DISCUSSION 



M. E. Solomon: If there is a high level of competition among small 

 plaice is it a good thing to have a mesh size which allows them to escape 

 fishing ? 



J. A. GuLLAND : There is little evidence of extreme competition among 

 fish of the sizes just escaping. 



R. P. Silliman: Watt seems to imply that intraspecific competition 

 does not apply to the populations he is considermg. 



L. B. Slobodkin: In a population which is being held down by climate 

 or predation, intraspecific competition will be unimportant for most of the 

 time. But if and when a large increase occurred, it would become important. 



W. Edmondson: His Table compares population dynamics, but are the 

 data really comparable? Are the nutrient incomes, for example, the same in 

 all cases ? 



L. B. Slobodkin: It is the yield per unit of nutrient income which is 



being sought for. 



W. Edmondson: In the algal populations he uses, the yield per unit of 

 nutrient is itself related to the harvesting system used. 



L. B. Slobodkin: I think that the figure in the Table will still be of the 

 right order of magnitude. 



G. C. Varley: I am a little troubled by Watt's use of the term 'regula- 

 tion in relation to chmatic factors. Isn't this using the word synonymously 

 with mortahty'? A correlation may be obtained between mortality and 

 climate, but this is not a proof of regulation in the strict sense of the word. 



L. B. Slobodkin: Watt uses 'regulation', for example, of a population by 

 cUmate under extreme conditions, to mean that in these conditions the size 

 of the population is predictable from meteorological observations when 

 nothing is known of the population itself His usage differs from that often 

 adopted in that it does not include a feed-back concept. 



J. G. Skellam: There are two aspects to the use of mathematical models 

 in biology — the quahtative and the quantitative. In general I rarely expect 

 much more from a model than that it should represent Nature in its broad 

 features and yield a vahd pattern of qualitative results. When a model 

 contains several parameters there are great difficulties in estimating them 

 separately, as is easily seen by anyone who attempts to estimate « and p 

 simultaneously in the case of the binomial distribution when p is small. 



