320 TRANSURANIC ELEMENTS IN THE ENVIRONMENT 



accurately quantify them before other model parameters are considered. We know the 

 least about critical transfers to and from soil because they are the most difficult to 

 quantify. For example, in this model three of the preceding transfers required some 

 parameter fitting to calibrate the performance of the model against field measurements. 

 Although potentially important to animal components because of the radiotoxicity of 

 plutonium, the addition of plutonium transfers to animals had a minor influence on the 

 major plutonium reservoiis in the forest. 



Sources of variation in model predictions include spatial variation, temporal 

 variation, and measurement error in transfer coefficients. These sources are independent 

 of any bias due solely to model structure or coupling. COMEX simulations of plutonium 

 behavior in the model forest exclude temporal (annual and seasonal) variation in transfers 

 because the model transfers are considered time invariant. In this sense our simulations 

 differ from the time variant or stochastic modeling of plutonium behavior in 

 pinon-juniper forests in New Mexico (Wlieeler, Smith, and Gallegos, 1977), where 

 ecosystem behavior in response to random variations in climate was considered. Variation 

 in model predictions from the COMEX simulations can be interpreted as a consequence 

 of spatial variation in transfers, measurement error, or both, depending on perspective. 

 Intraforest variation in transfer coefficients resulting from different edaphic or 

 microclimatic conditions within a single forest will produce local differences in the 

 amount of plutonium (picocuries per square meter) in biota. Intraforest variation, 

 however, could be negligible relative to differences in transfer coefficients between 

 distinct forest stands (e.g., in different counties). Therefore the average predicted amount 

 of plutonium in each forest, given a uniform soil contamination, could vary, depending 

 on site conditions and forest species composition. 



An example of how such variation between forests could bear on the assessment of 

 the environmental impact of plutonium is provided by considerations of fire. Assume that 



TABLE 4 Correlations Between Varying Transfer Coefficients and Predicted 

 Steady-State Values for Each Forest Model Compartment* 



Compartment 



Ground Tree Tree Tree Forest Soil 



Transfer Soil vegetation Consumers roots wood leaves litter fauna 



Soil to tree roots -0.67 0.68 0.54 0.51 



Soil to litter ' 0.68 0.45 



Tree roots to soil 0.69 -0.69 -0.49 -0.52 



Soil to ground vegetation 0.71 



Litter to soil -0.65 -0.39 



Soil fauna to soil -0.57 



Ground vegetation to litter -0.74 



Tree roots to wood 0.40 0.35 



Consumers to litter -0.77 



Tree wood to leaves -0.50 



Litter to soil fauna 0.34 



Tree leaves to litter -0.48 



Soil to consumers 0.52 



*Data are based on 300 independent simulations. Transfers were varied simultaneously before each 

 simulation, using a Monte Carlo random-n.umber generator. Only correlations greater than 0.30 are 

 reported. 



