Table 1.
Discriminant analysis of principal components for the presence of specific prey types using the morphological data
| Prey type | DAPC | GLM for prey type presence (22 taxa) | Best-fitting GLM for prey type selectivity (32) (7 taxa) | |||
| Discrimination (%) | Top quartile variable contributions | Sign | R2 | Sign | R2 | |
| Copepods | 95.4 | Total nematocyst volume | − | 67.8 | −* | 97.9 |
| Tentacle width | − | + | ||||
| Haploneme elongation | − | + | ||||
| Haploneme SA/V ratio | + | − | ||||
| Haploneme row number | + | + | ||||
| Cnidoband length | − | + | ||||
| Cnidoband width | − | − | ||||
| Cnidoband free length | + | + | ||||
| Fish | 68.1 | Total haploneme volume | − | 45.8 | + | 96.0 |
| Heteroneme volume | + | − | ||||
| Total nematocyst volume | − | + | ||||
| Total heteroneme volume | − | − | ||||
| Cnidoband length | − | − | ||||
| Cnidoband free length | + | + | ||||
| Involucrum length | − | − | ||||
| Pedicle width | + | + | ||||
| Large crustaceans | 81.8 | Involucrum length | +* | 73.2 | + | 98.7 |
| Total heteroneme volume | − | − | ||||
| Elastic strand width | − | +* | ||||
| Rhopaloneme length | + | + | ||||
| Heteroneme volume | + | − | ||||
| Haploneme elongation | − | + | ||||
| Desmoneme length | − | − | ||||
| Tentacle width | + | + | ||||
Top quartile variable (character) contributions to the linear discriminants are ordered from highest to lowest. Logistic regressions and GLMs were fitted to predict prey type presence and selectivity, respectively. The sign of the slope of each predictor is reported, marked with an asterisk if significant (P < 0.05), and highlighted in bold if it differs between prey presence in diet and prey selectivity. Pseudo-R2 (%) approximates the percent variance explained by the model.