Overwintering management of Osmia lignaha 



JESO Volume 139, 2008 



conditions in an unheated screened building), 2) location of origin (Nova Scotia population 

 versus Utah population), and 3) sex. In addition, interactions between each factor and among 

 all three factors were studied. The proportion of bees surviving for each nesting block was 

 subject to arcsine transformation (Zar 1999) to achieve normality and homoscedacicity of 

 variance, which was successful. All analyses were therefore conducted in the transformed 

 form, although the reported population parameters are of non-transformed data. 



Supercooling Point Determination and Weight Loss 



At two intervals throughout the winter (January and March), sub-samples of 

 cocoons were removed from the nesting blocks. These were re-weighed to determine 

 percent weight loss within the interval of wintering. Bees were then removed from 

 their cocoons, placed into a gelatin capsule, and were attached with petroleum jelly to a 

 thermocouple within a NalgeneTm 2.0 ml cryovial. The cryovial was then submerged in 

 coolant in an FTS-Systems ultra-low temperature bath (Stone Ridge, NY) interfaced with a 

 computer using virtual instrumentation (VI) Lab View 5 (National Instruments, Austin, TX) 

 software and hardware (Nubuss port and A/D boards). From an initial bath temperature 

 between 10°C and 15°C, the temperature was dropped at a rate of 1°C /min (after Salt 

 1966), and supercooling points were measured to the nearest 0.1 °C; the supercooling point 

 is the lowest body temperature recorded prior to the increase in temperature due to the latent 

 heat released during crystallization (Lee 1991). 



Nonparametric analysis (based on ranked data) was used on both supercooling 

 point and weight loss data sets as they were not normally distributed and variances were 

 unequal. For non-parametric analysis of supercooling points, each datum was transformed 

 to a positive value, T, using the following formula T = (a°C) (-1), where a represents the 

 supercooling point datum (which is always a negative number). This transformation was 

 done as more negative supercooling points are indicative of enhanced cold-hardiness, and it 

 allowed the most negative values to have the highest rank instead of the lowest. 



Results 



Overwintering Survival 



Populations overwintered outside had significantly higher mean survival rates 

 than those kept inside (F = 16.59; df = 1; p < 0.001), and bees reared in Nova Scotia had 

 significantly higher mean survival rates than populations imported from Utah (F = 29.78; df 

 = 1; p < 0.001) (Fig. 2). Significant interactions were observed between origin x wintering 

 site (F = 5.28; df = 1 ; p = 0.027), and the main effects plot (Fig. 2) indicated that origin had 

 a larger main effect than the wintering site (indicated by the slightly steeper slope). Sex 

 did not significantly influence wintering survival (F = 0.04; df = 1; p = 0.851) or the two- 

 way interactions of 'origin x sex' and 'wintering site x sex', although these did approach 

 significance (F = 3.15; df- 1; p = 0.083 and F = 3.22; df = 1; p = 0.080, respectively). The 

 three-way interaction of 'wintering site x origin x sex' was not significant (F = 0.28; df = 1 ; 

 p = 0.60). 



7 



