of 3, 9, or 12 months, with one exception being Coreopsis, which was stored up to only 9 months.
Humidity treatments were achieved by placing 100-seed replicates in 15 x 25-cm petri dishes on wire
screens supported by segments of tubing 1 cm above 50 ml of saturated lithium chloride, magnesium
chloride, magnesium nitrate, or sodium chloride to maintain 11%, 32%, 52%, or 75% RH, respectively
(Copeland 1976). Incubators were maintained at constant temperatures during seed storage or
germination. Four 100-seed replications were germinated at the recommended temperature and light or
dark regime immediately following storage at each duration and RH level, while another four replications
were desiccated in forced-draft ovens for 48 hours at 75C to determine the moisture contents of stored
seed. Analysis of variance and response surface regression procedures were used to analyze the data.
Within each storage temperature, a randomized complete block design was used with a factorial treatment
structure (4 x 4) between relative humidity and storage period.
The data were analyzed by fitting multiple regression equations using PROC REG of SAS Institute
(1985) for seed of each of the 12 species. Both storage period and relative humidity were studied at four
levels each (except for Coreopsis which had only three levels of storage period), and the model fitted
initially was a complete third-degree polynomial of the form:
Response = Bo+B1SP+B2RH+B3SPxRH+B4SP2+B35RH2+B6(SPxRH2)+ (1)
B7(SP2xRH) +B8SP3+89RH3+e
In the above model the terms B1SP and B2RH represent the linear effects of storage period (SP) and
relative humidity (RH) on the response; B3SPxRH represents a linear by linear interaction effect between
storage period and relative humidity, and B4SP2 and BgRH2 represent the curvilinear effects (quadratic)
of storage period and relative humidity. The third-degree terms BgSPx RH2 and B7SP2xRH represent the
linear x quadratic interaction effects, with, 86SPxRH2 representing the linear effect of storage period on
the quadratic effect of relative humidity, while 1BSP3 and B9RH3 represent the cubic effects of storage
period and relative humidity, respectively, on the response.
The model fitting exercise consisted of testing the significance of the coefficient estimates of the