Ent 442/542 L Coop Lab 1: HOWTO
Degree-days and phenology models
How to determine developmental requirements from lab and field data
Methods for doing assignment 1,
http://pnwpest.org/ent442/phenmod442_06.html
Jan. 2006

Purpose of lab: to derive a simple degree day model, with 2 parameters, the 
lower threshold and DDs for development, for the egg stage of a moth pest. 
We have two types of data: constant temperature lab rearing data, and field 
data from caging studies. We will use the x-intercept technique for lab data, 
and lowest C.V. technique for field data, and compare. Note that these data 
are not the same as your actual assignment (and are in degrees F rather than C).

1) Lab data:
starting with constant temperature data such as these:

Temp (F)       Days
-------       ------
50		40
60		20
70		13
80		 9

Step 1: Compute inverse of Days (1/days) in a new column:

1/Days
-------
0.025
0.05
0.0769
0.1111

Step 2: Do linear regression of Temp (as x-range) on 1/Days (as y-range)

Excel: use menu items: Tools, Data Analysis, Regression
Place results in same sheet (highlight an area such as below here):

Shown here are a portion of the results:


		      coefficient	SEM	etc
a =	intercept	-0.1196
b =	x-variable	0.00285


These results indicate that the regression equation is Y = a + bx = -0.1196 + 0.00285 * x


Step 3: Obtain the Tlow and Deg Days for development from the equation:

	Tlow = x-intercept = equation solved for when Y = 0, so:
		0 = -.1196 + 0.00285 * x
	rearranged:
		x = -a/b = .1196/0.00285 =	41.96	so Tlow = 42 degrees F


	Deg Days = 1/slope = 1/b =		350.8	degree-days for development

	So we now have used the x-intercept method on lab data
	After you are familiar with the lowest CV method for field data, you 
	should also apply it to the lab data.


2) Field data:
	From a field study, we had four replicates, with these (made-up) results:
		Start	days	DDs to hatch for Tlow values (computed with a DD calculator):
	Rep	date	to hatch	39	42	45	48
	-------------------------------------------------------------
	1	6/23	38	        351	332	327	310
	2	7/12	27	        333	325	303	300
	3	7/31	24	        340	327	310	294
	4	8/22	26	        314	305	288	270

Step 1: Compute means of DDs for each Tlow (use function button Fx in Excel)

			average=x=	334.5	322.25	307	293.5

Step 2: Compute st. dev. of DDs for each Tlow (use function button Fx in Excel):

			St.dev=s=	15.54	11.87	16.18	17

Step 3: Compute coefficient of deviation (C.V.) of DDs for each Tlow:

		C.V. = s/x*100 =	4.65	3.68	5.28	5.79

Step 4: Select "best" model for threshold with lowest C.V.:

		Tlow =		42	(because the C.V. was lowest for a threshold of 42 F)
		Deg days =	322	(the average DDs for the Tlow=42)



3) Finally: compare results from the two methods and discuss:

	The lab (x-intercept) method resulted in a Tlow = 42, DDs = 351, while the field data,
	which are likely to be more appropriate for a working IPM model, 
	gave us Tlow = 42 and DDs = 322. So, for this case, the lab and 
	field agree for Tlow=42, and I would use the field value of 322 DDs for development time.

	To implement this model, we might monitor females with light traps, and start
	computing DDs at Tlow=42, as soon as we catch a significant number (or see a 
	peak), and expect peak egg hatch to occur after about 322 DDs. This is based on the understanding
        or assumption that peak egglaying corresponds to peak catch.