Summary1 Much ecological theory is based on the characterization of ecological habits of species as 'generalist' or 'specialist', but standard measures for placing species along a generalist-specialist gradient do not exist. 2 We introduce a method for quantifying habitat specialization (i.e. relative niche widths) using species co-occurrence data. Generalists should co-occur with many species, whereas specialists should co-occur with relatively few species, given equal plot occurrences. We quantify this concept using a generalist-specialist metric ( θ ) derived from a beta diversity statistic. 3 We evaluate the ability of our generalist-specialist metric to correctly rank species according to simulated (known) niche widths. Our technique is generally robust to a wide variety of niche distribution structures and sampling designs, but surveys strongly biased toward certain habitats can undermine the ability of θ to accurately describe niche widths for underrepresented species. 4 We apply our technique to three spatially nested surveys of the large woody flora (> 1 cm d.b.h.) of the south-eastern USA. For each dataset we rank the generalistspecialist tendencies of all species of non-trivial occurrences, including 113 species across the Southeast, 71 species of southern Appalachian forests, and 44 species of the 6800-ha Joyce Kilmer-Slickrock Wilderness Area (NC and TN, USA). 5 Rankings of species' θ -values were generally consistent among datasets of different spatial extent. Generalist species (e.g. Ilex opaca , Ulmus rubra , Morus rubra , Prunus serotina , Acer rubrum ) were often those with large geographical ranges, particularly for θ -values from the largest dataset, and overall were more likely to be bird-dispersed, deciduous, and shade tolerant. South-eastern specialist species (e.g. Taxodium spp., Abies fraseri , Quercus laevis , Pinus pungens , Pinus palustris ) were those associated with stressful or unusual conditions, such as a long duration of flooding, high fire frequency, or extreme cold or dry climates. 6 Our study demonstrates that increasingly available, large-survey datasets can contribute niche-related species information in the absence of detailed environmental or habitat measurements. Applications include new assessments of relationships between species traits, ecological and environmental tolerances, and species packing in different assemblages.

The reintroduction of pre-European fire regimes has allowed the entry of many invasive plant species into fire-dependant ecosystems of North America. However, the environmental factors that favor the post-fire establishment of these species across complex landscapes are not well understood and the initial establishment of invasive species does not necessarily result in long-term persistence. To evaluate the post-fire establishment and persistence of disturbance-dependent invasive plants, we studied the invasion of Paulownia tomentosa (princess tree, an early-successional species introduced from Asia) across three burns in the southern Appalachian Mountains. Based upon classification tree analysis, the presence/absence of P. tomentosa 2 years after burning was most strongly related to the cover of residual vegetation, topographic shading, and moisture availability. Spatial application of classification tree models to repeated survey data showed that P. tomentosa established across a wide range of microsites 2 years after burning. However, predicted habitat for P. tomentosa decreased by 63% 4 years after fire and by 73% 6 years after fire. Following its initial widespread establishment, P. tomentosa only persisted on xeric and exposed topographic positions that experienced high intensity burning. However, the sites where it persisted include rare community types that contain two endangered plant species that depend upon fire for successful reproduction. The control of P. tomentosa on these ecologically important sites may require special attention from land managers.

No abstract

No abstract