Evolutionary mismatch along salinity gradients in a Neotropical water strider

Abstract The evolution of local adaptation is crucial for the in situ persistence of populations in changing environments. However, selection along broad environmental gradients could render local adaptation difficult, and might even result in maladaptation. We address this issue by quantifying fitness trade‐offs (via common garden experiments) along a salinity gradient in two populations of the Neotropical water strider Telmatometra withei—a species found in both fresh (FW) and brackish (BW) water environments across Panama. We found evidence for local adaptation in the FW population in its home FW environment. However, the BW population showed only partial adaptation to the BW environment, with a high magnitude of maladaptation along naturally occurring salinity gradients. Indeed, its overall fitness was ~60% lower than that of the ancestral FW population in its home environment, highlighting the role of phenotypic plasticity, rather than local adaptation, in high salinity environments. This suggests that populations seemingly persisting in high salinity environments might in fact be maladapted, following drastic changes in salinity. Thus, variable selection imposed by salinization could result in evolutionary mismatch, where the fitness of a population is displaced from its optimal environment. Understanding the fitness consequences of persisting in fluctuating salinity environments is crucial to predict the persistence of populations facing increasing salinization. It will also help develop evolutionarily informed management strategies in the context of global change.

However, making inferences about local adaptation in the context of broad environmental gradients remains challenging. First, classical studies of local adaptation tend to focus on populations that show (prior to experiments) trait divergence between alternative environments, where local adaptation is most likely to occur (Hereford, 2009;Schluter, 2000). Yet, a priori trait divergence is generally unknown for nonmodel species persisting along broad environmental gradients. Second, the contexts in which local adaptation is most often estimated represent highly divergent-yet binary-environmental gradients (Hereford, 2009) that impose stable (and perhaps predictable) selection pressures. Examples include low-and high-predation sites (Endler, 1980(Endler, , 1991Reznick & Endler, 1982), benthic and limnetic zones of lakes (McPhail, 1993;Schluter & McPhail, 1992), or high and low salinity environments (Defaveri & Merila, 2014;Kozak et al., 2013;Wrange et al., 2014). By contrast, most natural environmental gradients are likely to be broad and highly variable, resulting in variable (and perhaps unpredictable) selection pressures. Thus, selection imposed by variable conditions could hinder local adaptation along broad environmental gradients, but this expectation remains understudied.
Salinity gradients provide a good model to test for the adaptive consequences of variable conditions on natural populations. Salinity levels experienced by freshwater organisms can vary anywhere from nearly fresh (<0.5 ppt) to brackish (0.5-30 ppt) and saline (30-50 ppt) water, and the exposure to these salinity levels can vary temporally, from hours to years (Gomez-Mestre & Tejedo, 2003;Kozak et al., 2013). Previous studies have found evidence for local adaptation to high salinity levels in plants (Al-Gharaibeh et al., 2017;Busoms et al., 2015), fishes (Defaveri & Merila, 2014;Kozak et al., 2013), and amphibians (Gomez-Mestre & Tejedo, 2003), but most of these studies have been limited to narrow salinity gradients-generally comparing fresh versus brackish water populations, and only a few compare multiple salinity levels (Defaveri & Merila, 2014;Kozak et al., 2013).
Furthermore, studies on the consequences of salinization for freshwater organisms have been limited to coarse taxonomic levels (i.e., above genus level), and to geographic regions historically affected by salinization (reviewed in (Castillo et al., 2017)). In fact, there are virtually no studies on the effect of salinization in Neotropical regions (Castillo et al., 2017), which contains a large portion of the planet's freshwater biodiversity (Abell et al., 2008).
The fluctuating nature of salinization could render local adaptation difficult if salinity changes overcome the adaptive potential of populations. That is, if populations lack phenotypic or genetic variation to cope with current changes in salinity, they are likely to undergo local extinction (Lewontin, 1974;Sinervo et al., 2010). In addition, even if populations manage to persist in newly salinized environments, their local fitness might be lower than expected in the ancestral freshwater environment, effectively rendering populations maladapted (Brady, 2013;Crespi, 2000;DeWitt & Yoshimura, 1998) to saline environments. In this case, maladaptation could be relative (Brady et al., 2019;Geladi et al., 2019;Hendry & Gonzalez, 2008;Hendry & Taylor, 2004;Rolshausen et al., 2015) or "partial," in the sense that populations are able to persist, albeit with suboptimal fitness. This contrasts with absolute maladaptation (Geladi et al., 2019;Hendry & Gonzalez, 2008), where populations are unable to persist. Consequently, selection pressures imposed by fluctuating salinization could result in an "evolutionary mismatch" whereby the fitness of a population is displaced from its optimal environment (Hale et al., 2016;Lloyd et al., 2011;Negrin et al., 2019;Robertson et al., 2013;Schlaepfer et al., 2002). Understanding the persistence of populations along broad and variable environmental gradients requires a better understanding of the magnitude of adaptation and maladaptation along those gradients. Here, we use a combination of field surveys and common garden experiments to examine fitness trade-offs along a salinity gradient in the Neotropical water strider Telmatometra withei in Panama.

| Study organism
Telmatometra withei (Bergroth, 1908) is a common water strider distributed from Ecuador to México (Molano et al., 2017;Pacheco, 2012;Padilla-Gil, 2012), including the islands such as Puerto Rico and Trinidad and Tobago (Molano et al., 2017). Although T. withei is considered a freshwater species (Pacheco, 2012;Padilla-Gil, 2012), we have found several populations inhabiting in a broad range of salinities, ranging from fresh to brackish water along the two slopes of the Isthmus of Panama, as well as on Coiba Island (Figure 1a). Our preliminary molecular analyses based on Mitochondrial COI found low genetic variation among populations (Figure 1e), which is consistent with the presence of a single species across salinity gradients in Panama. While some genera of saline-adapted water striders are known (e.g., Genus Halobates) (Cheng, 2005;Harada, 2005), the potential for adaptation in typically freshwater species remains unexplored. For example, the Japanese water strider, Aquarius paludum (Kishi et al., 2006(Kishi et al., , 2009, and Gerris thoracicus from Finland (Kaitala, 1987;Vepsäläinen, 1978) are sometimes found in brackish waters, but their degree of local adaptation to high salinity environments has not been tested.

| Study sites and experimental setting
Individuals of T. withei were collected from two sites located in Llano de Catival on the Western Azuero Peninsula on the Pacific coast of Panama (Figure 1a) (Castillo et al., 2020). This site is influenced by both seawater intrusion (due to daily tidal fluctuations) and precipitation (during the rainy season), resulting in salinity levels that can range from 0.4 to 11 ppt (Figure 1b) oxygen (TDS), pH, and salinity. These water parameters were also recorded in each experimental box weekly (Table 1 and 2).

| Morphological identification and DNA barcoding
Adult specimens were identified using a standard taxonomic key  (Ebong et al., 2016). Multiple alignments were made using the ClustalW algorithm, according to the default settings (Ebong et al., 2016). We then ran a Randomized Axeelerated Maximum Likelihood analysis, using the nucleotide model GTR+G+I, with 1,000 bootstrap replicates and parsimony random seed set to 1 (Ebong et al., 2016

| Common garden experiments
To estimate fitness trade-offs along a salinity gradient, we per- Shown are initial (i) and final (f) values for T° (temperature in °C), SPC (specific conductivity), TDS (total dissolve oxygen), pH, and ppt (salinity).
TA B L E 2 Environmental parameters in experimental treatments for freshwater and brackish populations F1 juveniles, before wing development). After this period, surviving adults were removed from the experimental boxes, but we continued to monitor offspring survival (F1) to maturity to get an estimate of longevity until 90 days. We also estimated egg size for a subset of the eggs from the FW (n = 27) and BW (1 ppt; n = 30) populations, using digital photographs and ImageJ v1.51 (Rasband, 1997(Rasband, -2012.

| Magnitude of local adaptation and maladaptation
We used all fitness-related traits (survival, fecundity, oviposition rate, and number of offspring [F1]) from common garden experiments to quantify local adaptation for both FW and BW populations in each of their home environments with the following equation from (Hereford, 2009).
where W represents the mean fitness of the native and the foreign population at the native population's site, and avg (W) represents the mean fitness across both populations at that site (Hereford, 2009).
Positive and negative values indicate local adaptation and maladaptation for the focal native populations, respectively (Hereford, 2009).
As a complementary approach, we then inferred the magnitude of maladaptation by estimating the proportional fitness difference between the ancestral freshwater population and the derived brackish population. To quantify this parameter, we used the following formula: representing the difference between the mean fitness of the ancestral (reference) population in its home environment standardized to 1.0 (W ideal ; here, the freshwater population) and the fitness of the derived population in its home environments (W realized ; here, the brackish water population), with MA between 0 and 1 indicating 0% and 100% maladaptation, respectively. These estimates assume that the ancestral population experiences an optimal fitness in its home environment, which is a simplified assumption, given that the environment may change constantly, and thus, populations might not always be near the optimum. In addition, even if the derived population shows lower fitness values in the novel environment, this difference may still be adaptive.
However, comparing the proportional fitness difference between both populations under similar experimental condition will give an indication of the magnitude of fitness loss in the derived population in the novel environment.

| Data analysis
To estimate salinity tolerance for both fresh and brackish water populations, we performed logistic regressions between survival and salinity. Survival was estimated as the ratio between the number of survival individuals and the initial number of individuals in each ex-

| Salinity tolerance experiments
Salinity had a significant effect on survival of T. withei, with both FW and BW populations reaching 50% mortality (LC 50 ~48 hr) at relatively low salinity levels (Table 3). Interestingly, LC 50 tended to be lower for the FW (8.69 ppt) than BW (10.58 ppt) populations, although this difference was not statistically significant (Table 3;  (Table 3).
Our ANCOVA also showed a significant effect of salinity on LC 50 , but there was no effect of population of origin or their interaction (Table 3).

| Common garden experiments
Salinity had a significant effect on fitness correlates (Figure 3; (2) MA = W ideal − W realized treatment in which fitness decreased to nearly 0% for both populations, although survival in the BW population was ~5% (Figure 3; Table 4). When comparing both populations across salinity levels, we found higher fitness overall in the FW population in its home FW environment than either the FW or BW population across any of the salinity treatments. However, the BW population tended to show higher fitness than the FW population in high salinity treatments (1-11 ppt), and this difference was consistently significant at 3 ppt ( Figure 3; Table 4). A similar pattern was observed by sex, with both males and females from the FW population showing overall higher survival in their home environment, and the BW population showing higher survival (for both males and females) at higher salinities (3 and 5 ppt). Interestingly, only males from BW population tended to survive at 11 ppt (Figure 4; Figure S1a,b).
The number of immatures in the BW population showed a twofold increase when they were raised in the foreign FW environment, although this increase was not as high as that of the FW population in the same environment (Figure 3d; Table 4). In addition, there were no statistical differences in egg size between both populations (t (51) = −1.73, p = .08; Figure S2). Overall, these results were confirmed by our GLMEMs, which showed significant differences in the four fitness correlates across treatments, as well as an interaction between treatment and population of origin. In addition, the number of immatures showed significant differences between populations of origin, and survival showed a significant effect of sex (Table 5).
These results were supported by our Kaplan-Meier analysis that showed that 50% of the individuals from the FW population were likely to survive for at least 30 days in their home FW environment,

BW FW
but only ~5-10 days at 1-5 ppt, and 2 days at 11 ppt ( Figure S4a). By contrast, 50% of the individuals from the BW population were likely to survive for at least 20 days in their home BW environment, up to 18 days at other salinities (FW and 3-5 ppt), and ~4 days at 11 ppt ( Figure S4b).

| Local adaptation and maladaptation
We found evidence for local adaptation (LA) in both FW and BW populations. For the FW population, we found strong LA in its home en- Error bars show mean and standard error. Inner plots show the ideal fitness (W ideal ) of the freshwater population (i.e., the ancestral FW population in its home FW environment) and the realized fitness (W realized ) of the brackish water populations (i.e., the derived BW population in its home BW environment), with the difference between the two values representing the degree of maladaptation for the brackish water population (see section 2)

| D ISCUSS I ON
Salinization due to sea-level rise is an increasing challenge for freshwater biodiversity. However, the extent to which freshwater organisms might be able to adapt to these changes is not well understood, particularly in Neotropical environments (Castillo et al., 2017). We explored this issue by quantifying fitness tradeoffs along a salinity gradient in two populations of the Neotropical water strider T. withei. We observed a strong effect of salinity on survival and reproductive traits for both FW and BW populations. The FW population showed strong fitness trade-offs along salinity levels, with evidence for local adaptation to its home FW environment, but not to high salinity levels. The BW population also showed fitness trade-offs along salinity levels, with evidence for weak local adaptation (for survival only) across salinity levels (1-5 ppt). However, the overall fitness of the BW population was only a fraction of that of the FW population in its home FW environment, indicating a high magnitude of maladaptation in the population persisting in BW environments. A similar pattern was observed when examining survival by sex, although males tended to show higher survival than females. In the following, we discuss the implication of these findings.
Thus, the selective role of salinity (i.e., its fitness consequences) has been less explored in coastal freshwater organisms (Gomez-Mestre & Tejedo, 2003;Kishi et al., 2006Kishi et al., , 2009Kozak et al., 2013). Here, we showed that salinity has a strong effect on survival of T. withei, with both FW and BW populations experiencing 50% mortality at salinities as low as 4 and 5 ppt, respectively. This is consistent with studies of temperate water striders (A. paludum, (Kishi et al., 2009;Kishi et al., 2006;Kishi et al., 2007); Gerris latiabdominis, (Kishi et al., 2013)), which are often found in similar salinity levels. This indicates that freshwater water striders are, in general, able to cope with some degree of salinization, with some species even inhabiting the open ocean (Halobates; (Cheng, 2005;Harada, 2005) were no significant differences in LC 50 values between FW and BW populations. This suggests a degree of "mismatch" (i.e., environmental mismatch) between the osmotic tolerance of populations and the range of salinities they experience in natural environments.
This also suggests that salinity tolerance experiments are good indicators of the upper osmotic tolerance of populations (here 11 ppt), which can inform experimental settings to explore adaptation to saline environments. However, the short-term nature of tolerance experiments and their focus on immediate survival rather than life-long reproductive success is likely to underestimate the fitness consequences of salinization in typical freshwater organisms (see the following section).

| Magnitude of adaptation and maladaptation
Selective pressures imposed by divergent environments often result in local adaptation, where populations evolve higher fitness in their own "home" environment than in the alternative "foreign" environment and vice versa (Endler, 1986;Hereford, 2009;Kawecki & Ebert, 2004;Schluter, 2000). However, the evolution of local adaptation along broad (and sometimes fluctuating) environmental gradients is likely more challenging (Gomez-Mestre & Tejedo, 2003;Polechová et al., 2009). This is because fluctuating environments are likely to result in variable strength and direction of selection (Grant & Grant, 2002), which could overcome the adaptive potential of populations (Brady, 2013;DeWitt & Yoshimura, 1998;Fox & Harder, 2015;Sinervo et al., 2010), particularly if migration is not an option (Atkins & Travis, 2019;Kleynhans et al., 2016).
In addition, previous work suggests that in variable environments, plasticity is more likely to evolve than a fixed trait (Ashander et al., 2016;Burggren, 2018;Ghalambor et al., 2007;Hadfield, 2016;Via & Lande, 1985). In the case of salinization, populations might experience variable levels of salinity, ranging from fresh to highly saline F I G U R E 4 Probability of survival in Telmatometra withei along a salinity gradient based on common garden experiments. Panels show logistic regressions across the entire data set (a), and by sex (b). Shaded area represents 95% confidence intervals waters (e.g., ~0.22-11 ppt at Playa Reina lagoon), which could result in periodic "mismatches" between the fitness of a population and its optimal osmotic niche (Gomez-Mestre & Tejedo, 2003;Negrin et al., 2019). Thus, populations seemingly persisting in specific salinity levels might in fact be maladapted, following drastic changes in salinity.

Salinity (ppt) Predicted probability of suvival
We explored this issue by quantifying the magnitude of local adaptation (i.e., fitness trade-offs along salinity levels; (Hereford, 2009)) as well as the "magnitude of maladaptation" (i.e., fitness differences between the ancestral FW population and the derived BW population in their home environments). Using these metrics, we found that the BW population showed apparent local adaptation to saline environments (1-5 ppt), but only for survival. However, its overall reproductive success was ~60% lower than that of the ancestral FW population in its home environment, suggesting a high magnitude of maladaptation in the BW population. Indeed, its overall life-long fitness (based on the number of offspring) was significantly higher when it was raised in the FW treatment ( Figure S3a), perhaps suggesting that the BW population is persisting away from the species' optimal osmotic niche. Thus, the physiological challenges imposed by osmoregulation in saline environments (Kozak et al., 2013;Potts & Parry, 1964;Rivera-Ingraham & Lignot, 2017;Sutcliffe, 1961) are likely to constraint the evolution of local adaptation in those environments. This pattern is consistent with an evolutionary mismatch (Hale et al., 2016;Lloyd et al., 2011;Marshall et al., 2010;Negrin et al., 2019;Robertson et al., 2013;Schlaepfer et al., 2002), whereby drastic environmental disturbances might overcome the adaptive potential of populations (Polechová & Barton, 2015;Polechová et al., 2009). In the case of T. withei, adaptation to saline environments could be limited by potential trade-offs between reproduction and survival. This was indicated by the fact that the BW population showed substantial survival in the high salinity treatments, but its overall fecundity and number of offspring were extremely low in the same treatments.
Similarly, the temperate water strider G. thoracicus is known to show high longevity (a trait associated with survival), but low reproductive output in treatments with low food supply (Kaitala, 1987), suggesting that water striders can effectively trade-off reproduction for survival when faced with stressful environments. This also suggests that our

F I G U R E 5
Patterns of local adaptation of fresh and brackish water populations of Telmatometra withei along a salinity gradient. Each panel shows the fitness advantage for the freshwater population (blue) in its home (FW) and foreign (1, 3, and 5 ppt) environment, and conversely, for the brackish water population (green) in its home (1, 3, and 5 ppt) and foreign (FW) environment cross fitness-related traits: survival (cross), fecundity (circle), oviposition rate ( observation of high adult survival in saline environments (both in the field and in the common garden experiments) may reflect phenotypic plasticity, rather than local adaptation. However, more work is needed to confirm this possibility.
Another possibility is the existence of preadaptation of the BW population to the ancestral FW environments (Geladi et al., 2019).
This could occur if the BW population is able to retain genetic variation associated with survival in the FW environments. In addition, given that the BW environment is highly variable, the BW population is likely to experience a broad range of salinities, including freshwater. At a broader scale, although freshwater salinization due to climate change is expected to increase globally (Courchamp et al., 2014;IPCC, 2007;IPPC, 2000), salinization could also decrease in areas with high precipitation (Gomez-Mestre & Tejedo, 2003;Short et al., 2016;Wrange et al., 2014). Therefore, retaining ancestral polymorphism associated with FW environments (i.e., preadaptation) could facilitate persistence of populations in these fluctuating environments. However, preadaptation to ancestral environments could also be costly, and it could compromise the evolution of local adaptation in novel environments (Atkins & Travis, 2019). Another possibility is gene flow, which could constrain local adaptation in novel environments (Farkas et al., 2015;Hendry & Taylor, 2004;Hendry et al., 2002;Kawecki & Ebert, 2004). In this case, gene flow from the FW population could swamp adaptation to high salinity environments-a likely possibility in our system, given the proximity between populations and the downstream location of the BW population.
An important question is how can maladapted (or partially adapted) populations persist in the face of increased salinization?
Maladaptation to a stressful environment could be overcome ex situ (Bolnick & Nosil, 2007;Farkas et al., 2016;Lenormand, 2002) if populations are able to disperse to less stressful environments (Defaveri & Merila, 2014;Farkas et al., 2015). This is certainly a possibility for T. withei, given that we have observed in the field a high frequency (11%) of winged individuals in the BW population (Figure 1d), in contrast to FW populations (<1.5%; Figure 1c). In other water strider species (A. paludum, (Kishi et al., 2007;Kishi et al., 2013)), wing development has also been associated with changes in salinity (Kishi et al., 2006(Kishi et al., , 2007(Kishi et al., , 2009, which could allow for dispersal to less saline environments (Kishi et al., 2006(Kishi et al., , 2007. Thus, perhaps a combination of partial adaptation and dispersal and recolonization is a likely mechanism promoting persistence of populations in these fluctuating environments. Another possibility is phenotypic plasticity rather than genetic adaptation. For instance, similar to other systems (Ashander et al., 2016;Burggren, 2018;Crispo et al., 2010), plasticity could facilitate persistence of populations along salinity gradients, which could buy time for adaptation to evolve, a possibility that requires further research.

| Future work
Although we showed evidence for both adaptation and maladaptation in T. withei, we consider these results as preliminary, given that only two populations were included in our analyses. Thus, several questions remain to be explored. For instance, what are the physiological consequences of salinization as well as the plastic or genetic mechanism underlying local adaptation in T. withei. In addition, what is the extent of gene flow across FW and BW populations, and how it might promote or constraint adaptation (Farkas et al., 2015;Hendry & Taylor, 2004;Kawecki & Ebert, 2004) in this system is an open question. Finally, the role of demographic factors such as population size in mediating population persistence (Bell & Gonzalez, 2011;Gomulkiewicz & Holt, 1995) in T. withei needs to be considered.
Overall, although more work is clearly needed, our analysis of fitness trade-offs along a salinity gradient revealed several aspects of local adaptation that are difficult to observe in studies of discrete environments. First, adaptation to extreme salinities in T. withei may be limited, given that both FW and BW populations failed to survive at salinities beyond 5 ppt. Thus, persistence of populations in high salinity environments may be facilitated by phenotypic plasticity rather than local adaptation. Second, if it occurs, local adaptation to broad and fluctuating environmental gradients is costly (Hereford, 2009), and could result in maladaptation to those environments. Third, preadaptation to ancestral environments is important in determining the magnitude of local adaptation in novel-disturbed environments.
Finally, dispersal ability could facilitate persistence of seemingly maladapted populations along variable environmental gradients.

| CON CLUS ION
In summary, our results based on two populations of the Neotropical water strider T. withei suggest that variable conditions along environmental gradients such as salinization of coastal freshwaters are likely to result in evolutionary mismatch, where the fitness of a population is periodically decoupled from its optimal environment. From a theoretical perspective, quantifying the magnitude of adaptation and maladaptation along environmental gradients will inform the role of adaptive evolution in the persistence of biodiversity in variable environments. From a practical perspective, it will allow the development of "evolutionary-informed" management strategies to address biodiversity issues in the context of global change. Overall, however, further work along a broad range of taxa and populations is needed to confirm the generality of our findings.

ACK N OWLED G M ENTS
This work was supported by the Secretaría Nacional de Ciencia, Tecnología e Innovación (SENACYT, Panamá), in the form of a doctoral fellowship to AMC and research grants (No. ITE12-002 and FID16-116) to LFD. Additional support was provided by Instituto para la Formación y Aprovechamiento de los Recursos Humanos in the form of a doctoral fellowship to AMC (No. 270-2013-284).
AMC was also supported by the Sistema Nacional de Investigación (No.152-2018, SNI, Panamá). LFD is supported by the University of Massachusetts Boston. We thank D. Sharpe for providing feedback during the design of the experiments.