GABB: A global dataset of alpine breeding birds and their ecological traits

Defining alpine habitat and mountain regions

We defined alpine habitat as the area above climatic treeline, including the nival belt, where temperature, wind, drought, snow, or nightly frost limit vegetation growth to shrubs, krummholz, or fragmented tree patches less than 3 m in height^3,23,24. Realized treeline can be markedly lower than the climatic treeline due to the absence of continuous forest at lower elevations, or human activities such as logging, burning, and livestock grazing²⁵. While anthropogenically influenced treeline produces habitat reminiscent of alpine meadows, these habitats are not climatically representative of alpine ecosystems and thus they were not included when assembling this dataset. Climatic treeline elevation varies globally based on latitude, topography, aspect, and proximity to the coast (i.e., oceanic influence)¹¹. Therefore, we defined alpine habitat separately for each mountain region based on local climate and published accounts of alpine vegetation. While alpine habitats usually occur above at least 1,500 m elevation globally, at high latitudes (>55°N or 41°S) this elevation can be as low as ~400 m²⁶ (Fig. 2).

The alpine habitats we identified broadly align with the ‘lower alpine’, ‘upper alpine’, and ‘nival’ belts mapped by Korner et al.⁹ and made available by the Global Mountain Biodiversity Assessment project (http://www.mountainbiodiversity.org/explore)^27,28. However, certain areas, such as the Sierras de Córdoba, Argentina or the Isthmian Páramo on volcanoes in Central America were classified as ‘upper montane’ by Korner et al.⁹ based on thermal belts alone. For the purposes of this dataset, we considered these regions alpine habitat based on published measurements of treeline and distinct alpine plant communities facilitated by a mixture of temperature, precipitation, nightly frost, and wind constraints. For example, the Drakensberg range in South Africa was identified as ‘upper montane’ only, but botanical studies have characterized the region as Themeda-Festuca grassland from 1,900–2,800 m and alpine heathlands above 2,800 m¹³, representing extensive habitat above treeline. As a result, our definition of alpine habitat expands on the thermal belts mapped by Korner et al.⁹. In this way, the avian communities we identified retain species lineages that are confined to cooler high elevation habitats, representing remnants of more extensive alpine ecosystems from the last glaciation event.

We grouped mountain ranges into 12 global regions and 38 subregions based on similar climates and alpine vegetation stemming from shared geographic position (Tables 2, 3; Fig. 1). The ‘Islands’ category represents very limited alpine habitat on four disparate islands that do not easily fit within any other major region, but nevertheless occur in subtropical or tropical realms: Hawaii, Sumatra, Borneo, and the Canary Islands. Alpine breeding birds and life-history traits were identified for each individual region so that future analyses can either include or remove mountain ranges depending on their definition of alpine habitat. This approach also promotes comparisons of avian communities at a finer scale across the full diversity of alpine habitats.

Alpine breeding bird species

For each region described in Tables 2 and 3, we assembled a list of alpine breeding species from published literature, environmental assessment reports, regional monitoring schemes, bird atlases, and expert knowledge following the most recent International Ornithology Committee taxonomy, version 12.1²⁹. An alpine breeding bird is any species that nests above treeline, regardless of how frequently, such that all or a certain proportion of a species is dependent on alpine habitat during the breeding season. Due to certain data-deficiencies underlying existing species range estimates above treeline, using knowledge from regional experts was the most accurate method to assemble a global list of alpine breeding birds for most mountain regions. See the Technical Validation section for specifics on how we validated the use of expert knowledge when assembling species and their traits for the Global Alpine Breeding Bird list.

Species traits

We included species traits that fall under three general topics: 1) alpine breeding propensity, 2) ecological traits, and 3) conservation value. Alpine breeding propensity includes breeding habitat specialization and alpine breeding status, ecological traits include migration behaviour and nest traits, while conservation value encompasses mountain endemism and conservation status. Together, these variables broadly reflect alpine habitat use during the breeding season globally, as well as provide the basis for evaluating the conservation potential and risks for alpine bird communities. We recorded general trait specifications for each species using available resources such as Birds of the World³⁰, the IUCN Red List³¹, and AVONET²¹. We then solicited region-specific traits from regional experts and the same review process was conducted for these traits as for alpine breeding evidence (see Technical Validation). All traits were specific to alpine breeding birds whenever possible. The global distribution of each species trait can be visualized in Fig. 3.

Specialization for breeding in alpine habitats (hereafter ‘breeding specialization’) and the propensity to breed in alpine habitats (hereafter ‘breeding status’) form a tiered estimate of alpine breeding behaviour. First, we classified each species into one of three breeding specialization categories to differentiate among species that predominantly breed above treeline (alpine specialists), breed both above and below treeline (elevational generalists) or are limited to high latitude tundra habitats (tundra specialists). The latter includes alpine-Arctic or alpine-Antarctic transition zones, where species nest in higher, drier tundra (approximately >400 m elevation) but may also breed in wet tundra at lower or coastal elevations. In this way, we clearly identified species that breed in alpine tundra habitat, but where tundra is the primary driver of breeding presence, not necessarily selection for high elevation. Under breeding status, we quantified the likelihood of breeding above treeline relative to below treeline as common, uncommon, or rare. Alpine specialists are always common alpine breeders (regardless of their density and distribution), but generalists or tundra specialists can be common, uncommon, or rare breeders in alpine habitats depending on whether they are found breeding consistently above treeline, more often breeding below treeline, or only incidentally breeding in the alpine, respectively. Together, these variables identify a species’ relative probability of breeding along the elevational gradient and with respect to the treeline (Fig. 3).

We used two nest traits to identify the general breeding niche of each species: nest type and nest site. Nest type included three primary category levels (open cup, cavity, domed nest), while nest site was subdivided into seven levels (ground, bank, shrub, tree, rock, cliff, and glacier). Brood parasite species, which will use a range of nest types and sites depending on the host species, were placed in an additional ‘brood parasite’ category for each nest trait. A species with an open cup or domed nest is limited to placing the nest on the ground, in vegetation (e.g., a shrub or stunted tree), or on a cliff, while cavity nesters may be in a bank (i.e., burrow/tunnel), in a rock (e.g., crevice), or in a tree (e.g., natural or excavated cavity). If nest traits were undescribed for a certain species, we inferred nest traits from the most closely related species in similar high elevation habitats (see Data uncertainty).

Species were assigned to three migration categories based on their predominant behaviour: resident, short-distance, and long-distance migrants. Resident species remain near their breeding habitat year-round, allowing for occasional, short-term movements in response to extreme weather events. Short-distance migrants conduct seasonal altitudinal migrations, short latitudinal migrations, or nomadic movements where the species remains within the general breeding region (e.g., within the temperate zone). Long-distance migrants travel extensive distances to winter in an entirely different region than their breeding habitat (e.g., temperate breeders to tropical habitats). A general threshold of 3,000 km was used to distinguish between short- and long-distance migrants because it approximates the distance traveled from the Himalayas to the southern coast of India, Northern Europe to the Mediterranean coast, or Alaska to California. In other words, this distance represents a relatively consistent reference across global regions. While there are finer-scale migration designations that could be made, such as partial or altitudinal migration, we lack detailed movement data for most species and regions. Although a global list of potential altitudinal migrants exists that can be incorporated with this alpine breeding bird dataset if desired³², altitudinal migration often co-occurs with short-distance latitudinal movements and there are considerable differences in migration behaviour among subspecies, populations, and even individuals³³. We therefore chose to use established migration categories that align with other global trait databases. In fact, our migration designations were largely congruent with those in AVONET²¹, with the primary difference being between resident and short-distance migrants. We identified ~200 short-distance migrants that were considered sedentary (resident) under the AVONET classification. This difference is to be expected given that we defined migration behaviour for alpine breeding populations compared to global trait values for all populations. For many species, alpine breeding birds will depart higher elevations during winter to avoid severe weather conditions, even though low elevation populations of the same species may be predominantly resident³⁴. Therefore, the three broad categories chosen here are intended to balance available information with sufficient accuracy to provide data useful for large-scale life-history and biogeographic analyses of alpine breeding birds.

Mountain endemism refers to a species whose breeding range is restricted by physical, environmental, or biological barriers to a general mountain region and the surrounding low elevation habitat. For example, a species breeding only on the Tibetan Plateau was classified as an endemic species, but a species that breeds across the Tibetan Plateau, the Himalayas, and the Altai Mountains was classified as non-endemic. When possible, we also classified endemism for defined subspecies. Species endemism is a more conservative metric, while subspecies endemism attempts to estimate additional cryptic endemism given that species differentiation is not well-defined for many high elevation birds. For example, the Caucasus Mountains support several distinct subspecies isolated from their primary distributions, including the Great rosefinch (Carpodacus rubicilla rubicilla), Dunnock (Prunella modularis obscura), and Güldenstädt’s redstart (Phoenicurus erythrogastrus erythrogastrus).

Finally, conservation status refers to the IUCN Red List designations, version 2022-1³¹. In addition to the traditional IUCN categories (e.g., Least Concern, Near Threatened, Vulnerable, etc.), we also included a Not Assessed (NA) category that generally occurred when a species was recently split. See Table 4 for complete definitions of all traits.

Data uncertainty

Globally, there is significant variation in accessibility to alpine habitats and funding for alpine research. As a result, uncertainty in alpine breeding status may differ among regions and species. For example, in New Guinea, mist-net surveys and point counts across elevation have identified species that frequently use alpine habitat, but a dearth of breeding biology studies means that there are few nest records above treeline. It is thus necessary to codify this level of uncertainty for each species.

To this effect, we included a variable termed ‘Data reliability’, which is a four-level categorical variable from 0 to 3 that is based on the number of reported nests that have been found and described for each species. We used the presence of nest descriptions to evaluate uncertainty because active nests are the must fundamental form of evidence for breeding above treeline, and therefore it is reasonable that a species with less existing knowledge about nest traits or nesting behaviours will have considerably more uncertainty around its designation as an alpine breeding species. For this variable, 0 indicates that nest traits are undescribed for a given species, 1 means less than five nests have been described, 2 indicates more than five nests have been described, but all from a single population, and therefore there is limited understanding of geographic variation, while 3 occurs when nests have been described from multiple populations or regions. If nest traits were undescribed for a species (data reliability = 0), then nest type and site were inferred from the most closely related species with available data, and whenever possible, a congener was selected that also breeds at high elevations or in alpine habitats. While the nest traits of most species have been sufficiently described, there is a significant proportion of alpine breeding birds with less available data (27.0%; Fig. 3i). The relative number of described nests was derived from Birds of the World³⁰. We recognize that these data may not reflect true knowledge of nest traits given that not all species accounts have been recently updated. However, it does represent a consistent data source that allowed us to approximate data reliability sufficiently for our purposes.

In combination, data reliability and alpine breeding status fully characterize alpine breeding uncertainty. For example, a species considered a rare alpine breeder with a data reliability of 3, means that there is strong evidence for breeding above treeline, but only incidentally under very specific circumstances. However, a rare alpine breeder with a data reliability of zero (i.e., nest undescribed), means that the likelihood of breeding above treeline may be probable based on behavioural observations, but further confirmation is required. When using this dataset for analyses, one must decide whether to use a conservative approach or consider all potential alpine breeding species with the appropriate caveats (see Usage Notes).

Source: Ecology - nature.com