined at 30°C in accordance with Racker (1955) by following the decrease in absorbance at 340 nm (Δε = 6.22 mM⁻¹ cm⁻¹). The reaction mixture contained, in 160 mM potassium phosphate buffer (pH 7.4), 0.5 mM oxidized glutathione, 0.11 mM NADPH and 1 mM EDTA. The activities of catalase and glutathione reductase were normalized for protein concentration.

Total glutathione concentration was measured by using a colorimetric end point coupled enzymatic assay (Cappiello et al., 2013), based on the measurement of cysteine produced from glutathione by γ-glutamyltransferase and leucyl aminopeptidase. Briefly, the standard incubation mixture (250 µL final volume) contained 8 mM MgCl₂, 0.2 mM MnCl_2,2 mM dithiothreitol, 40 mM Gly-Gly, 50 mU/mL γ-glutamyltransferase and 50 mU/mL leucyl aminopeptidase in 32 mM Tris HCl pH 8.5. The reaction was started by the addition of limpet extracts. After 30-min incubation at 37°C, the reaction was stopped with the addition of 12.5 µL of 100% (w/v) trichloroacetic acid and the incubation mixture was centrifuged at 12,000 ×g for 1 min in a Beckman Microfuge E. The cysteine formed was then evaluated spectrophotometrically. An aliquot of 200 µL of the supernatant was added to 200 µL of glacial acetic acid and 200 µL of a reagent, prepared by dissolving 250 mg of ninhydrin in 10 mL of glacial acetic acid/4 M HCl (3:2). The mixture was placed in a boiling bath for 4 min. Under these conditions, cysteine specifically reacted with ninhydrin to give a pink colored complex. After cooling on ice, 300 µL of the mixture were diluted with 300 µL of 95% ethanol and the absorbance at 560 nm measured. A reference standard curve constructed with known concentrations of cysteine was used for the detemination of cysteine in samples.

Protein concentration was estimated by the Coomassie Blue binding assay (Bradford, 1976), with bovine serum albumin as the standard. Total glutathione concentration was normalized for wet weight of each individual limpet analyzed.

Data analyses

Values of catalase, glutathione reductase, total glutathione concentration and growth rate of limpets were analysed by means of mixed-effect models. Growth rate was calculated as (Length_FINAL−Length_INITIAL)/ Length_INITIAL. Areas and plots nested in areas were included as random effects into the model. A first set of analyses examined possible artefacts associated with the experimental procedure, by comparing limpets from fenced plots in the low-shore habitat and those relocated inside or near open plots at the same height on the shore (i.e., marked and disturbed in the low-shore habitat; PC, procedural control) (Fences vs. PC, fixed effect). As no significant difference emerged for any response variable (Table 1), these replicates were merged into a single low-shore level treatment, to increase sample size. Data were then analysed by including height on the shore as a continuous predictor variable (i.e., with the low-, mid- and high-shore treatments corresponding to values of 0, 15 and 30 (cm above MLWL), respectively) (Height, fixed effect), (Table 2). The newly introduced limpets compensating for missing ones remained exposed to experimental conditions for shorter periods than the original animals. To control for this effect, the previous analysis was also performed by including duration to treatment conditions as a covariate (Permanence, fixed effect). For each variable, this new model was compared to the reduced model without the covariate, by means of a log-likelihood ratio test. The effect of the covariate was never significant, nor did it significantly increase the goodness of fit of any of the models examined (Table 3). Analyses were run using the “lmer” function (able to deal with unbalanced data) from “lme4” package (R v2.15.3; R Development Core Team, 2013). Normality and homoscedasticity of data was assessed by normal QQ-plots and by plotting residuals vs. fitted values.

An index of ‘per capita interaction strength’ quantifying the effect of limpets on their macroscopic resources (filamentous algae) was calculated for each height on the shore. As P. ulyssiponensis is generalist species, we assumed the consumption of filamentous algae was a good estimate of the general feeding activity of these limpets. We used a modified version of the index proposed by Paine (1992). In the original formula, the relative change in abundance of a resource in presence or absence of a consumer is standardized against consumer density. In our case, the index was calculated by comparing enclosure and natural plots, separately for each height on the shore. Our study system, however, is characterized by an extremely patchy distribution of algae and limpets, particularly in the high-shore habitat, making estimates of mean consumption rates from natural plots quite challenging. To take into account this feature, we opted for using long term data from ancillary studies conducted at the same site, to derive expectations of the mean values of abundance of the resource and density of consumer in control plots (C and d_C, respectively) at each tidal height. We then calculated the relative change in percentage cover of filamentous algae between enclosures (E) and natural plots in the surrounding environment (C), divided by the difference in density of grazers between the two conditions (d_E, equal to 2 individuals of P. ulyssiponensis in each fence, and d_C, lower than 2 individuals at all heights): ${\displaystyle \frac{E-C}{C\left(d_E-d_C\right)}}$ The lower the interaction strength, the greater the impact of an individual limpet within enclosures compared to the surrounding open rock surface. Mean estimates and 95% confidence intervals of per capita effects were estimated by bootstrapping the original data 10,000 times separately for each tidal height. Analysis was run using the “boot” package (R v2.15.3, R Development Core Team, 2013).