Family-level detection probabilities by method
In BC, models supported an effect of methodology on detection for only 6
of the 14 families examined. These were: finches (Fringillidae),
sparrows (Passerellidae), warblers (Parulidae), kinglets (Regulidae),
thrushes (Turdidae) and corvids (Corvidae; Figs S1-S3). All six families
were better detected by ARUs. Kinglets, warblers and thrushes showed an
interaction between methodology and hours after sunrise: detection
probability declined over the morning for point counts but remained
consistently high for ARUs (Fig. S2). Warblers, thrushes and corvids
showed an interaction between methodology and canopy cover: detection
probability was more variable over the range of canopy cover for point
counts than for ARUs (Fig. S3).
In Chile, detection models for all 11 families examined supported a
methodology effect, with a higher detection probability in point counts
than in ARUs (Figs. S4-S6). Of these, six families showed an interaction
between methodology and date: ARU detection probability for swallows
(Hirundinidae), hummingbirds (Trochilidae), woodpeckers (Picidae), wrens
(Troglodytidae), and parakeets (Psittacidaewas either lower early in the
monitoring period or exhibited a mid-season dip. The detectability of
ovenbirds (Furnariidae), showed a mid-season dip in point counts but not
ARUs (Fig. S4). Ovenbirds additionally had lower detection probability
for ARUs under conditions of high canopy cover. Swallows were better
detected by point counts in the mid-morning; wrens were more poorly
detected by ARUs in the early morning (Fig. S5).