849 
A polyphasic growth function for the 
endangered Kemp's ridley sea turtle, 
Lepidochelys kempii 
Milani Chaloupka 
Queensland Department of Environment 
PO Box 155, Brisbane Albert Street, Queensland 4002, Australia 
E-mail address: m.chaloupka@mailbox.uq edu.au 
George R. Zug 
Department of Vertebrate Zoology 
National Museum of Natural History 
Washington, D.C. 20560 
The Kemp’s ridley, Lepidochelys 
kempii, is the smallest of the seven 
extant species of sea turtle ( Marquez, 
1994) and is endemic to the Gulf of 
Mexico and Atlantic coast of the 
United States (Pritchard, 1989). It 
has been subject to extensive human 
exploitation and is the most endan- 
gered sea turtle species in the world 
(Marquez, 1994). Seasonal trawl 
and pound-net fisheries are major 
hazards, posing a serious risk to the 
long-term population viability of 
the Kemp’s ridley (Epperly et al., 
1995; Caillouet et al., 1996). Al- 
though the Kemp’s ridley sea turtle 
is endangered, the somatic growth 
and population dynamics of this 
species are not well known (Cha- 
loupka and Musick, 1997) despite 
several important growth studies 
that have been carried out for cap- 
tive or head-started stocks (Cail- 
louet et al., 1986; Caillouet et al., 
1995b). We propose a new growth 
model for the endangered Kemp’s 
ridley sea turtle that is based on a 
skeletochronological data set de- 
rived recently by Zug et al. (1997) 
from wild stock sea turtles stranded 
along the Atlantic Bight and Gulf 
coasts of the United States. The 
growth model presented provides a 
basis for improving our under- 
standing of sea turtle growth dy- 
namics in general and for modeling 
Kemp’s ridley population viability. 
Materials and methods 
Data set 
The data set used here comprised 
70 size-at-age records for Kemp’s 
ridley sea turtles — 69 records from 
stranded turtles plus the inclusion 
of known mean hatchling size (see 
Marquez, 1994). The data set 
(n = 70) also comprised growth 
records spanning the postnatal de- 
velopment phase (from 4 to 72 cm 
straight carapace length, SCL) and 
including the mature adult phase, 
but the records were not distributed 
evenly over this size range. The age 
estimates were derived from a 
skeletochronological analysis of 
wild Kemp’s ridley sea turtles 
stranded along the Atlantic coast of 
the United States and in the Gulf 
of Mexico (see Zug et al., 1997). 
Straight carapace length (SCL) was 
measured to 0.1 cm and age to 0.1 
yr. The original sample of stranded 
turtles comprised 73 individuals, 
but age estimates for 4 individuals 
were not possible because of either 
1) a lack of discernible growth rings 
or 2) uninterpretable irregular 
growth. Records for these 4 individu- 
als were discarded, yielding the 69 
individual turtles used in this study. 
The data set also included strand- 
ing location, with 79% of the sample 
comprising turtles stranded on the 
Atlantic coast. Sex was recorded for 
37% of the strandings; no propor- 
tional difference was evident be- 
tween the Atlantic and Gulf of 
Mexico subsamples. Further details 
of the strandings data set and the 
skeletochronological methods used 
for age estimation can be found in 
Zug et al. (1997). 
The limitations of skeletochron- 
ological ageing techniques and the 
need for caution in interpreting 
such age estimates for sea turtles 
have been well discussed elsewhere 
(Zug et al., 1986; Zug, 1990; Zug et 
al., 1997). Chaloupka and Musick 
(1997) have also provided a critical 
review of sea turtle skeletochron- 
ological studies and have discussed 
the limitations of such studies in 
terms of age validation, length 
back-calculation, growth estima- 
tion, layer loss adjustment proto- 
cols, and implications of the specific 
time-dependent sampling design 
implicit in the data set. For in- 
stance, the implicit sampling de- 
sign in the current study was mixed 
cross-sectional because only the ter- 
minal age-size estimate was avail- 
able for each of the 69 stranded 
turtles. This sampling design con- 
founds age and cohort effects and 
thus only an expected or mean 
growth function can be estimated 
(see Chaloupka and Musick, 1997). 
Statistical modeling approach 
The functional relation between 
size (cm SCL) and estimated age for 
the 70 Kemp’s ridley sea turtles 
was modeled with a two-stage ap- 
proach: 1) exploratory data visual- 
ization including nonparametric 
smoothing (see Cleveland, 1993) to 
Manuscript accepted 4 April 1997. 
Fishery Bulletin 95:849-856 (1997). 
