Matching native plant counts reduced the measured winter richness loss on 15 Portal plots in 1989–2005

We tested whether lower native winter annual plant richness at Portal, Arizona, remained when samples had equal native plant counts. We used PortalData 6.117.0, 15 fixed plots, and two sets of six censuses from 1989–1995 and 1998–2005. The panel contained 258,510 native individuals. For each same-plot early-year and late-year pair, we used random thinning to match the lower native count. We kept quadrat labels and calculated expected richness exactly. On five rodent removal plots, raw richness changed by −8.47 taxa per 4 square metres sampled. The matched change was −1.74 taxa (95% plot-and-year bootstrap interval −4.73 to +1.58). The count-related difference was 6.73 taxa (2.64 to 10.37), or 79.5% of the raw loss in the point estimates. On ten rodent-access plots, raw and matched changes were −4.37 and +1.10 taxa. Count matching reduced the measured loss at all four sampled spatial scales. Taxon, plot, and year selection checks gave the same main pattern. The residual removal loss had an interval below zero when years were fixed and only plots were resampled, so uncertainty depends on the target scope. This is a reanalysis of a published period comparison. It does not identify an invasion cause, show no change in the native species mix, or make the fall in native plant density unimportant. The package contains source tables, hashes, code, results, figures, and offline repeat instructions.

biodiversity · plant richness · rarefaction · biological invasion · Erodium cicutarium · Portal Project · density · spatial scale