Morphological Covariance and Onset of Foot Prehensility as Indicators of Integrated Evolutionary Dynamics in the Herons (Ardeidae)

Synopsis The ultimate form an organism attains is based, in part, on the rate and timing of developmental trajectories and on compensatory relationships between morphological traits. For example, there is often an inverse correlation between the relative size of an organism's head and the length of its legs. Avian examples with a disproportionately small head and long legs include ostriches (Struthionidae), flamingos (Phoenicopteridae), cranes (Gruidae), and stilts (Recurvirostridae). To determine whether a possible compensatory relationship exists between relative head size and hind-limb length in a typically long-legged family of birds—the Ardeidae—we measured and analyzed skull dimensions (length, width, and height of cranium, and bill length) and skeletal hind-limb dimensions (femur, tibiotarsus, and tarsometatarsus) of the 12 North American species (north of Mexico) and of 12 additional taxa, including the morphologically divergent Agamia and Cochlearius. We found that Ardea species exhibit the smallest relative head sizes associated with the longest legs, while Butorides, Nycticorax, Nyctanassa, and Cochlearius have among the largest heads relative to hind-limb length. Furthermore, both positive and negative allometries occur in paired comparisons between the three hind-limb bones, expressed in tall morphotypes having disproportionately short femurs while short-legged morphotypes exhibit disproportionately long femurs; we show that this relationship has implications for foraging behavior. Moreover, the nestlings of short-legged herons exhibit functional precociality of the hind limbs through an early onset of prehensile ability of the feet to grasp branches, which is later expressed in adult foraging mode. This developmentally accelerated prehensile function in small-bodied species may be attributed, in part, to selection for predator avoidance in the early nestling stage.


Introduction
In recent decades, advances in evolutionary developmental biology, or evo-devo ( Hall and Olson 2003 ;West-Eberhard 2003 ;Carroll 2005 ), have shed light on the origin and maintenance of morphological diversity in many taxa. A central theme has been that of heterochrony-changes in the rate and timing of development in relation to ancestral developmental trajectories ( Gould 1977( Gould , 2002McKinney and McNamara 1991 ). The evolution of birds from theropod dinosaurs has been attributed to heterochronic outcomes, specifically paedomorphosis ( Bhullar et al. 2012( Bhullar et al. , 2016, as has the evolution of toe orientation ( Botelho et al. 2015 ) and even modifications of avian behavior ( Andersson 1999 ).
Traits, however, do not develop in isolation but as integrated units, what Richter and Wirkner (2014) call coherence morphology. Accordingly, compensatory relationships, or trade-offs, exist between and among distinct morphological structures as well as among individual modules ( Gatesy and Dial 1996 ). Thus, "whole morphologies… evolve in a correlated fashion" ( Zelditch and Moscarella, 2004 , p. 274). This dynamic system of interrelationships was early on articulated by Goethe-the founder of the science of morphology-in his principle of compensation ( Lenoir 1987 ;Riegner 2013 ). More recently, gene networks have been proposed in which alleles, even at distant loci, exhibit influences across the entire genome, thus more or less affecting all traits ( Davidson and Erwin 2006 ;Yukilevich et al. 2008 ). In other words, due to widespread pleiotropy, traits lack independence and thus exhibit varying degrees of covariance. Accordingly, West-Eberhard (2003 , p. 310) asks: "How often does the hypertrophy of an appendage or organ provoke, or even require, the reduction of some other organ or activity?" Thus, while traits may covary, they often do not covary in a one-to-one, or isometric, fashion. For example, an inverse correlation between the size of an organism's head and the length of its legs is evident across animal phyla. Among extant birds, a spectrum between small head/long legs and large head/short legs morphotypes is evident. For example, ostriches (Struthionidae) and hummingbirds (Trochilidae) not only occupy opposite regions of avian morphospace with respect to body mass ( Pigot et al. 2020 ), or regarding relative leg vs. wing investment ( Heers 2016 ), but also in cranial and hind-limb proportions. Relative to body size, ostriches have a diminutively small head and exceptionally long legs with two short toes on each foot, while hummingbirds have a disproportionately large head and atrophied legs with four-toed feet.
In contrast, relatively large-headed birds, such as owls (Strigiformes) and parrots (Psittaciformes), typically have short hind limbs and robust feet, which are associated with enhanced foot prehensility (expressed, for example, in the ability to grasp and manipulate prey or other food items; see Backus et al. 2015 ). Overall, the Passeriformes also have relatively large heads (for their body size), as well as short to mid-length legs and long, prehensile toes, including an elongated hallux especially adapted for grasping perches ( Fjeldså et al. 2020 ). These examples suggest that, while morphological traits-e.g., head size and leg length-typically covary, the nature of this covariance is such that it often deviates from isometry or one-to-one scaling.
Although considered "long-legged wading birds," there is a wide span ( > 4.5x) of leg lengths exhibited across the phylogeny of the Ardeidae, from the Zigzag Heron ( Zebrilus undulatus ) to the Goliath Heron ( Ardea goliath ). In addition, even a cursory comparison reveals that cranial dimensions vary from, for example, the slender-headed Tricolored Heron ( Egretta tricolor ) to the relatively robust-headed Yellow-crowned Night-Heron ( Nyctanassa violacea ). Accordingly, the ardeids (hereafter also referred to as "herons") are ideal subjects to compare and contrast such morphological features to gain insight into the evolutionary dynamics of this taxon, which may hold broad applicability to other taxa.
From an ecomorphological perspective, body size and morphological diversity among herons is associated with variation in foraging behavior and niche occupation ( Kushlan 1978 ). According to Watanabe (2018Watanabe ( , p. 2645, the Ardeidae show "prominent interspecific variation in the length of the distal leg even among closely related species, which is associated with foraging behavior…" and their "apparently unique ontogenetic trajectory might have facilitated diversification of distal leg length and hence foraging habitat segregation in this family." For example, taller birds are able to forage at greater water depths ( McKilligan 2005 , p. 34;Morales 2018 ) and capture larger prey than relatively short-legged species. Furthermore, while taller species, for example, the Reddish Egret ( E. rufescens ), regularly hunt by active wading ( Koczur et al. 2020 ) or, as in the Great Egret ( A. alba ), may walk slowly across wide expanses, shorter-legged taxa, such as the Green Heron ( Butorides virescens ), often perch on roots or branches and wait for prey to come within striking range, a relatively low-energy hunting strategy ( Meyerriecks 1960 ;Kushlan 1978 ;Kushlan and Hancock 2005 ).
Since hind-limb length is a critical determinant of ardeid foraging ecology, the drivers of differential limb growth, and associated morphological and behavioral adaptations, are worthy of investigation. For example, there is evidence the mechanism for differential growth of ardeid hind-limb bones can be attributed to varying rates of cell proliferation, resulting in "morphological heterochrony" ( Cubo et al. 2000 ). Furthermore, Ávila (2011 , 2017 ), in studies of growth rates in seven species of ardeids, identified heterochrony or, more specifically, hypermorphosis, as "the main evolutionary process in… Ardeidae," resulting from a relatively extended postnatal growth period ( Ávila 2011 , p. 771), which culminates in Ardea . For example, tarsometatarsus growth in the Green Heron terminates already on Day 20 while in the Great Egret termination is delayed until Day 43 ( Ávila 2011 ). Between these extremes are the Black-crowned Night-Heron ( Nycticorax nycticorax ), at Day 23, and the Reddish Egret, at Day 36 ( Ávila 2011 ). Thus, differential limb growth durations are expressed in paedomorphic, or relatively "underdeveloped," and peramorphic, or relatively "overdeveloped," heterochronic patterns evident in short-legged vs. long-legged herons, respectively.
The behavioral elements, and certainly the morphological components, of heron foraging behavior presumably originate in early ontogeny, in some cases perhaps in the very early nestling developmental phase of these mostly semi-altricial birds. Kushlan and Hancock (2005) identified the "brancher" stage as the age at which heron nestlings first clamber from the nest-an early developmental marker indicating when the prehensile ability of the feet is first exhibited. As discussed later, we propose that foot prehensility plays a significant role in the foraging repertoire of various heron species.
In this study, we test whether morphological characters of herons deviate from isometry-perhaps indicative of morphological trade-offs-and whether this deviation is related to the foraging behaviors of adults and to the timing of first appearance of nascent foraging behavior components in nestlings. Specifically, we address the following questions: (1) Do heron species with longer legs have disproportionately smaller heads (negative allometry) while, conversely, species with shorter legs have disproportionately large heads? (2) Do the skeletal components of the hind limbs scale isometrically as overall leg length among species increases or is differential growth evident (positive or negative allometry)? (3) Does the degree of deviation from isometry among the three hind-limb bones correlate with foraging behavior? (4) Do heron species with relatively short legs exhibit an enhanced foot prehensility (as alluded to earlier in regard to short-legged owls and parrots), perhaps already evident during the nestling stage?
Our working definition of morphological trade-off, as implied by the measurement of negative allometries, presupposes that an organism is a dynamic integrated whole, and (assuming constant body mass) that "an increase in the magnitude of one [trait] means a decrease in the magnitude of one or more others" ( West-Eberhard 2003 , p. 296). In other words, "the simplest type of trade-off occurs when one trait cannot increase without a decrease in another" ( Garland et al. 2022 , p. 83). Furthermore, "most traits can be considered to involve a trade-off with every other trait, since the development or performance of every trait involves some cost in terms of time, material, or energy and therefore may detract from others" ( West-Eberhard 2003 , p. 303). Acknowledging that "trade-off" defies clear definition, Garland et al. (2022) identify six nonmutually exclusive categories of trade-offs and argue for a pluralistic perspective that embraces multiple levels of biological organization across disciplinary boundaries. By attempting to integrate morphology, behavior, development, and evolution, our study lends support to Garland et al.'s perspective. And finally, why test for covariance and potential morphological trade-offs specifically between the limb module and the cranium module? More typically, comparisons are made between/among avian "locomotor modules," such as the forelimbs, hind limbs, and tail (e.g., Gatesy and Dial 1996 ;Eliason et al. 2023 ). However, just as one can apply a logical conceptual model of "locomotor modules" to avian anatomy, so too can one identify "foraging modules," which, especially in the herons, would embrace both cranial and hind-limb morphologies.

Mor pholog ical measurements
We took morphological measurements of 24 species of Ardeidae from museum skeletal specimens (37.5% of 64 extant species; Winkler et al. 2015 ), including all 12 North American species north of Mexico ( Table 1 ). Additionally, 12 other taxa were selected to represent a broad range of sizes and morphologies and had accessible specimens housed in United States natural history collections (see Acknowledgments for list of institutions). We attempted to measure a minimum of 10 samples per species but were not always able to do so due to the rarity of some specimens. We measured a total of 265 adult specimens, though 15 specimens were incomplete (e.g., a skull may have been intact, but a femur was missing). We measured all skull dimensions (width, height, and length of cranium, bill length, and total skull length ( = cranium length + bill length; Table S1) and skeletal hind-limb lengths (femur, tibiotarsus, and tarsometatarsus; Table S2) using a Mitutoyo 4LB11 digital caliper ( ± 0.01 mm). We used a ruler ( ± 1.0 mm) to measure some hind-limb bones of the largest specimens (e.g., A. goliath ) that were too long for the caliper ( > 160 mm). Cranial length was measured from the nasofrontal hinge along the occipital crest to the greatest distance of the occipital bone, and cranial width was measured as the distance between the distal ends of the two post-orbital processes. Bill length was measured from the nasofrontal hinge to the distal tip of the bill. Hind-limb dimensions were measured as follows: femur, from the tip of the trochanter to the distal medial condyle; tibiotarsus, from the cnemial crest to the distal medial condyle; and tarsometatarsus, from the calcaneal ridge to the distal trochlea for digit III. One person (M.F. Riegner) took all the measurements.
Body mass data for each of the 24 species were obtained from the literature ( Dunning 2008 ; Table S3); when multiple values were given, such as for males and females, means were calculated (following Zeffer et al. 2003 ).

Behavioral indices
Foraging behavior diversity and activity-level categories were determined from publications, especially Kushlan (1978) , but also Martínez-Vilalta and Motis (1992) and Kushlan and Hancock (2005) , and Cornell Lab of Ornithology Birds of the World online, as well as from M.F. Riegner's field observations, over several decades, of 18 of the 24 species included herein. Four categories were estimated and scored, ranging from category 1, representing minimal behavioral diversity and generally stationary perching/crouching, to category 4, representing relatively high behavioral diversity and high activity, including running and turning quickly (column E, Table 2 ). We then compared foraging activity level with allometric scaling of the femur, which was proportionately the most variable hind-limb bone.
Regarding the age when nestlings first clamber from the nest (the "brancher" stage), data were taken from the literature, as well as provided by various heron field researchers. There is a range from approximately 5-21 Days, and possibly beyond, for the onset of this developmental marker ( Table 3 ). We compared the age of the brancher stage with allometric scaling of leg length and, when descriptions were available, we identified which species as adults tend to forage by grasping branches, reeds, mangrove roots, etc. We considered these species as expressing a relatively high degree of foot prehensility in their adult foraging repertoire, especially compared with those species that typically forage by walking on firm substrate, wading in shallow water, standing on mud flats, etc. Of course, all ardeids are capable of perching on branches to roost, etc.

Statistical analyses
We first tested whether the relationship between body mass and other traits (leg length, cranium length, cranium width, cranium height, bill length, and total skull length) differed from what is expected based on isometric scaling of a volumetric measurement (using mass as a proxy for volume) and length measurements of the other traits. Such an approach is appropriate because we are interested in the covariation between body size, morphological traits, and foraging behaviors and not how morphological traits and foraging behaviors are related independently of the effect of body size. We log 10 transformed both body mass and the other variables and included log 10 body mass as a predictor in a general linear model (GLM) with the other traits, Zigzag Heron n/a n/a n/a n/a 1 Least Bittern 4 1 n/a 1.0 1 Green Heron 11 4 13 1.1 1 Squacco Heron n/a n/a n/a n/a 2 Malayan Night-Heron n/a n/a n/a n/a 2 Whistling Heron n/a n/a n/a n/a 3 Capped Heron n/a n/a n/a n/a 2 Cocoi Heron n/a n/a n/a n/a 2 Great Blue Heron 13 4 16 1.7 3 Goliath Heron 3 1 n/a 1.3 2 Column A: foraging behavior diversity tabulated from Kushlan (1978 , Table 1) Column B: foraging behavior diversity adjusted to include only ambulatory behaviors Column C: additional foraging behavior diversity data from Kelly et al. (2003) Column D: foraging (i.e., feeding) activity indexes from Kushlan (1978 , Fig. 2) Column E: estimated foraging activity-level categories: (1) perching and "passive;" (2) standing, crouching, "inactive," and "motionless;" (3) slow wading, slow walking, standing, and patrolling large area; (4) "active," running, and turning quickly. Estimates tabulated from data and accounts from Kushlan (1978) , Martínez-Vilalta & Motis (1992) , and Kushlan & Hancock (2005) , Cornell Lab of Ornithology Birds of the World online, and personal observations; terms in quotes are actual descriptors taken from the preceding sources from individual species accounts. Species ordered by increasing total leg length. Shaded boxes indicate the three most diverse foraging behaviors or foraging activity indexes within a column (except Column E where only the highest category is shaded), which tend to be displayed in mid-sized species; n/a = not available.
one at a time, as dependent variables. Based on allometric scaling expectations, volumetric traits (V) and length traits (L) are predicted to scale to the one-third, based on V = aL 3 or L = ( 1 a V ) 1 / 3 . On the log 10 scale, this equation is log 10 L = −lo g 10 a + 1 3 lo g 10 V , meaning Table 3 Approximate age at which nestlings first stand and clamber from the nest (the "brancher" stage)

Common name Age in days References
American Bittern 7-14 1 (n/a) Lowther et al. (2020 ); Baicich & Harrison (1997 ) Zigzag Heron n/a 1 (n/a) Mathews & Brooke (1988 ) Least Bittern 4-5 1 (16%) Nero (1950 ); Weller (1961 ); Baicich & Harrison (1997 ); Corman (2005 ) Great Blue Heron 21 (40%) Pratt (1970 ) Gray Heron 2 25-27 (54%) Voisin (1991 ) Goliath Heron 21 (50%) Kushlan & Hancock (2005 ) Purple Heron 2 20 (45%) Tomlinson (1975 ) Great that isometry between overall size and length measurements are expected to result in slope parameters that are not different than one-third. We tested for this relationship using lo g 10 -transformed linear measurements of the leg lengths and cranium dimensions as dependent variables and lo g 10 -transformed body mass as a predictor variable in a GLM. We asked whether the estimated relationship between length and mass on the log 10 scale differed from one-third using a t -test. Slope parameters that were significantly lower than one-third were considered to show negative allometric relationships and those that were significantly greater than onethird were considered to show positive allometric relationships. We centered the analysis on the mean log 10 mass across the species. Next, to test the hypothesis that heron species with longer legs have disproportionately smaller heads (negative allometry), we conducted a similar analysis to the one above using overall leg length as an independent variable and cranium length, cranium width, cranium height, bill length, and total skull length as dependent variables. Each of these traits is a length measurement and thus an isometric relationship would be indicated by slope parameters of 1. We asked whether the estimated relationship between these traits on the log 10 scale differed from 1 using a GLM and tested whether the estimated slope parameter differed from 1 using a t -test.
We then asked whether there was evidence for allometric relationships between the constituent bones of the leg. We did so by including total leg length as a predictor in the GLM and the length of the individual leg bones (femur, tibiotarsus, and tarsometatarsus) as dependent variables. As in the relationships between leg length and cranium dimensions, isometric scaling is expected to produce a slope parameter of 1 between the individual bones of the leg.
To ask whether the degree of deviation from isometry of the femur to the overall leg length relates to foraging behavior, we subtracted the expected femur length based on an isometric relationship from the observed femur length of each species and used this as a measure Table 4 Parameter estimates, standard errors, and significance from nonph ylogenetic and ph ylogenetic reg ressions of log 10 -transf or med character lengths against log 10 -transf or med bod y mass. Ph ylogenetic reg ression assumed Brownian motion. Degrees of freedom for all tests was 22 We used a similar approach to ask whether herons with relatively shorter legs exhibit functional precociality through an accelerated foot prehensile ability, that is, an early expression of the brancher stage. Here, we used the deviation from the allometric relationship between total leg length and mass as a dependent variable and used age (days post-hatching) at brancher stage as a continuous predictor.
We performed both nonphylogenetically informed and phylogenetically informed analyses using program R ( R Core Team 2020 ). Nonphylogenetic analyses were conducted using the lm function ( R Core Team 2020 ).
For the phylogenetically informed analyses, we based our phylogeny on Hruska (2018) . However, a few species in Ardea were not included in Hruska's phylogeny so we supplemented the information for this genus from Huang et al. (2016 ; Fig. S1). We then performed the analyses using the gls function ( Pinheiro et al. 2020 ) and assumed Brownian motion with correlation structures calculated using corBrownian ( Paradis and Schliep 2019 ).
To increase sample size, as well as to include incompletely labeled specimens, data for adult males and females were combined (following Zeffer et al. 2003 ;Maccarone and Brzorad 2016 ).
Do nestlings of small heron species with relatively short legs exhibit an early onset of foot prehensility, that is, an early expression of the brancher stage? Species with shorter than expected legs relative to body mass, compared with isometric scaling, tended to be Those species that exhibit a delayed development of foot prehensility, for example, Day 21, are among the tallest of herons (e.g., Ardea herons, dark blue points, Fig. 5 ). The Yellow-crowned Night-Heron (rightmost light blue point) has shorter legs than expected for a delayed brancher stage. Finally, the onset of foot prehensility is relatively delayed until up to 40-54% of the total nestling period has transpired in the large Ardea herons, while this developmental marker is accelerated to only 16% of the total nestling period in the Least Bittern and 32% in the Green Heron ( Table 3 ). In other words, not only does the brancher stage first appear in relatively fewer days in the smaller ardeids, but also it appears proportionately much sooner in their respective overall nestling periods compared with the longerlegged herons.

Cranium and hind-limb mor pholog ical relationships
We found significant departures from isometry between cranial dimensions and length of hind limbs in ardeids. The tallest species generally exhibit the smallest relative head dimensions, with A. alba as an extreme morphotype in this regard ( Fig. 6 A). In contrast, Cochlearius exhibits among the most robust head dimensions relative to leg length ( Fig. 6 A) and in this respect represents the "opposite" ardeid morphotype to A. alba . In other words, Cochlearius and A. alba can be considered as occupying divergent extremes of ardeid morphospace. Clearly, however, the overall morphological trajectory can be somewhat diverted, presumably due to strong selection pressure resulting in specialized adaptation and niche occupation, as evident in the exceptionally long, slender bill and total skull length of Agamia (green points in Fig. 1 E and F and Fig. 2 D and E; Fig. 6 B) or the heavy, robust cranium to accommodate the broad, scoop-like bill of Cochlearius (yellow points in Figs 1 C and 2B; Fig. 6 A).

Intra-hind-limb mor pholog ical relationships
Regarding possible trade-offs between hind-limb elements, the two distal hind-limb segments-tibiotarsus and tarsometatarsus-are responsible for the greatest contribution to increase in total leg length in herons, whereas the femur becomes disproportionately shorter in longer-legged species (Table S2, Fig. 6 C). Similarly, in a study of 323 diverse species of birds, increase in leg length was attributed mainly to the distal bones, i.e., tibiotarsus and tarsometatarsus, and not to the femur ( Zeffer et al. 2003 , p. 467): "relatively long-legged birds have relatively shorter femurs compared with short-legged species." Compared with the norm for all birds (that is, the 323 species measured in their study), waders have a significantly lower mean value for femur index but significantly higher values for tibiotarsus and tarsometatarsus. Barbosa and Moreno (1999) maintain that relative lengthening of the tibiotarsus and tarsometatarsus results in increased stride length in "waders" (i.e., shorebirds), while increase in relative femur length translates into increased stride frequency. Moreover, Zeffer et al. (2003) measured proportionately longer femurs in birds of prey and hanging and climbing species, both of which have relatively well-developed prehensile feet; in our study, disproportionately longer femurs were associated with small, short-legged heron species (e.g., Least Bittern), which exhibit a functional precociality of the grasping ability carried over into the adult morphofunctionality (see next section). As in our study, Stoessel et al. (2013) found a positive correlation between tibiotarsus and tarsometatarsus lengths among 236 species of birds sampled across all major avian subclades, but an "inverse proportionality" between the femur and tarsometatarsus (as also stated for flamingos by Eliason et al. 2023 , p. 348); notably, in addition, an inverse proportional covariation between the femur and tibiotarsus was determined only in "rather long-legged birds" (p. 489), as our data indicate for 24 heron taxa. Overall, the avian hind-limb bones clearly exhibit a high degree of integration for efficient terrestrial locomotion ( Stoessel et al. 2013 ).

Onset of foot prehensility (the "brancher" stage)
As shown in this study, foot prehensility develops earlier in species with relatively shorter hind limbs (and thus longer femurs), that is, in smaller morphotypes, which typically employ perching and crouching in the adult foraging repertoire. What may be a driver for the evolution of functional precociality of the hind limbs in smaller species of herons, subsequently displayed in typical foraging modes in adults? In general, small species with accelerated foot prehensility, e.g., Green Heron, are mostly solitary or loosely colonial nesters ( Davis and Kushlan 2020 ), and/or nest close to the ground or just above water (e.g., Least Bittern nests in emergent vegetation only 15-76 cm above the water's surface: Poole et al. 2020 ; Zigzag Heron, of which there is little known, is also a solitary nester, as is the American Bittern: HeronConservation 2020 ). Thus, nestlings of these species are more vulnerable to predators (e.g., lack "safety in numbers") compared with densely colonial species and/or those that nest higher in trees; they also typically have smaller body sizes and consequently Fig. 4 Relationship between the de viation of f emur length from isometry with leg length and behavioral mode without ( A ) and with ( B ) phylogenetic corrections. Points are means and bars are 95% confidence intervals. See Table 2 , column E, for estimation of foraging activity level and species assigned to each category; numbers in parentheses indicate sample size in each f oraging activity-le vel category. Representative species from each foraging activity-level category (from left to right): Least Bittern ( Ixobrychus exilis ), Blackcrowned Night-Heron ( Nycticorax nycticorax ), Cattle Egret ( Bubulcus ibis ), and Reddish Egret ( Eg retta ruf escens ). Bird illustrations reproduced with permission from Lynx Edicions.
would be susceptible to a broader array of predators. Therefore, it behooves nestlings of these species to accelerate their ability to clamber from the nest at an early stage of development in order to escape predators. Researchers who band nestling herons can attest that, as soon as young birds are capable of scrambling from the nest, they will climb away when approached. Selection may therefore act on the early nestling stage to accelerate a behavioral ability that is then retained and expressed in the foraging behavior of the adult. Carrier (1996 , p. 480), investigating the ontogeny of locomotor behavior in vertebrates, maintains that, due to "higher predation pressure on juveniles, the locomotor phenotype of adults may, in some cases, be more a reflection of selection acting on the performance of juveniles than a direct result of selection acting on the adults" (see also Herrel and Gibb 2006 ;Young et al. 2022 ), what Gignac and Santana (2016) call "ontogenetic inertia." For example, in Mallards ( Anas platy rhy nchos ), "hindlimbs retain characteristically juvenile morphology and performance" ( Dial and Carrier 2012 , p. 3707). Thus, the proclivity of certain species of herons-e.g., Least Bittern and Green Heron-to forage while grasping a perch, may be attributed to mostly solitary nesting near the ground (or water surface), concomitant greater vulnerability to predation, and the evolution of functional precociality of foot prehensility to enable early-stage nestlings to clamber from the nest to safety. In other words, "vulnerability to predation would lead to selection for accelerated development" ( Carrier 1996 , p. 480). While the early onset of foot prehensility in early-stage nestlings is an expression of ontogenetic acceleration, this juvenile feature subsequently remains imprinted on the foraging behavior repertoire and is expressed throughout the adult life stage. And finally, although the brancher stage has not been previously studied systematically in herons, or in most avian taxa, Holcomb (1966) examined the development of the grasping reflex in seven species of passerines, a taxon noted for the evolution of enhanced prehensility resulting in specialized perching ability ( Fjeldså et al. 2020 ). The American Goldfinch ( Carduelis [ Spinus ] tristis ) showed the earliest expression of prehensility at Day 5 post hatching while the other species ranged to Day 9 ( Holcomb 1966 ). Accordingly, among ardeids, the Least Bittern converges with passerines in this early developmental marker ( Table 3 ).

Mor pholog ical trade-offs?
The evolution of the avian Bauplan has been marked by repeated morphological trade-offs and compensations. For example, the forelimbs evolved aerodynamic ability at the expense of the hind limbs, as evidenced in relative limb lengths and associated muscle mass ( Heers andDial 2012 , 2015 ;Heers 2016 ). Furthermore, as proavian forelimbs lengthened for flight and hind limbs shortened, the former relinquished grasping ability, while the latter evolved prehensility ( Heers and Dial 2012 ). Among herons, our results show that shorter-hind-limbed species exhibit an accelerated development of the grasping reflex and a frequent Fig. 5 Relationship between the deviation of total leg length from isometry with body mass against age at brancher stage, with fitted line (data for brancher stage from Table 3; if range was reported, then mean was calculated). Those points below the horizontal dashed line indicate species that have shorter than expected total leg lengths for a given body mass in contrast to points above the dashed line that indicate species with longer than expected total leg lengths for a given body mass. The latter tend to take relati vel y more days to attain the brancher stage.
application of prehensility in adult foraging behavior. This trend, in fact, may echo the evolution of perching in the transition from nonavian maniraptoran dinosaurs to birds. Dececchi and Larsson (2013) note that the hind limbs of Archaeopteryx were relatively shorter (by at least 30%) than those of similar-sized Microraptor (the approximate proportional difference between, respectively, Green Heron [ B. virescens ] and Little Blue Heron [ E. caerulea ]), and that "long limbs would hinder perching and movement along tree branches" (p. 2750). Apparently, the adaptive shortening of avian hind limbs permits a concomitant enhancement of prehensility.
In other examples of compensator y evolutionar y responses in birds, Harell (2016) identified a negative correlation within the hind-limb module regarding the lengths of the cnemial crest vs. the patella, the former shorter and the latter longer in cormorants in comparison with anhingas. Bhullar et al. (2016) proposed a morphological and functional compensation driven by heterochrony between the loss of hand functions of dinosaurs, as hands were integrated into the avian wing, and the enhanced kinematic functions of the avian beak; however, kinematic grasping function of feet was not considered in their formula. Regarding morphological integration and compensations specifically between and among avian hind-limb elements, Stoessel et al. (2013 , p. 483) summarize it nicely: "Different de-mands on posture, locomotor mechanics, or behavior are known to influence differences in length between intra-limb elements, particularly overall proportions." As mentioned earlier, we consider the avian hind limbs and cranial features as "foraging modules," and thus the negative allometric relationships between these two modules, identified in this study, can be considered expressions of the integrated nature of the whole organism, an expanded evolutionary perspective garnering renewed attention these days (e.g., Schwenk et al. 2009 ;Huneman 2010 ;Nicholson 2014 ). Even in early studies, reciprocal relationships were demonstrated between what were initially thought to be independent structures, such as comparisons of tarsus length and pointedness of bill in icterids ( James 1982 ). More recently, correlated evolution was demonstrated between, for example, avian sternal keel length and ilium length ( Zhao et al. 2017 ), thus indicating that the avian organism expresses a high degree of integration (see Orkney et al. 2021 ), with morphological compensations occurring at various levels of organization. Relationships between apparently disparate organs, perhaps through pleiotropic effects, seem to be the rule rather than the exception across organisms, as evidenced, for example, in a single set of genes controlling both limb and horn development in Onthophagus beetles ( Moczek 2005 ). Our study was limited in its focus to only skeletal features; however, especially regarding foot prehensility, avian muscle distribution would also need examination ( Backus et al. 2015 ), as well as accommodations of the nervous system, blood supply, etc.
Although neck length was not measured in our study, Böhmer et al. (2019 , p. 5) found that avian "neck length showed an isometric scaling in relation to total leg length," and thus the same can be assumed for ardeids: that is, longer-legged herons should have longer necks, and thus longer-necked herons should have disproportionately smaller heads. This relationship is evident in a cursory comparison of, for example, relatively largeheaded, short-necked Nycticorax and Nyctanassa nightherons with relatively small-headed, long-necked Ardea and Egretta day-herons. Accordingly, longer-necked herons would be expected to have disproportionately shorter femurs; in fact, Böhmer et al. (2019 ) identify the Gray Heron ( A. cinerea ) as having an exceptionally long neck in relation to femur length.
Brancher stage, hind-limb length, relative femur length, and foraging behavior Species that display the earliest development of the brancher stage, e.g., the small-bodied Least Bittern and Green Heron ( Table 3 ), and, we propose, most likely the Zigzag Heron, possess short legs with longer than expected femurs (Table S2, Figs 3 A, 5 , and 6 C) and are those in which the adults tend to employ a minimal diversity of low-energy foraging behaviors (category 1, Table 2 , Figs 4 , and S1C and D), such as crouch-/perch-and-wait, by grasping branches, reeds, mangrove roots, snags, etc. ( Meyerriecks 1960 ;Kushlan and Hancock 2005 ). Species that exhibit the most delayed development of foot prehensility, from Day 21 and beyond, i.e., the tall Ardea herons ( Table 3 ), exhibit long legs with shorter than expected femurs (Table S2, Figs 3 A, 5 , and 6 C) and are those species in which the adults, compared with the smallest heron species, display relatively more foraging behaviors, including low-energy behaviors, such as stand-and-wait/peering over, as well as slightly more active behaviors, including slow wading while patrolling broad swaths of habitat in search of prey (categories 2 and 3, Table 2 , Figs 4 , and S1C, D; Kushlan and Hancock 2005 ). Notably, unlike, for example, the Green Heron, such tall morphotypes, specifically A. alba and A. cocoi , "rarely use perches to forage" ( Pinto et al. 2013 , p. 793). Species that develop prehensility at intermediate ages (Day 10-20 posthatching), e.g., the Egretta herons ( Table 3 ), possess mid-length legs ( ∼260-300 mm) with shorter than expected femurs ( Table 3 , Figs 3 A, 5 , and 6 C) and are those in which adults display the most diverse and most active foraging behaviors (mostly category 4, Table 2 , Figs 4 , and S1C, D), such as run- Fig. 7 Diagrammatic representation of ardeid hypothetical morphospace inf er red from interrelationship of total leg length, leg heterochronic growth pattern, propor tional f emur length, propor tional cranium size, and relative foraging activity. Representative species are (from left to right) Green Heron ( Butorides virescens ), Tricolored Heron ( Egretta tricolor ), and Great Blue Heron ( Ardea herodias ). Bird illustrations reproduced with permission from Lynx Edicions.
ning, turning quickly, foot-stirring, wing-flicking, etc. ( Kushlan 1978 ;Kushlan and Hancock 2005 ). Morales (2018) also found that Egretta species display more active foraging behaviors than Ardea species (regarding feeding behavior diversity and body size, see also Meyerriecks 1960 , p. 149, andKushlan 1978 , p. 255). Accordingly, in our sample, the intermediate-sized herons exhibit the highest species richness as well as the highest diversity of foraging behaviors, including the most active foraging behaviors ( Table 2 , Fig. S1C, D). Thus, by considering relative timing of hind-limb functional development, intra-hind-limb morphological relationships, proportional relationships of cranium size, and elements of foraging behavior within an ecomorphological context, an approximation of ardeid hypothetical morphospace can be inferred that defines the possibilities and boundaries of potential form within this avian clade ( Fig. 7 ).

Conclusions
In conclusion, our study interrelated (1) potential morphological trade-offs between cranial and hind-limb modules, (2) variation in the relative proportions of intra-hind-limb elements, (3) association of foraging behavior with hind-limb morphology, (4) variation in the early development of the hind-limb grasping reflex, and (5) possible selection for predator avoidance in early ontogeny in smaller-bodied taxa. Accordingly, these ecomorphological relationships may contribute to integrated evolutionary dynamics that ultimately influence ardeid morphospace, morphofunctionality, and adaptation to a broad range of habitats and foraging circumstances in this widely distributed group of wading birds.