Table 4.

Linear discriminant analysis (LDA) prediction of majority sample component in 28 dust samples comparing both the individual component samples or mixture samples as training sets. Prior probabilities for LDA prediction on mixture samples was excreta mix (12.5%), feed mix (25%), and rice mix (62.5%). Prior probabilities for LDA prediction on individual component samples was excreta (5.4%), feather (15.4%), feed (38.5%), rice hulls (15.4%), and wood litter (15.4%).

Training set	Dust sampleN	Bedding material of flock of origin of the dust sample	LDA prediction of majority component from the training set
13 individual component samples (Feed, wood shavings, sawdust, excreta)	7	Wood shavings	Excreta (7/7, 100%)
	7	Sawdust	Excreta (7/7, 100%)
	7	Rice hulls	Excreta (6/7, 85.7%) Feed (1/7, 14.3%)
	7	Straw	Excreta (6/7, 85.7%) Wood shavings (1/7, 14.3%)
8 mixture samples (See Table 2 for mixture details- Mix 3 ≥50% excreta)	7	Wood shavings	Excreta mix (4/7, 57.1%) Rice hull mix (3/7, 42.9%)
	7	Sawdust	Excreta mix (7/7, 100%)
	7	Rice hulls	Excreta mix (7/7, 100%)
	7	Straw	Excreta mix (6/7, 85.7%) Rice hull mix (1/7, 14.3%)