Loss of transcriptional plasticity but sustained adaptive capacity after adaptation to global change conditions in a marine copepod

Abstract

Adaptive evolution and phenotypic plasticity will fuel resilience in the geologically unprecedented warming and acidification of the earth's oceans, however, we have much to learn about the interactions and costs of these mechanisms of resilience. Here, using 20 generations of experimental evolution followed by three generations of reciprocal transplants, we investigated the relationship between adaptation and plasticity in the marine copepod, Acartia tonsa, in future global change conditions (high temperature and high CO₂). We found parallel adaptation to global change conditions in genes related to stress response, gene expression regulation, actin regulation, developmental processes, and energy production. However, reciprocal transplantation showed that adaptation resulted in a loss of transcriptional plasticity, reduced fecundity, and reduced population growth when global change-adapted animals were returned to ambient conditions or reared in low food conditions. However, after three successive transplant generations, global change-adapted animals were able to match the ambient-adaptive transcriptional profile. Concurrent changes in allele frequencies and erosion of nucleotide diversity suggest that this recovery occurred via adaptation back to ancestral conditions. These results demonstrate that while plasticity facilitated initial survival in global change conditions, it eroded after 20 generations as populations adapted, limiting resilience to new stressors and previously benign environments.

Fig. 1. Schematic of the experimental design.

Blue lines are ambient (AM) pH and temperature that represent current conditions (pCO₂: 400 ppm; temperature: 18 °C). Red lines are simulated future warming and acidification (OWA) conditions (pCO₂: 2000 ppm; 22 °C). Adult Acartia tonsa were collected from the wild and reared in the lab for three generations. Six hundred laboratory-acclimated adults seeded each of four replicates at AM and OWA conditions where they were reared for 20 non-overlapping generations. At generation 20, each replicate was split in two and transplanted into the same conditions as the previous 20 generations (AM_AM, OWA_OWA) and to the opposite condition (AM_OWA, OWA_AM). These transplanted lines were reared for three additional generations and sampled for life-history traits at the first generation and genomics at the end of each generation.

Fig. 2. Allele frequency and gene expression divergence after 20 generations of selection.

Principal component analysis of (a) genome-wide variation in allele frequencies (322,595 SNPs) and (b) gene expression (24,927 genes) at the F1 generation. c, d Gene ontology enrichment results from one-sided Mann–Whitney U-test using P-values from Cochran–Mantel–Haenszel tests (allele frequencies, c) and DESeq2 Wald tests (gene expression, d). Gene categories are collapsed for visualization purposes with the number of categories indicated in parentheses. See Supplementary Data 3, 4 and Supplementary Figs. 9 and 10 for full results. Source data are provided as a Source Data file.

Fig. 3. Plasticity in gene expression in response to reciprocal transplant across three generations in transplant conditions.

a Gene expression plasticity for AM and OWA transplanted lines as defined by differentially expressed genes within a line following transplant (AM_AM vs. AM_OWA; OWA_OWA vs OWA_AM). b Comparison of plastic changes between AM and OWA lines. Color and shape indicate differentially expressed genes by one or both lines. If gene expression plasticity in response to the environment is equal between the two lines, the slope of the relationship would be 1 (dashed black line). The observed slope is shown as the solid black line. c Plastic vs. genetic changes in gene expression. Bar plots show the relative number of plastic vs. genetic changes in expression between AM_AM and OWA_OWA where the dashed line indicates equal numbers. AM to OWA indicates AM lines moving to OWA conditions (Forward adaptation), OWA to AM is the opposite (Reverse adaptation). Inset plots show the counts of the genes that went into the barplot where PO = plasticity only and GC = genetic change. All proportions of PO to GC significant at P < 0.001, two-sided G-test of independence. Source data are provided as a Source Data file.

Fig. 4. Genome-wide variation in gene expression and allele frequencies across transplant generations.

Discriminant analysis of principal components across three transplant generations for a gene expression and b allele frequencies. The x-axis shows the discriminant function space that maximizes differences between lines in their home environments and the background shading represents the “home” discriminant function space for each line. Shape and color of points indicate selection line and treatment condition, respectively. Small points are individual replicates while the mean change for each group is represented by the large transparent points; arrows connect home to transplant means. Significantly different shifts between lines are represented by black arrows and asterisks. Source data are provided as a Source Data file.

Fig. 5. Changes in genetic diversity.

a Median nucleotide diversity (π) for all treatments and generations following transplant (n = 4 biologically independent replicates per treatment and generation). Boxes represent the upper and lower 25% quantiles and median while whiskers are min and max. Letters above each point show significance from Tukey post hoc comparisons following an ANOVA where P < 0.05; b Mean change in π from after transplant for adaptive and non-adaptive SNPs with 95% confidence intervals (n = 9177 π estimates per replicate). Changes relative to AM_AM F1. Letters indicate significance from two-sided Wilcoxon tests with a Bonferroni correction and 0.05 significance threshold. c Gene ontology enrichment for the loss of genetic diversity of OWA_AM at transplant F3; AM_OWA showed no enrichment. Source data are provided as a Source Data file.

Fig. 6. Life-history traits after 20 generations of selection in ambient (AM) and warming and acidification (OWA) conditions and one generation of reciprocal transplant.

For all panels, shape indicates selection line, color indicates environmental condition. Data are presented as mean values and error bars represent 95% confidence intervals. a Egg production rate under ad-libitum and food-limited conditions (n = 6–12 independent mate pairs per food condition per treatment). b Survivorship under ad-libitum and food-limited conditions (n = 3–4 independent experiments per food condition per treatment). c Lambda (net reproductive rate) calculated from combined fecundity, development rate, sex ratio, and survivorship data where a values >1 indicate positive growth (n = 24–28 independent estimates per treatment per food condition). Capital letters in a and c represent statistically separate groups (P < 0.05) from two-way anovas with a post hoc two-sided t-test to correct for multiple testing. Groups with multiple letters are included in multiple statistical groups. Source data are provided as a Source Data file.