Porch: Estimating velocity and diffusion from tagging data 



705 



area 1 , season 1 



o 



o 



50 100 150 200 250 300 



area 1 , season 2 



50 100 150 200 250 300 



area 2, season 1 



100 150 200 250 



area 2, season 2 



100 150 200 250 300 



Sample size 

 Figure 4 



Coefficients of error of the estimates for the parameters of the discrete model under weak 

 diffusivity. The three curves in each graph correspond to zone B recovery probabilities of 0.0 

 (squares), 0.1 (triangles), and 1.0 (crosses). 



based methods must prescribe a velocity model, but 

 the latter must also specify models for a great many 

 other processes. Even if the velocity model is correct, 

 abundance-based estimates may still be biased if any 

 of the models for the other components are mis- 

 specified. Furthermore, because abundance-based 

 methods estimate many parameters, they tend to 

 require a large amount of data. In the absence of such 

 a large data base, trajectory-based approaches may 

 be the only reasonable option. 



Another attractive feature of trajectory -based ap- 

 proaches is that they naturally accommodate data from 

 archival tags. The chronological sequence of n position 

 updates from each tag can be treated as though it were 

 a sample of n independent tags with short liberty times. 

 Conventional abundance-based methods cannot take 

 advantage of this additional information. 



Trajectory-based approaches do have their limita- 

 tions. One is that they are not useful for assessing 

 aspects of the population dynamics other than ve- 

 locity. This point is especially important when the 

 behavior of the population is being investigated 

 within a management context. Under these circum- 

 stances abundance-based methods would seem to be 

 the more viable option because they can, at least in 



principle, be formulated to estimate any relevant 

 parameter. The enormous number of parameters re- 

 quired of such models, however, compromises the effi- 

 ciency of the parameter search. Moreover, abundance- 



1600 



-1000 1000 3000 



Position (km) 



Figure 5 



Distribution of 2,000 recovery positions generated by us- 

 ing the discrete advection model and weak diffusivity. The 

 legend refers to the probability (P) of recovering a tag in 

 zone B (right side of dashed vertical line). The solid verti- 

 cal line marks the boundary between areas 1 and 2. 



