An evolutionary innovation for mating facilitates ecological niche expansion and buffers species against climate change

Significance

Determining the factors that govern species’ ecological limits is fundamental to understanding the processes that generated the diversity of life over historical timescales and to forecasting how organisms will respond to global change over the coming decades. Although traits involved in mating have traditionally been excluded from consideration of ecological niche access, the results from our experimental, comparative, biogeographic, and citizen-science analyses collectively show that a trait that facilitates male mating activity has expanded species’ tolerances of climatic conditions and buffered them against contemporary climate change. This study therefore demonstrates that traits involved in mating can be just as influential to species’ climatic limits as traits involved in growth and survival.

Abstract

One of the drivers of life’s diversification has been the emergence of “evolutionary innovations”: The evolution of traits that grant access to underused ecological niches. Since ecological interactions can occur separately from mating, mating-related traits have not traditionally been considered factors in niche evolution. However, in order to persist in their environment, animals need to successfully mate just as much as they need to survive. Innovations that facilitate mating activity may therefore be an overlooked determinant of species’ ecological limits. Here, we show that species’ historical niches and responses to contemporary climate change are shaped by an innovation involved in mating—a waxy, ultra-violet-reflective pruinescence produced by male dragonflies. Physiological experiments in two species demonstrate that pruinescence reduces heating and water loss. Phylogenetic analyses show that pruinescence is gained after taxa begin adopting a thermohydrically stressful mating behavior. Further comparative analyses reveal that pruinose species are more likely to breed in exposed, open-canopy microhabitats. Biogeographic analyses uncover that pruinose species occupy warmer and drier regions in North America. Citizen-science observations of Pachydiplax longipennis suggest that the extent of pruinescence can be optimized to match the local conditions. Finally, temporal analyses indicate that pruinose species have been buffered against contemporary climate change. Overall, these historical and contemporary patterns show that successful mating can shape species’ niche limits in the same way as growth and survival.

Much of the Earth’s biological diversity has arisen from organisms gaining access to underused habitats and resources (1–3). Although many processes can facilitate exploitation of novel ecological niches, a common trigger of niche expansion is thought to be the evolution of traits or trait combinations that allow organisms to interact with the environment in a new way (4–6). So-called “evolutionary innovations” indeed appear at the base of many spectacularly diverse clades, including the feathers of birds (7), the water vessels of terrestrial plants (8), and the complete metamorphosis of holometabolous insects (9). Research to date on the identity and impact of evolutionary innovations has focused exclusively on those traits that improve aspects of growth and survival like foraging, locomotion, and defense (5, 10, 11). By contrast, characters that mainly function in mating have traditionally been ignored or actively discounted from discussion of evolutionary innovations because of the assumption that mating activities drive lineage diversification without generating much or any ecological differentiation (3, 6, 7). However, without successful mating, an environment will not be able to support a sexually reproducing species irrespective of how well it grows and survives there (12, 13). Moreover, organisms often experience an environment’s most stressful conditions while engaged in mating activities (14). Thus, despite a longstanding delineation between mating and ecological interactions, the evolution of traits that permit mating behaviors in a new environment could actually be a vital feature of niche expansion and the ecological diversification of life on Earth. Nevertheless, it is unclear whether traits that evolve for mating functions can subsequently determine a clade’s ability to access new or underused ecological niches.

We used North American dragonflies to explore whether species’ niche access can be driven by a character that facilitates mating activity—a hydrophobic ultra-violet (UV)-reflective pruinescence that covers the bodies of males (15, 16). Comprised of an unusual mix of long-chain methyl ketones and aldehydes, pruinescence is found in several arthropod clades, but it is typically produced only as a fine coating (17, 18). By comparison, the males of many dragonfly species produce a dense covering of pruinescence that can cover most or all of their bodies (Fig. 1A). Among dragonflies, it is found in two of the most speciose families: the Libellulidae and, to a lesser extent, the Gomphidae (19–21). Prior research has implicated this pruinescence in mating for two reasons. First, expression of the waxy substance is limited almost entirely to mature males, with, for example, only one North American species normally exhibiting female pruinescence (19, 20). Second, behavioral studies have shown that males leverage pruinescence’s UV-reflectance to intimidate rivals and court potential mates (22). In light of pruinescence’s biomaterial properties, sex-specific production, and behavioral function, we investigated whether this mating-related trait in dragonflies confers physiological benefits that expand species’ ecological niches and influences how they respond to climate change.

Fig. 1.

The diversity of male pruinescence. (A) Examples of male pruinescence. Top: Leucorrhinia intacta (PC: S. Zundell) vs. Leucorrhinia frigida (PC: T. Rickett); Middle: Libellula needhami (PC: S. Baran) vs. Libellula luctuosa (PC: N. Walther); and Bottom: Erythemis vesiculosa (PC: C. Bentley) vs. Erythemis collocata (PC: A. Parsons). (B) Thermohydrically stressed percher species are more likely to possess pruinescence than flier species. Top: Numbers below silhouettes represent number of Nearctic species. Bottom: Squares are estimated-marginal means ± 95% CIs from 1,000 bootstrap replicates. (C) Species that defend mating territories at open-canopy sites are more likely to possess pruinescence than those that use forested sites. Top: Numbers below silhouettes represent number of closed-canopy and open-canopy Nearctic species. Bottom: Squares are estimated-marginal means ± 95% CIs from 1,000 bootstrap replicates, and letters represent significantly different groups. Clip art from Vecteezy.

Given pruinescence’s light reflectance and hydrophobic properties, we began by experimentally testing whether pruinose dragonflies are physiologically buffered against 1) overheating when exposed to direct light and 2) water loss when faced with dry conditions. We captured male blue dashers (Pachydiplax longipennis) and common whitetails (Plathemis lydia) and then gently removed the pruinescence from half of the individuals in each species. We first assessed whether pruinescence’s light reflectance properties could defend males against overheating. In this experiment, we heated males under a lamp and assessed if individuals with intact pruinescence heated less than males without pruinescence. Consistent with our prediction, pruinose males heated to lower maximum temperatures than non-pruinose males in both species (P. longipennis = −2.14 ± 0.97 °C, 95%CIs: −0.25 to −4.02; P. lydia = −1.69 ± 0.22, 95%CIs: −1.26 to −2.12; Fig. 2 A and C and SI Appendix, Table S1). We next explored whether pruinescence’s waxy, hydrophobic properties could buffer males against water loss. In this second experiment, we maintained live males in a drying chamber at 20% relative humidity for 10 h and tested whether males with intact pruinescence lost less mass than males without pruinescence. Indeed, non-pruinose males lost more mass than pruinose males in both species (P. longipennis = 38.7 ± 6.7%, 95%CIs: 25.5 to 52.1%; P. lydia = 57.6 ± 8.2%, 95%CIs: 42.0 to 73.1%; Fig. 2 B and D and SI Appendix, Table S1). These physiological tests in two different genera suggest that pruinescence can prevent water loss and overheating.

Fig. 2.

Physiological benefits of male pruinescence. (A and B) Males that had their pruinescence removed (gray) reached higher maximum temperatures after 5 min of heating than males with intact pruinescence (blue). Points are the estimated maximal temperatures gained above the starting temperature for each individual, and squares are the estimated marginal means ± 95% CIs. (C and D) Males that had their pruinescence removed (grey) lost proportionally more mass after 10 h at 20% humidity than males with intact pruinescence (blue). Points are the proportions of mass that each individual male lost, and squares are estimated-marginal means ± 95% CIs.

Although our experiments show that pruinescence offers physiological advantages that could benefit both sexes, it is produced almost exclusively by males (19–21). One potential explanation for this sex-specific expression could be the divergent reproductive behaviors of male vs female dragonflies (16, 19, 20). Males spend most of their time each day defending a mating territory and might miss out on rare mating opportunities if they leave the reproductive site to thermoregulate, drink, or eat (23). By contrast, female dragonflies are thought to be far less thermohydrically stressed than males because they spend their time foraging in cooler microhabitats, only briefly and infrequently visiting the reproductive site to copulate and oviposit (16). Thus, pruinescence’s main advantage might be that it protects males against the hot and dry conditions they endure while defending mating territories.

We next assessed whether pruinescence might have evolved to help males engage in thermohydrically stressful mating activities. To test this hypothesis, we specifically compared male pruinosity among species based on the behavioral mating strategies that males adopt at the reproductive site. In the “percher” strategy, males typically remain perched in their territories and fly mainly to ward off rivals or to court mates (24). Males of percher species risk losing their territory if they leave their perches to thermoregulate or drink (16, 23). Conversely, in the “flier” strategy, males always remain in flight while defending their territory, which allows them to convectively cool and drink water without having to abandon the reproductive site (16, 23, 24). We conducted a phylogenetic logistic regression across 319 Nearctic species and, indeed, found that perchers are more likely to be pruinose than fliers ( $\beta$ = 2.221 ± 0.705, 95%CIs: 1.121 to 3.909; Fig. 1B). This pattern suggests either that 1) pruinescence evolved in response to males adopting the thermohydrically stressful percher strategy or 2) prior evolution of pruinescence permitted males to subsequently adopt percher behaviors. To estimate the likelihood of each scenario, we conducted Pagel’s test for correlated evolution (25). The best-supported model of evolution showed that male pruinescence evolved after species began using the percher strategy (Q = 0.034 transitions/My) >4 times more frequently than before (Q = 0.008; SI Appendix, Tables S2 and S3). Moreover, pruinose species that transitioned to the flier strategy almost invariably lost male pruinescence (Q = 0.111) and never gained it (Q = 0.000). This finding implies that selection during physiologically stressful mating activities favored the evolution of male pruinescence, perhaps as a way for males to more effectively withstand the inclement conditions at the reproductive site.

Our results are consistent with the hypothesis that male pruinescence evolved to ease physiological stress while defending territories. However, the physiological advantages of male pruinescence could be beneficial in several other contexts as well. For example, defense against overheating and water loss might allow pruinose species to exploit warmer and/or drier microhabitats when defending territories and courting mates. Species that are capable of using those stressful abiotic conditions for mating should gain a reproductive advantage because they will have greater access to exposed, sunlit microhabitats that aid the detection and evaluation of mating signals (26). Indeed, despite preferring to mate in exposed microhabitats, many animal species are restricted from using these portions of their current environments because the risks of surpassing physiological limits outweigh the ease of reproductive opportunities (14). Nevertheless, when males use these microhabitats for mating, it can affect many aspects of the species’ ecology. For example, since female dragonflies typically oviposit their fertilized clutches into the water or vegetation directly below the sire’s territory (16), shifting the mating microhabitat will also alter the natal environment of any developing offspring. Consequently, by broadening the range of abiotic conditions that species can tolerate during mating, a mating trait like male pruinescence could actually give species access to microhabitats that would otherwise be relatively inaccessible.

To examine whether pruinescence allows dragonflies to exploit more exposed, sunlit microhabitats, we tested whether species that use open-canopy water bodies for mating are more likely to possess male pruinescence than those that use closed-canopy water bodies. For the 319 Nearctic species, we compiled data on whether each species forms breeding aggregations in open-canopy water bodies, closed-canopy water bodies, or both (19, 20). We conducted a phylogenetic logistic regression and, consistent with the hypothesis, found that species that breed in open-canopy microhabitats are more likely to possess male pruinescence than those that breed in forested habitats ( $\beta$ = 1.04 ± 0.41, 95%CIs:0.257 to 1.775; Fig. 1C and SI Appendix, Table S4). To test whether the evolution of male pruinescence preceded access to these microhabitats or whether male pruinescence mainly evolved after species already began using them, we conducted another Pagel’s test (25). The best-supported set of models indicates that pruinose species expanded into open-canopy microhabitats (Q = 0.111) >4 times more frequently than open-canopy species subsequently evolved male pruinescence (Q = 0.027; SI Appendix, Tables S5 and S6). Moreover, non-pruinose species transitioned from open-canopy back to closed-canopy microhabitats (Q = 0.095) >2.4 times more frequently than pruinose species did (Q = 0.039), which suggests non-pruinose taxa had difficulty persisting in those microhabitats. Finally, a supplemental analysis shows that species’ male mating strategies evolved independently from the use of these microhabitats (SI Appendix, Table S7). Collectively, the evolutionary patterns uncovered here indicate that prior evolution of male pruinescence in the mating context granted species lasting access to new ecological niches.

Exploitation of a greater portion of the local environment is a hallmark of evolutionary innovations in general, and such a pattern has frequently been observed for innovations that affect foraging, growth, and/or survival (5, 6). Our results emphasize that mating-related traits too can alter the ecological conditions that species use for reproduction and, subsequently, offspring development. However, by broadening the range of suitable mating conditions, male pruinescence might do more than just improve access to microhabitats within the current environment. For instance, evolutionary innovations that facilitate growth and survival have enabled species to colonize entirely new geographic areas (27, 28). Since recent research reveals that species’ geographic ranges can be more strongly shaped by physiological limits on mating than by those on survival (12, 13), traits that permit mating under more stressful physiological conditions could also expand species’ geographic ranges (Fig. 3A). In the case of male pruinescence, the physiological advantages of pruinose species might have enabled them to colonize and persist in parts of North America that became particularly warm and dry after the Last Glacial Maximum.

Fig. 3.

Biogeographic patterns of male pruinescence among species. (A) Predicted patterns of Nearctic re-colonization after the Last Glacial Maximum if male pruinescence expands species’ climatic niches. Arrow thickness reflects the frequency of a pruinose (blue) or non-pruinose (grey) species re-colonizing a given area as a function of its temperature and aridity. (B) Local dragonfly assemblages across the Nearctic. The color of the local area represents the percent of species in the area that possess male pruinescence. Black = 0% pruinose species, and light blue = 100% pruinose species. (C and D) Species with warmer (C) and drier (D) ranges are more likely to exhibit male pruinescence. Lines are estimated-marginal trends, and bands are 95% confidence ranges from 1,000 bootstrap replicates. Tick marks represent individual species that are either pruinose (Top) or non-pruinose (Bottom).

To test whether pruinose species occupy hotter and/or more arid geographic regions, we examined current biogeographic patterns of pruinescence across the Nearctic. We estimated the mean temperature and aridity that Nearctic species encounter across their ranges using >387,900 GBIF occurrence records (Fig. 3B) (SI Appendix). We then conducted phylogenetic logistic regression to test whether species with warmer and/or drier ranges are more likely to possess male pruinescence. Consistent with this prediction, male pruinescence is most common in species with the warmest ranges ( $\beta$ = 0.416 ± 0.232, 95%CIs: 0.017 to 0.846; Fig. 3C) and driest ranges ( $\beta$ = −0.373 ± 0.194, 95%CIs: −0.715 to −0.018; Fig. 3D). Importantly, there was no relationship between a species’ male mating behavior and either environmental variable, suggesting that the geographic distribution of perchers does not underlie the distribution of male pruinescence (SI Appendix, Table S8). The biogeographic patterns were also notably pronounced for species with pruinose abdomens (SI Appendix, Table S9), which is the primary location of the respiratory spiracles through which water is lost (16). Thus, in this current interglacial period, species that have persisted in some of the Nearctic’s most physiologically demanding regions are also those that are most likely to possess pruinescence. Though more research remains necessary, the strong relationship between abdominal pruinescence and aridity further suggests that defense against water loss may have been an especially important function of male pruinescence as species dispersed into their current ranges. Ultimately, by expanding the breadth of conditions under which males can engage in breeding activity, an evolutionary innovation for mating has determined the geographic regions across the landscape that species are able to occupy.

Our results show that the evolution of male pruinescence has allowed species to mate in hotter and drier habitats. However, the evolutionary rates of gaining and losing male pruinescence are quite slow (1 gain every ~55.5 My, 1 loss every ~7.5 My; SI Appendix, Table S10), suggesting that male pruinescence evolved long before species moved into their current geographic ranges. Thus, as with classic evolutionary innovations for ecological characters, it seems likely that prior evolution of male pruinescence exaptively facilitated expansion into species’ current niches (4, 10, 29). In addition to exaptive functions, many ecological innovations nevertheless exhibit considerable adaptive potential as well—that is, once the innovation is gained, it continues to diversify to match local conditions. For example, the evolution of lizard toe pads aided access to vertical substrates; but after lizard clades gained toe pads, the pads have continued to evolve to meet the demands of the local environment by increasing in size or re-arranging the adhesive structures (30, 31). We therefore wanted to test whether a mating innovation like male pruinescence could also be adaptively fine-tuned to the local environment. A common approach to answering such a question is ancestral character reconstruction. Unfortunately, ancestral reconstructions are fraught when estimating geographic ranges and climate-related traits in ancient continental assemblages because continental drift and glaciation cycles have decoupled species’ current ranges from the conditions under which the traits likely originated (32). An alternative method to addressing whether an evolutionary innovation can be fine-tuned to the local environment is to examine intraspecific variation in the trait (29, 33). In this approach, if male pruinescence has the adaptive capacity to match local conditions, then we should observe that populations of pruinose species produce more male pruinescence in warmer and/or drier parts of their geographic ranges (Fig. 4A).

Fig. 4.

Geographic variation of male pruinescence within P. longipennis. (A) Potential patterns of the additional male pruinescence that would suggest exaptation (Top) or adaptation to local conditions (Bottom). Arrow thickness reflects the frequency of occurrence. (B) Actual geographic occurrences of males with extra pruinescence. Points represent iNaturalist observations of P. longipennis males that either had abdominal pruinescence only (light blue) or abdominal and thoracic pruinescence (dark blue). (C and D) Males were more likely to have abdominal and thoracic pruinescence in the driest parts of the species range (D), but there was no relationship with temperature (C). Lines are estimated-marginal trends and bands are 95% confidence ranges from 1,000 bootstrap replicates. Tick marks are individual iNaturalist observations of males that either had abdominal pruinescence only (Bottom) or both abdominal and thoracic pruinescence (Top).

P. longipennis is well suited to exploring geographic patterns of pruinosity within species because 1) it is distributed across a wide array of environments across southern Canada, the United States, and northern Mexico (34, 35); and 2) some males produce pruinescence only on their abdomens, whereas others produce it on both their abdomens and thoraxes (20). To characterize geographic patterns in pruinosity, we thus examined 436 georeferenced iNaturalist observations from across the species’ range and determined whether the observed male had pruinescence only on its abdomen or on both its abdomen and thorax (Fig. 4B). We then used a generalized linear mixed-effects model to test whether males living in the warmest and/or driest parts of the species’ range typically exhibit greater pruinosity (35). Our results show that males are more likely to possess additional pruinescence in the driest part of the species’ geographic range ( $\beta$ = 3.941 ± 0.640, 95%CIs: −5.375 to −2.813; Fig. 4D). However, there was no relationship between pruinosity and temperature ( $\beta$ = −0.430 ± 0.361, 95%CIs: −1.185 to 0.291; Fig. 4C). These patterns indicate that male P. longipennis produce more pruinescence in places with the lowest local moisture levels. Since developmental studies have not yet been conducted to assess whether warmer and drier temperatures induce the production of more pruinescence, it remains to be seen whether these results are due to phenotypic plasticity, local adaptation, or some combination of the two (36). Moreover, additional work remains necessary to uncover the specific physiological function of thoracic pruinescence and to assess whether similar patterns occur in other species. Nonetheless, the intraspecific pattern of male pruinescence in P. longipennis suggests that this mating innovation can be adaptively optimized to match the local conditions as is seen in other ecological innovations.

Through a combination of both exaptive ecological filtering and, possibly, local optimization, male pruinescence has expanded species’ climatic niches over historical timescales and improved their ability to mate in underused habitats. An evolutionary innovation for mating has therefore granted species’ access to new climatic niches in a manner similar to that seen for traits involved in growth and survival (5, 6). However, global warming is currently upending species’ historical interactions with their environments (37). Recent work has shown that these novel climatic conditions are disrupting mating and fertilization far more than they are hindering growth and survival (12). As a result, species that have difficulty mating in a warmer and drier future could be most vulnerable to local extinction (14). Conversely, those taxa that have previously evolved traits that permit mating under more stressful conditions, like male pruinescence, might be buffered from climate change.

We assessed how contemporary changes in temperature and precipitation have affected the likelihood of local extinction for pruinose and non-pruinose species across the United States. We leveraged an existing dataset on the local persistence and extinction of 60 widespread species within 100 km × 100 km grid cells across the contiguous United States (38). For each of the 385 geographic areas with estimates of local extinctions, we calculated how temperature and precipitation have changed between the 1980s and the present (39). We then used a generalized linear mixed-effects model to assess the relationship between species’ probabilities of local extinction and climatic shifts in each geographic area. Our results show that non-pruinose species were less likely to persist in areas with greater increases in temperature ( $\beta$ = −0.338 ± 0.122, 95%CIs: −0.557 to −0.096) whereas pruinose species have been unaffected ( $\beta$ = −0.365 ± 0.224, 95%CIs: −0.804 to 0.075; Fig. 5A and SI Appendix, Table S11). Similarly, although pruinose and non-pruinose species display similar probabilities of extinction under the most extreme levels of aridification (Fig. 5B), the extinction risk for pruinose species rapidly declines with less severe drying ( $\beta$ = 0.661 ± 0.293, 95%CIs: 0.087 to 1.234; SI Appendix, Table S11). These findings indicate that the last 40 y of climate change have been less detrimental to pruinose species than to non-pruinose species. Thus, in addition to shaping species’ interactions with their historical climates, an evolutionary innovation for mating has also affected how species are responding to contemporary climate change.

Fig. 5.

Sensitivity of pruinose (blue) versus non-pruinose (grey) species to changes in temperature and precipitation over the last 40 y. Lines are estimated-marginal trends, and bands represent 95% confidence ranges from 1,000 bootstrap replicates. (A) Greater increases in local temperatures led to lower probabilities that a non-pruinose species will have persisted there but had no effect on the persistence of pruinose species. (B) Pruinose species were less vulnerable to extinction than non-pruinose species at all changes in precipitation except the most severe levels of drying.

Overall, our physiological, comparative, biogeographic, citizen-science, and temporal analyses show that a trait related to mating is an evolutionary innovation that has expanded species’ ecological niches and is currently buffering them against global change. A species’ access to an ecological niche can therefore be determined just as much by traits that facilitate mating activity as by traits that enable foraging, locomotion, and defense (5, 10, 11). To be sure, sexual selection can operate independently from other ecological interactions (40), but our results argue that it has been shortsighted to exclude mating-related traits from all consideration about the sources of niche expansion. Organisms often encounter the most challenging conditions they will ever face when engaged in mating activities (14), and the exploitation of new ecological niches may therefore depend just as much on evolutionary innovations for mating as those for growth and survival.

Although reproductive traits related to gamete characteristics or parental care have long been integrated into discussions about evolutionary innovations (8, 41), mating characters have not received similar attention despite a similarly lengthy appreciation of their ecological advantages. However, ecologically beneficial mating characters are thought to often be co-opted by the other sex (42, 43)—a pattern that has been extremely rare among Nearctic dragonflies (19–21). Since female dragonflies typically spend little time at thermohydrically stressful reproductive sites (16), they are unlikely to benefit as much from pruinescence as males do. Nevertheless, it would be valuable for future research to assess whether the energetic costs of production are sufficient to negate any small physiological advantages that females may gain. While the co-option of ecologically advantageous sexual traits is a commonly invoked hypothesis (43), many species also experience dramatically divergent environmental pressures owing to the mating system (14, 44). If the territorial sex cannot remain at the reproductive site, the other sex may have difficulty finding mates when they arrive, and the subsequent breakdown of the breeding system will limit the species’ ability to persist in the habitat (44, 45). Sex-specific innovations, like male pruinescence, should resolve this problem and allow species to exploit niches that otherwise may be hard to access. Our findings argue that future investigation into the taxonomic distribution of these mating innovations will undoubtedly help us better understand the ecological diversification of life.

Biologists as far back as Darwin have drawn distinctions between “ecological” and “sexual” interactions (46). Even today, to the extent that biologists consider interactions between these two components of an organism’s life, we typically focus on the ways that ecological interactions shape and constrain the evolution of species’ sexual traits (26, 47). Our findings here suggest that the opposite pattern can also occur: The success of sexual interactions places limits on species’ ecological characteristics. Such a result indicates that the long-assumed boundaries between ecological and sexual evolution are much more porous than assumed by traditional models (14). Moreover, in light of the speed and intensity with which our planet is changing (37, 48), the consequences of this oversight are no longer semantic. Discounting the role of mating activities on species’ niches can obscure the historical drivers of ecological diversification and inhibit forecasts for where and how species will live over the coming decades. When we consider the processes that shape historical and future niches, the findings of our study ultimately show that we must broaden our focus to include the full range of organismal activities throughout the life cycle (3, 7).

Methods

Physiological Trials.

We captured wild P. longipennis and P. lydia males, anesthetized them overnight at 4 °C, and then randomly assigned each male to an experimental manipulation or to a control group. In the experimental manipulation group, we used a scalpel to gently scrape off as much of the abdominal pruinescence as possible. In the control group, we mimicked the scraping motion with the scalpel but did not remove any of the pruinescence. It is plausible that the removal has other physiological effects on the individuals. We then measured each male’s body mass (g) and subjected it either to a heating trial or a desiccation trial. For heating trials, we inserted a thermal probe into each male’s thorax, placed the male 20 cm below an illuminated 60W lamp, and measured how their internal temperature changed every 30 s for 5 min. We then compared patterns of temperature gain through time between manipulated males and control males using an asymptotic non-linear mixed-effects model [“nlme” (49)]. For desiccation trials, we secured live males inside 10 cm × 13 cm mesh bags, placed them in a 20% relative humidity chamber, and measured how much mass that they lost after 10 h. We then compared mass loss between manipulated and control males using linear mixed-effects models [“lme4” (50)] (SI Appendix).

Behavioral Mating Strategy, Breeding Habitat Use, and Interspecific Ranges.

For our comparative analyses, we first created a database of species’ male pruinescence (y/n; location on the body), behavioral mating strategies (percher/flier), breeding microhabitats (closed-canopy, open-canopy, both), and climatic conditions across their geographic ranges. Data on species’ pruinescence, behavioral strategies, and breeding microhabitats were gathered from two comprehensive field guides of North American dragonflies and species reports (19–21). Data on the average temperature and aridity across each species’ range came from previously reported estimates of each species’ biogeographic ranges and climatic conditions [(34) see SI Appendix for Global Biodiversity Information Facility datasets, methods, and rationale). In all analyses, we used the phylogeny of Nearctic dragonflies produced by Rocha-Ortega and colleagues (51), which we pruned to include only those species for which we had data.

Mating strategy.

We used a phylogenetic logistic regression to test whether species that employ the percher mating strategy are more likely to be pruinose than species that employ the flier strategy [“phylolm” (52)]. We next used Pagel’s test for correlated evolution to estimate the transition rates between the four character states (25): pruinose-percher, pruinose-flier, non-pruinose-percher, and non-pruinose-flier (SI Appendix).

Breeding habitat.

We used a phylogenetic logistic regression to test whether species that defend breeding territories in open-canopy habitats were more likely to possess male pruinescence. We next used Pagel’s test for correlated evolution to estimate the transition rates between the character states. Because Pagel’s test uses binary characters, we pooled species that use both habitats into the “open” category, giving us the following character states: pruinose-open, pruinose-closed, non-pruinose-open, and non-pruinose-closed (SI Appendix).

Interspecific ranges.

We used phylogenetic logistic regression to test whether species with warmer and/or more arid ranges are more likely to possess male pruinescence. We also conducted a supplemental phylogenetic logistic regression to examine whether certain climatic conditions favored pruinescence on particular body segments (abdominal pruinescence [y/n]; thoracic pruinescence [y/n]) (SI Appendix).

P. longipennis Geographic Variation.

We leveraged iNaturalist observations of P. longipennis to examine intraspecific variation in male pruinosity. To choose and download useable images, we followed validated protocols described elsewhere [(33, 35, 53), see also SI Appendix]. For each useable observation, two observers independently scored whether the male had pruinescence on its abdomen or on both its abdomen and thorax. We also downloaded the mean annual temperature and the precipitation in the driest quarter for the location of each observed male (39). We then used a generalized linear mixed-effects model to test whether additional male pruinescence (abdomen only = 0, abdomen + thorax = 1) was associated either with the local temperature or aridity. We included a random effect for municipality (county or its equivalent) nested within state (USA, Mexico) or province (Canada) to account for non-independence of observations of males from within the same population, which previous work has shown provides similar results to using a spatially explicit logistic regression (35) (SI Appendix).

Responses to Contemporary Climate Change.

We combined our dataset on species’ pruinescence with published data on local extinctions for 60 species across the United States [(38); see also SI Appendix for datasets). Local extinction was indicated by a species failing to persist within a 10,000 km² grid cell between a historical (1980 to 2002) and contemporary timepoint (2002 to 2021). For each grid cell, we also calculated the change in the average monthly maximum temperature and precipitation (mm) between the timepoints (54). We then used a generalized linear mixed-effects model to examine whether male pruinescence (y/n) affected a species’ probability of local extinction in response to rising temperatures and declining precipitation (50). We included each grid cell’s historical temperature and precipitation as covariates. We also modeled the grid cell as a random effect to account for any additional non-independence associated with measures of local extinction taken from the same location. We also included a random effect of species nested within genus nested within family to account for the possibility that some species and higher taxonomic groups were more likely to go extinct than others. This structure fit as well as a model that directly modeled the phylogeny (see SI Appendix for full methods and Global Biodiversity Information Facility data).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All data and code used for statistical analyses are available on Zenodo (55).