FISHERY BULLETIN: VOL. 74, NO. 2 



compound microscope at approximately 30 

 diameters. One postnatal layer was considered to 

 consist of an opaque subunit and a translucent 

 subunit (Figure 3). The layers in most of the teeth 

 examined were not as well-defined or as regular 

 in thickness as those illustrated by Kasuya (1972) 

 for Stenella coeruleoalba or by Klevezal' and 

 Kleinenberg (1969) for Delphinus delphis. Teeth 

 from 39 of the 442 animals were completely un- 

 scorable, being heavily worn or showing no dis- 

 crete layers in the sections examined. All the 

 teeth were scored several times, over a period of 

 several months, without referring to specimen 

 numbers or to values obtained previously, until the 

 scorer felt confident of the results. The values 

 used in the analyses are those obtained in the 

 final round of scoring. The teeth were scored to 

 the nearest postnatal layer when possible, or a 

 range, e.g., "8 to 10 layers," was estimated. Aver- 

 age accuracy is estimated at ±1 layer for teeth 

 with 5 layers or less and ±2 layers for teeth with 

 5 to 12 layers. Convoluted secondary dentine was 

 present in most of the teeth vdth more than 12 

 layers, making counts very difficult and of dubi- 

 ous reliability. We feel that the counts for many of 

 these teeth are probably underestimates. Teeth in 

 which the pulp cavity was entirely closed in all 

 sections examined were scored as "occluded." 



The NORMSEP computer program was used to 

 define modes in the length-frequency distri- 

 butions for fetuses. The program was written by 

 Hasselblad (1966) and modified by Patrick K. 

 Tomlinson, Inter-American Tropical Tuna Com- 

 mission. The program separates the mixture of 

 normal length distribution into its components, 

 assuming that the lengths of individuals within 

 age groups are normally distributed and that an 

 unbiased sample of the length distribution was 

 obtained. 



GROWTH 



Length at Birth 



Average length at birth of 82.5 cm was obtained 

 from a linear regression line based on 3-cm group- 

 ings of fetuses and neonatals (Figure 4). The 

 largest fetus of the 461 examined was 904 mm 

 long. The smallest neonatal animal was 780 mm 

 long. Eighty-six calves and fetuses between 73 and 

 94 cm were measured in random samples. As- 

 sumptions inherent in the method used to arrive 

 at this estimate are that pregnant females and 



calves are 1) equally vulnerable to capture in the 

 purse seine, 2) equally likely to die once captured, 

 and 3) equally represented in the sample of dead 

 animals measured. For example, if neonates were 

 less likely to be included in the samples than were 

 pregnant females, average length at birth would 

 be overestimated. Other potential sources of error 

 are differential rates of prenatal and postnatal 

 natural mortality and premature births caused by 

 stresses imposed by pursuit and by capture in the 

 purse seine. 



Gestation Period and Fetal Growth 



The most commonly used method for estimating 

 the gestation time of cetaceans is that of Huggett 

 and Widdas ( 195 1). They showed that for a variety 

 of mammals of widely different orders, a plot of the 

 cube root of fetal weight on age is linear except 

 during the first part of pregnancy, when growth is 

 exponential. Their model can be expressed in the 

 general formula W" = a(t - to), where W = 

 weight, t = age, a = the "Specific Fetal Growth 

 Velocity," andto = "the intercept where the linear 

 part of the plot, if produced backwards, cuts the 

 time axis." Laws (1959) applied the method of 

 Huggett and Widdas to fetal length/time data for 

 three odontocetes [Physeter catodon, Delphinap- 

 terus leucas , and Phocoena phocoena) and obtained 

 estimates of gestation periods (15, 14, and 11 mo, 

 respectively). He assumed that weight is propor- 

 tional to the cube of length and used the form L = 

 aitg - to),-whereL = length. This assumption is not 

 entirely correct (see length-weight results below), 

 but is a close enough approximation of the real 

 relationship between length and weight to allow 

 its use in estimating gestation period. Laws' esti- 

 mates corresponded closely with other estimates 

 obtained by more direct methods. Laws' version of 

 Huggett and Widdas' method is used here. 



A gestation period of 11.5 mo was obtained from 

 an analysis based on 281 fetal and postpartum 

 specimens collected in January, February, March, 

 April, May, and October 1972 (Figure 5). The 

 January-May samples comprised all of the fetuses 

 of all of the females examined. The postpartum 

 samples in these months were not random and are 

 therefore not included. The October samples were 

 random over all age-classes in the catch; therefore, 

 all specimens less than 160 cm long, approxi- 

 mately the length at onset of puberty (Harrison et 

 al. 1972), are included in the plot. Obvious modes 

 are present in the length distributions (seasonal- 



234 



