Determination of optimal riparian forest buffer dimensions for stream biota–landscape association models using multimetric and multivariate responses

The dimensions of riparian buffers selected for stream biota–landscape association models determine correlation strength and subsequent model interpretation. Efforts have been made to optimize buffer dimensions incorporated into models, but none has explicitly determined a single optimum based on both longitudinal and lateral buffer dimensions. We applied partial correlation and multivariate linear regression on functional fish community response attributes and the index of biotic integrity using stream samples (N = 107) from the Eastern Corn Belt Plain Ecoregion of Indiana, USA. Land-cover data in digital format were processed in geographic information systems for an area covering 300 m on either side of selected streams and within 2000 m longitudinally. The optimal buffer dimension for the study area was 30 m laterally and 600 m longitudinally, with a partial correlation of 0.29 (P = 0.002), and there was agreement in the partial correlation and multiple regression models. The longitudinal dimension was more conclusively determined, but the lateral dimension was optimum only with respect to the resolution of the land-use data used. Based on these results, we propose the use of this approach to optimize the riparian buffer parameter in landscape models.