Results
We autonomously searched 75,858 hours of ambient sound for potential gobbles. The CNN identified 324,236 potential gobbles of which 194,655 (60%) were true gobbles (Table 1). Mean gobbles per ARU from 1 March - 31 May was highest on SRS (937 ± 326, mean ± SD), 11% less on CWMA (838 ± 404), 52% less on the Webb WMA complex (443 ± 120), 46% less on BFG (507 ± 38), and 61% less on CCWMA (369 ± 130, Table 1).
The state-space model accurately predicted gobbling activity compared to our observed gobbling activity (Figure 2) and R-values indicated model convergence (Table 2). Results from the state space model indicated the occurrence of rain most impacted (negatively) gobbling activity (Table 2). Where the mean expected number of daily gobbles would be 21 (CrI = 15, 30) without rain, compared to 12 (CrI = 7, 22) if rain occurred. Conversely, an increase of barometric pressure from one day to the next was positively associated with gobbling activity (Figure 3, Table 2). We found gobbling activity was negatively influenced by increased temperatures (Figure 4, Table 2), and by greater wind speed with the largest effect occurring when wind speeds exceeded 10 kilometers per hour (Figure 5, Table 2). Humidity had no effect on the average predicted rate of change in gobbles across all study sites and years (Table 2).