Fig. 3: Incident rate ratios for predicted plant abundance.
From: Seed dispersal by Martu peoples promotes the distribution of native plants in arid Australia
Dot and whisker plots on the left show the strength of each predictor variable as an incident rate ratio with 95% CI for the best abundance model for each species (see values in Table 6). Red values are negative (below 1), blue are positive (above 1). Coefficients have been standardized to 2 sd, so that an IRR of 0.5 means that a 2 sd increase in the value of the predictor decreases stem counts by two times; an IRR of 2 increases stem counts by two times. Confidence intervals for IRRs that cross the line at 1 are non-significant. IRRs for interaction terms (e.g., Season × Fire Freq) are to be interpreted with caution. Since this analysis predicts stem counts only when plants are present in the plot, sample size (plots present) for each species varies: SDIV(n) = 109, SCENT(n) = 709, ERAG(n) = 617, SCAV(n) = 177. The inset line graph (right) shows the predicted abundance (with 95% CI) and interactions for the strongest predictor.
