The effect of spatiotemporal sampling strategies and data acquisition accuracy on the characterization of dynamic ecological systems and their behaviours

This study examined the effects of spatial sampling strategies on the description of complex environmental systems in the presence of measurement error. A simulation experiment approach was used since field experiments are prohibitively expensive and analytical solutions are generally intractable for realistic ecological complexity.; A two-species patch occupancy model was constructed that included a spatially structured substrate, climatic influence on resources, and species-independent response to resources and disturbances. The model was executed using a broad range of resource patterns and species characteristics. Software instruments simultaneously collected data from the simulation world according to a prescribed set of sampling strategies and degrees of measurement accuracy. Resource inventories and species frequencies were estimated using the simulated “field” data and then compared to the “true” state of the simulation world.; The spatial sampling pattern (random or gridded) only affected estimates when the analysis method explicitly used spatial information, in which case gridded samples produced lower magnitudes of error but random samples produced the smallest, most consistent bias. For aspatial analyses, the mean difference between estimates using each sampling pattern were within 3% of each other.; Aside from spatial sampling pattern, the spatial sampling density was the most influential design factor for resource inventories. The errors in individual measurements were strongly reduced upon aggregation and had only a minimal effect on the overall inventory estimates.; A threshold effect appeared in census estimates at a spatial sampling density of 0.05, relative to grid cell size. Estimates from sparser samples (typical of most census efforts) were dominated by spatial density itself. Denser samples were predominantly influenced by classification accuracy. At the greatest spatial densities (0.15 or greater), an increase in density resulted in a larger census error if not accompanied by an increase in classification accuracy.; The experiment results suggest that field exercises should be planned to maximize mobility and coverage density, with less emphasis placed on obtaining the most accurate measurements possible. Inexpensive sensors that allow rapid measurement with moderate accuracy, combined with mobile computing and GPS units for rapid spatial data acquisition, should produce the most effective data for characterizing overall system state.