The evolution of crocodilian nesting ecology and behavior

Abstract Crocodilians comprise an ancient and successful lineage of archosaurs that repeatedly raises questions on how they survived a mass extinction and remained relatively unchanged for ~100 million years. Was their success due to the change‐resistant retention of a specific set of traits over time (phylogenetic conservatism) or due to flexible, generalist capabilities (e.g., catholic diets, phenotypic plasticity in behavior), or some combination of these? We examined the evolution of reproductive ecology and behavior of crocodilians within a phylogenetic perspective, using 14 traits for all 24 species to determine whether these traits were phylogenetically constrained versus (ecologically) convergent. Our analysis revealed that the ancestral crocodilian was a mound nester that exhibited both nest attendance and defense. Nesting mode exhibited 4–5 transformations from mound to hole nesting, a convergence of which habitat may have been a driving factor. Hole nesters were more likely to nest communally, but this association may be biased by scale. Although there were exceptions, mound nesters typically nested during the wet season and hole nesters during the dry season; this trait was relatively conserved, however. About two‐thirds of species timed their nesting with the wet season, while the other third timed their hatching with the onset of the wet season. Nest attendance and defense were nearly ubiquitous and thus exhibited phylogenetic conservatism, but attendance lodging was diverse among species, showing multiple reversals between water and burrows. Collectively, our analysis reveals that reproductive trait evolution in crocodilians reflects phylogenetic constraint (nest attendance, nest defense), ecological convergence (seasonal timing of nesting, nest attendance lodging), or both (mode of nesting). Some traits (e.g., communal nesting and mode of nesting) were autocorrelated. Our analysis provides a framework for addressing hypotheses raised for why there has been trait convergence in reproductive ecology and behavior in crocodilians and why some traits remained phylogenetically conserved.


| INTRODUC TI ON
A modern goal of evolutionary biologists is to bridge the conceptual divide between microevolution and macroevolution (Bock, 1970;Hansen & Martins, 1996;Reznick & Ricklefs, 2009;Gavrilets & Losos, 2009;Rabosky, 2013). Although this goal is fraught with challenges, the stage may be set for re-evaluation; according to Erwin (2000) and Reznick and Ricklefs (2009), it is the integration of information from the fossil record, the population and evolutionary dynamics of extant organisms, and phylogenetics that will provide the ultimate test of the bridge between microevolution and macroevolution (but see Tobias et al., 2014). Regardless, macroevolution can inform microevolution: Information about the historical patterns of diversification of lineages can now be mined from molecular phylogenies, potentially shedding light on the underlying causes of these patterns (Reznick & Ricklefs, 2009).
If we define success as the longevity of a lineage, the clade that comprises the living crocodilians must rank as highly successful. The Crocodylia as we know them are ~100 million years old and carry the legacy of their archosaur cousins, the dinosaurs, as the largest reptiles on earth. They present modest species richness (25 species), generalist habits, and wide-ranging, amphibious capabilities. In any case, their success raises questions on what facets of their biology might explain their longevity as a group-for example, how did the crocodilian lineage survive the K-T extinction when their cousins the dinosaurs did not? Was it due to the change-resistant retention of a specific set of natural history traits over geological time (phylogenetic constraint) or due to their generalist capabilities (e.g., catholic diets, phenotypic plasticity in behavior), or some combination of these?
The most recent review of crocodilian biology seems to argue for the former; Grigg and Kirshner (2015) repeatedly highlight similarities in the physiology, ecology, and behavior within the group.
For example, they note that differences in the ecology and behavior between the Crocodylidae and Alligatoridae are "surprisingly few," given their existence for ~100 million years. Indeed, all living crocodilians are aquatic generalist (strict) carnivores, engage in mating vocalizations, possess temperature-dependent sex determination, and exhibit considerable parental care to their eggs and neonates (reviewed in Webb & Manolis, 1989, and in Grigg & Kirshner, 2015).
Further, when considering long-lived iteroparous species, parental behavior, nesting habitat, and size-based reproductive output vary greatly among species. Marked variation in egg mass, clutch size, and clutch frequency with respect to maternal size among crocodilian species (reviewed in Thorbjarnarson, 1996) lends to evolutionary lability in behavioral and reproductive traits as an explanation for their historical success. According to Grigg and Kirshner (2015), "no explanation has yet emerged that accounts for this striking dichotomy in behavior." The opportunity for understanding the evolution of reproductive ecology and behavior has arguably never been better. First, a recent robust phylogenetic analysis (Oaks, 2011) provides a framework for tracking reproductive trait evolution in crocodilians. Second, there has been a major surge in new information on the biology of crocodilians in recent years and decades (Grigg & Kirshner, 2015). For example, of the ~1,100 publications cited in Grigg and Kirshner (2015), 90% have been published since 1970 and 45% of them since 2000.
Naturally, reviews and syntheses have lagged behind, including those on reproductive ecology and behavior (but see Magnusson, Lima, & Sampaio, 1985;Thorbjarnarson, 1996, for exceptions). Data on reproductive ecology are now available for all 24 species; however, and collectively, these prospects make crocodilians a good choice for studying reproductive trait evolution.
Here, we examine the evolution of reproductive ecology and behavior of crocodilians from a phylogenetic perspective, using 14 traits for all 24 species. Specifically, we pose the following questions: (a) Which traits associated with crocodilian reproductive ecology and behavior are phylogenetically constrained versus (ecologically) convergent? (b) How do these findings accord with previous hypotheses on reproductive ecology and behavior in crocodilians? (c) What were the reproductive traits (character states) of the hypothetical ancestor(s) of the Crocodylia? We discuss interrelationships between different traits within the context of the ecology and behavior of crocodilians and pose further questions for investigation.

| ME THODS
The evolutionary relationships of crocodilians have been plagued by incongruence among studies (Brochu, 2003;Poe, 1996). Aside from questions pertaining to the timing of diversification, there has been no consensus on the placement of the false or Sundha gharial (Tomistoma sp.) as sister to Gavialis or to the family Crocodylidae (Gatesy, Amato, Norell, DeSalle, & Hayashi, 2003). However, a recent robust analysis (Oaks, 2011) yields a monophyletic Gavialidae (comprised of Tomistoma and Gavialis) as sister to Crocodylidae, and provides an opportunity to determine the taxonomic distribution, and thus reproductive trait evolution, in crocodilians.
Character states for 14 characters were collected for all 24 species from published literature ( Table 1). The suite of characters chosen attempts to characterize all the variable modes of behavioral and ecological investment allocated to reproduction among crocodilians.
The character states overwhelmingly found in the literature were assigned to that taxon. For example, the character state for nest defense in the American crocodile (Crocodylus acutus) was assigned as "absent" (obtained from Charruau & Hénaut, 2012) despite a note of present TA B L E 1 Characters and cited published literature sources

Species
Character Literature source(s)

No data
Hatching stimulus (signal calling, vibration)

No data
Nesting season (wet, dry) Stuebing et al.

| Mating vocalizations
Accelerated transformations (AT) and delayed transformations (DT) indicate a mating roar as the ancestral state among crocodilians (Figure 1a,b). This plesiomorphy was followed by the independent and convergent evolution of bellowing vocalizations in Alligator as well as Caiman latirostris, and Crocodylus mindorensis of the Pacific Philippine Islands. Additionally, the independent evolution of no vocalizations, a "buzz," and "moan" occurred in the monotypic genera Tomistoma, Gavialis, and Osteolaemus, respectively (Figure 1).
Discrepancies among AT and DT result from an algorithmic assignment of the ancestral character state for Gavialidae in which AT recovers an equivocal ancestral state for the family and DT recovers "roar" as the ancestral condition for Gavialidae.

| Nesting mode and communal nesting
Both AT and DT suggest mound nesting as the ancestral state among crocodilians, with the independent evolution of hole nesting in New World Crocodylus (C. acutus and C. intermedius) and the following Old World species: Crocodylus palustris, C. johnstoni, C. suchus, C. niloticus, and Gavialis (Figure 2a,b). The ancestral character state for African and New World Crocodylus is the only discrepancy among AT and DT, resulting in either a reversal to mound nesting in C. moreletii or the convergence to hole nesting in the rest of African and New World Crocodylus, as C. rhombifer remains polymorphic for nesting mode.
One enigmatic apomorphy is the use of termite mounds (coded as mounds) by Paleosuchus trigonatus, an alligatorid. Furthermore, the independent evolution of communal nesting behaviors has occurred five times, within Alligatoridae (Caiman latirostris), and in one Australian, both African, and one New World member of the genus Crocodylus (C. johnstoni, C. suchus and niloticus, and C. acutus, respectively), as well as Gavialis. Results were unequivocal among transformation types.

| Maternal nest attendance and lodging, nest defense, and hatchling attendance
Every species of crocodilian aside from Crocodylus johnstoni exhibits nest attendance, defined here as the maintenance of active positioning of the mother in close proximity to the nest; therefore, its origins may be in an earlier lineage (Figure 3a-e). Nest defense, or the active aggressive defending of the nest by the mother, is also characteristic of most crocodilians and their hypothetical ancestors. This behavior has apparently been lost independently in  (2015) Nest attendance (yes, no) Grigg and Kirshner (2015) Hatchling attendance (yes, no) Lang and Kumar (2013) Attendance lodging (water, burrow, wallow) Lang and Kumar (2013) Hatching stimulus (signal calling, vibration) Lang and Kumar (2013) Nesting season (wet, dry) Lang and Kumar (2013) Clutch frequency (1, 2) Lang and Kumar (2013) Clutch size (small, medium, large) Thorbjarnarson (1996) Clutch mass (small, medium, large) Thorbjarnarson (1996) Mating vocals (none, roar, bellow, buzz, moan) Dinets (2013) Wet season hatching (yes, no) (2013) Incubation duration (short, medium, long, extralong) Thorbjarnarson (1996) Note: Bolded character states indicate the character state for the listed species.

| Clutch frequency
The ability to produce multiple clutches within the same nesting season has only been documented in two species: Crocodylus mindorensis and C. palustris (Figure 4a,b). Naturally, all analyses indicate one clutch per nesting season as the ancestral condition for Crocodylia, with discrepancies between transformation types resulting from missing data for C. novaeguineae.

| Seasonal timing of hatching, hatching stimulus, and nesting season
Timing hatching with the onset of the wet season theoretically provides better cover and resource access to emerging hatchlings (e.g., Doody, Georges, & Young, 2001); however, as hypothesized by AT, this trait has evolved only within the Crocodylidae with reversals in C. novaeguineae, C. palustris, and C. porosus (Figure 5a-e).
Additionally, P. trigonatus shows independent convergence on this reproductive timing. Delayed transformations indicate that hatching timed for the wet season only characterizes New World Crocodylus with three additional convergences within the genus as well as P.
trigonatus. It is worth noting that eggs of Alligator mississippiensis hatch with the onset of the wettest months at the northernmost periphery of the distribution (Wilkinson, 1983), so variation in this character may exist in wide-ranging taxa.

| Incubation period, relative clutch mass, and clutch size
Accelerated transformations revealed high variation in duration of the incubation period (Figure 6a-f)

| D ISCUSS I ON
We used a contemporary phylogenetic hypothesis and an improved knowledge of species-specific reproductive behaviors and life-history strategies to infer reproductive trait evolution in crocodiles. By tracking traits through crocodilian evolution, our analysis provided a first-ever complete look at when those traits evolved, facilitating informative speculation on why they evolved.
Our analysis revealed that the ancestral crocodilian was a mound nester, with 4-5 transformations to hole nesting; habitat may have driven mode of nesting (mound vs. hole), with mound nesters occupying mainly swamps and marshes and hole nesters inhabiting large rivers and lakes, at least during the nesting season. Hole nesters were more likely to nest communally, but this association may be biased by scale. Although there were exceptions, mound nesters nested during the wet season and holes nesters during the dry season; this trait was relatively conserved, however. Nest attendance and defense were nearly ubiquitous, and thus, our analysis exhibited phylogenetic constraint for these characters, but attendance lodging was diverse among species, showing multiple reversals between water and burrows. Finally, about twothirds of species timed their nesting with the wet season, while the other third timed their hatching with the onset of the wet season.
Collectively, our analysis provides a framework for understanding reproductive trait evolution in crocodilians.
All crocodilians vocalize, but there is some variation in the types of calls they make (reviewed in Vergne, Pritz, & Mathevon, 2009;Grigg & Kirshner, 2015). In the present study, we analyzed mating calls, which were overwhelmingly described as bellows or roars in all but three species (Figure 1). Although it makes distress calls (Bonke, Whitaker, Roedder, & Boehme, 2015), T. schlegelli is not known to make mating calls. Its sister species, Gavialis gangeticus, makes a loud "buzz," while the call of Osteolamus tetraspis has been described as a "moan." Our analyses thus suggest that at least the general sounds audible to human ears are conserved among crocodilians, with the recent loss or modification of calls in the three abovementioned species. However, the contexts of these calls, including habitat, sex, and season, along with associated behavioral traits (e.g., head slapping, back vibrations-see Vergne et al., 2009 and references therein) can be species-specific and thus the evolution of call repertoires in crocodilians requires further observations and comparative analysis. Ultimately, a comparative study using sonogram signatures is required to truly understand vocalization evolution in crocodilians.
Most crocodilians excavate and backfill a nest either in a large mound constructed of mud and/or vegetation, in the ground (Campbell, 1972;Greer, 1970;Thorbjarnarson, 1996). After reviewing nesting mode in crocodilians, Greer (1970) hypothesized that the "common ancestor of the gavialids and crocodylines, and therefore of all living crocodilians, was probably a hole nester." Grigg and Kirshner (2015) countered that the ancestor was likely a mound nester, based on newer phylogenetic hypotheses and additional data on mode of nesting. Our analysis suggests that the ancestral crocodilian was indeed a mound nester, with 4-5 F I G U R E 4 Character optimization of clutch frequency using accelerated (a) and delayed (b) transformations independent transitions to hole nesting, depending on transformation type (Figure 2; 3-4 within Crocodylidae and one within Gavialidae). To be fair, crocodilian relationships were poorly known at the time of Greer's contention; for example, T. schlegelli, which is now recognized as the sister to G. gangeticus (and part of Gavialidae), was at the time considered to be a monotypic species within the Crocodylidae, and some of his species pairs are no longer recognized as such (e.g., novaeguineae-porosus, niloticus-palustris). Greer (1970) noted apparent support for his hypothesis in other reptiles: Hole nesting is common in turtles, lizards, and the tuatara. However, we now have a more accurate understanding of the relationships among turtles, lepidosaurs, birds, and crocodilians. Among archosaurs, mound nesting occurs in megapode birds and possibly occurred in an extinct crocodilian (Diplocynodon darwini, Hastings & Hellmund, 2015). Dinosaurs very likely exhibited a diversity of nesting types given the length of their tenure; however, it is very difficult to determine (from fossils) if their eggs were laid on the ground, buried in the ground, or in a mound of vegetation (Carpenter, 1999). Although vegetation from any dinosaur mound nests would have long rotted and disappeared, leaving only clutches of eggs in rocks, there are hypotheses with some circumstantial evidence for mound nesting in Oviraptor and other species (Coombs, 1989;Horner & Makela, 1979;Sabath, 1991;Carpenter, 1999). There is, however, good evidence for hole nesting in dinosaur fossils: Both multilayered and erect eggs must have been at least partially buried in order to keep their positions F I G U R E 5 Character optimization of hatching timed for wet season using accelerated (a) and delayed (b) transformations; nesting season using accelerated (c) and delayed (d) transformations, and hatching stimulus (e). Accelerated and delayed transformations recovered unequivocal results for this character (Cousin, Breton, Fournier, & Watte, 1994;Carpenter, 1999). Hole nesting is nearly ubiquitous in turtles, with only two species exhibiting mound nesting (Manouria spp.). Collectively, then, early crocodilians may have evolved mound nesting from a hole-nesting lineage, but the uncertainty of the distribution of nesting types among fossil reptiles, along with ambiguity on the phylogenetic position among turtles, archosaurs, and lepidosaurs (Field et al., 2014), prevents confident assignment.
Are there any clues to why mode of nesting evolves in extant crocodilians? Greer (1970) found no clear association between mode of nesting and the general ecology or behavior of species, suggesting instead that mode of nesting is an ancient trait in crocodilians. In contrast, Neill (1971) and Campbell (1972) contended that the nesting mode of crocodilians was related to habitat, with mound nesting in marshy environments and hole nesting on banks. Two opposing camps thus attempted to explain mode of nesting in crocodilians; the phylogenetic camp (Schmidt, 1924;Greer, 1970; but see also Greer, 1970) and the ecological camp (Wermuth, 1953;Campbell, 1972;Neill, 1971;Ouboter & Nanhoe, 1987;Thorbjarnarson, 1996). Our analysis demonstrated repeated F I G U R E 6 Character optimization of incubation duration using accelerated (a) and delayed (b) transformations; sizeadjusted clutch mass using accelerated (c) and delayed (d) transformations, and size-adjusted clutch size using accelerated (e) and delayed (f) transformations convergence to mound nesting and, thereby, could not falsify an association between mode of nesting and ecological factors such as habitat type. All hole nesters seemingly include rivers in their breeding habitat types, even if the species occupies a diversity of habitats; conversely, all mound nesters seemingly include marshes or swamps in their breeding habitat types (as per Campbell, 1972).
While the evolution of nest site in association with lentic or lotic habitats was not directly tested here, our analysis provides reasonable evidence for habitat type driving mode of nesting. However, tracking mode of nesting through the lineage also reveals conservation of nesting mode: The ancestral character state of mound nesting has persisted in most species, with hole nesting appearing much more recently. Hole nesting has also persisted for some time within a group comprising four species of crocodylids, however (C. niloticus, C. suchus, C. intermedius, and C. acutus). Unfortunately, we cannot readily and reliably retrieve habitat types for ancestral (fossil) crocodilians to confirm that habitat has driven mode of nesting in crocodilians. However, based on our analyses it is unequivocal that mound nesting is the plesiomorphic and conserved condition and that hole nesting is a derived convergent behavior.
If nesting type can be confidently tied to nest habitat, then at minimum mound nesting appears conserved because where crocodilians live is a conserved trait.
The apparent presence of both mound and hole nesting in some species generates the hypothesis that mode of nesting reflects (individual) behavioral plasticity. This is further supported by intraspecific hole nesting, mound nesting, and intermediates (low, poorly differentiated mounds) in C. acutus in Florida and Belize (Campbell, 1972;Platt & Thorbjarnarson, 2000, respectively) and intermedius, Figure 2b). However, these differences may be due to a bias of scale. Communal nesting, defined as the deposition of eggs or young with those of conspecifics (Doody, Keogh, & Freedberg, 2009), can be conspicuous in hole nesters, which tend to aggregate their nests on sand banks and may not defend them (e.g., C. johnstoni), while mound nesters are in more direct attendance which would prevent multiple mothers from laying in the same mound.
Accordingly, multiple clutches are rarely found within one nest [but see Larriera (2002) for an exception in C. latirostris or P. Wilkinson

(personal communication) for an exception in A. mississippiensis).
However, communal nesting at the larger scale of a cluster of mound nests in a given area of a swamp or marsh can occur and may be more common than appreciated. It is quite possible that clumping of mound nests at such a larger scale has gone undetected in many species. Regardless, communal nesting occurs in other archosaurs (definitely in birds and probably in dinosaurs; Carpenter, Hirsch, & Horner, 1994;Carpenter, 1999) and thus is an ancient trait that can be explained by social behaviors, limited nest sites, predation guarding efficiency, or some combination of these (Doody et al., 2009).
Our analyses of nest attendance exhibited phylogenetic constraint ( Figure's 3a-c). However, attendance lodging was diverse F I G U R E 7 Principal components analysis by which clutch mass, clutch size, incubation period, and deposition site (categorical) arrange species in ordination space. This elucidates that hole-nesting crocodilians had larger clutch sizes than mound nesters among species, showing multiple reversals between water and burrows under one transformation type, along with recent origins of wallow use (Figure 3d). Thus, the behavior of utilizing a particular microhabitat appeared particularly evolutionarily labile. There appeared to be no relationship between attendance lodging and mode of nesting. Nest attendance interpreted from the fossil crocodile Diplocynodon darwini was somewhat expected, because of the species' nested phylogenetic position within the crown group Crocodylia (Hastings & Hellmund, 2015). The discovery of nest attendance in three species of theropod dinosaurs from the late Cretaceous (Norell et al., 1995;Dong & Currie, 1996;Varricchio, Jackson, Borkowski, & Horner, 1997) demonstrates that nest attendance is an ancient trait that extended into the nonavian theropod lineage (Hechenleitner, Grellet-Tinner, & Fiorelli, 2015).
Fifteen of the 24 species in our analysis nest during the wet season (Figure 5c, (Figure 5a,b). As with nesting season, reversals apparently did not occur (Figure 5c,d). Thus, collectively, two-thirds of crocodilian species time their nesting with the wet season and onethird times their hatching to the wet season, and we hypothesize that benefits accrue to eggs and hatchings, respectively. A caveat here is that the geographic distribution (or part thereof) of some species does not include a distinct wet/dry season. For example, A.
mississippiensis is abundant in Louisiana, which has very little seasonal variation in rainfall (Keim & Faiers, 1996). In other words, in some cases timing of nesting or hatching in crocodilians may not be an adaptation to synchronize some part of the life cycle to the wet season. Of course, alternative zeitgebers may be driving the timing of these events.
Clutch frequency is remarkably conserved in crocodilians; only two species are known to lay two clutches per year, both of which are tropical (Figure 4; C. palustris, C. mindorensis). Not surprisingly, our finding is consistent with higher reproductive frequency in crocodylids compared to alligatorids (Thorbjarnarson, 1996). Relative clutch mass appeared to be ancestrally small in crocodilians, with 6-7 independent transitions to mainly medium-sized clutch masses, but with one transition to large clutch mass in the Caimaninae, followed by a reversal to medium in C. yacare and M. niger (Figure 6c,d).
Relative clutch mass was higher in alligatorids than in gavialids and crocodylids, supporting the findings of Thorbjarnarson (1996).
Size-adjusted clutch size was highly variable among species, with 11 medium-sized, six large, and five small clutch sizes (Figure 6e,f).
Although there were some clear transitions (e.g., medium to small clutch size in C. johnstoni; large to small clutch sizes in Paleosuchus), the lack of a clear ancestral character state resulted in ambiguity ( Figure 6e,f). Moreover, we did not consider egg size and the potential for a tradeoff between egg size and clutch size. However, quantitative analyses have found no such tradeoff among crocodilian species (Platt, Rainwater, Thorbjarnarson, & McMurry, 2008;Thorbjarnarson, 1996;Thorbjarnarson & Hernández, 1993).
Our results indicate that the hypothetical ancestor for Crocodylia was a mound nester that exhibited nest attendance and active defense from nearby water or burrows. Nesting occurred in the wet season and mating "roars" were used as social copulatory strategies. This animal was likely not a communal nester and deposited one reproductive output annually. Additionally, this species had long egg incubation periods with an average or large number of relatively small eggs. Our results also suggest that the evolution of clutch mass, clutch size, incubation period, and nest site are correlated, with lineages commonly exhibiting small clutch sizes and larger clutch masses, mound nesting, and short incubation periods. An ad hoc principal components analysis was performed using species as the response variables by which clutch mass, clutch size, incubation period (all on a continuous distribution), and nest site (categorical) arranged species in ordination space to visualize the relative influence of these variables on one another (Figure 7). Hole nesting species had larger clutch sizes than mound nesting species. Incubation period appears to be independent of any other variable, a result that is supported by the strong relationship between incubation temperature and incubation period (Ewert, 1979(Ewert, , 1985. The relationship between clutch morphometrics and nest type was predicted by Seymour and Ackerman (1980) from the perspective of respiratory constraints on developing embryos. They posited that decomposing organic material around a clutch of eggs imposes more extreme gas tensions than soil or sand, through which gas readily diffuses and nest gas tensions, varying as a function of nesting material, provide constraints on clutch size. We corroborate this finding using the clutch morphometrics and nest sites of all extant crocodilian species.
Here, we present data that diagnose the evolution of nesting ecology within the highly successful Crocodylia. Results diagnose a combination of phylogenetic constraint and convergence, yielding a clade characterized by high evolvability (Brown, 2013), a potential contributor to the success of the lineage over geologic time.
In the face of contemporary threats such as climate change, high evolvability of traits directly involved in offspring success may serve well in maintaining the success of the lineage, as it did during previous periods of climatic change. Despite advancements in our understanding of the evolution of nesting ecology and behavior in crocodilians (recent and herein), many questions remain. First, reproductive behavior is still understudied in crocodilians, owed at least in part to their nocturnality, their secretive habits in an aquatic medium, and the difficulties imposed by their large size relative to humans. Second, we have a poor understanding of geographic or other interpopulation variation in reproductive traits within species. For example, we lack sonogram signatures across populations and species; these are needed to fully understand vocalization evolution in the group. Finally, a more careful and robust classification of relative habitat use, in tandem with transplant or common garden experiments, could help clarify the role of habitat in the evolution of reproductive behaviors (e.g., mode of nesting) in this ancient lineage.

ACK N OWLED G M ENTS
We thank Z. Nikolakis and A. Westfall for their resourcefulness and J. Warner for critical review. The project was partially funded by the Edward G. Schlieder Foundation.

CO N FLI C T O F I NTE R E S T
We have no competing interests to report.

AUTH O R CO NTR I B UTI O N S
CMM conceived and designed the study; acquired, analyzed, and interpreted the data. JSD conceived and designed the study, and acquired and interpreted the data. BIC conceived and designed the study, and analyzed and interpreted the data.

DATA AVA I L A B I L I T Y S TAT E M E N T
Raw data used in character optimizations are provided as a table within this manuscript.