Limited capacity of landscape features to buffer grassland bird declines

Grassland birds are among the most rapidly declining groups of species, largely due to historical habitat losses that have led to extinction debts. Beyond direct habitat manipulations (e.g., prescribed fire), conservation strategies for grassland birds often adopt a landscape-scale perspective, particularly when guiding decisions on reserve acquisition, prioritizing management across existing reserves, and restoring habitat. Additionally, climate is increasingly recognized as a critical consideration for grassland bird conservation, since temperature influences survival and nest success, while precipitation has an outsized effect on primary productivity and habitat structure in grassland ecosystems. Here, we used a 16-year breeding season avian point count dataset from some of the highest-quality patches of remaining grassland habitat in Minnesota, USA, to 1) estimate occurrence trends for 33 species that use grasslands and 2) assess the influence of two traditional habitat variables (amount of grassland habitat, grassland edge density) and two climate factors (temperature precipitation anomaly) on baseline occurrence and trends. Our model reveals declines across bird species that use grasslands; 4 out of 8 grassland-obligate species declined (with ≥95% confidence; 6/8 declined with ≥68% confidence) whereas 6 out of 25 facultative grassland species declined (with ≥95% confidence). The traditional habitat variables were associated with baseline occurrence of relatively few species (e.g., proportion of grassland habitat showed a clear relationship with baseline occurrence of 4 species, including 2 obligates) and buffered declines for only a handful of species, none of them obligates (e.g., amount of grassland habitat showed a buffering effect for Red-winged Blackbird). In contrast, breeding-season climate anomalies showed strong relationships with occurrence of more species (temperature: 7 species; precipitation: 11 species). However, climate effects on trends were less common (e.g., wetter breeding seasons buffered declines of Savannah Sparrows). Considered holistically, our results suggest that grassland bird declines are apparent even in the highest-quality remaining habitats of the upper midwestern US. Habitat factors may offer limited buffering against future grassland bird declines, especially as temperature and precipitation anomalies exert increasingly strong impacts on population trajectories and climate conditions continue to change. $~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~$

• 01_calculate_grassland_vars.R Calculate proportion grassland and edge density from contiguous grassland shapefile
• 01_calculate_weather_anomaly.R Calculate weather anomalies from long-term conditions
• 02_format_data.R Format data for analysis
• 03_check_linear_quadratic.R Quick check of linear vs. quadratic effects of climate variables
• 03_fit_models_hpc.R Fit models
• 04_figure_01.R Create Figure 1 (study area map)
• 04_figure_02.R Create Figure 2 (trend estimates)
• 04_figure_03.R Create Figure 3 (effects of ecological covariates on occurrence and trends)
• 04_figure_04.R Create Figure 4 (species heatmap showing covariate effects)
• 04_figure_s01.R Create Figure S1
• 04_figure_s02.R Create Figure S2
• 04_figure_s03.R Create Figure S3
• 04_figure_s04.R Create Figure S4
• 04_table_s01.R Create Table S01

• spatial Folder containing spatial data files
  • SPICE_contig_grass_NoHoles_20260213.shp Shapefile for contiguous grassland habitat (.cpg, .dbf, .prj, .sbn, .sbx, and .shx components also present in this folder)
  • allsites_SPICE_20201228_multipart.shp Shapefile with reserve boundaries (.cpg, .dbf, .prj, .sbn, .sbx, and .shx components also present in this folder)
  • ecs_provinces_of_mn_v99a.shp Shapefile with ecoregions of Minnesota (.cpg, .dbf, .prj, .sbn, .sbx, and .shx components also present in this folder)
• MN_prairie_bird_data.xlsx Grassland bird survey data (format received from collaborators); included here since it is used in scripts to calculate weather anomalies and habitat variables.
• habitat_vars.csv Point-level habitat variables

• list_of_data_lists3.RData List of data lists for modeling. 8 top-level elements (1 for each model). Each element is a list of six, containing model constants (loop control, etc), data, model code, parameters monitored, initial value function, and model label.
• mn_prairie_bird_data_clean.csv Grassland bird survey data. Data dictionary provided below.

• sp_key.csv Key with species identifiers

  Column name	Type	Meaning
  sp	double	species index used in model
  code	character	4-letter species code
  common	character	species common name

• weather_anomaly.csv Weather anomalies

  Column name	Type	Meaning
  point	character	name of survey point nested within reserve
  year	double	year
  panom_br	double	breeding season (May and June) precipitation anomaly
  tanom_br	double	breeding season (May and June) temperature anomaly
  panom_nb	double	nonbreeding season (Jan-Apr) precipitation anomaly
  tanom_nb	double	nonbreeding season (Jan-Apr) temperature anomaly

• figure_01.png Figure 1
• figure_02.png Figure 2
• figure_03.png Figure 3
• figure_04.png Figure 4
• figure_s01.png Figure S1
• figure_s02.png Figure S2
• figure_s03.png Figure S3
• figure_s04.png Figure S4

• 01_pgrass_250_tanom_br_panom_br.RData Model information and model output for model with pgrass (250 m) and breeding-season climate variables. RData with the following objects:
  • constants: constants (loop control, indices) for model
  • data: data used to fit model
  • out: model output with posterior distributions of model parameters
  • settings: MCMC settings used to fit model
  • model_code: Nimble model code
• 02_pgrass_250_tanom_nb_panom_nb.RData Model information and model output for model with pgrass (250 m) and nonbreeding-season climate variables
• 03_pgrass_3000_tanom_br_panom_br.RData Model information and model output for model with pgrass (3000 m) and breeding-season climate variables
• 04_pgrass_3000_tanom_nb_panom_nb.RData Model information and model output for model with pgrass (3000 m) and nonbreeding-season climate variables
• 05_ed_250_tanom_br_panom_br.RData Model information and model output for model with edge density (250 m) and breeding-season climate variables
• 06_ed_250_tanom_nb_panom_nb.RData Model information and model output for model with edge density (250 m) and nonbreeding-season climate variables
• 07_ed_3000_tanom_br_panom_br.RData Model information and model output for model with edge density (3000 m) and breeding-season climate variables
• 08_ed_3000_tanom_nb_panom_nb.RData Model information and model output for model with edge density (3000 m) and nonbreeding-season climate variables
• prairie_bird_trends_global_tanom_panom_br22026-02-01.RData Model information and model output for model with original covariates but quadratic effects of breeding-season climate variables to assess linear vs. quadratic effects