Hereafter, our use of language such as population ‘declines’ or species ‘responses’ refers to inferred changes resulting from ant invasion, and is shorthand for differences in measured densities between invaded and uninvaded

plots. At each site, we installed eight 5 by 5 m sampling plots into randomly selected habitat patches that contained all of the dominant shrub or tree species at the site (defined as the two to four most common shrub or tree species, see below), at a distance of 100–175 m behind the ant population boundaries. The longer distances were used at sites where invasion rates were faster; based on observed rates of spread, invaded plots were estimated to have been invaded for at least 4 years at all sites. These eight invaded plots were then AZD5153 supplier matched with eight uninvaded plots in randomly selected habitat patches located 120–175 m in front of the expanding Rabusertib mw ant population boundaries, and were placed such that percent covers of the dominant plant species in the uninvaded plots deviated from those in matched invaded plots by less than 15%. Methods for installing plots are elaborated in Krushelnycky and Gillespie (2008). To quantify arthropod densities in each

plot we employed three standardized sampling techniques, chosen to target the majority of species likely to interact with ants in these habitat types. First, we placed three pitfall traps (300 ml plastic cups half-filled with a

50:50 propylene glycol:water Orotidine 5′-phosphate decarboxylase solution), separated by at least 2 m, in each plot, with one randomly chosen trap baited around the rim with blended fish and the other two unbaited. These traps were left open for 2 weeks. Second, in each plot we collected leaf litter from three different areas, mixed it together and removed 1 liter, and placed this in a Berlese funnel for 24 h. Third, in each plot we beat each of the dominant shrub or small tree species at the site. These plant species were: Ahumoa—Dubautia linearis, Dodonea viscosa; Pohakuloa—Myoporum sandwicensis, Sophora chrysophylla, Chenopodium oahuensis; Huluhulu—Leptecophylla tameiameiae, Vaccinium reticulatum, Coprosma ernodiodes; Puu O Ili—Dubautia menziesii, L. tameiameiae, V. reticulatum, S. chrysophylla; Kalahaku—D. menziesii, S. tameiameiae. Each plant species received five beats, spread among multiple individual plants in the plot if possible, over a 1 m2 beating sheet. Sampling occurred from August to September, 2002 at Ahumoa and Pohakuloa; June, 2003 at Kalahaku; July, 2003 at Puu O Ili; and August, 2003 at Huluhulu. Dataset We sorted all vegetation beating samples collected, but due to time constraints only sorted samples from five of the eight matched pairs of plots at each site for the pitfall and litter sampling techniques.