Scripts from: Remotely sensed microhabitat characteristics associated with Haematopus palliatus (American Oystercatcher) nest-site selection can inform beach habitat restoration along the U.S. Atlantic Coast
Data files
Aug 15, 2025 version files 28.96 KB
-
ORNITH-APP-25-026.zip
26.30 KB
-
README.md
2.66 KB
Abstract
Shorebird populations are in steep decline and have lost almost 70% of their North American populations on average since 1970. The declines are accelerating for most species, in part because coastal environments face significant threats from anthropogenic activities, leading to habitat loss or degradation. Haematopus palliatus (American Oystercatchers) have begun to recolonize areas in their northern breeding range where they were previously extirpated; however, threats persist due to habitat loss, human disturbances, and high predation risk. Coastal habitat restoration, particularly the use of dredged material to enhance beach habitats, has been shown to be a viable conservation strategy. To inform restoration design, this study aims to quantify the microhabitat characteristics influencing H. palliatus nest-site selection across sandy beaches along the U.S. Atlantic coast, utilizing high-resolution lidar data. We analyzed 1,349 nest locations compiled from targeted surveys, comparing them to randomly generated pseudo-absence points. We calculated 17 geomorphological variables representing microhabitat from public domain, lidar-based digital elevation models and employed boosted regression trees to model their relationship to nest presence. Results indicated that nest elevation, distance to mean sea level and mean high water, and shoreline sinuosity were the most important predictors of nest presence, with differences observed between northern and southern regions. In the northern states from New Jersey to Connecticut, low shoreline sinuosity was most important, likely due to its role in reducing predation risk, whereas in the southern states from Virginia to Florida, nest elevation emerged as the dominant factor, suggesting a regional adaptation to the more rapid increases in sea level rise along the southern Atlantic coast. These findings highlight the importance of microhabitat characteristics in the design, restoration, and management of shorebird nesting habitats, and can contribute to best practices for conservation efforts aimed at sustaining H. palliatus populations amidst escalating coastal threats.
Dataset DOI: 10.5061/dryad.6wwpzgnb2
Description of the data and file structure
Summary
This directory contains the scripts used in the Ornithological Applications publication titled Remotely sensed microhabitat characteristics associated with Haematopus palliatus (American Oystercatcher) nest-site selection can inform beach habitat restoration along the U.S. Atlantic Coast (Grand et al. In Press).
The nest location data associated with this publication are owned by multiple institutions/agencies and require permission to use; therefore, the dataset is not included here. All data owners are listed in the acknowledgements section of the journal article.
Files and variables
File: ORNITH-APP-25-026.zip
Nest coordinates used in the analysis require permission from the data owners; however, the scripts provided can be applied to any set of nest coordinates and random points.
Code/software
Python scripts for calculating nest and random point covariates
1) Covariates_tidal_and_elevation_1.py: Run this script for each site (uncomment one site at a time)
2) Covariates_landform_2.py: Run this script for each site.
3) Covariates_combine_site_results_3.py: This script should be run just once, after all site-specific calculations are complete.
Please refer to additional data preparation instructions at the beginning of each script.
R scripts for Running Boosted Regression Trees
1) BRT_functions.R: These functions iteratively determine learning rate, bag fraction, and tree complexity and select the best full model while ensuring the model was built with >1000 trees. The file also contains a function that generates the variable importance plots.
2) BRT_script_AMOY_1.R: This script cleans up the input file, adds latitudinal and longitudinal folds for spatial cross validation, checks for correlations among covariates and for covariates with no variation, and calls the functions in BRT_functions.R to model presence/absence of AMOY nests.
3) BRT_script_AMOY_2.R: This script summarizes variable importance, deviance explained, cross-validated AUC, TSS, and Moran’s I. It also produces draft response plots using built in functions.
4) BRT_script_AMOY_3.R: The script produces customized response plots more suitable for publication than the built in function.
