A bird in the hand: Global‐scale morphological trait datasets open new frontiers of ecology, evolution and ecosystem science

Abstract

The recent prominence of functional traits in ecological analyses is based on the premise that measurable attributes of an organism's phenotype can take us beyond simple lists of species and closer to valid tests of mechanisms and processes (Cadotte et al., 2011). However, the full potential of trait-based ecology and evolutionary biology is ultimately constrained by incomplete coverage and completeness, particularly in the case of morphological traits (Etard et al., 2020). Filling these gaps in data coverage has proved challenging, with even the best-sampled major taxonomic groups—such as vascular plants—still lacking comprehensive morphological measurements for well over 50% of species worldwide (Hietz et al., 2021; Kattge et al., 2020; Violle et al., 2014). A major step has now been taken towards addressing this challenge with the completion of datasets containing multiple morphological traits for all 11000 bird species (Tobias et al., 2022). The goal of this special issue is to present these data for wider use alongside a series of studies summarising recent advances based on morphological analyses, highlighting their potential application to research and policy. The most widely used functional traits in macroecological and macroevolutionary analyses are categorical variables, mainly including information on habitat, life-history or diet (Jones et al., 2009; McLean et al., 2021; Wilman et al., 2014). These datasets have been highly influential, yet overall progress has been impeded because many categorical traits are relatively crude and uninformative, reducing their utility as indices of ecological function (Kohli & Jarzyna, 2021). Moreover, they offer an imperfect framework for some statistical models and phylogenetic analyses since many species are assigned the same values and the distance between categories is arbitrary. An obvious solution is to use continuous morphological variables, as these vastly improve the resolution of evolutionary models (Chira et al., 2018) and metrics of community assembly (Blonder et al., 2018; Ricklefs & Travis, 1980). To date, the availability of complete continuous morphological trait datasets has been largely restricted to body mass (Wilman et al., 2014), which is only weakly connected to ecological function (Pigot et al., 2020). A hawk and a duck may share the same body size, for example, but this tells us very little about their functional role in the ecosystem. Analyses based on more detailed compilations of morphological traits have not been possible outside a few well-studied families, leading to a variety of problems including sampling bias and inaccurate evolutionary models (Chang et al., 2020; Mouillot et al., 2021; Tobias et al., 2020). Birds offer the best opportunity to address the challenge of comprehensive trait coverage for a number of reasons. First, overall species richness (~11,000 species) is far lower than plants, for instance, offering a more achievable target. Second, birds are distributed worldwide across all oceans and terrestrial biomes, where they perform a range of key ecological services (Şekercioǧlu, 2006). Third, because of their visibility and appeal, they are the best-studied clade at this global scale, with extensive datasets now available on distribution, abundance, ecology and life history for almost all species (Bird et al., 2020; Callaghan et al., 2021; Sullivan et al., 2014; Tobias et al., 2020; Tobias & Pigot, 2019; Wilman et al., 2014). Fourth, bird morphology offers a classic system for investigating a range of novel ecological questions because their beaks, legs and wings provide insight into trophic interactions, locomotion and dispersal respectively (Dehling et al., 2016; Pigot, Trisos, et al., 2016; Sheard et al., 2020). Indeed, birds are unique in that specific combinations of traits have been shown to predict key functional characteristics, including dietary niche and foraging behaviour, with far greater accuracy than body mass alone (Kennedy et al., 2020; Pigot et al. 2020). The power of morphological traits to predict ecology was initially established by a series of papers on bird communities from 1960 onwards (e.g. Miles & Ricklefs, 1984). Although these analyses were based on relatively small samples of species (see Tobias et al., 2022, Figure 1), they provided the conceptual foundation for the field of ‘ecomorphology’ (Bock, 1994; Wainwright & Reilly, 1994) which in turn drove the subsequent (post-2000) development of avian functional ecology based on continuous variables. Over the last two decades, several research groups compiled and analysed bird trait datasets of gradually increasing size, initially targeting manageable samples of a few hundred species (e.g. suboscines: Claramunt, 2010; corvides: Kennedy et al., 2016) or local assemblages (e.g. Manu National Park, Peru: Dehling, Fritz, et al., 2014; Pigot, Bregman, et al., 2016), and more recently spanning thousands of species worldwide (e.g. Cooney et al., 2017; Kennedy et al., 2020; Phillips et al., 2018; Pigot et al. 2020). However, these resources have until now been fragmented, with raw data largely incompatible and unpublished. To provide an integrated resource with broad utility, managers of different bird trait datasets have joined forces to merge their work into AVONET, a compendium of morphological, ecological and geographical data for all bird species published as the flagship article of this special issue (Tobias et al., 2022). AVONET was inspired by the success of the TRY plant trait database, a potent catalyst of high-impact research in ecology and ecosystem science over the last decade (Kattge et al., 2020). To maximise the likelihood of a similar positive impact, and to align with Open Science principles (Gallagher et al., 2020), AVONET is released as individual measurements of specimens as well as species averages, without restrictions on data access. To some degree, the publication of AVONET marks an endpoint a personal journey. My fascination with bird traits began in the 1980s as a schoolboy walking the tidelines and powerlines of Northumberland in search of corpses for dismembering. I owe a belated debt of thanks to my mother for abiding with bedroom shelves full of skulls and cabinets loaded with malodorous wings and tarsi. But the story of AVONET extends far wider than that, and deeper in time. The completion of this first iteration—AVONET 1.0—is a truly international effort, with vital expertise and data contributed by 115 authors based at 106 institutions in 30 countries. The most important shifts in momentum occurred when the project was joined by colleagues managing their own extensive trait datasets, including Santiago Claramunt (Uruguay), Matthias Schleuning and Susanne Fritz (Germany), Carsten Rahbek (Denmark), Gavin Thomas (United Kingdom) and Gustavo Bravo (Colombia). A common denominator among these major datasets is their reliance on museum specimens. Across AVONET as a whole, most specimens were measured at the Natural History Museum, London and the American Museum of Natural History, New York, with smaller samples from a further 76 collections (see Tobias et al., 2022, Fig. 4). Indeed, the project would not have been possible without the contributions of countless museum curators, field assistants and specimen collectors since the mid-1800s, some luminaries among them, including Charles Darwin, Alfred Russell Wallace, Ernest Shackleton and John James Audubon, all of whom prepared specimens subsequently measured for trait data. Ultimately, given the key importance of well-preserved specimen material for trait-based ecology, AVONET is a monument to the museum community and the crucial service it provides to scientific research and human society in general (Suarez & Tsutsui, 2004). Many sources of information were distilled to provide the first detailed summary of morphological, ecological and geographical data contained in AVONET. Using this resource, anyone can now extract traits, ecology and spatial context for any avian taxon or assemblage—indeed, even for the entire radiation of extant birds. The data can be used to fit models, test hypotheses, or to calculate biodiversity metrics, including various dimensions of functional diversity. Comprehensive data improve the validity of these methods and increase the scale at which they can be applied. For example, tests of evolutionary models can be executed not only on well-sampled clades (e.g. Drury et al., 2018; Tobias et al., 2014) but also across far wider tracts of the avian phylogenetic tree (Crouch & Tobias, 2022). Similarly, methods using traits to quantify niche differences among species are no longer limited to smaller samples (e.g. Pigot & Tobias, 2013) and can now be applied across all birds (Drury et al., 2021; Freeman et al., 2022; Pigot et al., 2018). A unique feature of AVONET is that trait data are presented in alignment with three alternative taxonomic treatments: BirdLife International, Clements and BirdTree (Tobias et al., 2022). In theory, this will be a major time-saver for users, facilitating integration with published geographical range maps and IUCN Red List data, eBird citizen-science data (Sullivan et al., 2014) and the global bird phylogeny (Jetz et al., 2012). Interoperability across these datasets allows an array of research questions to be addressed in novel ways. The following sections summarise recent progress in applying AVONET data across different research fields along with a horizon-scan of emerging opportunities. Although the current global BirdTree (Jetz et al., 2012) is far from perfect and urgently requires an update, it has nonetheless provided a valuable tool for phylogenetic analyses. Recent studies combining AVONET traits with phylogenies downloaded from BirdTree or elsewhere have explored an array of evolutionary topics, including the role of species interactions (e.g. McEntee et al., 2018), ecology (e.g. Crouch & Tobias, 2022) and geographical context (e.g. Benítez-López et al., 2021) in driving phenotypic evolution. With the rapid ongoing improvement of avian phylogenies and the associated toolkit of evolutionary models, AVONET trait data offer an unparalleled template for future analyses of this kind. In particular, there is scope for a new wave of studies focused on intraspecific variation and sex differences, both of which are made possible by the open release of underlying raw measurements for over 90,000 individual birds. Avian genomics is another advancing frontier of evolutionary research, with efforts to sequence the genomes of all extant bird species now well underway (Jarvis, 2016). At the current rate of progress, whole-genome assemblies will soon be sampled for all extant avian genera (>2000), putting birds at the forefront of comparative genomics (Stiller & Zhang, 2019). AVONET data can play a key role in the next phase of this research programme, both in terms of providing traits for genome-wide association studies (GWAS) and predictors in models exploring the drivers of demographic patterns and responses over deep time (Nogués-Bravo et al., 2018). The availability of AVONET trait data allows variation in morphological traits to be mapped and analysed at global scales with reduced sampling bias. The first phase of such analyses included tests of geographical gradients in dispersal-related traits (Sheard et al., 2020) and the role of island colonisation as a driver of predictable trajectories of trait evolution—the so-called ‘island rule’ (Benítez-López et al., 2021). Further studies are needed to explore numerous other putative ecogeographical patterns, such as Bergmann's and Allen's rules, in greater detail. For example, AVONET data open up the possibility of partitioning these effects across different components of phenotype, including trophic, locomotory and dispersal traits. Quantification of niches via morphology may help trait-based analyses to illuminate the complex mechanisms driving community assembly (McGill et al., 2006; Trisos et al., 2014). In particular, the well-established connection between morphological traits and trophic niches in birds (Pigot et al. 2020) suggests that ecological patterns and processes can be inferred from the trait structure of bird communities. Until recently, a key challenge has been that most approaches for estimating community structure are sensitive to gaps and biases in trait datasets (Blonder et al., 2018; Bregman et al., 2016). This challenge has now been addressed with complex morphometric data based on 3-d scanning of bird beaks for several thousand species (Hughes et al., 2022), while AVONET also provides data sufficiently extensive and comprehensive to estimate the trait structure of communities at any scale, from local sites (e.g. Cannon et al., 2019; Chapman et al., 2018; Trisos et al., 2014) to continental or global assemblages (e.g. Sol et al., 2020; Stewart et al., 2022). Movement—or dispersal—is another important cross-cutting theme with relevance to many biological questions. The most promising dispersal trait in birds is the hand-wing index (HWI), a metric of wing-shape related to flight efficiency and dispersal ability (Claramunt, 2021; Sheard et al., 2020). AVONET provides calculations of HWI based on two linear wing measurements (wing chord and first secondary length). The first phase of analyses using earlier versions of AVONET data demonstrated the key role of dispersal in shaping patterns of allopatric speciation (Claramunt et al., 2012) and the build-up of alpha diversity worldwide (Pigot et al., 2018; Pigot & Tobias, 2015). Other analyses have used the same HWI data to test ideas in multiple fields, from evolutionary biology (Menezes & Palaoro, 2022; Stoddard et al., 2017) to conservation (Thaxter et al., 2017). The updated summary of HWI for all bird species released in AVONET may prove useful in any phylogenetic model or comparative analysis testing hypotheses related to dispersal, or wherein variation in dispersal needs to be accounted for. For example, HWI can now be used as an index of dispersal to improve the accuracy of models forecasting geographical range shifts of species under future climate change scenarios (Stewart et al., 2022). Zooming in from assemblage-level analyses to species interactions brings a further set of opportunities into focus. The relationships between morphological traits and key dimensions of avian ecological niches validates the use of trait divergence in studies of range expansion and invasion success among related species. The results of previous analyses are inconclusive, suggesting that trait similarity may either constrain (Pigot & Tobias, 2013) or promote et al., 2022) on Further of this issue is because trait-based metrics of niche similarity and dispersal ability may help us to predict the of future range shifts and a key goal of models in and (Tobias et al., 2020). and forecasting geographical range shifts is a major the morphological traits in AVONET have numerous other potential to global change studies sampling across or thousands of bird species can now traits predict responses to climate change (e.g. et al., 2021). Similarly, studies applying in community ecology have to explore the trait structure and functional diversity of bird assemblages are by climate change et al., 2019; Stewart et al., 2022), et al., 2020), and expansion et al., 2019; Chapman et al., 2018; et al., 2020). Using a trait-based can also important into functional between system For example, the functional diversity of bird species on not the of functional diversity et al., 2021). questions are for in numerous the first phase of this research has focused on the of change on the functional diversity of bird there is increasing that change can also in the morphological traits of species et al., 2021). is the that climate change has effects on avian morphological including for smaller body and longer wings at et al., 2020). With the of further specimen sampling and the extensive intraspecific sampling of traits for many species in AVONET a more of these particularly because morphological can be over the last two using museum specimens. promising of research the of these are sufficiently trait-based approaches may be used to functional to future or communities et al., 2022; et al., 2015). approaches be applied to bird trait datasets, or a of bird and plant traits, to provide a across trophic can also be used to and predict the effects of change on species interactions key to ecosystem function et al., 2020). The first towards this were taken using datasets to shifts in the functional relationships between and birds under climate change et al., 2019). analyses are now at a global analyses have to AVONET data to the of under different an that can help to that ecological et al., 2020). At a global scale, similar approaches an between avian functional diversity and suggesting that to maximise trait diversity in current may be in et al., 2022). of morphological traits can even be into ecosystem models to the of change on ecosystem (e.g. et al., 2013) and the of ecosystem services (e.g. et al., AVONET also the for a about biodiversity For example, trait diversity in can now be to functional for overall assemblages or into different trophic groups key such as dispersal and The morphological diversity of bird communities also provides a foundation for functional of and ecosystem with potential in or for as metrics to the and of biodiversity et al., 2018) or to climate change et al., 2021). that have the data in a phase of these methods and testing their is the most for trait-based ecology are integration of highly functional trait data to provide indices of ecosystem structure and function (Cadotte et al., Mouillot et al., 2021; et al., For example, functional traits with and may provide the for a general of biodiversity that can be up to and predict ecosystem function et al., 2015). These can in be into closer with comprehensive morphological traits. set of as functional the of and function across trophic et al., 2014). trait among species to explore key ecosystem processes by various of including trophic interactions, such as (e.g. and (e.g. analyses have demonstrated or trait at the local scale et al., 2016; Dehling, et al., 2014; et al., 2016), while AVONET data now allows such patterns to be explored (e.g. et al., 2022). as ecological and plant trait datasets combining these resources with AVONET data provides an opportunity to and new methods for trait-based patterns and over spatial scales in a context et al., 2018; et al. 2020). AVONET is an endpoint in some it is also a towards AVONET is with several and life-history trait datasets is also of scope for morphological trait data. intraspecific sampling from a wider geographical is along with of for museum to and to over and time. Second, future of AVONET a wider of traits, including with different dimensions of ecological such as by et al., 2021) and by or et al., 2018). these data will a from the research community and field is to the next phase of trait by using and the AVONET (see Tobias et al., 2022, to the morphological trait data in AVONET are First, the traits are using relatively linear measurements for example, may the major of and without for more The of beaks is to most macroecological studies but is a key in some for example, between beaks and et al., 2022). on and other can be using a resource also published in this special the containing information based on numerous of 3-d (Hughes et al., 2022). Second, AVONET only extant and recently with many progress in addressing this has been by traits of island birds (e.g. et al., 2021) but further efforts are needed to between trait data and extant on traits (e.g. and other The traits of extant species compiled in AVONET have been used as predictors in such as in analyses best variation in the importance of birds to et al., 2019). Other will no integration with data sources a range of is a citizen-science providing to of bird of species and et al., 2018; Sullivan et al., 2014). In the between trait-based ecology and biology can now be explored by AVONET data with global on and (Bird et al., 2020; et al., 2018). A little further on the but nonetheless is the of an ecosystem of and trait datasets which the of trophic between and can be and The morphological data presented in AVONET is based on over many by thousands of in specimen and Similarly, the ecological and geographical information is distilled from published of thousands of field on these AVONET provides a template for research and in or The sections some to the of questions that can now be addressed by and research recent when field field and were all on by the bird trait data into research and to from and morphological trait data provided a for and statistical in and many possible can test evolutionary or ecological hypotheses using phylogenetic and community assembly models, or even for functional diversity. AVONET now open to research and for anyone with access. advances in avian and were by the of global maps of all bird et al., and a phylogenetic tree (Jetz et al., which general to be from The release of new morphological trait data for all bird species has similar potential to models of and community and to provide a more toolkit for and forecasting the of to change (McGill et al., 2006; Tobias et al., 2020). the first complete summary of morphological trait variation across a global AVONET an important step towards more metrics of ecosystem function (Cadotte et al., Violle et al., 2014). In with a range of other trait datasets (e.g. Kattge et al., 2020), these metrics can help to efforts to more ecology et al., and to provide biodiversity indices with in research, international and et al., The thanks Susanne Fritz, Pigot, Matthias Sheard and Gavin Thomas for and

Topics & Concepts