The National Vegetation Classification (NVC) is a standardized taxonomy for vegetation in the United States adopted by the Federal Geographic Data Committee. Many vegetation associations have been described by modifying existing regional or state classification models, but little attention has been given to the range of floristic variability (taxonomic resolution) allowed by the levels of the hierarchy.

We used National Park Service (NPS) data to test the comparability of western United States and eastern United States NVC association models in two ways: (1) linear regression to compare the Whittaker’s beta (β) and gamma (γ) diversity metrics calculated from over 40 individual NPS site plot data sets with the number of associations reported for these plots and (2) we conducted in situ tests of the results of field keys in areas that were both spatially constrained (< 20 hectare) and floristically homogeneous at higher description levels (NVC alliance, NatureServe Ecological System, and project map class levels) at 19 western and 19 eastern NPS sites, representing a wide diversity of vegetation, keying each of 34 sample units of 0.5 hectare in size to NVC association.

Results/Conclusions

Linear regression (method 1) showed significant differences between the regional models, with western NPS projects reporting about twice as many associations per unit of β and about 3 times as many associations per unit of γ as eastern projects. The field tests (method 2) reported over twice as many keyed associations and higher evenness (Shannon diversity) among associations at western sites, even though the spatial, environmental and floristic gradients were similarly constrained. We conclude that the difference is likely either an artifact of more small-scale processes being recognized by the NVC in western models or an artifact of greater rates of key failure at western sites because of taxonomically finer models being employed in the West.

We conclude that regional model bias creates cost, logistical, and performance difficulties for NVC users in both field identification and mapping of vegetation, as it has within the NPS national Vegetation Inventory. Users may expect one regional model to perform as does the other model for equivalent levels of the hierarchy. Such a bias also would be expected to create problems in consistency in using the NVC for national (and possibly international) biodiversity assessment. We hypothesize that the effects of broad scale climate patterns on stand and plot scale diversity patterns have been not well appreciated by the NVC effort.