Landscapes are often managed to maximize connectivity and limit or reduce habitat fragmentation in order to promote ecosystem functions and the abundance and persistence of desired species.  However, theory and models suggest that habitat fragmentation may slow the spread of an invasive species when long distance dispersal is rare.  We tested this empirically.  In central Texas landscapes, herbaceous vegetation is often fragmented by patches of woody plants.  The invasive Eurasian perennial bunchgrass Bothriochloa ischaemum (King Ranch bluestem) has relatively limited dispersal and does not grow under woody canopies.  Accordingly, we hypothesized that B. ischaemum would occur less often in landscapes with higher degrees of fragmentation.  We recorded the occurrence of B. ischaemum in randomly-located 1 m radius circular plots in herbaceous patches at four sites on the eastern Edwards Plateau.  Aerial photographs of the sites were converted to binary maps representing woody and herbaceous habitat.  Fragmentation of the herbaceous habitat, i.e., potential B. ischaemum habitat, was measured at five scales (circles 10m to 100 m in radius centered on each plot).  Logistic regression was used to estimate the frequency of B. ischaemum presence as a function of various measures of fragmentation (e.g. number of patches, total habitat edge, contiguity index). 

Results/Conclusions

Bothriochloa ischaemum occurrence was significantly negatively related to many measures of fragmentation.  This supports our initial hypothesis about this system and the more general hypothesis that fragmentation can sometimes slow the spread of invasive species.  However, we also found that there was considerable significant site-to-site and scale-to-scale variation in the relationships between different measures of fragmentation and B. ischaemum presence.  Different aspects of fragmentation appear to affect B. ischaemum spread at different scales.  The site-to-site differences are probably related to differences in management practices among sites.