74 EIGHTH PACIFIC SCIENCE CONGRESS 



related in that both Scomberomorus spp. and Chirocentrus spp. feed 

 voraciously on Clupea spp. and Stolephorus spp. These latter groups, 

 in turn, feed on zooplankton (especially copepods) and phytoplankton. 

 The physical and chemical elements of the environment play an im- 

 portant part in determining the availability of plankton as well as the 

 various elements of the fish populations. It is clear, therefore, that 

 the various elements of the marine eco-system in Singapore Straits are 

 very closely integrated. 



In an attempt to study further the close relationship between the 

 fish population on the one hand and the environmental conditions, viz: 

 rainfall, windforce, temperature, salinity, phosphate, phytoplankton 

 and copepods, on the other, the method of multiple regression is used. 

 The multiple regression equation obtained may be expressed as fol- 

 lows:— 



X,^ = — 49,903 — 395.61100 X, — 1.77415 X3 — 1,105.75115 X^ 

 — 1, 269.82920 X- —' 0.47599 X, — 72.13537 Xg 

 v;here X-^4 = the monthly total catch of three kelongs in pounds, 



Xo = total r.ainfall in inches for that month, 



X3 = total wind speed in metres per second for that month, 



X^ = average temperature of the sea water in °C for that month, 



X^ = average salinity in Voo ^^r that month, 



X- = average number of phytoplankton cells and/or chains of cells 

 per cubic metre for that month, 



Xg = average number of copepods per cubic metre for that month. 



On testing the significance of the multiple correlation by working out 

 the value of "z" it is found to be 0.87977 whilst the 0.1% point is 

 0.8657. The multiple correlation is therefore clearly significant. The 

 adjusted multiple correlation coefficient squared is R^ = 0.5566. The 

 standard error of estimate S is = 2316 lbs. This is 14.44% of the mean 

 monthly total catch for the two years under study. 



The total catch for each month for the two years 1948 and 1949 

 has been calculated and compared with the corresponding actual total 

 monthly catches in Table I. In the above regression equation it is 

 assumed that all the relationships are linear. If it were further re- 

 fined statistically by taking into account the curvilinear nature of some 

 of the relationships a much closer estimate would, in all probability, 

 be obtainable. The results of the multiple regression shows that if 

 predictions of the independent variates could be made with accuracy 

 a prediction system of sufficient accuracy for catches in Singapore Straits 

 is within reach. It would appear also from these results that, in the 

 exploration of new grounds, an intensive study of the marine eco-sys- 

 tem should yield a comprehensive picture of the habits and migration 

 of the fish stocks in a comparatively short time. 



