Species diversity indices
In BC, at 97% predicted community coverage, point counts and pooled ARUs obtained equivalent species richness (q=0) in both the alpine and the subalpine (Fig. 1). Pooled ARUs obtained higher richness scores than point counts in the upper montane. When species were weighted by their frequency of occurrence in either dataset (q=1), the methods performed equivalently in the subalpine, but pooled ARUs outperformed point counts in the alpine and upper montane (Fig. 1). Thus 3 of 6 comparisons in BC showed equivalent performance for the two methods and 3 indicated ARUs were superior, particularly in the upper montane.
In BC, ARU detections were more likely to intersect with our lower estimate of the entire community than point counts in the alpine and subalpine (Fig. 2). In the upper montane, predicted final species richness for both methods did not approach our lower estimate of the entire community. On average, pooled ARUs were predicted to capture 92% (range: 83-100%) of the entire community across all habitats. Point counts were predicted to capture 73% (62-79%).
In Chile, at 97% predicted community coverage, point counts obtained greater species richness (q=0) values than pooled ARUs in the subalpine and alpine (Fig. 1). In upper montane forest, the richness obtained by both methods was equivalent. When first order diversity (q=1) was assessed, point counts continued to be better than pooled ARUs in detecting species diversity, outperforming ARUs in the upper montane as well. Thus 5 out of 6 comparisons in Chile indicated that point counts outperformed ARUs; the sixth showed a bias toward point counts, but was non-significant (Fig. 1).
For both methods in Chile, the predicted asymptotes of the species accumulation curves did not approach our lower estimate of the entire community (Fig. 2). On average, pooled ARUs were predicted to capture 58% (range: 50-68%) of the entire community across all habitats. Point counts were predicted to capture 70% (57-78%).
In both regions, over the morning, dawn ARU counts detected lower or equivalent richness to counts later in the morning (q=0, Fig. 1). The only exception was in the BC alpine, where dawn counts detected more species than counts two hours after dawn (q=0, Fig. 1). Although dawn recordings were less rich, in Chile they detected two owl and one nightjar species that were not detected later in the morning (seeTable S1(B) and below).