A continuous study of the total drift of freshwater shrimps, Gammarus pulex, in a small stony stream in the English Lake District

1. The objective was to determine the major factors affecting the downstream dispersal (drift) of freshwater shrimps, Gammarus pulex. Sample replication and frequency are major problems in the quantification of drift. For the first time, these problems were avoided by sampling the whole stream continuously so that all the shrimps drifting downstream at the sampling point were caught in a net emptied at dusk and dawn in 1966, and every 3 days in 1967. 2. There was no consistent seasonal pattern in drift rates, but a high proportion of annual drift was taken in only a few samples. There was a nocturnal diel pattern of drift with peaks soon after dusk and just before dawn. A power function described the significant (P < 0.001) relationship between drift and flow, and was used to neutralise the dominant effects of flow by standardising total drift over 24 h, nocturnal drift and diurnal drift (drift per 50 m³). These were all significantly (P < 0.001) related to benthos density, but not to date, temperature, or length of the night or day. 3. The relationship between drift values and the independent variables, flow and benthos density, was well described (P < 0.001) by a multiple‐regression model. Adding temperature, date, and/or the length of the night or day did not improve model fit. Variations in flow and benthos density explained 94% of the variation in total drift over 24 h, 97% of the variation in nocturnal drift, but only 44% of the variation in diurnal drift. A power function described (P < 0.001) the relationship between total drift and the volume of water sampled over 3‐day periods in 1967. Flow explained 95% of this drift variation; it was unnecessary to add another variable such as benthos density. 4. The significance of this study is that it avoided the problems associated with the quantification of drift samples. Therefore, the conclusions are more robust than those of many previous studies. A high proportion of the annual drift losses would have been undetected by intermittent sampling. Temperature, season, night or day length had no significant effect on drift densities, and the relationship between drift and benthos densities was proportional, not density dependent. The nocturnal increase in drift could not be interpreted as an antipredator behaviour. The dominance of flow and benthos density was apparent but the quantitative relationships posed further questions, especially those related to drift distances at different velocities.