Analysis of animals that occur in groups

Analysis of animals that occur in groups

Distance for Windows exercise Solution

Project Browser results

Below is an abbreviated set of Project Browser results of the 8 fitted models.

Striking features from the Results Browser is the strong preference (in \(\Delta\)AIC score) for models that contain cluster size in the detection function model. Contrast the adjusted cluster size estimates (whether by regression or by use of cluster size as a detection function covariate) with the average size of detected clusters of 7.2. Observe the knock-on effect of the estimate average group sizes upon the estimates of abundance of individuals.

Analysis	Num par	Delta AIC	Adjusted ClusSize	Adjusted CS CV	N(indiv)	N(indiv) LCL	N(indiv) UCL	Bootstrap
hn cos average group size	2	22.4	NA	NA	31655	25889	38705
unif cos size regression	2	24.9	6.3	0.026	26106	21665	31457
hr poly size regression	2	23.7	6.1	0.026	27749	20917	36813
hn size bootstrap	2	3.3	6.5	0.020	25449	22787	27868	bootstrap
hr size bootstrap	3	0.0	5.7	0.183	27704	23343	34598	bootstrap
hr poly average size	2	23.7	NA	NA	32673	24666	43278
hn cos group size regression	2	22.4	6.2	0.026	27142	22150	33258
unif cos average size	2	24.9	NA	NA	29858	24837	35893

Interval estimates of abundance

All confidence intervals for estimates of individual abundance include the true value, however, each of the estimates using the average cluster size of detected clusters as an estimate of average cluster size in the population are positively biased. Positively biased to the extent that the lower confidence limits only just include the true value of 26000.

For further experimentation

A DistWin project with the 8 completed analyses is available for download if you wish to further explore the data set and conduct additional analyses.