Matskási István (szerk.): A Magyar Természettudományi Múzeum évkönyve 86. (Budapest 1994)

Kemenes, I. ; Demeter, A.: Uni- and multivariate analyses of the effects of environmental factors on the occurrence of otters (Lutra lutra) in Hungary

kiri2 square of the National Grid of Hungary. Representative survey points at all the major streams and big lakes and at most of the smaller streams and ponds were chosen. Survey points were frequently located at bridges, piers etc. that provided good access to the water. In areas with few roads or tracks, other features were chosen, such as lake inflows/outflows and stream junctions, to enable future surveyors to locate the same sites with precision. Only spraints and footprints were accepted as proof of the otters' presence as these are clearly distinguishable from the similar signs of any other species. At each survey point a maximum of 600 meters of watercourse was checked. Within this distance, the search was terminated as soon as the first spraint or footprint was found and the area was considered as inhabited by otters and a score of 1 was assigned to the site. On the other hand, if no signs of otter were found within 600 meters, the area was considered as uninhabited by otters, and accordingly, a score of 0 was assigned to the site. 2. The environmental parameters used in the study At each study site (n=369) three scalable environmental parameters were recorded and scaled as shown in Table 1. The presence (1) or absence (0) of the following non-scalable parameters was also recorded: - any kind of disturbance of human or natural origin (0 rated only if none of the factors detailed below was present) - agricultural use of the land surrounding the study area - activities related to animal husbandry in the immediate vicinity of the study area (grazing, etc.) - vicinity of towns - other factors (e.g. visible water pollution, blocked or temporarily dry waterway, etc.) Note that whereas there was only one condition, namely the lack of any disturbance, when a score of 0 disturbance was recorded at a site, there were many combinations of factors (each of them either present (1) or absent (0) at a particular site) which resulted in a general disturbance score of 1 (one or more factors present). Table 1. Scalable environmental variables recorded during the survey of the distribution of otters in Hungary Parameter Water depth Density of the bank vegetation Steepness of the water bank Scale 1 = 0 - 30 cm 2 = 30 - 100 cm 3 = over 100 cm 0 = no vegetation 1 = sparse vegetation 2 = dense vegetation with thin patches 3 = dense vegetation covering large areas 0 = plain (0°) gently sloping (<45°) 2 = steep (>45°) 3. Correlation/association analyses Cramiir's statistics were calculated for measuring the association between the binary variable for otter presence and the habitat variables, and Spearman's rank correlation coefficients were calculated among the scaled ordinal habitat variables (DANIEL 1990). 4. Multiple logistic regression analysis A variety of multivariate statistical techniques can be used to predict a dependent variable from a set of independent variables. However, the most commonly used multivariate techniques pose difficulties when the dependent variable can have only two values - like in our case, the presence (1) or absence (0) of otters at a particular site. In this case we cannot assume that the distribution of errors is normal, therefore the criteria necessary for hypothesis testing in regression analysis, the most commonly used multivariate method, are not fulfilled. This extreme case also violates the assumptions necessary for other well-known multivariate techniques, such as linear discriminant analysis.