The Saltmarsh Sparrow (Ammospiza caudacuta) is a tidal marsh bird facing rapid population decline throughout its range, largely caused by degradation and loss of breeding habitat. Thus there is a need to preserve tidal marshes in the northeastern United States, but to do so requires an understanding of the habitat features that support robust populations. Previous studies have shown Saltmarsh Sparrow abundance increases with marsh size, but in similar bird species, area sensitivity is more directly linked to edge avoidance. Whether additional landscape features affect the abundance of Saltmarsh Sparrows is unknown. We explored how the height of objects on the horizon, an index of habitat openness, affected abundance of Saltmarsh Sparrows. Our primary goal was to determine whether the angle to the highest point on the horizon (“angle to maximum horizon”) predicted abundance better than marsh area or distance to the marsh edge. We used N-mixture models to evaluate the combination of spatial factors that best predicted Saltmarsh Sparrow abundance while also accounting for survey-level variables that could influence detection probability. We found that the interaction between distance to edge and angle to maximum horizon best predicted abundance. Taller objects on the horizon were negatively correlated with bird abundance, and this effect was strongest within 50 m of the marsh edge. When we considered the predictive powers of patch area, distance to edge, and angle to maximum horizon individually, angle to maximum horizon was the best single predictor. We found the highest abundance of Saltmarsh Sparrows at point locations where the angle to maximum horizon was 0.0°, and at angles greater than 12° the predicted abundance fell below one bird per survey point. We propose that managers should prioritize marsh openness and experimentally test the effect of marsh edge manipulations when making conservation decisions for this rapidly
declining species.

This dataset includes the abundance counts of Saltmarsh Sparrows and the site and survey covariates for all surveys included in the analysis