6 March 2010 - Size of bowl and amount of water

Location:Beltsville Agriculture Research Center , MD
Conditions:  Sunny, upper 80's
Start: 7:15 a.m. April 6
End: 5:00 p.m. April6
Habitat:Mature deciduous woodland with Spring Beauty understory
Bees: Andrena, Osmia, Lasioglossum, Ceratina, Nomada
Experiment: 3.25 ounce white solo cups  were alternated with a completely filled 12 ounce Party store white (on outside and inside) plastic drink cup and the same drinking cup filled half full.  Bowls were spaced 5 m apart and filled with soapy water (unscented laundry detergent and tap water). 

Results:

Site Number 1

	1	2	3	4	5	6	7	8	9	10	Mean
3.25	2	4	2	3	2	20	6	2	0	3	4.40
12 Full	0		3	3	4	9	9	3	1	5	4.11
12 Half	0	1	3	1	2	10	2	1	1	2	2.30

Pretending the data were normally distributed resulted in an F=0.719, P=.4814.  A Kruskal-Wallace test yielded a P=0.14.

Site Number 2

	1	2	3	4	5	6	7	8	9	10	Mean
12 Full	4	1	5	2	3	5	6	2	5	15	4.80
12 Half	3	4	0	1	5	5	0	0	0		2.00
3.25	4	3		6	0	6	6	11	7	2	5.00

ANOVA: F=0.2454, P=0.1064.   KRUSKAL-WALLACE: P=0.06

Combining both datasets (without getting fancy and adding a factor for site)

ANOVA: F=2.842, P=0.06706.  KRUSKAL-WALLACE: P=0.02

Note that Site Number 2 was the same site run the previous day under almost identical weather conditions, but now with many more captures...almost surely due to putting in sufficient laundry detergent this time....

Note also that when we indicate that a bowl or cup is full that it is actually a few mm shy of the rim...it just is not practical to fill them to the rim given the unevenness of the ground.

Results indicate that there is no statistical difference between the white 3.25 ounce bowls and the 12 ounce cups when both are full, but that catch is decreased by half when the cup is half full and that is statistically different from the two full bowls (paired tests not presented here). 

The practical ramifications here is that when running traps that use deep cups it is best to fill them to the top even if rain dilutes the mix.