For other tallies of a tally sequence, only the incremental number of trees 

 becoming established on the plot during that time period is added to the tree 

 list. 



The number of species is also conditional on the plot being stocked. Logistic 

 regression equations predict the probability of 1, 2, 3, 4, 5, and 6 species on 

 the plot. The number of species cannot exceed the number of trees chosen 

 in step 5. 



A uniformly distributed pseudo-random number is used to make a discrete, 

 but unbiased, choice of the number of species on the plot. The probabilities 

 for each number of species are calculated and totaled. The total is divided 

 back into the probabilities so that the sum of the probabilities equals 1.0. The 

 adjusted probabilities are accumulated within the interval [0,1], and a uni- 

 formly distributed pseudo-random number is compared to the accumulated 

 probabilities. The number of species on the plot is the one for which the accu- 

 mulated probability first exceeds the pseudo-random number. 



There are 10 species in the regeneration model and each can occur in any 

 of three ways: as advance best trees, as subsequent best trees, and as excess 

 regeneration. Advance trees germinated more than 3 years prior to the har- 

 vest date. Subsequent trees germinated after the cutoff date for advance 

 trees. Excess trees were established on the plot, but were not aged. 



First, the probabilities of advance and subsequent best trees are predicted 

 by species (probabilities can be 0.0 when certain species do not occur on some 

 habitat types or in some geographic locations). The advance and subsequent 

 distributions are accumulated within the interval [0,1], and pseudo-random 

 numbers are used to choose the same number of species on the plot as was 

 chosen in step 6. 



Next, the probability of excess regeneration is predicted by species. Species 

 of remaining trees on the plot are determined from the excess distribution. 

 Probabilities in the excess regeneration distribution can differ considerably 

 from the combined advance and subsequent distribution. 



Heights of best trees are predicted from linear regression equations. Separate 

 equations were developed by species for advance and subsequent best trees. 



Heights are predicted for each best tree on the plot. Tree heights are var- 

 ied by adding or subtracting a random proportion of the standard error of 

 the estimate to predicted tree height. A pseudo-random number determines 

 whether to add or subtract. A second pseudo-random number in the inter- 

 val [0,1] determines the proportion of the standard error to add or subtract. 



Excess tree heights are determined from Weibull cumulative distribution 

 functions that were developed from the heights of best trees. A pseudo- 

 random number is used to assign heights for excess trees between the mini- 

 mum establishment height and the height of the best tree of the same species 

 on the plot. 



The regeneration model predicts natural regeneration. Planted trees are 

 an addition to predicted natural regeneration. Model users specify species, 

 density, year of planting, and survival of planted trees. The default option 

 is for trees to be planted uniformly throughout the stand, but options allow 

 for planted trees to vary in density in relation to residual overstory basal 

 area (see Ferguson and Crookston 1991). 



27 



