ERRORS ASSOCIATED WITH OTTER LUTRA-LUTRA FECAL ANALYSIS .1. ASSESSINGGENERAL DIET FROM SPRAINTS

Although frequency of occurrence (either as percentage or relative fre quency) is the most common method of expressing the content of otter L utra hutra faeces (spraints), the accuracy of the method, and the effe cts of varying sampling procedures (e.g. inter-collection interval) an d sample sizes, have not been quantified. The validity of the techniqu e was assessed in the present study by feeding trials involving captiv e, tame otters and computer simulation of various spraint sub-sampling regimes. Four animals were fed known quantities (numbers and biomass) of a total of nine fish species, two bird species and one mammal over a 28-day period. Most prey remains were passed in spraints within 24 h, although perch Perca fluviatilis scales appeared up to 10 d after c onsumption Remains from single meals of perch were recorded in 60 subs equent spraints from two otters, and remains of individual eels Anguil la anguilla were recorded in up to 11 spraints. Some single spraints c ontained the remains of up to seven individual salmonids, Salmo spp. M innows Phoxinus phoxinus placed within the body cavities of larger rai nbow trout Oncorhynchus mykiss were easily identified in spraints, as were the remains of Dytiscus spp. beetles which were not included in t rial meals. The latter confirms that otters actively consumed large fr ee-swimming insects. Spraint analysis accurately determined the rank o rder of prey groups for individual otters and for all four combined. H owever, few of the true proportions consumed fell within the 95% confi dence limits calculated from spraints. Over the month-long trial, the overall picture of otter diet was altered little by increased inter-sa mpling period for spraints. But as samples were reduced, coefficients of variation for the mean estimates of each prey group increased and w ere often too large for estimates to be meaningful. It is not possible to quantify otter diet accurately by frequency of occurrence methods, and the results of previous studies attempting to quantify the amount of a specific prey item consumed by otters using this method must be treated with caution. Diet could be estimated more accurately from spr aint analysis by concentrating on the main prey species and using keyb ones, which are resistant to digestion, to determine relative size-fre quency distributions.