Spatio-temporal dynamics of UK moths

• Main Author: Bell, Ewen D
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01-01-1998
• Subjects: Sample variance; Spatial analysis

The Rothamsted Insect Survey (RIS) light-trap network has been running for nearly 30 years and has amassed a vast quantity of spatial and temporal data for over 600 species of moths. The RIS has played an central role in the development of Taylor's Power law which states that for a given species sampled under similar circumstances, sample variance will be a power function of the mean. Using 13 years data from the RIS to produce TPL parameters, Taylor & Woiwod (1982) predicted that the addition of a second decade of data would not change the parameter estimates. This prediction was tested and a small but significant number (8.7%) of species were found to change parameter values. It is unlikely that this disproves Taylor's Power Law and is probably indicative of environmental change. The range of means used to estimate the TPL parameters was discovered to be of crucial importance and should be greater than 0.4. Traditional measures of spatial pattern used in ecology use only the mean and variance of samples, ignoring the spatial relationship of the counts. SADIE (Spatial Analysis by Distance IndicEs) is a novel method of spatial analysis which uses all available spatial information within the data to produce two indices, Ia and Ja. Before applying SADIE to the RIS data, the performance of SADIE given a variable sampling density was examined. Ia was found to systematically underestimate spatial pattern under variable sampling density. Ja was less affected by sample density although it is incapable of detecting pattern when more than one cluster is present. A new SADIE-like methodology was devised called Multiple Foci SADIE (MF-SADIE) which extends the Ja concept to data sets with multiple clustering and produces a new index Fa. MF-SADIE was then applied to 23 species of moth from the RIS to look for temporal pattern in their spatial distribution. A range of spatio-temporal patterns was found, from cyclical to almost uniform with good correlation between species sharing food plant types.