A systematic re-sampling approach to assess the probability of detecting otters Lutra lutra using spraint surveys on small lowland rivers

Assessing and monitoring populations of elusive species frequently rely on the identification of indirect signs such as faeces. The absence of signs does not necessarily denote the absence of a species, thus, the ability to determine the presence/absence is susceptible to false negative results. The probability of detection is central to the interpretation and utility of data from field sign surveys. A low probability of detection may introduce considerable error into distribution patterns, resulting in inaccurate ecological conclusions.We used a systematic resampling approach, based on sequential spatial replication of spraint surveys, to investigate the probability of detecting Eurasian otters (Lutra lutra L.) with different survey designs. This included the standard otter transect survey methodology, which is widely used in conservation and scientific studies. In particular, we focus on the impact of applying broad scale population assessment techniques at smaller spatial scales. Fortnightly catchment-level otter surveys were undertaken on four lowland rivers in South Wales, over a period of two years. GIS was used to construct binary vectors for each survey, denoting the presence (1) or absence (0) of otters at each 50 m section of river. Vectors from all study rivers were pooled and resampled to test the different survey designs. The mean probability of detecting otters based on the standard protocol of a single 600 m transect survey was very low (0.26 ± 0.01 SE). The best way of obtaining a detection probability of 0.8 was to undertake three repeat surveys at two separate sites, using a transect of 800–1000 m.We demonstrate how sequentially collected spatial data can be analysed to determine the reliability of field sign surveys. Increasing the number of visits and study sites was a more efficient means of improving detection power than increasing transect length alone. The study emphasises the importance of determining detection probabilities and designing field sign surveys according to study scale and objectives. Our findings question the value of survey designs that aim to provide an instantaneous assessment of species presence/absence.

► We systematically resampled field data to determine the power of field surveys.
► The standard otter survey design has a very low detection power of 0.26 (± 0.01 SE).
► Distribution at small scales is significantly underestimated by the standard design.
► Repeat visits at multiple sites were necessary to achieve sufficient detection power.
► Surveys designed for broad scale objectives should not be used at smaller scales.