Table 4. The effects of habitat (a) and patry (b) on call characters, NRR, ND and DF.
| d.f. | Mean square | F | P | ||
| (a) Habitat (n = 324) | |||||
| NRR | Between group | 1 | 0.000 | 0 | 1 |
| Within group | 322 | 0.241 | |||
| ND | Between group | 1 | 0.000 | 0 | 1 |
| Within group | 322 | 0.000 | |||
| DF | Between group | 1 | 0.000 | 0 | 1 |
| Within group | 322 | 16075 | |||
| (b) Patry (n = 127) | |||||
| NRR | Between group | 1 | 0.000 | 0 | 1 |
| Within group | 125 | 0.251 | |||
| ND | Between group | 1 | 0.000 | 0 | 1 |
| Within group | 125 | 0.000 | |||
| DF | Between group | 1 | 0.000 | 0 | 1 |
| Within group | 125 | 15951 | |||
We performed a MANCOVA with predictor variables including barrier, locality, SVL, temperature and humidity (see Methods). One-way ANOVA was conducted on residuals of MANCOVA for three call characters. Habitat that was either rice paddy or non-rice paddy was a predictor variable in one-way ANOVA using all samples. Patry was either allopatry or sympatry. Localities only in the western group were included to test the effect of patry.