Abstract
Most ecosystems lack long-term extensive data that can support analysis of what determines their inherent biodiversity patterns. The spectral variation hypothesis (SVH) suggests that between-plot differences in remotely sensed signal (spectral variation) is a proxy for environmental heterogeneity and therefore can act as an indicator of plant beta-diversity. Spatial process also controls plant betadiversity patterns. It is unclear how much of variation in beta-diversity that is explained by spectral variation can also be explained by spatial distances and vice versa. We evaluate the potential of spectral variation of Landsat-TM data and spatial autocorrelation in predicting plant beta-diversity in a fragmented landscape. We use Mantel correlograms to investigate the correlation between spectral distances and beta-diversity and the presence of spatial dependency in spectral distances and betadiversity. We fit a partial least squares regression (PLSR) to predict the variance in the plant betadiversity explained by spectral and spatial distances and partition the variances due to pure spectral, spatial dependency in spectral distances and purely spatial autocorrelation. Mantel correlograms
revealed an exponentially decaying correlation between beta-diversity and spectral distances that
was significant between plots with short spectral distances and at the farthest lag distances. Therefore dissimilarity in beta-diversity was lower at shorter spectral distances but increased with
increasing spectral distances. We detected significant spatial autocorrelation in beta-diversity and spectral distances suggesting that the variation in beta-diversity was influenced by spatial structure directly and indirectly through spatially dependent environment (spectral distances). The PLSR model
explained 37% of total variance in beta-diversity. Both environmental and spatial processes have significant control on beta-diversity patterns in this landscape though the former explained cumulative higher variance (66%) than spatial autocorrelation (34%). This suggests that conservation initiatives should aim at enhancing habitat diversity
revealed an exponentially decaying correlation between beta-diversity and spectral distances that
was significant between plots with short spectral distances and at the farthest lag distances. Therefore dissimilarity in beta-diversity was lower at shorter spectral distances but increased with
increasing spectral distances. We detected significant spatial autocorrelation in beta-diversity and spectral distances suggesting that the variation in beta-diversity was influenced by spatial structure directly and indirectly through spatially dependent environment (spectral distances). The PLSR model
explained 37% of total variance in beta-diversity. Both environmental and spatial processes have significant control on beta-diversity patterns in this landscape though the former explained cumulative higher variance (66%) than spatial autocorrelation (34%). This suggests that conservation initiatives should aim at enhancing habitat diversity
Original language | English |
---|---|
Pages | s1-s16 |
Publication status | Published - 11 Feb 2014 |
Event | Netherlands Annual Ecology Meeting 2014 - Congrescentrum De Werelt, Lunteren, Netherlands Duration: 11 Feb 2014 → 12 Feb 2014 |
Conference
Conference | Netherlands Annual Ecology Meeting 2014 |
---|---|
Abbreviated title | NAEM 2014 |
Country/Territory | Netherlands |
City | Lunteren |
Period | 11/02/14 → 12/02/14 |
Keywords
- METIS-302754