Date of Award

1996

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

First Advisor

James D. Barbour

Abstract

Studies were conducted in Louisiana to determine the spatial and temporal distribution of rice water weevil (RWW), Lissorhoptrus oryzophilus Kuschel, adults, eggs, larvae and pupae in drill-seeded rice. In addition, three sequential sampling plans, Kuno's fixed precision sequential sampling plan, the sequential probability ratio test (SPRT), and 2-SPRT, were developed and evaluated to establish an accurate and economically efficient sampling plan for RWW larvae in drill-seeded rice in Louisiana. Values obtained for the aggregation indices, b, from Taylor's power law, $\beta$, from Iwao's patchiness regression and, k, from the negative binomial distribution indicated that RWW larvae were nearly randomly distributed regardless of sample date. Iwao's patchiness regression and the negative binomial distribution modeled larva populations better than Taylor's power law. A common k (kc) of 13.63 was determined from the larva data. Monte Carlo simulations of Kuno's sampling plan provided actual precision levels that were higher than those specified for the simulation. Kuno's sampling plan required $\approx$6 and 14 samples to estimate RWW larva economic threshold at the specified precision levels of D = 0.20 and 0.30, respectively. Monte Carlo simulation of the SPRTs indicated that the 2-SPRT generally required fewer samples to make terminating decisions for RWW larvae management compared to the SPRT, however, only the SPRT maintained Type I and II error rates below the specified error rate of 0.10. The SPRT and 2-SPRT required an average of 2.43 and 2.59 samples to make terminating management decisions at RWW larva economic threshold. The SPRTs required the least sampling effort and would substantially decrease sampling effort compared to larva sampling programs currently used to make RWW management decisions in Louisiana. Spatial autocorrelation analysis indicated that all RWW developmental stages exhibited significant spatial dependence. Spatial correlograms and spatial density maps suggested that 2 to 13 m$\sp2$ patches were exhibited by all RWW life stages. This spatial phenomenon was not detected by Taylor's power law, Iwao's patchiness regression or k from the negative binomial distribution.

Pages

100

Share

COinS