Data from: Habitat structure and an introduced predator limit the abundance of an endangered ground-nesting bird
Data files
Aug 06, 2024 version files 34.51 KB
Abstract
Understanding the factors that limit the abundance of threatened species is critical for the development of effective conservation strategies. However, gaining such knowledge from monitoring programs and using it to inform decision making for rare species can be difficult due to methodological issues posed by the problems of distinguishing true zero counts from false zero counts and analysis of datasets dominated by zero counts. The plains-wanderer (Pedionomus torquatus) is a critically endangered ground-nesting bird that occurs in grasslands of south-eastern Australia. Decline of the plains-wanderer has been attributed to habitat modification but little emphasis has been placed on the role of introduced predators, such as the red fox (Vulpes vulpes), which have had a devastating effect on small ground-dwelling vertebrates in dryland regions of Australia. Here, we use a 9 year-time series of spotlight counts to investigate the impact of vegetation structure and fox presence on plains-wanderer occupancy and abundance. We used distance sampling to determine the effective strip width for sighting plains-wanderers during spotlight surveys. We then used a hurdle model approach whereby binomial generalised additive models were fitted to presence/absence data within the effective strip-width across all sites and negative-binomial models were fitted to an index of abundance at sites where plains-wander were observed. Plains-wanderer occupancy and abundance fluctuated markedly through time. Where foxes were absent, occupancy (but not abundance) of plains-wanderers showed a humped relationship with grass height with an optimal height between 50-150 mm. Where foxes were present however, this relationship broke down and plains-wanderers were rarely recorded. Our results suggest that plains-wanderers should benefit from management strategies that maintain grass height at optimal levels and exclude foxes or effectively suppress their populations. A key message from this study is that if statistical analyses of data generated by monitoring programs for rare species are intended to inform management decisions by identifying relationships between threatened species and drivers of their abundance there should be consideration of analytic approaches that account for true and false zeroes, high prevalence of zeroes and the possibility of non-linear responses.
README: Habitat structure and an introduced predator limit the abundance of an endangered ground-nesting bird
https://doi.org/10.5061/dryad.t1g1jwtb9
Spotlight surveys for plains-wanderers and foxes were conducted 1-3 times per year (Autumn, Winter, Spring) at each site giving a total of 249 surveys. When detected, the perpendicular distance from the survey vehicle to the plains-wanderer or fox was estimated visually. The total transect length varied between sites dependent on the extent and spatial arrangement of suitable plains-wanderer habitat (mean 15.13 ±SD 3.52 km length: range 7 – 22 km length).
Distance sampling was used to identify the effective strip width within which we had a high (≥75%) probability of sighting plains-wanderers and could be confident that zero counts represented true zeroes. Specifically, we used a detection function with a half-normal distribution to identify the distance from the survey vehicle when plains-wanderer detectability reached 75 %. This assessment showed that plains-wanderer detection probability decreased with distance from vehicle, with the 75 % detectability threshold reached at a distance of 5.2 m from the survey vehicle. Given this, we truncated the plains-wanderer dataset to only include observations ≤ 5 m from the survey vehicle. This truncated dataset was then used to calculate plains-wanderer relative abundance within each survey transect (n= 247). To account for differences in line-transect length across sites, plains-wanderer observations were standardised to represent the number of observations per 100 km during each sampling transect.
Fox relative abundance was calculated on each transect as the number of observations per 100 km; Distance sampling was not used because foxes were easily detected within the open grassland habitat. Fox relative abundance provided a qualitative tool to visualise temporal differences in fox activity across the survey period. However, we used fox presence / absence in our statistical analyses (see below). This was because the relatively small size of the survey transects may not have allowed a representative estimate of fox relative abundance, given their large home-ranges.
We measured grass height at each site immediately following each spotlight survey. Maximum grass height was assessed by measuring the maximum modal height (90th percentile of grass heights) of all grasses within 1 x 1 m quadrats at 200-m intervals on each transect. Mean maximum grass height was then calculated by averaging maximum grass heights across all quadrats for each transect.
Description of the data and file structure
The attached excel file has two data files in separate worksheets, one worksheet for the whole dataset on which distance samplimg was performed. the second worksheet is for the truncated dataset after distance sampling. Meta-data is provided in separate worksheets for each dataset.
Methods
Data from spotlight transects for plains-wanderers. Data were first subject to distance sampling then the dataset was truncated to only include observations of birds within the effective strip-width. metadata in attached excel file.