Table 2.
Results of negative binomial generalized linear models predicting the number of pest control studies in 2015/16 as a function of insecticide use intensity and world production metrics per crop. Significant terms are in bold. n = 12 (model 1, ‘other vegetables’ excluded; see Methods) and 13 (model 2) separate crop types, respectively. The combined effects of insecticide use, world yield and world production (model 1; Figure 4) and the converse effects of world harvested area only (model 2; Figure S3) are shown. Terms were examined in separate models due to correlation between harvested area and other metrics (Methods and Figure S2).
| Deviance | Residual Deviance | Estimate | Confidence Interval | p-Value | |
|---|---|---|---|---|---|
| Model 1 | 49.7 | ||||
| (Intercept) | 3.31 | 2.64–3.97 | <0.001 | ||
| Insecticide use (kg/ha) | 13.9 | 35.75 | −0.74 | −1.4–0.1 | 0.022 |
| World yield 2014 (t/ha) | 20.6 | 15.15 | −0.03 | −0.05–0.02 | <0.001 |
| World production 2014 (kt) | 0 | 15.15 | −1 × 10−9 | −0.71 × 10−6−0.73 × 10−6 | 0.986 |
| Model 2 | 26.86 | ||||
| (Intercept) | 1.5 | 0.88–2.14 | <0.001 | ||
| log(World harvested area 2014 (million ha)) | 11.1 | 15.8 | 0.28 | 0.11–0.45 | 0.002 |