Cooper and Mangel: Metapopulation structure in the conservation of salmonids 



217 



Results 



When habitats are constant over the simulation pe- 

 riod, exponential growth occurs DS'vf of the time. 

 Because no density dependence was incorporated in 

 the model, denies in the source habitats increased 

 exponentially, as did the number of strays from these 

 sources. Such a situation could occur if, for instance, 

 harvesting pressure was decreased on a deme that 

 had been severely overharvested in the past (lead- 

 ing to low abundance) but whose habitat was rela- 

 tively pristine (maintaining high productivity). This 

 results in all dynamics in the sink populations being 

 obscured by the massive number of immigrants. 



It is possible that all the streams constitute sinks 

 (all H;,0) < 1.05); this occurred in 1.3% of the simu- 

 lations. In this case, there is still the possibility that 

 the metapopulation as a whole could persist for de- 

 cades before all populations began to decrease (Fig. 

 2). The reason for this persistence is that individu- 

 als that stray are lost only to their natal stream, not 

 to the metapopulation as a whole. The case shown in 

 Figure 2 results when the losses due to some habi- 

 tats (where r(/,n<l) are nearly offset by production 

 in other habitats (where r(z,n>l). Forty percent of 



the all-sink metapopulations produced populations 

 that did not decrease over the course of the 100-year 

 simulation. If we had incorporated a parameter to 

 represent decreased reproductive success of strays, 

 the incidence of nondecreasing, all-sink meta- 

 populations would have been lower. 



In scenario 2 (Fig. 3), the source populations 

 (demes 3, 4, and 5) initially increased exponentially 

 but eventually began to decrease as the habitat deg- 

 radation increased. Habitat degradation leads to 

 peaks in local deme abundance (Fig. 4A). The mode 

 of this distribution was around year 20 despite the 

 fact that habitat degradation began in year 5. Fur- 

 thermore, in over 3% of the sources, deme sizes in- 

 creased throughout the 100-year simulation. 



The population trajectories for the sink habitats 

 were less intuitive ( Fig. 3 ). None of their habitat com- 

 ponents changed, yet some demes increased (deme 1) 

 or stayed constant (deme 9) over a number of decades, 

 whereas others decreased. The result depends on the 

 proximity to sources (noting that deme number trans- 

 lates to the location of the deme along the water- 

 way), the per-capita rate of reproduction in those 

 sources, and the sink's own per-capita rate of repro- 

 duction. Over 'd'^c of the sink habitats never attained 



