Effet de la trophie et de la taille du substrat sur le spectre de taille des invertébrés benthiques en ruisseaux.

Abstract

Five different substrates (sand, fine gravel, coarse gravel, cobble and boulder) in nine streams forming a trophic gradient in the Ottawa Valley have been sampled in June 1992 to determine the effect of trophic level and substrate size on benthic communities. Total biomass of invertebrates (B, mg dry mass m$\sp{-2})$ and periphyton are strongly related to substrate particle size $(\Phi$, $-$Log$\sb2$diameter in mm) and trophy, represented by total phosphorus concentration (P, $\mu$g L$\sp{-1}).$ The model for periphyton biomass (Log$\sb $Peri = $-$0.08 $-$ 1.68Log$\sb $P + 1.08Log$\sb $P$\sp2\ -$0.13$\Phi\sp2)$ explains 67% of the observed variability and the model for invertebrate biomass (Log$\sb $B = 2.69 + 0.55Log$\sb $P + 0.3Log$\sb $Peri $-\ 0.01\Phi\sp2\ -$ 0.02Z) explains 68% of the observed variability (Peri = periphyton biomass in mgChla m$\sp{-2}$ and Z = depth in cm). The highest invertebrate biomass is observed on intermediate-size substrate (gravel) in phosphorus-rich streams. Periphyton biomass peaks on coarse substrates (boulder and cobbles) in streams where total phosphorus concentration is high $(>$50$\mu$gL$\sp{-1}).$ Individual body mass (W$\sb{\rm moy})$ is not affected by substrate size but is positively correlated with total phosphorus concentration (Log$\sb $W$\sb{\rm moy}$ = 1.67 $-$ 1.71Log$\sb $P + 0.96Log$\sb $P$\sp2,$ R$\sp2$ = 0.44). The effect of phosphorus and substrate on total biomass and mean individual body mass, are reflected in the size distribution of the invertebrates. Although the shape of the distribution is constant (unimodal), abundance of medium and large organisms increases in phosphorus-rich streams on intermediate-size substrates. A polynomial model including phosphorus concentration and substrate particle size is presented to quantify those variations of the size spectra. The predictions of this model are close to the observed values (R$\sp2$ = 0.89, residual mean square = 0.075) and this model could be very useful for a fast estimation of invertebrate size spectra.