Data from: A spatial kernel density method to estimate diet composition of fish
Data files
May 02, 2019 version files 5.54 MB
-
Bhattacharyya Distance.R
295 B
-
Creating S Predict Grids.R
2.02 KB
-
Cross Validation using Grid Search.R
8.34 KB
-
Cross Validation using Optimization.R
8.59 KB
-
Data Simulation.R
1.87 KB
-
Estimating Diet using NonSpatial Estimator.R
1.35 KB
-
Estimating Diet using Spatial Estimator (SDKE).R
2.25 KB
-
Monte Carlo Simulation Study and Sample Size Comparison.R
40.39 KB
-
Pam_Lat_Longs.csv
15.45 KB
-
Pamlico_Waters_Extended_wgs.zip
5.35 MB
-
s_predict.csv
95.81 KB
-
spot_trawl.csv
9.25 KB
Abstract
We present a novel spatially-explicit kernel density approach to estimate the proportional contribution of a prey to a predator’s diet by weight. First, we compare the spatial estimator to a traditional cluster-based approach using a Monte Carlo simulation study. Next we compare the diet composition of three predators from Pamlico Sound, North Carolina to evaluate how ignoring spatial correlation affected diet estimates. The spatial estimator had lower MSE values compared to the traditional cluster-based estimator for all Monte Carlo simulations. Incorporating spatial correlation when estimating the predator’s diet resulted in a consistent increase in precision across multiple levels of spatial correlation. Bias was often similar between the two estimators but when it differed it mostly favored the spatial estimator. The two estimators produced different estimates of proportional contribution of prey to the diets of the three field-collected predator species, especially when spatial correlation was strong and prey were consumed in patchy areas. Our simulation and empirical data provide strong evidence food habits data should be modeled using spatial approaches and not treated as spatially-independent.