The decline of biodiversity from anthropogenic landscape modification is among the most pressing conservation problems world-wide.  In North America, long-term population declines have elevated the recovery of the grassland avifauna to among the highest conservation priorities.  Because the vast majority of grasslands of the Great Plains are privately owned, the recovery of these ecosystems and bird populations within them depend on landscape-scale conservation strategies that integrate social, economic, and biodiversity objectives.  The Conservation Reserve Program (CRP) is a voluntary program for private agricultural producers administered by the United States Department of Agriculture that provides financial incentives to take cropland out of production and restore perennial grassland.  We investigated spatial patterns of grassland availability and restoration to inform landscape-scale conservation for a comprehensive community of grassland birds in the Great Plains.  The research objectives were to 1) determine how apparent habitat loss has affected spatial patterns of grassland bird biodiversity, 2) evaluate the effectiveness of CRP for offsetting the biodiversity declines of grassland birds and 3) develop spatially explicit predictions to estimate the biodiversity benefit of adding CRP to landscapes impacted by habitat loss.  We used the Integrated Monitoring in Bird Conservation Regions program to evaluate hypotheses for the effects of habitat loss and restoration on both the occupancy and species richness of grassland specialists within a continuum modelling framework.  We found the odds of community occupancy declined by 37% for every 1 Standard Deviation (SD) decrease in grassland availability [log_e(km²)] and increased by 20% for every 1 SD increase in CRP land cover [log_e(km²)].  There was 17% turnover in species composition between intact grasslands and CRP landscapes, suggesting grasslands restored by CRP retained considerable, but incomplete representation of biodiversity in agricultural landscapes.  Spatially explicit predictions indicated absolute conservation outcomes were greatest at high latitudes in regions with high biodiversity, whereas the relative outcomes were greater at low latitudes in highly modified landscapes.  By evaluating community-wide responses to landscape modification and CRP restoration at bioregional scales, our study fills key information gaps for developing collaborative strategies, and balancing conservation of avian biodiversity and social well-being in agricultural production landscapes of the Great Plains.

Sampling design

The study area corresponded to the Great Plains sampling frame from the Integrated Monitoring in Bird Conservation Regions (IMBCR) program (Pavlacky et al. 2017).  The sampling frame was developed by superimposing a 1 km ×1 km grid over four BCRs in the study area, stratified by state and partner defined-regions, and 1-km² sampling units were selected within each stratum using Generalized Random-tessellation Stratified (GRTS) sampling (Stevens and Olsen 2004).  We sampled Bird Conservation Regions (BCR) 11, BCR 17 and BCR 18 in eastern Colorado every year from 2010 through 2018, but except for a small number of isolated strata, sampling in the greater BCR 18 and 19 began in 2016.  We sampled the set of sampling units in successive years, but because annual sampling intensity within strata varied, some units were not sampled in successive years.  We sampled 4,140 1-km² sampling units within the study area from 2010 through 2018.  The IMBCR design sampled vegetation types in proportion to availability within strata (Pavlacky et al. 2017), and we included all data in the analysis.

The sampling protocols for avian monitoring involved a two-stage design with systematic sub-samples of 16 point count plots located 250 m apart and ≥125 m from grid cell boundaries (Pavlacky et al. 2017).  We monitored the occurrence of bird species at 44,849 point count plots on one visit per year from 2010 through 2018 using 6-min counts from one-half hour before sunrise to five hours after sunrise at each accessible point count location (Pavlacky et al. 2017).  Field technicians measured distances to each bird detection using a laser rangefinder and we truncated distances < 125 m to specify 4.9-ha, non-overlapping point count plots (Pavlacky et al. 2012).  We used a removal sampling protocol to estimate incomplete detection (MacKenzie et al. 2018), and binned the 6-min point count intervals into three, 2-min time occasions to maintain a constant detection rate in each occasion and ensure a monotonic decline in the detection frequency through time (Pavlacky et al. 2012).

Landscape covariates

We measured 3 continuous landscape composition covariates in 3 km × 3 km (9 km²) square landscape buffers surrounding the 1-km² sampling units using remotely sensed data.  We selected a 3 km × 3 km landscape buffer based on the eight 1 km² grid cells neighboring the IMBCR sampling unit (Pavlacky et al. 2017) to allow a design-based hierarchal structure for predictions.  The 9 km² landscape buffer was similar in size to a grid of point counts buffered by the mean of the best-supported landscape radii for 6 grassland bird species (10 km²) studied by Niemuth et al. (2017).  We quantified the area of grassland and shrubland vegetation in the 9-km² landscapes using the LANDFIRE Existing Vegetation Type (EVT) spatial data layer (USGS 2016) using a Geographic Information System (GIS; ArcGIS Version 10.1, Environmental Systems Research Institute, Redlands, CA, USA), and the raster and spatialEco packages in the R statistical computing environment (R Version 3.5.2, www.r-project.org).  We classified landscape composition as grassland or shrubland vegetation according to the EVT System Group Physiognomy field, except we reclassified three grassland types, two conifer-hardwood types and one hardwood type as shrubland based on review of the vegetation types.  The grassland vegetation was composed of native grassland vegetation, as well as agricultural grasslands such as pastures and hay fields.  In addition, we measured the area of CRP in the 9-km² landscapes using Common Land Unit spatial data (USDA 2014).  We included only the CRP conservation practices that involved grassland or wetland cover types, and removed practices involving tree cover and parcels containing missing practice information across all years.  For missing practice information within a particular year, including all CRP raster data from 2008 - 2010, we updated values with data from the closest available year, with the exception of CRP parcels with an expiration date > 15 years after the data year or parcels with a missing expiration date.  When possible, we replaced missing parcel data at the county or state level with data from the closest available year.  We intersected the annual CRP land-cover data and replaced the intersected land cover with CRP to arrive at seamless annual vegetation mosaics composed of grassland, shrubland and CRP land cover.  In addition to the landscape composition covariates, we used GIS to calculate latitude and longitude for the centroid of the 1-km² sampling units.

Statistical analysis

We log_e transformed the land cover covariates [log_e(1 + km²)] to allow non-linear and threshold responses to landscape features, and centered and standardized all covariates using the z-transformation (Schielzeth 2010).

Literature Cited

Pavlacky D. C. Jr., J. A. Blakesley, G. C. White, D. J. Hanni, and P. M. Lukacs. 2012. Hierarchical multi-scale occupancy estimation for monitoring wildlife populations. Journal of Wildlife Management 76:154-162.

Pavlacky, D. C., Jr., P. M. Lukacs, J. A. Blakesley, R. C. Skorkowsky, D. S. Klute, B. A. Hahn, V. J. Dreitz, T. L. George, and D. J. Hanni. 2017. A statistically rigorous sampling design to integrate avian monitoring and management within Bird Conservation Regions. PLOS ONE 12:e0185924.

Stevens, D. L., Jr., and A. R. Olsen. 2004. Spatially balanced sampling of natural resources. Journal of the American Statistical Association 99:262-278.

MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2018. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Second edition. Academic Press, London, UK.

Niemuth, N. D., M. E. Estey, S. P. Fields, B. Wangler, A. A. Bishop, P. J. Moore, R. C. Grosse, and A. J. Ryba. 2017. Developing spatial models to guide conservation of grassland birds in the U.S. Northern Great Plains. The Condor: Ornithological Applications 119:506-525.

Schielzeth, H. 2010. Simple means to improve the interpretability of regression coefficients. Methods in Ecology and Evolution 1:103-113.

United States Department of Agriculture [USDA]. 2014. Common Land Unit geospatial data.  Memorandum of understanding between the USDA and Bird Conservancy of the Rockies, signed 4 August 2014. USDA, Farm Service Agency, Economic and Policy Analysis, and Commodity Credit Corporation, Washington, D. C., USA.

United States Geological Survey [USGS]. 2016. Landfire 1.4.0: existing vegetation cover layer. United States Department of the Interior, Geological Survey, Sioux Falls, South Dakota, USA. <http://landfire.cr.usgs.gov/viewer>. Accessed 23 January 2018.

Two data files are provided:

point-count_data_dryad.csv contains presence-absence data at point-count plots for 44 grassland species in the study area. 

landscape_covariate_data_dryad.csv contains landscape covariate data.

We omitted Conservation Reserve Program (CRP) data for private lands enrolled in confidential Farm Service Agency agreements.  In addition, we omitted coordinates for primary sampling units from the Integrated Monitoring in Bird Conservation Regions program that require a data sharing agreement through Bird Conservancy of the Rockies.

A shapefile for species richness predictions is provided: 

sampling_frame_sr_pred_dryad.zip contains a shapefile for species richness predictions to 9-km² landscapes in the study area

Field "sr_all" is estimated species richness of 44 grassland specialist bird species

Field "sr_p1_all" is estimated species richness from adding 1-km² of CRP to cultivated land 

Field "sr_p1_all_" is predicted proportional increase in species richness from adding 1-km² of CRP to cultivated land

Field "sr_p1_all1" is predicted percentage increase in species richness from adding 1-km² of CRP to cultivated land