CLIMATE AND THE DISTRIBUTION OF MARINE ANIMALS 



299 



Since 1930, January tcniporaturcs liavc increased 

 about 2° while July temperatures have deereased 

 about 1°. These opposite trends account, of 

 course, for tlie magnitude of the phenomenon to 

 be observed in fifjure 8. 



Recordsof surface water temperatures, smoothetl 

 by 5-year moving averages, for each month over 

 the period 1905-49 (fig. 9) show significant upward 

 trends for the months of January, February, No- 

 vember, and December. To establish the degree 

 and significance of these trends, we have (livi(U'd 



< 

 > 



1920 1928 1936 1944 1952 

 YEAR 



FiGfRE 7. — Annual deviations from the mean surface 

 water temperatures, 1921 to 1951, at St. Andrews, 

 New Brunswick. (The solid line is a 5-year moving 

 average.) 



1925 1930 1935 

 YEAR 



FifiiRE 8. — .Ianu:iry-.luly differences in surface water 

 temperatures as deviations from the mean difference, 

 1906 to 1949. at Boothbay Harbor, Maine. (The curve 

 is a 5- .year moving average.) 



tile records into two approximately equal periods 

 of 22 and 23 years, calculated tlic mean for each 

 month for each period, tiie standard error of the 

 difference between these means, and Student's 

 t (table 1). January, February, and December 

 show highly significant increases in mean tem- 

 perature, while a statistically significant increase 

 has occiu-red for the November period. March, 

 July, August, and September siiow slight down- 

 ward trends, but these trends are not statistically 

 significant. 



Table 1. — Coinparison of mean monthly nurface tempera- 

 tures at Boothbay Harbor, Maine, for the first and second 

 halves of the period 190o-J,9 



I Periods A and B for January and February are 1906 to 1927 and 1928 to 

 1949; for March to July, 1905 to 1927 and 1928 to 1949; for -August to December. 

 1905 to 1926 and 1927 to 1948. respectively. See also appendi.x table 1. p. 344. 



•5-percent level of significance. 



'•1-percent level of significance. 



Water-temperature records for Woods Hole, 

 Mass., for the period 1881-1914, 1932-42, and 

 1945-52 are given in appendi.x table 2. The 

 means for various periods are as follows : 



The trend in surface water temperatures for 

 January at Woods Hole is similar to that for 

 Boothbay Harbor for comparable periods. Except 

 for the period 1933-41, liowever, there does not 

 appear to be any trend at Woods Hole toward 

 declining July temperatures comparable to that 

 for Boothbay Harbor. T1h> annual mean for the 

 period since 1945 is considerably higlier than tlie 

 annual means for the earlier periods. This in- 

 crease is statistically significant, 'i'lie data sliow 



