Florida Lake Management Society Annual Conference, Naples, Florida, June 4 – 7, 2007
Eureka Manta multiparameter system. In vivo chlorophyll and in vivo phycocyanin fluorescence were measured using a Turner Designs Aquafluor fluorometer. For this study, data from August 3, 2006 are presented. Sub-samples of the total 70-pond data set were randomly selected representing 5, 10, and 20% of the total. Additionally, the data was ranked and the 5% of the ponds with the lowest and highest values were then non-randomly selected. Mean and 90% confidence intervals values for the sub-samples were calculated.
Figure 1. Means and 90% confidence intervals for temperature, dissolved oxygen, phycocyanin fluorescence, and chlorophyll fluorescence.
Overall the 5, 10, and 20% sub-samples generally did a good job of predicting the mean values of the entire population of lakes for all of the parameters. As would be predicted by statistical theory, the larger the sub-sample the better the agreement with the overall population measure. In general, the more biologically influenced the parameter, the less the contraction of the 90% confidence interval with increased sample size. Of the 120 sample:parameter combinations tested only 6 did not have confidence intervals that overlapped with the population mean. In the sub-samples that were intentionally selected to have all high or all low values, the confidence intervals never overlapped the population mean. While these extreme combinations will only occur once out of >22 million times, significant errors can arise when they do.
Session 4 – Page 6