HOHN and HAMMOND: POSTNATAL CROWTH OF SPOTTED DOLPHIN 



described above with the following results: 



Fi.xed length at birth 



(estimated in this paper) 

 Estimated time of birth 

 Estimated length at 1 yr 

 Estimated length at 2 yr 



Rates of growth as centimeters per month for the 

 two fitted curves adjusted for length at birth are as 

 follows: 



Southern Population 



Thble 3 shows the estimates of mean length of the 

 fitted normal distributions for each month. For these 

 data it is clear, apart from the mean lengths of 105.0 

 cm in January and 127.5 cm in May, that there is 

 only one cohort born each year in the southern sum- 

 mer. As a result of this and the much smaller sam- 

 ple sizes, distributions of length could only be 

 distinguished up to about 18 mo. The two final 

 columns of Tkble 3 show the mean lengths of the two 

 distributions to the right of the length-frequency 

 plots. These are quite consistent from month to 

 month, as with the northern data. 



Figure 6 shows the Gompertz model of growth fit- 

 ted to the mean lengths from columns 1 and 3 of 

 Tkble 3. Time at birth and length at 1 yr were 

 calculated as described above with the following 

 results: 



Fixed length at birth 



(estimated in this paper) 

 Estimated time of birth 

 Estimated length at 1 yr 



83.2 cm 

 6 January 

 127.9 cm 



Rates of growth for this fitted curve do not 

 decrease from birth as they do for the northern 

 population because, the curve has a point of inflec- 

 tion at approximately 50 mo. The rates of growth 

 at 0, 6, 12, and 18 mo after birth are 3.29, 3.72, 4.12, 

 and 4.47 cm/mo, respectively. 



Table 3.— Mean lengths of the fitted normal distributions for 

 the southern offshore spotted dolphin. 



There are several sources of variability in the 

 estimates of mean length by month to which the 

 growth models have been fitted. There is individual 

 variation in time of birth, length at birth, and growth 

 rate The calving season may vary from year to year 

 and area to area. The specimens which were 

 measured are subject to the usual sampling varia- 

 tion. Sampling in a particular year may not have been 

 random with respect to time in each month. Given 

 these sources of variability, it is interesting that the 

 results should appear so consistent. 



The growth curves were fitted to the unweighted 

 mean lengths. If the variation in the mean length 

 of a distribution is considered to be due largely to 

 sampling error, then there is a justification for a 

 weighted regression. We believe that this is not 

 necessarily the case and that the unweighted regres- 

 sions represent the best descriptions of growth for 

 these data. When weighted regressions were per- 

 formed the fitted curves changed negligibly. 



The most important potential problem is that the 

 method relies upon being able to analyze a sample 

 of data in which reproduction is seasonal and in 

 which the timing of seasonality is constant. This 

 analysis has shown that this may be difficult to 

 achieve. Only by stratification of the data by area 

 could consistent results be obtained. Stratification 

 of the data by area improves the consistency of the 

 series of mean lengths because offshore spotted 

 dolphins appear to have different calving seasons 

 depending upon the area of capture In probability, 

 this seasonality is not actually a function of area but 

 of schools or groups of schools which tend to inhabit 

 different areas with different environmental condi- 

 tions. Thus, even with the best stratification scheme, 

 there may always be asynchronous seasonal elements 

 in a sample of data from any given area affecting 

 the estimation of the mean lengths of the cohorts. 



In this analysis we pooled the data from several 

 years for our monthly samples, rather than attempt- 



561 



