Historical interactions moderate species' fitness response to environmental change.

Species must simultaneously adapt to climate stressors and other species, though available genetic variation may constrain this adaptation. Although evolutionary responses to climate can alter interactions among species, it is unknown how the intensity of selection by natural enemies influences species’ ability to withstand (i.e. survive and reproduce following) climate extremes like heat shock, and whether genetic diversity moderates these eco-evolutionary processes. Here we test whether impacts of heat shock on Drosophila simulans (host) fitness depend on their population’s history of interactions with a parasitoid or on the available host and parasitoid genetic diversity (manipulated by inbreeding). We exposed hosts to parasitoid populations over 11 host generations, then exposed their offspring and control hosts to experimental heat shocks. Heat shock more negatively affected the fitness of host populations with a history of high parasitism rates. Surprisingly, less-inbred hosts suffered more severely from heat shock, particularly when they had high historical parasitism rates. However, historically low parasitism rates were associated with a significantly reduced impact of heat shock on fitness relative to no or high parasitism, particularly for less-inbred hosts. Together these results suggest that genetically diverse host populations may retain heat-shock-vulnerable genotypes at high densities (perhaps due to a competition-tolerance trade-off), whereas lighter parasitism (at the approximate rates seen in nature) may prevent this accumulation of genotypes with low tolerance. The intensity of trophic interactions can therefore moderate species’ fitness responses to environmental change in non-linear ways.

Anthropogenic climate change threatens many species with extinction 1 . Although evidence exists for rapid evolutionary change in response to climate 20,21, extreme climate events are expected to increase in frequency and intensity globally 3, intensifying species’ need to adapt. However, evolutionary responses to abiotic changes may be constrained by competitive 4 .Similarly, predator-prey interactions provide great potential for both evolution ( i.e . genetic changes within species over time) and coevolution ( i.e . reciprocal evolution through selection by species on each other 5–7 ), particularly when predators show preferences for specific prey genotypes (e.g., in host-parasitoid interactions 8,9 ). Consequently, high diversity and frequency of interactions with antagonists can increase selection pressure 10 and impact the evolution of phenotypic traits 11 . Conversely, climatically-driven evolution of interaction partners can alter predator-prey dynamics 12, potentially driving a strong selection for traits relevant to their interaction 13 . Antagonistic interactions may drive evolutionary changes 5, though interaction-trait selection may generate other costs, such as a resistance-competitive ability trade-off 14 and reciprocal phenotypic changes 15 . For example, the coevolution of host-resistance and parasitoid-virulence traits is influenced by the net fitness costs associated with increased resistance and counter-resistance 15,16 . Therefore, the high evolutionary costs of resistance may influence responses to environmental changes and vice versa. For example, hosts artificially selected for increased parasitoid resistance present lower and higher heat tolerance at the adult and larval stages respectively 17 . Likewise, hosts selected for heat resistance can experience an increase in population abundance 12, or lower survival 18 under predation. However, species engage in numerous interactions with others, the strength of which can vary from rare, opportunistic interactions to heavy population limitation. It remains unclear how the strength of interactions across trophic levels shapes their fitness responses to climate. Previous evidence that heat-shock proteins may be involved in parasitoid encapsulation 19 would suggest positive pleiotropy, such that protection against parasitoids and heat tolerance would be selected for simultaneously, by either or both of these stressors. Yet, the influence of parasitoid-host interactions on heat tolerance likely depends on a balance between the extent to which the same genes control parasitoid encapsulation and heat shock protection versus the amount of host genetic diversity available for adaptation (which will decline if parasitism is genetically non-random 8 and if high parasitism rates essentially create a host population bottleneck and genetic drift). Therefore, we hypothesise that this balance may tip further towards reduced available genetic variance (and greater vulnerability to heat shock) when parasitism rates are extremely high. Species’ adaptation to climate extremes depends on their genetic diversity 2, which constrains evolutionary responses environmental changes and species interactions 22,23 . Although higher genetic diversity increases a population’s probability of harbouring a genotype that succeeds under a future climate 24,25, the ability of such a genotype to remain in the population will depend on whether it is outcompeted by other genotypes in the current environment (i.e. a competition-tolerance trade-off 26 ). Thus, the impacts of standing genetic diversity may not be easily predicted. Moreover, high attack rates by enemies could decrease this standing genetic diversity in their prey/hosts, either through selection of specific genotypes 27 or reduction of population size (which can correlate with genetic diversity through drift). Although there are many mechanisms through which species interactions could modify evolutionary trajectories of species, it remains unknown how trophic interactions among species with different standing genetic diversities shape their fitness response to environmental stressors. Here we explore how fitness responses to climate extremes (heat shocks) are influenced by the intensity of historical antagonistic interactions, and the extent to which these effects are moderated by inbreeding. We test how interactions between populations of a host–parasitoid system, inbred at different levels to create a gradient of genetic diversity, affect host fitness (i.e. reproduction and survival) under heat shock events. Specifically, we hypothesise that [i] higher host population genetic diversity (i.e. lower inbreeding) increases their potential to reproduce and survive under heat shock, based on the notion that that more genotypes increases the probability that one will be present that can tolerate heat shock 28,29 . Alternatively, if there is a competition-tolerance trade-off 26, then higher genetic diversity could increase the probability of having a highly competitive genotype with low tolerance. [ii] Long periods of host-parasitoid interaction and high parasitoid genetic diversity (i.e. less-inbred parasitoid populations) will increase host fitness under heat shock (relative to hosts with no recent history of parasitism) by maintaining host genetic diversity through reduced host competition (i.e. successful host genotypes – with high abundance - would be a higher fitness reward for the parasitoids to attack, thereby preventing dominance by any higher fitness host genotypes). [iii] However, high attack rates by parasitoids would impose a further bottleneck on host populations, reducing their genetic diversity and fitness following heat shocks. Thus, we hypothesise that the strength of species interactions will determine selection pressure and combine with standing genetic variation to determine species’ fitness response to environmental change.

To understand the long-term consequences of host-parasitoid interactions and their standing genetic diversity on the host fitness response to heat shock, we performed a sequential experiment shown in Fig. 1 and more details from each of these steps follow below in the methods. Fig. 1 : Experimental design to determine how host-parasitoid interactions and genetic diversity interactively influence host survival and reproduction after a heat shock. Timeline (from left to right) shows the experimental steps. First, wild hosts and parasitoids were sampled and brought to the laboratory (i.e. ‘ Source populations ’). To produce a gradient of genetic diversity, source populations were subjected to a gradient of ‘ Host and parasitoid inbreeding treatments ’, each used as predictors in our final model. These inbred populations were then used in a ‘ Host-parasitoid interaction experiment ’ for 11 host generations. This experiment exposed (or not in the case of controls) host populations to the parasitoid (both of which had a level of inbreeding from the previous step). During this experiment, we recorded the mean parasitism rates (i.e. ‘historical parasitism rate’) at fixed host and parasitoid densities to be used in our final model. Finally, for the ‘ Heat shock experiment ’, we used F2 offspring from the last generation of the host-parasitoid interaction experiment (i.e. hosts that were reared without parasitoids, but which came from F1 populations whose parental generations were involved in the host-parasitoid interaction experiment). In this final step, hosts were subjected to a heat shock during multiple stages of their life cycle (shown in red in the ‘Fitness measurements’ circle) and 10 measures of host fitness were recorded (across the five life stages shown in the figure) as response variables. Source populations Between 2018 and 2019, individuals of Drosophila simulans (hereafter ‘host’) and Leptopilina heterotoma (hereafter ‘parasitoid’) were sampled using 300 traps across 25 different urban and rural locations in Christchurch, New Zealand to maximise the genetic diversity of D. simulans and L. heterotoma populations for our experiment . Populations were maintained in the cage for 22 generations (>4,000 individuals per generation). In total, 110 wild-caught mated females from different sites/seasons were added to the host source population. The source population of parasitoids comprised 238 females and 139 males wild caught. Host and parasitoid inbreeding populations After 22 host generations, we established 16 host populations, with varying inbreeding levels to generate a gradient of host inbreeding (which can be negatively related to available genetic diversity 31 ). Four inbreeding levels were established via different bottlenecked population sizes. The two most inbred host populations (F3 and F2) were maintained at the population size of one pair for two (most inbred, F3) or one generation (F2) of full-sib mattings, respectively, while the least inbred had 8 and 32 pairs from the source population, all pairs were from virgin females and males. To increase the population size of all host populations, the 16 host populations (four at each of four levels of inbreeding) went through three more generations before the start of the Host-parasitoid interaction experiment . To account for both full-sib inbreeding and bottleneck size within a single currency, we calculated the host and parasitoid inbreeding coefficient based on the number of individuals used to found each host population and on the number of generations from the start of the populations to the day they were assigned to the treatment cages to start the Host-parasitoid interaction experiment . The formula used to calculate the inbreeding coefficient ( F ) was F =1-[1-1/(2 N e )] t , where N e is the effective population size (number of females and males which founded each population) and t is the number of generations before the experiment began 32,33 . The inbreeding coefficients (F) calculated for the four inbreeding levels of host were F = 0.76, F = 0.68, F = 0.09 and F = 0.02, from more to less inbred populations. Similarly to the hosts, we created eight parasitoid populations (because half of the 16 host populations were to become non-parasitoid controls) with inbreeding levels generated by varying bottleneck sizes (8 pairs (n= 2 populations) and 16 pairs (n= 6 populations). We used different numbers of the two different bottleneck sizes because two populations of the 16-pair bottleneck died, and therefore needed to be replaced using replicates of the 8-pair bottleneck. The offspring of both population bottlenecks were allowed to mate and after two days they were transferred to the experimental cages to start the Host-parasitoid interaction experiment. D uring the bottleneck process, we found that many populations of parasitoids went extinct; therefore to reduce the severity of inbreeding, we replaced two randomly-selected females with new females from the source population to increase the genetic diversity. Twice as many were added to the 16 pair population, to ensure that it would still have the highest genetic diversity. This parasitoid replacement happened twice; in the second and third generations out of the six parasitoid generations. Unlike the hosts, inbred parasitoid populations did not start the experiment with full-sib mating, since preliminary trials revealed that inbreeding a single pair over many generations produced inviable parasitoid populations. However, as some degree of inbreeding would have occurred during the one generation before being assigned to an experimental cage, we calculated the level of inbreeding after one generation, given the two bottleneck sizes. Thus, using the same inbreeding calculations as applied to hosts, the two parasitoid inbreeding levels were F = 0.031 and F = 0.015 for the 8- and 16-pair bottlenecked population size respectively. Finally, inbreeding coefficients were calculated (1 indicating fully inbred to 0 indicating fully outbred) across host and parasitoid populations as described above. However, to better relate the figures and analysis to our hypotheses regarding genetic diversity, we will use 1- inbreeding coefficient in all analyses, such that a score of 1 indicates fully outbred and a score of 0 indicates fully inbred (i.e. higher values equal greater genetic diversity). Therefore, the inbreeding levels from most to least inbred populations were F = 0.2373, F = 0.3164, F = 0.9091 and F = 0.9767 for hosts and F = 0.969 and F = 0.985 for parasitoids. Host and parasitoid interaction experiment We conducted a long-term host-parasitoid interaction experiment (~11 and ~6 host and parasitoid generations respectively). From the 16 host populations (4 different levels of inbreeding), eight were exposed to parasitoids (8 pair bottleneck n= 2; 16 pair bottleneck n=6) and eight were kept as controls (no parasitoids). Hosts were fed different yeast strains (VIN13 or EC1118), but this treatment was controlled for in the analysis since it is not central to the hypotheses. For each treatment/cage we transferred 50 females and 25 males from each of the host inbred lines and 4 females (with access to mate with all males produced by the bottlenecked population) from each parasitoid bottlenecked populations to 18L (30x30x30cm) plastic containers (replicate cages) with a mesh window for ventilation. During the experiment, hosts and parasitoids had access to a jar with cotton soaked in 10% water-honey solution and fresh YPD medium inoculated with yeast, both of which were provided fresh every new host generation. We collected data on parasitism rates at every host generation and calculated the mean of this parasitism rate across generations (hereafter historical mean parasitism rate), because we hypothesised that this historical parasitism pressure could have imposed constrains on the host responses to heat shock (hypothesis iii). To collect parasitism rate data, we sampled 50 host eggs and exposed to a fix density of parasitoids in a way that is not confounded by density-dependence (please see supplementary material for more details). In total, 1,196,934 host and parasitoid individuals interacted along 11 host generations during the host-parasitoid interaction experiment. By the end of the 11th host generation, before the parasitoids drove hosts to extinction, 10 randomly chosen gravid host females were selected from each of the 16 cages and transferred to a new cage within the same treatment with the same resource to start the heat shock experiment. To minimize any behavioural effect of the presence of the parasitoid in the heat shock experiment, all 16 host populations (cages) spent two more generations in the absence of parasitoids before the start of the next experiment. Therefore, any changes to the host population, as a result of interaction with parasitoids, needed to be heritable to influence host response to heat shock in the next experiment. Heat shock experiment To determine the effect of heat shock on host fitness, six pairs (one virgin female and male) were sampled from each of the 16 host populations. Each pair was individually placed into a glass vial with YPD medium inoculated with yeast (VIN13 and EC1118 yeast strains). The six glass vials per host population were immersed in a circulating water bath. Three vials (from each of the 16 host populations) were exposed to 25°C (control temperature) and three to 38°C (heat shock treatment) for 45 minutes (following Klepsatel et al., 2016). The water-bath temperature was controlled using a digital thermostat Optima TM GD100. The heat shock was performed three times during the ~12 day host life cycle. The first heat shock event was performed on the parental adult hosts immediately after being assigned to the glass vial, and the second and third heat shock events were performed during the second larval instar and the pupal stage, respectively, of the F1 generation. In total, 10 response variables were measured. Nine response variables were collected during the host life cycle, and the host offspring biomass measurements were taken after all F1 offspring had emerged. The response variables were: longevity of the (1) male and (2) female parental generation, recorded from the day hosts were collected as virgin pairs until the day they died in the vial. Note that all the parental individuals died before the F1 offspring had emerged. (3) Number of eggs produced by the parental generation, as a proxy of adult host fitness. We counted the number of eggs laid by each female in the 2 nd day (after being assigned to vials) to allow 24h of female oviposition. (4) The number of third instar larvae was counted in the 7 th day (no pupae had formed) and (5) the number of pupae counted in the 11 th day. Data on host adult offspring were also collected: (6) the total number of offspring emerged from pupae (as a measure of all the surviving offspring produced during the parental female adult lifetime) and number of these offspring that were (7) male and (8) female. Finally, the (9) number of days taken to the emergence of the first offspring individual from each vial was recorded as a proxy for differences in host development time. Vials were checked daily for emerged adult offspring, and these were stored at -20°C in Eppendorf tubes for later weighing. To estimate (10) individual biomass (mg), all the offspring individuals from each vial were dried in the oven at 40°C for 48h and weighed (Mettler Toledo UMX2 Ultra-microbalance). The total biomass was divided by the number of total offspring individuals per vial. The effects of heat shock on reproduction, survival, and biomass were assessed at different life stages. For the parental adult stage, heat shock effects were measured by the number of eggs laid by females (heat shock effect on reproduction). For the second instar larval stage, survival was reflected by the number of third instar larvae and pupae formed. At the pupal stage, heat shock effects were measured by the number and biomass or emerged offspring. Except during heat shock events, all populations were kept under controlled laboratory conditions (humidity >65%, 16:8 L:D light cycle, 25°C). Statistical analysis To investigate the effects of heat shock, host and parasitoid inbreeding and historical parasitism rates (and all their possible interactions) on 10 host fitness variables (described in the Heat shock experiment section), we used generalized linear mixed-effects models (GLMMs) with the package lme4 35 in the R environment 36 . All models had a Poisson error distribution, except for one with the response variable host offspring biomass (Gaussian distribution). Continuous predictors were scaled to facilitate comparison of effect sizes and to improve model convergence. Less-inbred parasitoid populations are likely to have greater genetic diversity and, thus, higher attack rates (because they can attack more host genotypes), potentially making these predictors collinear (Maia et al. unpublished). Consequently, we first investigated if parasitoid inbreeding levels and historical parasitism rates were related using linear models. The predictors were positively correlated (see Fig. S1), however models of host fitness (corrected for overdispersion by adding an observation random level effect) with both predictors fitted better than either model with only one predictor (see Supplementary Table S1). Therefore, we included both predictors, but emphasize that their results are partial coefficients capturing their independent effects. Yeast strain was added to the model as a main effect to account for any difference in host fitness due to the type of yeast used as resource, however we did not include any interactions with yeast strain, as these were not among the hypotheses tested here. Host population (cage) was included as a random effect to account for the fact we could not control for differences in host genotypes assigned to each cage. Because historical parasitism rates were estimated (with variance) from the previous Host-parasitoid interaction experiment, rather than being experimentally controlled, their use in a linear model would violate the assumption of fixed predictors. To account for this, we used a bootstrap procedure according to the following steps. First, for each observation that had parasitoids (the 48 replicates in the heat shock experiment with all of their original historical parasitism rate measures over ten host generations in the Host-parasitoid interaction experiment), we simulated a historical mean parasitism rate (hereafter mean parasitism rate). Because parasitism rate data are binomially distributed, we generated the distribution using a logistic regression (glm; binomial distribution) on the ten historical parasitism rate measures for each of our 48 replicates (as the response variable), and fitting only an intercept term. After fitting each glm, we used the function simulate ( stats package ) to simulate a mean parasitism rate from the binomial distribution (i.e. to randomly draw with replacement 8 values – one for each cage that had parasitism - from a binomial distribution with parameters derived from the intercept of our glm and its standard error, and then calculate the mean of the 8 values, repeating this process for 999 iterations). Second, we could not use each of these 999 simulated mean parasitism rates to fit our GLMM model and build distributions of the parameters directly from the 999 models, because the standard errors from the other predictors in the model would be calculated based on differences in the simulated mean value of sampled parasitism rates alone (rather than their own between-replicate variation), so could not give appropriate hypothesis tests. Thus, from the heat shock dataset containing the simulated parasitism rates from a single iteration, we randomly re-sampled rows with replacement to produce a bootstrapped dataset with the same number of observations as the original heat shock dataset (96 observations). By bootstrapping the dataset with simulated parasitism rates, we were able to test the significance of all the predictors (host and parasitoid inbreeding, and historical parasitism rates) in the model simultaneously. Finally, we used the bootstrapped dataset to fit the GLMM models using the 10 host fitness variables as responses, and repeated these steps 999 times (once per iteration of the parasitism rate simulation) for each response variable. The significance level of the main and interaction effects was tested against the bootstrapped distribution using a one-sample T-test (95%CI, with the t.test function in the base package of R). In terms of overall fitness, our primary interest was the number of surviving offspring. However, to understand how fitness was influenced by responses at individual life stages, we repeated the above analyses using a set of response variables that captured the proportion of individuals that survived from one life stage to the next. We used the same set of predictors (heat shock, host and parasitoid inbreeding, historical parasitism rates and yeast strain) and statistical approach (GLMMs) as described above, including cage as random effect, but with a binomial error distribution (because these analyses included proportion survival rather than abundances). In some cases, we recorded more individuals at one life stage than the preceding life stage (e.g., more pupae than larvae), because the parental hosts kept laying eggs after the egg count, and these eggs may have developed more quickly, adding more individuals to the next life stage. This resulted in a proportion survival greater than 1, so for these few cases (egg to larva= 4; larva to pupa= 20; pupa to adult= 2; each out of 96 observations) we assigned a proportion survival of 1. Finally, the adult offspring response variable, which measured the proportion that survived from pupal stage to adult offspring, considered survivors to be the number of offspring emerged in the first four days. The first four days represents the time it would take for the first eggs laid to emerge. Testing many different life stages as response variables can potentially inflate the type I error rate. Thus, to ensure that the significance level of the models would not be due to multiple testing, we used the Bernoulli equation (suggested by Moran, 2003) to evaluate the probability of one test being statistically significant due to chance alone. The Bernoulli formula accounts for the number of tests undertaken (here N=13) and number of tests that presented a significant three-way interaction (here K=8) below α=0.05. Note that the three-way interactions between heat shock, host inbreeding and either parasitoid inbreeding or attack rate were our main hypothesis tests of interest. Summaries of the raw dataset are shown in Fig. S2. All the statistical analyses were conducted in R 38 .

Several measures of host fitness responded significantly to the interactive effects of heat shock, parasitism rates and host and parasitoid genetic diversity. Overall, the interacting effects on host fitness were visible in all stages (e.g., number of: eggs laid, larvae, pupae and adult offspring) of the host life cycle (for t and p values and CI see Table 1). Heat shock (main effect) reduced the numbers of all non-adult life stages, relative to control temperatures. Heat shock more negatively affected the fitness of less-inbred host populations and those with a history of high parasitism rates (i.e. all two- and three-way interactions among heat shock, historical parasitism rate and 1-host inbreeding had negative coefficients; Table 1, Fig. 2, Fig. S3, S4 and S5). That is, under heat shock, high historical parasitism rates decreased the fitness of all hosts, but more strongly for less-inbred host populations. Fig. 2 : Effects of host inbreeding ( a - d ), heat shock ( b, d ), parasitoid inbreeding ( a, b ) and historical mean parasitism rates ( c, d ) on the number of eggs produced by the parental host generation (one of the host fitness measurements) holding all other effects in the model (Table1) constant. Panels show fitted values from GLMM’s. Host inbreeding coefficient is presented as 1-host inbreeding coefficient, ranging from more to less-inbred host populations. Mean historical parasitism rates and host inbreeding were scaled in the models, but presented here on original scales to facilitate interpretation. Note the differing y axis scales between the control and the heat shock panels due to the larger variance in fitness under the heat shock treatment. From the parasitism rates measured, we present in this graph one of the low (0.4) and high (0.57) parasitism rates. See Table 1 for confidence intervals and p values. Although high historical parasitism rates were associated with reduced fitness, some history of interaction with parasitoids (at either level of parasitoid inbreeding) was associated with a significant reduction in the negative impact of heat shock on fitness (relative to parasitoid-free controls), because both less- and more-inbred parasitoid populations had a positive effect on fitness of the less-inbred host population under heat shock (interaction between T and P 0.969 or P 0.985 in Table 1). The size of the two positive T:P coefficients relative to the effect of parasitism on heat shock (the negative T:R coefficient) shows that R (scaled parasitism rates) would need to have a value of approximately two before this benefit of interaction with parasitoids on heat-shock tolerance (Fig. 2a,b) is compensated for by the cost of high historic parasitism rates (Fig. 2c,d) (Table 1). Unscaled, this would equate to a parasitism rate of approximately 52%. The direction of the interaction effects also remained consistent for the total number of male and female offspring produced and for the number of individuals that survived from egg to larval and larval to pupal stage (Table S2 and S3). Although not our focus here, yeast strain also affected host fitness (Table 1, S2, S3 and S4). The Bernoulli equation shows a probability <0.00003 of finding eight significant tests by chance, suggesting that our results were likely not the result of Type I error associated with many response variables. Finally, there were no interacting effects of heat shock, host and parasitoid inbreeding and historical parasitism rates on the longevity of the parental female and male, suggesting that the changes to the number of offspring produced were not caused by a significant change of the female lifespan (Table S4). In addition, no interacting effects were found for the survival from pupal to adult stage and the number of days it took for the offspring to emerge (Table S2 and S4). Table 1 : Coefficients of models testing the impact of heat shock on number of surviving individuals at egg, larval and pupal life stages and adult F1 offspring (i.e. host fitness) depends on historical parasitism rates and host and parasitoid genetic diversity. The predictors were elevated Temperature (T) (vs. control temperature as the intercept condition), as a factor; host inbreeding levels (H = 1-inbreeding coefficient, such that higher values would be expected to have greater genetic diversity); historical parasitism rates (R); historical parasitoid presence at one of two inbreeding levels (P 0.969 and P 0.985 ) as a factor (parasitoid absent is the intercept); yeast strain (S) (EC1118 vs. the intercept condition, yeast VIN13), as a factor. Model coefficients (mean and 95% confidence interval, CI) were generated by bootstrapping 999 simulations for host fitness response variables. Continuous predictors are scaled to facilitate comparison of effect sizes. Significant results are shown in bold (p <0.05 with Poisson errors and a log link function).

Our study shows that hosts’ genetic diversity and their history of parasitoid interactions can impose selection pressures that limit their ability to survive environmental extremes. Contrary to our first hypothesis, less-inbred host populations (in the absence of parasitoids) produced fewer offspring under heat compared with the control temperature and their more-inbred counterparts. This indicates that initial host genetic diversity (at the beginning of the host-parasitoid interaction experiment) did not improve subsequent environmental tolerance. However, supporting our second hypothesis, a history of interacting with parasitoids (at both levels of parasitoid inbreeding) had a strong positive effect on host fitness under both heat shock and host inbreeding (i.e. there were strong positive interaction effects between either level of parasitoid inbreeding, heat shock and host inbreeding). Conversely, a history of strong interactions (i.e. high parasitism rates) had an opposite effect, offsetting the benefit of interaction with parasitoids and increasing the negative effect of heat shocks on host fitness and supporting our third hypothesis. In our study, the high parasitism rates needed to offset the benefit of interaction with parasitoids and produce a net negative effect on host fitness under heat shock (approx. 52%) are much higher than hosts usually experience in the field 39–41 . Therefore, natural parasitism rates may provide an evolutionary benefit to hosts. Host genetic diversity influences host responses to environmental change In the absence of historic interactions with parasitoids, high host genetic diversity (i.e. low inbreeding) was associated with stronger decline in host fitness under heat shock, compared with that shown by populations of low host genetic diversity maintained under control temperatures or heat shock. This finding was surprising, given that previous studies have indicated that genetic diversity is likely to improve species’ overall environmental tolerance 28,42 . Genetically diverse populations would be expected to maintain their positive fitness under environmental change because they have higher probability of harbouring a genotype that copes better (positive fitness) under those conditions 29 . However, for genetic diversity of a population to confer tolerance to environmental change, the availability of genetic diversity needs to persist in a population through time. Because our parasitoid-free populations continued for 11 generations before the heat shock experiment (Fig. 3.1), it is possible that intraspecific competition or drift altered the genetic diversity subsequent to our experimental population bottlenecks. In fact, if there is a trade-off between (intraspecific) competitive ability and heat tolerance (as has been shown before for Drosophila feeding rate and stress-resistant genes 43 ) these 11 generations may have been sufficient to select for low heat tolerance, unless parasitoids were present to prevent dominance by any single host genotype. In addition, random genetic drift could have also decreased host genetic diversity 47,48, especially in populations experiencing bottleneck events through small populations, either driven by demographic processes or high parasitism rates. However, it is unlikely that genetic drift alone could have played a major role decreasing genetic diversity of host populations, since the population sizes of more and less-inbred hosts were similar over time (Fig. S6). Parasitoid genetic diversity affects host population fitness We did not find differences between the two levels of the parasitoid’s genetic diversity, however, our study suggests that parasitoid populations may either be able to maintain host genetic diversity, potentially through preferential attack of the most common host genotypes preventing dominance 49, or select for heat tolerance through pleiotropic effects, eventually resulting in the observed positive fitness of host populations under heat shock. Predator-prey genotype specificity 8,50–52 may allow complementary host use by parasitoids 53, such that the parasitoid genotype frequencies (and attack rates) can respond to the frequencies of host genotypes. Therefore, parasitoid genetic diversity may maintain host genetic diversity, preventing a single host genotype from increasing disproportionately in frequency and out-competing others. Interestingly, the matching of parasitoid to host genotypes can even become stronger under warming conditions 8, potentially further strengthening the benefits of host-parasitoid interactions for maintaining host genetic diversity and adaptive capacity against climate extremes. However, because we exposed hosts to heat shocks after parasitism had ceased, this latter effect could not have driven our experimental results. Parasitism rates also influence host responses to environmental change Our results suggest that the maintenance of host genetic diversity by parasitoids could break down as attack rates increase. High parasitism rates increased the negative effect of heat shock on the fitness of the high genetic diversity host populations. If the history of intense species interactions (high parasitism rates) resulted in either a decrease in the genetic diversity of less-inbred hosts or selection for a host genotype that was associated with poor heat tolerance (i.e. antagonistic pleiotropy), this could have decreased the evolutionary potential of hosts to cope with environmental change, evidenced by reduced host fitness under extreme temperatures. Conversely, when attack rates were low, a history of parasitism may have maintained the high genetic diversity of less-inbred host populations, thereby maintaining the positive fitness that is expected from genetically diverse populations of hosts under heat shock events 54, or allowed for positive pleiotropy without the host population bottleneck caused by high attack rates. Although population size is kept constant in many empirical studies and models of co-evolutionary dynamics 55,56, in our experiment population size was free to change over time. Thus, high parasitism rates had the potential to decrease the overall population size of hosts and thereby may have decreased host genetic diversity either by driving preferred host genotypes to extinction 53 or by reducing host population size to exacerbate the influence of genetic drift (i.e. generating a population bottleneck). Moreover, a decrease in host genetic diversity, as a result of high parasitism rates, could result in a short-term fitness advantage for a particular host genotype competing for resources, and as discussed above, host traits being selected under high parasitism rates may have been correlated with heat shock tolerance through pleiotropic effects 57, though we were not able to test this hypothesis. Although species interactions mediating ecological and evolutionary responses to climate conditions have recently been shown in plant-animal and animal-animal interactions 4,58, our paper highlights the importance of genetic diversity and species interaction at two different trophic levels. Our results have shown that species’ antagonistic interactions affect their evolutionary potential to adapt to environmental change. Acknowledgments We thank all the researchers and Ōtautahi citizens who helped to distribute and return traps for hosts and parasitoids. We thank Matthew Goddard and Darren Ward (University of Auckland – Waipapa Taumata Rau) for kindly providing the yeast strains and help with parasitoid identification, respectively. I also thank Julian Kasper (Museum of New Zealand – Te Papa Tongarewa) for help with host identification, Daniel Stouffer (University of Canterbury - Te Whare Wānanga o Waitaha) for advice on the statistical analysis, Alan and Aynsley (University of Canterbury - Te Whare Wānanga o Waitaha) for their technical assistance regarding laboratory set up and Marcus and Charlotte for the help in the laboratory. L.F.M. scholarship and P.C., S.L.G. and J.M.T. were funded by the New Zealand Tertiary Education Commission via the BioProtection Centre of Research Excellence. AgResearch provided support in the form of salary for S.L.G., but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

1. Mann, M. E. et al. Influence of Anthropogenic Climate Change on Planetary Wave Resonance and Extreme Weather Events. Sci. Rep. 7, (2017).2. Norberg, J., Urban, M. C., Vellend, M., Klausmeier, C. A. & Loeuille, N. Eco-evolutionary responses of biodiversity to climate change. Nat. Clim. Chang. 2, 747–751 (2012).3. Perkins, S. E., Alexander, L. V. & Nairn, J. R. Increasing frequency, intensity and duration of observed global heatwaves and warm spells. Geophys. Res. Lett. 39, 1–5 (2012).4. Grainger, T. N., Rudman, S. M., Schmidt, P. & Levine, J. M. Competitive history shapes rapid evolution in a seasonal climate. Proc. Natl. Acad. Sci. U. S. A. 118, (2021).5. Paterson, S. et al. Antagonistic coevolution accelerates molecular evolution. Nature 464, 275–278 (2010).6. Lively, C. M. & Dybdahl, M. F. Parasite adaptation to locally common host genotypes. Nature 405, 679–681 (2000).7. Goldson, S. L. L. et al. If and when successful classical biological control fails. Biol. Control 72, 76–79 (2014).8. Lavandero, B. & Tylianakis, J. M. Genotype matching in a parasitoid-host genotypic food web: An approach for measuring effects of environmental change. Mol. Ecol. 22, 229–238 (2013).9. Vorburger, C. The evolutionary ecology of symbiont-conferred resistance to parasitoids in aphids. Insect Sci. 21, 251–264 (2014).10. Betts, A., Gray, C., Zelek, M., MacLean, R. C. & King, K. C. High parasite diversity accelerates host adaptation and diversification. Science (80-. ). 360, 907–911 (2018).11. Barbour, M. A., Greyson‐Gaito, C. J., Sotoodeh, A., Locke, B. & Bascompte, J. Loss of consumers constrains phenotypic evolution in the resulting food web. Evol. Lett. 4, 266–277 (2020).12. Harmon, J. P., Moran, N. A. & Ives, A. R. Species Response to Environmental Change: Impacts of Food Web Interactions and Evolution. Science (80-. ). 323, 1347–1350 (2009).13. Shields, M. W. et al. Behaviour drives contemporary evolution in a failing insect-parasitoid importation biological control programme. Front. Ecol. Evol. 10, 1–14 (2022).14. Kraaijeveld, A. R. & Godfray, H. C. J. Trade-off between parasitoid resistance and larval competitive. Nature 389, 278–280 (1997).15. Schulte, R. D., Makus, C., Hasert, B., Michiels, N. K. & Schulenburg, H. Multiple reciprocal adaptations and rapid genetic change upon experimental coevolution of an animal host and its microbial parasite. Proc. Natl. Acad. Sci. 107, 7359–7364 (2010).16. Fellowes, M. D. E. E. & Travis, J. M. J. J. Linking the coevolutionary and population dynamics of host – parasitoid interactions. Popul. Ecol. 42, 195–203 (2000).17. Takigahira, T., Suwito, A. & Kimura, M. T. Assessment of fitness costs of resistance against the parasitoid Leptopilina victoriae in Drosophila bipectinata. Ecol. Res. 29, 1033–1041 (2014).18. Hangartner, S., Dworkin, I., DeNieu, M. & Hoffmann, A. A. Does increased heat resistance result in higher susceptibility to predation? A test using Drosophila melanogaster selection and hardening. J. Evol. Biol. 30, 1153–1164 (2017).19. Xavier, M. J. & Williams, M. J. The Rho-Family GTPase Rac1 Regulates Integrin Localization in Drosophila Immunosurveillance Cells. PLoS One 6, e19504 (2011).20. Geerts, A. N. et al. Rapid evolution of thermal tolerance in the water flea Daphnia. Nat. Clim. Chang. 5, 665–668 (2015).21. Franks, S. J., Sim, S. & Weis, A. E. Rapid evolution of flowering time by an annual plant in response to a climate fluctuation. Proc. Natl. Acad. Sci. 104, 1278–1282 (2007).22. Lai, Y. T. et al. Standing genetic variation as the predominant source for adaptation of a songbird. Proc. Natl. Acad. Sci. U. S. A. 116, 2152–2157 (2019).23. Moya-Laraño, J. Genetic variation, predator–prey interactions and food web structure. Philos. Trans. R. Soc. B Biol. Sci. 366, 1425–1437 (2011).24. Boles, B. R., Thoendel, M. & Singh, P. K. Self-generated diversity produces ‘insurance effects’ in biofilm communities. Proc. Natl. Acad. Sci. U. S. A. 101, 16630–16635 (2004).25. Yachi, S. & Loreau, M. Biodiversity and ecosystem productivity in a fluctuating environment: The insurance hypothesis. Proc. Natl. Acad. Sci. U. S. A. 96, 1463–1468 (1999).26. Wettlaufer, J. D., Ye, A., MacMillan, H. A. & Martin, P. R. A test of the competitive ability–cold tolerance trade‐off hypothesis in seasonally breeding beetles. Ecol. Entomol. 48, 55–68 (2023).27. Dubuffet, A. et al. Genetic interactions between the parasitoid wasp Leptopilina boulardi and its Drosophila hosts. Heredity (Edinb). 98, 21–27 (2007).28. Dahlgaard, J., Krebs, R. & Loeschcke, V. Heat-shock tolerance and inbreeding in drosophila buzzatii. Heredity (Edinb). 74, 157–163 (1995).29. Huston, M. A. Hidden treatments in ecological experiments: Re-evaluating the ecosystem function of biodiversity. Oecologia 110, 449–460 (1997).30. Palanca, L., Gaskett, A. C., Günther, C. S., Newcomb, R. D. & Goddard, M. R. Quantifying Variation in the Ability of Yeasts to Attract Drosophila melanogaster. PLoS One 8, 1–10 (2013).31. Charlesworth, D. Effects of inbreeding on the genetic diversity of populations. Philos. Trans. R. Soc. B Biol. Sci. 358, 1051–1070 (2003).32. Falconer, D. S. & Mackay, T. F. C. Introduction to quantitative genetics . (Longman, 1996).33. Reed, D. H. & Frankham, R. Correlation between fitness and genetic diversity. Conserv. Biol. 17, 230–237 (2003).34. Klepsatel, P., Gáliková, M., Xu, Y. & Kühnlein, R. P. Thermal stress depletes energy reserves in Drosophila. Sci. Rep. 6, 1–12 (2016).35. Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting Linear Mixed-Effects Models Using lme4. J. Stat. Softw. 67, (2015).36. R Core Team. R: The R Project for Statistical Computing. (2021).37. Moran, M. D. Arguments for rejecting the sequential bonferroni in ecological studies. Oikos 100, 403–405 (2003).38. R Core Team. R: A language and environment for statistical computing. R Found. Stat. Comput. Vienna, Austria. (2015).39. Tylianakis, J. M., Tscharntke, T. & Lewis, O. T. Habitat modification alters the structure of tropical host-parasitoid food webs. Nature 445, 202–205 (2007).40. Frost, C. M. et al. Apparent competition drives community-wide parasitism rates and changes in host abundance across ecosystem boundaries. Nat. Commun. 7, 1–12 (2016).41. Peralta, G., Frost, C. M., Rand, T. a., Didham, R. K. & Tylianakis, J. M. Complementarity and redundancy of interactions enhance attack rates and spatial stability in host-parasitoid food webs. Ecology 95, 1888–1896 (2014).42. Franke, K. & Fischer, K. Inbreeding interferes with the heat-shock response. Heredity (Edinb). 114, 80–84 (2015).43. Foley, P. A. & Luckinbill, L. S. The Effects of Selection for Larval Behavior on Adult Life-History Features in Drosophila melanogaster. Evolution (N. Y). 55, 2493–2502 (2001).44. Tagg, N., Doncaster, C. P. & Innes, D. J. Resource competition between genetically varied and genetically uniform populations of Daphnia pulex (Leydig): Does sexual reproduction confer a short-term ecological advantage? Biol. J. Linn. Soc. 85, 111–123 (2005).45. Svanbäck, R. & Bolnick, D. I. Intraspecific competition drives increased resource use diversity within a natural population. Proc. R. Soc. B Biol. Sci. 274, 839–844 (2007).46. Bolnick, D. I. Intraspecific competition favours niche width expansion in Drosophila melanogaster. Nature 410, 463–466 (2001).47. White, P. S., Arslan, D., Kim, D., Penley, M. & Morran, L. Host genetic drift and adaptation in the evolution and maintenance of parasite resistance. J. Evol. Biol. 34, 845–851 (2021).48. Rich, S. S., E., B. A. & Wilson, S. P. Genetic Drift in Small Populations of Tribolium. Evolutuion 33, 579–584 (1979).49. Finke, D. L. & Snyder, W. E. Niche Increases Resource Partitioning by Diverse Communities Exploitation. Science (80-. ). 321, 1488–1490 (2008).50. Carius, H. J., Little, T. J. & Ebert, D. Genetic variation in a host-parasite association: Potential for coevolution and frequency-dependent selection. Evolution (N. Y). 55, 1136–1145 (2001).51. Dubuffet, A., Álvarez, C. I. R., Drezen, J. M., Van Alphen, J. J. M. & Poirié, M. Do parasitoid preferences for different host species match virulence? Physiol. Entomol. 31, 170–177 (2006).52. Salvaudon, L., Héraudet, V. & Shykoff, J. A. Genotype-specific interactions and the trade-off between host and parasite fitness. BMC Evol. Biol. 7, 1–10 (2007).53. Stireman, J. O., Nason, J. D., Heard, S. B. & Seehawer, J. M. Cascading host-associated genetic differentiation in parasitoids of phytophagous insects. Proc. R. Soc. B Biol. Sci. 273, 523–530 (2006).54. Kristensen, T. N., Dahlgaard, J. & Loeschcke, V. Inbreeding affects Hsp70 expression in two species of Drosophila even at benign temperatures. Evol. Ecol. Res. 4, 1209–1216 (2002).55. Bérénos, C., Schmid-Hempel, P. & Mathias Wegner, K. Evolution of host resistance and trade-offs between virulence and transmission potential in an obligately killing parasite. J. Evol. Biol. 22, 2049–2056 (2009).56. Greeff, M. & Schmid-Hempel, P. Influence of co-evolution with a parasite, Nosema whitei, and population size on recombination rates and fitness in the red flour beetle, Tribolium castaneum. Genetica 138, 737–744 (2010).57. Zhu, J. Y., Wu, G. X., Ye, G. Y. & Hu, C. Heat shock protein genes (hsp20, hsp75 and hsp90) from Pieris rapae: Molecular cloning and transcription in response to parasitization by Pteromalus puparum. Insect Sci. 20, 183–193 (2013).58. Valdés, A. & Ehrlén, J. Plant–animal interactions mediate climatic effects on selection on flowering time. Ecology 102, (2021).59. United Nations Development Programme. Biodiversity and the 2030 agenda for sustainable development . (2015).60. Nosil, P. et al. Natural selection and the predictability of evolution in timema stick insects. Science (80-. ). 359, 765–770 (2018).61. Ives, A. R. et al. Self-perpetuating ecological–evolutionary dynamics in an agricultural host–parasite system. Nat. Ecol. Evol. 4, 702–711 (2020).62. Schaffner, L. R. et al. Consumer-resource dynamics is an eco-evolutionary process in a natural plankton community. Nat. Ecol. Evol. 3, 1351–1358 (2019).63. Pauls, S. U., Nowak, C., Bálint, M. & Pfenninger, M. The impact of global climate change on genetic diversity within populations and species. Mol. Ecol. 22, 925–946 (2013).64. Nadeau, C. P. & Urban, M. C. Eco-evolution on the edge during climate change. Ecography (Cop.). 42, 1280–1297 (2019).65. Tomasetto, F., Tylianakis, J. M., Reale, M., Wratten, S. & Goldson, S. L. Intensified agriculture favors evolved resistance to biological control. Proc. Natl. Acad. Sci. 114, 3885–3890 (2017).66. Casanovas, P., Goldson, S. L. & Tylianakis, J. M. Asymmetry in reproduction strategies drives evolution of resistance in biological control systems. PLoS One 13, e0207610 (2018).67. Damien, M. & Tougeron, K. Prey–predator phenological mismatch under climate change. Curr. Opin. Insect Sci. 35, 60–68 (2019).68. Senior, V. L., Evans, L. C., Leather, S. R., Oliver, T. H. & Evans, K. L. Phenological responses in a sycamore–aphid–parasitoid system and consequences for aphid population dynamics: A 20 year case study. Glob. Chang. Biol. 26, 2814–2828 (2020).69. Tougeron, K., Brodeur, J., Le Lann, C. & van Baaren, J. How climate change affects the seasonal ecology of insect parasitoids. Ecol. Entomol. 45, 167–181 (2020).70. Henry, L. M., May, N., Acheampong, S., Gillespie, D. R. & Roitberg, B. D. Host-adapted parasitoids in biological control: Does source matter? Ecol. Appl. 20, 242–250 (2010).71. Lommen, S. T. E., de Jong, P. W. & Pannebakker, B. A. It is time to bridge the gap between exploring and exploiting: prospects for utilizing intraspecific genetic variation to optimize arthropods for augmentative pest control – a review. Entomol. Exp. Appl. 162, 108–123 (2017). Supplementary Material File (image3.emf) - Download - 54.24 KB Information & Authors Information Version history Copyright This work is licensed under a Non Exclusive No Reuse License.

Authors Metrics & Citations Metrics Article Usage 252views 164downloads Citations Download citation Lais Maia, Paula Casanovas, Amy Osborne, et al. Historical interactions moderate species' fitness response to environmental change.. Authorea. 18 February 2025. DOI: https://doi.org/10.22541/au.173989462.21758636/v1 DOI: https://doi.org/10.22541/au.173989462.21758636/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu.