Size at the onset of maturity (SOM) revealed in length–weight relationships of brackish amphipods and isopods: An information theory approach

In amphipods and other small-sized crustaceans, allometric relationships are conventionally analysed by fitting the standard model Y = a·Xb (X and Y are, e.g., body length and weight, respectively) whose scaling exponent b is assumed to be constant. However, breakpoints in allometric relationships have long been documented in large-sized crustaceans, ultimately determined by ontogenetic, abrupt variations in the value of b. Here, the existence of breakpoints in length–weight relationships was investigated in four amphipod (i.e., Gammarus aequicauda, Gammarus insensibilis, Microdeutopus gryllotalpa, and Dexamine spinosa) and three isopod species (i.e., Lekanesphaera hookeri, Sphaeroma serratum, and Cymodoce truncata) from three Mediterranean lagoons. The power of two candidate linear models fitted to log10-transformed data – a simple model assuming a constant exponent b and a segmented model assuming b to vary after a breakpoint - was compared using a parsimonious selection strategy based on the Akaike information criterion. The segmented model with a breakpoint provided the most accurate fitting of length–weight data in the majority of the species analysed; non-conclusive results were obtained only for D. spinosa and C. truncata, of which a limited number of specimens was examined. Model parameters were consistent for amphipod and isopod species collected across the three different habitats; the generality of the results was further supported by a literature search confirming that the identified breakpoints corresponded with ontogenetic discontinuities related with sexual maturation in all the species investigated. In this study, segmented regression models were revealed to provide a statistically accurate and biologically meaningful description of length–weight relationships of common amphipod and isopod species. The methodological limitations of the approach are considered, while the practical implications for secondary production estimates are discussed.