Fig 1.
Possible mechanisms leading to DRI violation.
Left column shows development rates of three hypothetical life stages, early (blue dotted line), intermediate (orange dashed line) and late (green line), and the overall rate of development (thick grey); grey rectangles highlight the thermal range (T1, T2) in which the overall rate of development increases almost linearly with temperature; time units are omitted. Right column shows the corresponding relative duration of each stage in the (T1, T2) range. (a) DRI; (b) violation of DRI due to shifted, stage-dependent temperature optima; (c) violation of DRI due to progressively limiting, temperature-dependent metabolic scope for growth in individual stages; (d) violation of DRI due to limiting metabolic scope for growth in the intermediate stage. Even more severe DRI violation outside the (T1, T2) range in panels (b)–(d) is omitted for clarity. See text for further details.
Table 1.
Summary of ΔAICc values for Dirichlet regression models for individual species of aquatic and semiaquatic insects.
Table 2.
Summary of ΔAICc values for Dirichlet regression models for individual species of terrestrial insects.
Fig 2.
Examples of results of Dirichlet regression for (a) Acilius and (b) Loxostege.
Acilius: best fitting model (f+tF, grey solid line) compared to DRI (model const, black dashed lines); Loxostege: dataset with average clutch data, 21–27°C and different population origin (B = Buryatia, K = Krasnodar, H = Hebei), best fitting DRI model (f, black dashed lines) compared to model f+t (black solid lines). Box and whisker plots of raw data: horizontal line = median, box = first to third quartiles, line = data within 1.5 times the interquartile range; dots = outliers. Non-feeding stages labelled with asterisk.
Fig 3.
DRI violation in the nine insect species studied.
(a) Average stage-specific DRI violation from Dirichlet regression. Stages on a relative scale (0 = egg, 1 = last pre-adult stage). Curve ± 95% confidence interval = best model describing the dependence of DRI violation VD on the relative developmental stage S (see main text for details). (b) Comparison of average stage-specific DRI violation from Dirichlet regression and standard ANCOVA analysis as described in Jarošík et al. (2002). Line ± 95% confidence interval = regression of VA on VD (see main text for details). Symbols represent values averaged across all levels of a factor and all datasets for each species included in Tables 1 and 2; small amount of horizontal jitter added to all data. Green fill = feeding stages of aquatic species, orange fill = non-feeding stages of aquatic species, grey fill = feeding stages of terrestrial species, no fill = non-feeding stages of terrestrial species; circles = Coleoptera, squares = Cloeon, diamond = Heteroptera, triangles = Loxostege.