Adaptation to simultaneous warming and acidification carries a thermal tolerance cost in a marine copepod

Abstract

The ocean is undergoing warming and acidification. Thermal tolerance is affected both by evolutionary adaptation and developmental plasticity. Yet, thermal tolerance in animals adapted to simultaneous warming and acidification is unknown. We experimentally evolved the ubiquitous copepod Acartia tonsa to future combined ocean warming and acidification conditions (OWA approx. 22°C, 2000 µatm CO₂) and then compared its thermal tolerance relative to ambient conditions (AM approx. 18°C, 400 µatm CO₂). The OWA and AM treatments were reciprocally transplanted after 65 generations to assess effects of developmental conditions on thermal tolerance and potential costs of adaptation. Treatments transplanted from OWA to AM conditions were assessed at the F1 and F9 generations following transplant. Adaptation to warming and acidification, paradoxically, reduces both thermal tolerance and phenotypic plasticity. These costs of adaptation to combined warming and acidification may limit future population resilience.

1. Introduction

The unprecedented rate of global climate change is expected to yield atmospheric CO₂ levels of 1000 µatm by the end of 2100 [1], and 2000 µatm by 2300 [2] generating temperature increases of 3–5°C and pH levels between 7.8–7.6 [1]. Evolutionary adaptation provides a mechanism to mitigate deleterious effects of climate change [3]. Yet, adaptation may incur costs. Here, we are concerned with adaptation costs arising from reduced phenotypic plasticity—the ability for phenotypes to respond to environmental change without changes to the genome—which ultimately may reduce fitness [4,5]. These adaptation costs can be demonstrated through reciprocal transplant experiments, which can identify potential fitness penalties in non-native environments [6–8]. Selection can limit phenotypic plasticity [9–12], either by genetic assimilation [13–15] or other mechanisms, but can also support the maintenance of phenotypic plasticity by phenotypic or genetic accommodation [16–18].

Thermal tolerance—the temperature that represents an organism's ability to handle thermal stress (i.e. CT_max, CT_min, LD₅₀ [19])—should reflect environmental temperature [20,21]. The projected combination of warming and acidification, however, may result in novel environments for marine species [22,23]. Expanding studies into multi-generational, multi-stressor experiments that provide a more comprehensive view of future conditions is necessary to assess climate change adaptation and its costs [24]. Copepods, likely the most abundant metazoans [25,26], provide a critical link in marine food webs between primary producers and upper trophic level consumers, contributing greatly to biogeochemical cycles [27]. They also have short generation times, making them excellent candidates for evolutionary adaptation studies [28]. Despite their experimental potential, most copepod studies separately evaluate evolutionary or plastic effects of temperature [11,29–31] and acidification [32–37]. Previous studies addressing both effects on metazoans are rare, limited to few generations, or have not considered potential costs of adaptation [24,38,39].

We took advantage of a separate study to examine transgenerational response of a copepod species (25 generations) to combined ocean warming and acidification (OWA) conditions [39,40]. Here, we extended the study to 65 generations (approx. 2 years) and evaluated thermal performance (the thermal reaction norm for survival across a temperature range [41]) relative to cultures raised for the same calendar time frame under average ambient, present-day conditions (AM). We then reciprocally transplanted both treatments and re-evaluated thermal performance to assess costs of adaptation to OWA and explore effects on developmental phenotypic plasticity. Treatments that were transplanted from OWA to AM, denoted as OWA_AM, were evaluated twice: once at the initial transplant generation (F1_T), and again after nine generations in the AM treatment (F9_T) to evaluate transgenerational effects and their potential for evolutionary rescue [3,42]. We previously identified costs to OWA adaptation with a loss of plasticity after three transplant generations and measured life-history traits to estimate overall population fitness [39]. Here, we repeated the transplant but extended it to nine generations to identify if costs were sustained past three generations and to evaluate costs separate from conventional life-history traits.

2. Material and methods

(a) . Study organism and culturing

Acartia tonsa (Copepoda: Calanoida) is an estuarine and coastal foundational species [25,26], used in previous experimental evolution studies [28,33,40]. Acartia tonsa was collected June 2016 from Long Island Sound CT USA (41.3 N, −72.0 W). Copepods were cultured in the laboratory for at least three generations prior to the start of the transgenerational experiment to minimize the influence of parental effects and acclimation to field conditions on observed patterns [43]. Copepods were fed every 48–72 h at food-replete concentrations of carbon (greater than 800 µg C l⁻¹) on a mixed prey phytoplankton diet [44]. Parental lineages were acclimated to one of two experimental conditions: (i) AM—ambient temperature (18°C), ambient CO₂ (400 µatm CO₂, pH ∼ 8.2), (ii) OWA—high temperature (22°C), high CO₂ (2000 µatm CO₂, pH ∼ 7.5). About 400 females and 200 males were used to seed each treatment, yielding approximately 15 000 eggs and newly hatched nauplii to start the F1 generation. During the first 30 generations, copepods for each treatment were separated into four replicate 10 l culture containers kept in separate temperature-controlled incubators per treatment. After generation 30, cultures were moved to a different location for long-term maintenance. Culture volume was increased to 18 l with containers kept in one of two large water baths. Temperatures in this new location were controlled with aquarium heaters (JBJ Aquariums, CA, USA) and temperature chillers (Aqua Logic, San Diego, CA, USA). Elevated CO₂ levels for all 65 generations were maintained with two-gas proportioners (Cole-Parmer, Vernon Hills, IL, USA) combining air with 100% bone dry CO₂ that was delivered continuously to the bottom of each replicate culture for the entire experiment. Target pH values were monitored daily using a pH probe (Orion Ross Ultra pH/ATC Triode with Orion Star A121pH Portable Meter; Thermo FisherScientific^®, Waltham, MA, USA). Continuous bubbling maintained dissolved oxygen levels at greater than 8 mg l⁻¹. Copepod water was changed every generation (approx. 10–12 days for OWA, approx. 14–16 days for AM) for the first 30 generations, and every two weeks thereafter. Media were replaced with pre-acidified, temperature-acclimated water for respective cultures.

(b) . Transplants

For transplants, eggs and newly hatched nauplii representing half of each culture were moved into new developmental environments at the start of a new generation. Each transplant contained greater than 1500 individuals per replicate, minimizing chances of genetic drift or bottlenecks on performance [45]. Transplant between AM and OWA conditions yielded four replicates for each of two new treatments: AM in OWA (AM_OWA) and OWA in AM (OWA_AM). A schematic of the experiment and transplant is highlighted in electronic supplementary material, figure S1. Copepods were fed and media replaced as described above.

(c) . Heat stress experiments

Acute heat stress experiments were done at 10 temperatures between 18 and 38°C at 2°C intervals following previous protocols [30,46,47]. Individual copepods were placed in 1.5 ml microcentrifuge tubes with 1.5 ml of 0.22 µm filtered Long Island Sound seawater. Copepods in OWA developmental environments were evaluated with pre-acidified water in a custom plexiglass enclosure with elevated atmospheric CO₂ to maintain concentrations of 2000 µatm CO₂ during the 24 h experiment. Copepods in AM developmental environments were evaluated at ambient pCO₂. For temperatures ranging between 22 and 32°C, 14 individuals (seven females and seven males) per replicate culture (N = 4) per treatment were evaluated. For temperatures ranging from 18 to 20°C and 34 to 38°C, 14 individuals (seven females and seven males) from two of the four replicates per culture were evaluated. The different number of individuals at different stress temperatures allowed us to use more temperatures across a larger range. OWA_AM transplants were evaluated twice: at the initial transplant generation (F1_T) and after nine generations (F9_T). The number of copepods evaluated totalled 2224 individuals. Survivorship was recorded under a dissection microscope as the response to physical stimuli or visible gut-passage movement and scored as binary data (1, survival; 0, mortality).

(d) . Analyses

All analyses were performed in R (v. 3.6.1). For survival, we constructed generalized linear mixed-effects models using the lme4 package [48] to test for effects of lineage, developmental environment, generation, sex and stress temperature on survivorship (survivorship∼lineage × dev.treatment × generation × sex × temp) with culture replicates included as a random factor. We created secondary models of specific data subsets based on lineage or developmental environment including the same fixed and random effects mentioned previously. We used post hoc ANOVAs on these secondary models to evaluate the effects of an environment on a lineage, or to compare survival across different lineages. Generational effects were tested in the OWA lineage-specific model. Comparisons across lineages (i.e. OWA to AM) tested for effects of adaptation. Comparisons within lineages across developmental environments (i.e. OWA_AM to OWA_OWA) tested for effects of developmental plasticity. Generalized linear models were constructed for each treatment/developmental environment/generation combination to calculate the LD₅₀ (temperature at which there is 50% mortality) and maximum survivorship (i.e. the highest survivorship at any temperature). ‘Thermal tolerance’ was estimated as LD₅₀ with accompanying standard error, as calculated using the MASS package [49].

3. Results

(a) . Costs linked to adaptation

Copepods adapted to OWA conditions (OWA_OWA; figure 1a), surprisingly, displayed a thermal performance curve shifted towards colder temperatures, which was significantly lower than AM_AM due to lowered high-temperature performance (figure 1a) (p < 0.01, d.f. = 1, χ² = 16.53). This is also reflected in the thermal tolerance of OWA_OWA, which is the lowest of all treatments (OWA_OWA LD₅₀ = 27.32 ± 0.31°C; AM_AM LD₅₀ = 29.34 ± 0.35°C; AM_OWA LD₅₀ = 28.13 ± 0.53°C; OWA_AM F1_T LD₅₀ = 28.12 ± 0.42°C; OWA_AM F9_T LD₅₀ = 29.84 ± 0.62°C; figures 1b and 2a). Thermal performance of copepods from ambient conditions (AM_AM), by contrast, was shifted furthest towards warmer temperatures out of any treatment (OWA_OWA: p < 0.01, d.f. = 1, χ² = 16.53; OWA_AM: p = 0.01, d.f. = 1, χ² = 6.38; AM_OWA: p < 0.01, d.f. = 1, χ² = 10.21; figure 1a) and experiences the highest thermal tolerance of any treatment (figure 2a).

Figure 1.

Thermal performance curves. (a) Thermal performance of AM (blue) and OWA (red) lineages among developmental environments during the F1 generation. AM_AM (blue solid line) is significantly different from all other F1 treatments (OWA_OWA (red dashed line): p < 0.01, d.f. = 1, χ² = 16.53; OWA_AM (red solid line): p = 0.01, d.f. = 1, χ² = 6.38; AM_OWA (blue dashed line): p < 0.01, d.f. = 1, χ² = 10.21). (b) Thermal performance comparing F1_T (yellow) and F9_T (grey) OWA_AM transplant treatments. Model for F9_T is significantly different from OWA_OWA and F1_T OWA_AM (p < 0.03, d.f. = 1, χ² = 5.08).

Figure 2.

Thermal tolerance and maximum survivorship for AM and OWA treatments at F1. (a) LD₅₀ stress temperatures evaluated for AM and OWA lineages in the respective treatments and reciprocally transplanted treatments. (b) Maximum survivorship across all temperatures for AM and OWA lineages. Colours represent lineage. Error bars represent standard error of the model. The low OWA_OWA thermal tolerance reflects the lowest thermal performance of any treatment (p = 0.01, d.f. = 1, χ² = 6.26), while AM_AM represents the highest (p = 0.01, d.f. = 1, χ² = 6.62).

(b) . Transgenerational effects

Thermal tolerance (LD₅₀) for the OWA treatment was partially recovered after development in AM conditions for multiple generations (figure 2b). After one generation, OWA_AM (figure 1a) copepods experienced no significant change in thermal performance relative to OWA_OWA (figure 1a; p = 0.11, d.f. = 1, χ² = 2.54). However, after nine generations, we observed a significant increase in high-temperature thermal performance relative to that of the OWA lineage (OWA_OWA and OWA_AM) or the AM_OWA transplant (p < 0.03, d.f. = 1, χ² = 5.08) but like that of AM_AM (p = 0.49, d.f. = 1, χ² = 0.46). Additionally, a significant stress temperature × lineage interaction effect for OWA_AM F9_T is observed relative to AM_AM (p < 0.002, d.f. = 9, χ² = 26.23), and a significant stress temperature × generation effect relative to OWA_AM (p < 0.0001, d.f. = 1, χ² = 26.82). This suggests a change in the shape of the survivorship curve, resulting from increased mortality at low temperatures.

(c) . Developmental phenotypic plasticity

The developmental environment effect for a lineage tests for developmental phenotypic plasticity. Copepods in the AM lineage showed developmental phenotypic plasticity—AM_OWA thermal performance was lower than AM_AM (p < 0.01, d.f. = 1, χ² = 10.21; figure 1a). Copepods from the OWA lineage did not—OWA_AM and OWA_OWA had similar thermal performance (p = 0.11, d.f. = 1, χ² = 2.54; figure 1a). Both transplant treatments (AM_OWA and OWA_AM) had the same thermal performance (p > 0.1, d.f. = 1, χ² = 0.16), with values significantly lower than AM_AM (p < 0.005, d.f. = 1, χ² = 8.01), but not significantly different from OWA_OWA (figure 2a). Moreover, AM_OWA showed lowered maximum survivorship relative to AM_AM (p < 0.01, d.f. = 75.49, t = 3.16; figure 2b), but the maximum survivorship for OWA_AM at the F1_T generation was not significantly different from OWA_OWA (figure 2b).

(d) . Sex effects

We observed no sex effects on the thermal performance of A. tonsa.

4. Discussion

(a) . Costs of adaptation

Copepods adapted to elevated temperatures and CO₂ should perform better at higher temperatures, as has been reported for populations adapted to warmer temperatures [50]. A decrease in thermal performance of these copepods relative to those in ancestral conditions (cooler temperature, ambient CO₂) reveals a cost to adaptation [7]. High-temperature thermal performance of copepods under OWA conditions decreased relative to AM_AM in all cases. We previously observed that copepods raised under warming conditions increase thermal performance [30,50]. Here, exposure to acidification and warming decreases thermal performance, likely due to increased energetic costs of acid–base regulation to maintain homeostasis in a lower pH environment [51]. Across the stress temperatures evaluated, copepods adapted to OWA conditions exhibit a decrease in maximum survivorship in addition to a shift towards cooler temperatures. The reduced thermal tolerance for OWA copepods could also be explained by antagonistic interactions of warming and acidification, which we observed after 25 generations of adaptation [40]. Copepods from OWA lineages (OWA_OWA and OWA_AM) and those that developed in OWA conditions (AM_OWA) show significantly lower high-temperature thermal performance relative to AM_AM, with the OWA_OWA treatment exhibiting the lowest LD₅₀ (figure 2a). Meanwhile, the AM_AM performance curve was the furthest towards warmer temperatures, which reflects a typical thermal tolerance for A. tonsa collected from Connecticut raised at 18°C [30].

(b) . Transgenerational effects

Transgenerational effects for OWA copepods returning to ambient conditions are evident via changes between F1_T and F9_T (figure 1b). F9_T cultures show an increase in thermal tolerance relative to the F1_T cultures and resemble the AM_AM thermal tolerance values. Moreover, comparing models for F1_T and F9_T reveals a significant temperature × generation effect, suggesting differential thermal performance across time with increased high-temperature performance at F9_T. This suggests a relatively fast recovery after returning to ambient conditions, consistent with some degree of evolutionary rescue. Previously, we observed signals of selection for OWA_AM transplants over multiple generations consistent with genetic changes rather than plasticity [39], which could explain the thermal performance patterns we observe here. Further analysis of F1_T and F9_T performance curves suggests lower maximum performance for F9_T cultures than for F1_T (figure 1b), which is consistent with a trade-off: LD₅₀ improves at the expense of maximum survivorship. Thus, while tolerance of warm temperatures post-transplantation may increase after nine generations of habitation, tolerance of cold temperatures decreases, which represents costs associated with OWA adaptation.

(c) . Loss of plasticity

Increased environmental temperature can increase thermal performance via both adaptation and plasticity [20,21]. Transplant experiments that include a return to ancestral conditions can be used to test for local adaptation [6–8], or maintenance of plasticity [52]. Comparisons among transplant treatments can also be used to assess for additional adaptation costs [7]. First, we consider how well populations maintained developmental phenotypic plasticity by comparing thermal performance between developmental conditions for each lineage. In our experiment, thermal performance changed only for copepods from the AM lineage exposed to OWA conditions. We have previously observed AM lineages maintain transcriptional plasticity when transplanted to OWA conditions [39]. This plastic response is also reflected here in the similar thermal tolerance and lowered maximum survivorship for the AM_OWA treatment relative to OWA_OWA (figure 2).

In contrast with the AM lineage, we observe no change in thermal performance for copepods from the OWA lineage. Other work has shown that as populations adapt to environments closer to their thermal limit, the potential for plasticity decreases [9–11,30,50]. If plasticity were sustained and thermal performance reflected developmental temperature, we would expect thermal performance of OWA_AM to decrease relative to OWA_OWA. Our results suggest that warming and acidification combined lead to both lower plasticity and lower thermal tolerance. These results are also consistent with genetic assimilation [13–15] because copepods from the OWA lineage do not change thermal performance with developmental environment [9–12,50]. Consequently, the thermal regime alone does not determine thermal tolerance or developmental plasticity in this species when CO₂ is elevated. Forecasting changes in thermal performance in a warmer, more acidic environment should account for this interaction between stressors; predictions based solely on temperature changes may overestimate a population's capacity to respond via both adaptation and plasticity. Future work should evaluate whether adaptation to alternate multiple stressors exacerbates this loss of plasticity.

5. Conclusion

We provide clear evidence that thermal performance in this species is affected by multiple environmental factors, not just temperature regimes. In this study, adaptation to warming and acidification reduces thermal tolerance. All lineages that were raised in OWA conditions exhibit lowered thermal tolerance relative to their counterparts in ambient conditions. Further, consistent with previous experiments, the OWA lineage loses phenotypic plasticity [39] and expresses constitutive performance consistent with genetic assimilation [13–15]. After nine generations in ambient conditions, the OWA lineage thermal performance is partially recovered, but this results in a trade-off of reduced survival at lower temperatures and a lower maximum survivorship. Thus, adaptation to OWA conditions is costly. We present evidence that (i) adaptation to warming and acidification incurs costs in the form of reduced phenotypic plasticity and thermal performance, (ii) developmental temperature does not predict thermal performance in the presence of acidified conditions, (iii) climate change conditions may, hence, limit evolutionary rescue. With future climate scenarios approaching rapidly, adaptation to simultaneous warming and acidification may limit population resilience.

Data accessibility

All data and code for analysing and visualizing the data are deposited into a Zenodo repository located at: https://doi.org/10.5281/zenodo.4480338 [53].

Authors' contributions

J.A.d. designed the study, analysed the data, created figures, drafted and critically revised the manuscript. A.G. performed the experiments and revised the manuscript. M.C.S. designed the study and critically revised the manuscript. H.G.D. conceived the study, co-drafted the manuscript and critically revised the manuscript. All authors gave final approval for publication and agree to be held accountable for the work performed therein.

Competing interests

The authors declare no competing interests.

Funding

This work was supported by NSF OCE-1559180, NSF (REU) 1658742, Connecticut Sea Grant R/LR_25, and the UConn Crandall Cordero Graduate Fellowship.