PARENT SESSION
Oral Session 13: Rare, Threatened, and Endangered Species I: Management I.
Presiding: H Balbach
Monday, August 2, 8:00 AM to 11:30 AM, Meeting Room D 135.
Incorporating spatial autocorrelation into experimental analysis for large conservation areas: Rethinking "pseudoreplication".
Pollock, Jacob*,1, Doak, Dan1, 1 University of California, Santa Cruz, CA, USA
ABSTRACT- For large conservation areas, limited money, and natural heterogeneity often make the replication demanded by standard agronomic-type experimental designs difficult or impossible to achieve This is true not only for observational experiments but for manipulative ones as well. Using analytical techniques for spatial autocorrelation (SAC) and simulation of data collection in correlated landscapes we propose when and how to conduct rigorous analyses of data from these common "pseudo-replicated" situations. SAC can be due to factors extrinsic to the object of study, such as soil moisture, or intrinsic to the system, as with the spatial distribution of seed-dispersed plants. Additionally, SAC can manifest as a gradient across a study area, or can be relative to many points within a study area. To test the efficacy of SAC-corrected data analysis, we simulated ecological variables including observation error and each of two types of SAC over a 1000 x1000 point lattice. On these landscapes, we simulated experiments with varying degrees of "pseudo-replication," and then used modified statistical tests that estimate and account for SAC. We also estimated and removed the effects of environmental gradients. Our results, reported as both Type I (alpha) error and 1-Type II error (power), describe many cases in which the estimation and corrections for SAC are highly successful. Overall, our results show that relatively simple estimation and statistical testing that accounts for SAC can dramatically improve statistical power in the many situations where "pseudoreplication" is unavoidable.
Key words: autocorrelation, conservation, spatial, pseudoreplication