Data and code for: Leveraging long-term data to improve biodiversity monitoring with species distribution models

Anselmetto, Nicolò 1 2 ; Garbarino, Matteo1; Weldy, Matthew J.3 2; Bell, David M.3; Daly, Christopher2; Epps, Clinton W.2; Ferrari, Nina2; Kim, Hankyu4; LaManna, Joseph A.5; Lesmeister, Damon B.3 2; Penaluna, Brooke E.3; Schulze, Mark2; Sutton, Madison5; Tosa, Marie I.2; Betts, Matthew G.2

Published Sep 12, 2025 on Dryad. https://doi.org/10.5061/dryad.j3tx95xqb

Data files

Sep 12, 2025 version files 6.12 MB

data_and_code.zip

6.10 MB
README.md

15.35 KB

Abstract

This dataset contains 10 years (2010–2019) of bird occurrence records and associated environmental predictors from the H.J. Andrews Experimental Forest (Oregon, USA). Bird distributions were monitored during the breeding season at 184 sites, resulting in presence–absence data for 37 species. These observations are paired with fine-scale environmental data, including hourly under-canopy temperature records from 184 microclimate sensors and LiDAR-derived vegetation structure metrics at 25-m resolution. The dataset also includes derived predictor variables (e.g., growing degree days, canopy cover) and R code used for model calibration, validation, and spatial prediction.

The data were assembled to support the development and evaluation of dynamic species distribution models (SDMs) that account for interannual variability and microclimatic heterogeneity. However, the resources have reuse potential for biodiversity monitoring, ecological forecasting, habitat management, and methodological studies of spatial and temporal model transferability.

Bird and microclimate data were collected as part of long-term monitoring programs at the H.J. Andrews Experimental Forest and do not involve animal capture or manipulation. All species are free-ranging and no additional ethical approvals were required.

When using this dataset, please cite the associated article:

Anselmetto N, Garbarino M, Weldy MJ, Bell D, Daly C, Epps CW, Ferrari N, Kim H, LaManna JA, Lesmeister D, Penaluna BE, Schulze M, Sutton M, Tosa MI, & Betts MG. (2025). Leveraging long-term data to improve biodiversity monitoring with species distribution models. Journal of Applied Ecology.

This repository contains data and R code associated with the manuscript.

The dataset provides a .csv dataframe with bird occurrence data (3 focal species) across 182 sampling locations for 10 years and associated microclimate and vegetation predictors, and R code to reproduce model calibration, validation, and spatial predictions.

Folder Structure

data_and_code.zip/open_data/

├── open_data.Rproj             # R Project file
├── .Rhistory                   # R session history
├── README.txt                  # Short internal readme
│
├── data/
│   └── data_raw/
│       ├── df/
│       │   ├── birds_microclimate_year-by-year.csv   # Main dataset
│       │   └── birds_traits.csv                      # Functional traits of bird species
│       │
│       ├── raster/
│       │   └── biomass_mg_ha.tif                     # Biomass raster (25 m resolution)
│       │
│       └── shp/
│           ├── HJA_aoi.shp/.dbf/.prj/.shx/.cpg       # Shapefile of study area boundary
│
├── data_output/   # Stores model outputs and figures (user-generated after running scripts)
│
├── pred_maps/
│   ├── HAWO_*_prediction_*.tif                     # Prediction rasters for Hairy woodpecker
│   ├── RECR_*_prediction_*.tif                     # Prediction rasters for Red crossbill
│   ├── WIWA_*_prediction_*.tif                     # Prediction rasters for Wilson’s warbler
│   └── legend_predictions.txt                      # Legend information for maps
│
└── scripts/
    ├── 00_define_packages_variables.R            # Package loading and global settings
    ├── 01_calibrate_models.R                     # Calls calibration functions
    ├── 02_validate_models.R                      # Calls validation functions
    ├── 03_lmem_valid_movement.R    # Mixed-effects model: performance ~ framework * movement
    ├── 04_lmem_valid_body_size.R   # Mixed-effects model: performance ~ framework * body size
    │
    └── functions/
        ├── 11_calibration_functions.R                # Functions for calibrating SDMs
        └── 12_validation_functions.R                 # Functions for validating SDMs

Recommended execution order:

00_define_packages_variables.R → 01_calibrate_models.R → 02_validate_models.R → 03_lmem_valid_movement.R → 04_lmem_valid_body_size.R

Main Dataset: birds_microclimate_year-by-year.csv

This file contains year-by-year data for 184 sites (2010–2019) for three focal species:

HAWO = Hairy woodpecker (Dryobates villosus)
RECR = Red crossbill (Loxia curvirostra)
WIWA = Wilson’s warbler (Cardellina pusilla)

Columns and Definitions

Column	Description	Units
ID	Unique record ID (site + year)	-
SITE_ID	Sampling site code	-
YEAR	Survey year (2010–2019)	-
HAWO, RECR, WIWA	Presence/absence of species	1 = present, 0 = absent
Apr_Jun_mean_max	Mean maximum temperature, April–June	°C
Jul_Sep_mean_max	Mean maximum temperature, July–September	°C
GDD_winter_5	Growing degree days (>5 °C threshold) during winter	°C·days
CDD_0_1_3, CDD_0_4_6, CDD_10_4_6	Cooling degree days over specified months and thresholds	°C·days
m_X_max/mean/min	Monthly max/mean/min temperature for month X (1 = Jan, …, 12 = Dec)	°C
max_1_3, max_10_12	Mean of monthly maximum temperatures (Jan–Mar; Oct–Dec)	°C
mean_1_3	Mean of monthly mean temperatures (Jan–Mar)	°C
SD_1_3, SD_4_6	Standard deviation of monthly mean temperatures (Jan–Mar; Apr–Jun)	°C
biomass_mg_ha	Aboveground biomass	Mg/ha
closure_2, closure_10, closure_40	Canopy closure at 2, 10, and 40 m height	%
density_0_2, density_2_10	Vegetation point density in height bins (0–2 m, 2–10 m)	%
height_mean	Mean canopy height	m
height_sd	Standard deviation of canopy height	m
veg_height_DEM	Vegetation height above DEM	m
x, y	UTM coordinates (EPSG:32610, WGS84 / UTM zone 10N)	meters

Additional Datasets

birds_traits.csv

Contains functional trait information for the 37 bird species monitored at H.J. Andrews Experimental Forest derived from AVONET available on https://figshare.com/articles/dataset/AVONET_morphological_ecological_and_geographical_data_for_all_birds_Tobias_et_al_2021_Ecology_Letters_/16586228.

Column	Description	Type / Units
species	4-letter species code	Text
common_name	English common name	Text
scientific_name	Scientific name	Text
sn_spec_code	6-letter species code	Text
family	Taxonomic family	Text
order	Taxonomic order	Text
cluster_group	Modeling cluster group ID	Integer (unitless)
migratory	Qualitative migratory status	{resident, part_migratory, migratory}
migration	Encoded migratory status	Integer (1=resident, 2=part_migratory, 3=migratory)
habitat	Broad habitat	Text (e.g., Forest)
habitat_density	Ordinal habitat density code	Integer (unitless; ordinal)
trophic_level	Trophic level	{Herbivore, Carnivore, …}
trophic_niche	Diet niche	{Granivore, Invertivore, …}
lifestyle	Ecological lifestyle	Text (e.g., Insessorial)
Inference	Whether trait values were inferred/imputed	{NO, YES}
Beak.Length_Culmen	Bill length from culmen base to tip	mm
Beak.Length_Nares	Bill length from anterior edge of nares to tip	mm
Beak.Width	Bill width at nares	mm
Beak.Depth	Bill depth at nares	mm
Tarsus.Length	Tarsus length	mm
Wing.Length	Flattened wing chord	mm
Kipps.Distance	Kipp’s distance (wing tip to first secondary)	mm
Secondary1	Length of first secondary	mm
Hand-Wing.Index	Hand-wing index = 100·(Wing.Length − Secondary1)/Wing.Length	%
Tail.Length	Tail length	mm
Mass	Body mass	g
body_size	Standardized body-size index (higher = larger)	-

biomass_mg_ha.tif

A raster map of aboveground biomass at 25 m resolution, used to construct spatiotemporal modeling blocks. Units: Mg/ha.

HJA_aoi shapefile

Boundary polygon of the study area (H.J. Andrews Experimental Forest, Oregon, USA).
Projection: UTM zone 10N (EPSG:32610). NB: keep the .shp, .shx, .prj, .dbf, .cpg in the same folder and load the .shp file in QGis or R to inspect it!

Prediction rasters (.tif)
Raster outputs of spatial predictions from boosted regression tree (BRT) models for each of the 3 focal species under different modeling frameworks.
Files are named according to:
Species_Year_Framework_prediction.tif
(e.g., HAWO_2015_BRT_prediction_RY.tif).
Units: probability of occurrence (0–1).

legend_predictions.txt

Provides classification legend and metadata for interpreting prediction rasters.

External Data Sources

Bird response data (full set of 37 species): Available at
- Environmental Data Initiative: https://doi.org/10.6073/pasta/161101744422c412802e8287dbd8b663
- H.J. Andrews LTER database (code SA024): https://andlter.forestry.oregonstate.edu/data/
Microclimate predictors: H.J. Andrews LTER database (code MS045): https://andrewsforest.oregonstate.edu/data
LiDAR data: Oregon Department of Geology and Mineral Industries (Oregon Lidar Consortium): https://www.oregongeology.org/lidar

Software

R (≥4.2.3) – statistical computing (https://cran.r-project.org)
Required R packages are listed in 00_define_packages_variables.R
Raster data and shapefiles can be opened with QGIS (https://qgis.org)

__________________________________________________________________________________________________________________

Usage Notes (formats & how to open them)

CSV files (*.csv)

Open with R (e.g., data.table::fread()), Python (e.g., pandas.read_csv()), or spreadsheet software (e.g., Microsoft Excel, LibreOffice Calc).
Text encoding is UTF-8; field delimiter is comma.

GeoTIFF rasters (*.tif)

Open with QGIS, SAGA, or other GIS softwares.
In R: terra::rast("path/to/file.tif")

In Python:

import rasterio
with rasterio.open("path/to/file.tif") as src:
    arr = src.read(1); print(src.crs, src.transform)

Shapefiles (*.shp)

A shapefile consists of multiple files: at minimum .shp, .shx, .dbf, .prj, .cpg. Keep all together in the same directory; do not rename only one component.
Open with QGIS (Layer → Add Layer → Add Vector Layer).

In R:

library(sf)
aoi <- sf::st_read("data/data_raw/shp/HJA_aoi.shp")