ULANOWICZ ET AL.: IDENTIFYING CLIMATIC FACTORS INFLUENCING MARYLAND LANDINGS 



Monte Carlo trials were run for each species 

 using different random combinations of test 

 data. In five of the seven species considered, 

 pairs of two variables were identified frequently 

 enough to warrant their being cited as potential 

 predictor formulae in Table 2. In one case (blue 

 crab) no variables appeared often enough in the 

 trials to warrant reporting a predictor equation. 

 By contrast several sequences of oyster predic- 

 tors appeared often, but no sequence predomi- 

 nated in the trials. About five separate sequences 

 appeared with almost equal frequency. Hence, 

 no formula for oysters is cited. 



In the clam regression, Cwl appeared as the 

 primary predictor in over 50% of the trials. In 

 roughly 20% of these instances CS2 was included 

 as secondary predictor. No predictor was chosen 

 12% of the time. When the two selected variables 



In Maryland, soft shell clams spawn in spring 

 and fall (Pfitzenmeyer 1962). However, the 

 spring set each year is almost totally eradicated 

 because of predation by benthic feeding fish and 

 crabs which migrate onto Maryland clam 

 grounds each spring and leave each fall (Holland 

 et al. 11 ). Factors influencing the strength of the 

 fall set (which occurs from October through De- 

 cember) and the ensuing survival of juveniles 

 have not been identified. It appears that these 

 factors are the ones most likely to have the great- 

 est effect on the magnitude of commercial clam 

 landings. Since Maryland is near the southern 



"Holland, A. F., N. K. Mountford. M. Hiegel, D. Cargo, and 

 J. A. Mihursky. 1979. Results of benthic studies at Calvert 

 Cliffs. Final Report to Maryland Power Plant Siting Pro- 

 gram. Ref. No. PPSP-MP-28. 229 p. Martin Marietta Labor- 

 atories, Baltimore, Md. 



Table 2.— Potential predictors of landings (in metric tons) of designated species; see 

 text for code to predictor variables. 



were finally regressed against the entire data 

 set, 60% of the total variance was explained, 49% 

 by Cwl alone. Environmental data were avail- 

 able to assess the predictive value of this equa- 

 tion for the year 1977. As can be seen in Figure 1, 

 this projected value has a large deviation from 

 the recorded measure, but this deviation falls 

 within the range of errors in the hindcast. 



Interpretation of this equation in terms of 

 causality is complicated by the absence of effort 

 data. For example, the effects of the rise in 

 number of licensed clammers (from 3 in 1952 to 

 100 in 1957 to 200 in 1979 [Richkus et al. 10 ]) 

 on catch cannot be accounted for, and they may 

 have been substantial. Still, the strong correla- 

 tion between cumulative low water temperature 

 lagged 1 yr and catch suggests a causal relation- 

 ship. 



I0 Richkus, W. A., J. K. Summers, T. T. Polgar, and A. F. Hol- 

 land. 1980. A review and evaluation of fisheries stock man- 

 agement models. Martin Marietta Laboratories, Baltimore, 

 Md., 177 p. 



03 

 O 



z 

 □ 

 z 

 < 



2 

 < 



2000- 



1000- 



i i I i | I I I — I | I I — I I | I I i i | i — n- 

 1950 1960 1970 



YEAR 



1980 



Figure 1.— Maryland soft clam landings in metric tons from 

 1952 to 1977 (solid line) and landings predicted using the re- 

 gression model (dotted line) (Table 2). (Landings for 1977 did 

 not enter into the derivation of the model.) Environmental fac- 

 tors were a cumulative low deviation in water temperature 

 (with a 1-yr lag behind the harvest figure), and a cumulative 

 high deviation in salinity (with a 2-yr time lag). 



615 



