Canadian Forest Service Publications
Non-linear mixed models for repeated data assessment of time and temperature effects on conidia production in the fungus Valdensinia heterodoxa. 2004. Magnussen, S.; Vogelgsang, S.; Shamoun, S.F. BioControl 49(1): 47-62.
Issued by: Pacific Forestry Centre
Catalog ID: 23739
CFS Availability: Order paper copy (free)
Valdensinia heterodoxa Peyr. has potential for application as a biological control agent of salal (Gaultheria shallon Pursh) in regenerating forests in the Pacific Northwest region of North America. Conidia production of two isolates of Valdensinia heterodoxa was examined at four temperature levels during a 12-day period. Conidia production was strongly influenced by both temperature (with an optimum between 16 and 19 °C) and time (peaked at day 9). Replicate (five) variance was highly heterogeneous and made a conventional approach to the statistical analyses unsuccessful. However, newly developed generalized nonlinear models succeeded in identifying significant differences between isolates in their temperature specificity. One model was based on a time-dependent Poisson process (logistic model) modified by an independent nonlinear temperature function (a scaled gamma function). The other was a nonlinear growth model with fixed (time, isolate) and random (replicate) effects. We believe that nonlinear mixed models offer an attractive framework for analysis of longitudinal count data common in growth and reproduction studies.