A trait dataset for Taiwan's breeding birds

Abstract Background Species traits affect how a species interacts with the environment and other species and thus determine the role of the species in an ecosystem. They affect not only population dynamics of a species across space and over time, but also community structure and function through their key role in the community assembly processes. Information on species traits is also highly relevant for conservation planning as they determine the adaptive ability of a species in the face of environmental changes. However, information on species traits is usually scarce and sparsely distributed amongst diverse types of literature and sources. Difficulty in accessing comprehensive information on species traits has formed an essential knowledge gap, limiting our understanding of biodiversity patterns and ecosystem functioning and preventing effective conservation. Even for birds, a well-studied taxon, comprehensive trait information is still unavailable or distributed across different sources for many species. New information In this study, we compiled information from a variety of sources on 23 traits for all breeding birds, including 157 resident and 14 summer visiting species, in Taiwan and surrounding islands. The 23 traits include those related to the movement patterns, morphology, geographic distributions, activity patterns, feeding behaviour, habitat use, and breeding behaviour and strategies of the species. The trait information was obtained, not only from published literature and datasets, but also from unpublished banding records and specimen measurements. The dataset also contains derived traits, including the elevation and temperature boundaries of species distribution ranges in Taiwan. In addition, structured information on nest characters, which is seldom compiled in trait datasets, has been made available, for the first time, for the breeding birds in Taiwan. Therefore, the most comprehensive trait dataset to date on breeding birds in Taiwan will allow trait-based research and applications in diverse topics and thus enhance our understanding of the patterns and dynamics of breeding bird diversity and its functions in Taiwan.


Introduction
Traits are morphological, physiological, demographic and behaviour characteristics of organisms (McGill et al. 2006, Violle et al. 2007. They are related to organisms' fitness, interactions with the environment and other organisms and ecological roles in an ecosystem (Reed et al. 2013, Kraft et al. 2015, Santini et al. 2016. They may also affect the community assembly process (Laureto et al. 2015) and thus determine the property of a community or even an ecosystem (McGill et al. 2006, Gross et al. 2017, Pigot et al. 2016. Moreover, traits are important in biodiversity conservation because they, with their movement, distributions, habitat use and breeding strategies, can be good indicators of species' adaptive ability and vulnerability to environmental changes (Pöyry et al. 2009, Pacifici et al. 2017) and species' threatened status and extinction risk Jetz 2010, Cooke et al. 2019).
Unfortunately, despite the importance of traits, comprehensive information on species traits is usually unavailable or is distributed across different sources, even for well-studied taxa, such as birds. EltonTraits (Wilman et al. 2014) is currently one of the most comprehensive (in terms of taxonomic coverage) trait datasets for birds. However, it lacks information on many important traits. For example, body mass is the only morphological trait in the EltonTraits dataset, although some other characteristics of morphology, such as tail length and wing shape, are related to the ecological role of a bird (Phillips et al. 2018, Hsu et al. 2013). In addition, the dataset also lacks information on the traits related to the breeding strategy, such as clutch size and nest structure, which critically determines the performance and fitness of birds (Griesser et al. 2007, DuRant et al. 2013, Altamirano et al. 2019, their adaptation capacity to changing environments (Deeming andReynolds 2015, Pacifici et al. 2017) and their extinction risk (Cooke et al. 2019). Some other datasets cover more traits, but they are usually limited in terms of geographic and thus taxonomic coverage (eg. Garnett et al. 2015, Myhrvold et al. 2015, leaving many trait data gaps. Taiwan has abundant biodiversity, including more than 600 bird species, thanks to the various climates and vegetation types along its almost 4000 m elevation gradient. Although bird traits have been found to be an important determinant of the diversity pattern and species composition of both native (Ding et al. 2005) and invasive birds in Taiwan (Su et al. 2016), a comprehensive dataset on bird traits is still lacking. Therefore, in this study, we compiled the information on 23 traits, including morphological, behaviour, distributional and life-history traits, for all breeding birds in Taiwan. This structured and relatively comprehensive trait dataset can support trait-based research and diverse applications on Taiwan's breeding birds. For example, several traits, related to thermal tolerance, diets and habitat use, have been associated with birds altitudinal migration tendencies to explore potential drivers of this behaviour (Tsai et al., in review). This dataset may also improve our ecological and evolutionary understanding of the community assembly in an island-like system (Mittelbach andSchemske 2015, Weber et al. 2017).

General description
Purpose: The aim of this study was to compile trait information from diverse sources and to provide a dataset containing structured and the most comprehensive trait information to date for all breeding birds in Taiwan.

Habitat
Habitat types Thirteen habitat types (IUCN Red List of Threatened Species; http:// www.iucnredlist.org/): 1. Forest: areas with continuous stands of trees, including both forested and wooded areas 2. Savannah: areas with grass ground cover with an overstorey of widely-spaced trees and shrubs 3. Shrubland: areas with scrub, bushland and thicket 4. Grassland: areas with grasses and broadleaved herbaceous plants, without or with very sparsely-distributed trees 5. Wetland: areas with marsh or water, where the water is either flowing or stagnant 6. Rocky area: including inland cliffs, mountain peaks, talus and feldmark 7. Neritic area: including submergence, nearshore, on or over the continental shelf or oceanic island shelf 8. Intertidal area: areas of the shore between the extremes of high and low tides 9. Coastal or supratidal area: coastal areas above the high tide mark 10. Terrestrial artificial area: human-made terrestial habitats 11. Aquatic artificial area: human-made wetland habitats 12. Other habitats 13. Unknown habitat  Our own calculations by overlapping bird occurrence records from eBird (Sullivan et al. 2009) and the environmental data layers of elevation (Global Digital Elevation Models GDEM; https://gisgeography.com/free-global-dem-data-sources) and temperatures (Climatologies at High Resolution for the Earth's Land Surface Areas CHELSA; Karger et al. 2017).
Below are the detailed approaches for obtaining and compiling the information on individual traits:

Movement
One trait, residency, belongs to the category of movement. According to the Checklist of Birds of Taiwan (CWBF 2017), the residency of birds can be classified into six types: residents, summer visitors, winter visitors, passage migrants, straggers and introduced species. Although our dataset focuses on the trait information for the resident and summer visitor species in Taiwan, some of those species also contain winter visiting or passage migratory populations. Therefore, we used four binary variables to indicate whether a species contains a population of each of the four residency types (Table 2).

Morphology
Our dataset contains six morphological traits, including body mass, body length, head length, natural wing length, tail length and tarsus length (Table 1). We used three methods to calculate or obtain the mean and/or standard deviation of each of the traits: (1) calculating the two statistics from measurements on individual birds, (2) calculating the mean as the average of the maximum and minimum values reported in the data sources and (3) obtaining the two statistics directly from the data sources. The first method was to calculate the values from the measurements on living individuals or specimens, which were obtained from the Avifauna of Taiwan, MAPS banding datasets and the museum specimens. The measurements of juveniles and subadults were excluded from the calculation if the information was available. Some clear outliers (e.g. having a different order of magnitude) were also excluded. When measurements on individuals were not available from the above sources, we calculated the average of the maximum and minimum trait values reported in the Avifauna of Taiwan or HBW Alive. When only the mean and/or standard deviation were available in data sources, including the Avifauna of Taiwan, HBW Alive and the EltonTraits dataset, we included the statistics directly. The information on the data source, sample size and calculation method for each morphological trait of each species is provided in Table 3. For the measurements on individual birds, the information on the individual identifiers (identifiers of museum specimens or series number of banding records) is provided in Suppl. material 1. For the measurements obtained from the Avifauna of Taiwan, the volume and the page where the measurements were extracted are also included in the file.

Activity and Feeding
We obtained the information on the activity time, diet and foraging stratum of the bird species from the EltonTraits dataset (Wilman et al. 2014) by subsetting the dataset for the 171 breeding bird species in Taiwan. Therefore, the information on the three traits in our dataset is the same as that in the EltonTraits dataset. The reason for including the * # information was for completeness. The original information sources and the approaches for compiling the information can be found in Wilman et al. (2014).

Habitat
We obtained information on habitat use of the bird species from the Avifauna of Taiwan (Severinghaus et al. 2010) and defined the habitat types following the habitats classification scheme used by the Red List of Threatened Species of the International Union for Conservation of Nature (IUCN; https://www.iucnredlist.org/resources/habitatclassification-scheme). Thirteen of the 18 IUCN first-level habitat types were used by the 171 breeding bird species in Taiwan and, thus, 13 binary variables were included in our dataset to indicate which habitat types were used by each of the species (Table 2).

Breeding
Eight traits, including the clutch size, incubation period, egg length, egg width, brood parasitism, nest structure, nest site and nest attachment, are in the trait category of breeding (  Table 3. We obtained the information on the nest structure, nest site and nest attachment approaches from the HBW Alive (Del Hoyo et al. 2018) and defined seven, seven and four types of the three traits (Table 1), respectively, following Fang et al. (2018). As a single species may build nests with more than one type of structure, site and/or attachment, we used seven, seven and four binary variables to indicate the types of the three traits, respectively, for each species (Table 2). Some bird species do not build their own nests, but lay their eggs in the nests of other species and rely on the hosts to raise their offspring. Therefore, we included another trait, brood parasitism, to indicate the brood parasite species, whose nest traits cannot be defined.

Geographic coverage
Description: Taiwan, including Taiwan's main island, Penghu Islands, Green Island, Lanyu Island, Guishan Island and the Three Northern Islands (Fig. 1).