Statistical analysis of nematode counts from interlaboratory proficiency tests.

摘要

A series of proficiency tests on potato cyst nematode (PCN; n=29) and free-living stages of Meloidogyne and Pratylenchus (n=23) were investigated to determine the accuracy and precision of the nematode counts and to gain insights into possible trends and potential improvements. In each test, each participating laboratory received ten samples of soil collected from naturally infested fields and well mixed before distribution. The variance between the counts from the ten samples, estimated per test and laboratory, was constant for PCN cysts and free-living stages of Meloidogyne spp. and Pratylenchus spp. during the period monitored. However, during the initial years, the variance between the samples showed a downward trend for PCN living juveniles. The variance between the laboratories, estimated per test, was constant for PCN cysts over the period 1994-2010. For PCN living juveniles the variance between laboratory means declined from 1994-2000 and from 2000 it was more or less constant, which coincided with the introduction of a well-described new methodology. For Meloidogyne spp. and Pratylenchus spp., from 1998-2005 the variance between laboratory means was more or less constant or increased. From 2006, all participating laboratories incubated the organic matter and the variance between the laboratories was stable and below the levels observed during 2003-2005. After discarding data from initial years with unstable variances of the nematode counts, a meta-analysis was carried out on the remaining data using HGLMs to model the mean and variance of the counts simultaneously. The within-laboratory variance for the ten samples and the interaction variance of test and laboratory were estimated and combined to percentage coefficient of variation (%CV) per laboratory. For PCN cysts, %CV per laboratory ranged from 12.6 to 41.6 and for PCN juveniles in sandy soil from 24.4 to 67.4 and in loamy soil from 37.9 to 102.0. For Meloidogyne spp., %CV ranged from 53.2 to 84.4 and for Pratylenchus spp. from 24.3 to 90.6. The quality of a proficiency test is largely defined by random distribution of the nematodes over the samples allocated, statistical analysis of the data and communication of the results to participating laboratories. This study provides insights into the accuracy and precision at laboratories in recent years and the effect of concerted actions during the period in which the proficiency tests were conducted.