574 
Fishery Bulletin 95(3), 1997 
(Brown et al., 1979); in principle it should be a more 
effective indicator of population substructure than 
nuclear loci. Given that more mtDNA than nuclear 
BNA divergence is expected, how can an allozyme 
locus show differentiation when mtDNA haplotypes 
do not? The lack of mtDNA differentiation in yellow- 
fin tuna does not appear to be the result of a lack of 
variation nor of a small sample size: although in- 
creasing haplotype diversities and sample sizes will 
increase statistical power, the mtDNA haplotype di- 
versities of our populations, assayed for just two re- 
striction enzymes, were quite high at around 0.65- 
0.70, and sample sizes were similar to those used in 
the allozyme analyses. Nuclear DNA differentiation 
can exceed mtDNA differentiation when either the 
migration rate or the breeding sex ratio is strongly 
biased towards females (because mtDNA is mater- 
nally inherited), but there is no evidence that either 
of these conditions holds for yellowfin tuna (e.g. 
IATTC, 1992). The explanation for the seeming dis- 
crepancy may be that several independent polymor- 
phic allozyme loci were screened, whereas haplotypes 
of mtDNA are best treated as alleles at a single, 
nonrecombining locus. In a situation of low overall 
genetic divergence (resulting from gene flow or re- 
cent separation), the stochastic nature of genetic drift 
means that if several allozyme loci are screened, and 
notwithstanding the expected higher rate of mtDNA 
evolution, divergence might be first detected at an 
allozyme locus before it is detected for mtDNA. An 
alternative explanation, as intimated earlier, is that 
the GPI-A* differentiation results from selection. 
The delineation of the four stocks of yellowfin tuna 
does not seem unreasonable given what we know of 
their distribution and movements. Yellowfin are 
found circumglobally, but only in tropical and sub- 
tropical oceanic waters, approximately between the 
latitudes 40°N and 40°S (Collette and Nauen, 1983). 
Spawning occurs throughout the year in all core ar- 
eas of distribution, peaking in the warmer months 
(Collette and Nauen, 1983). Waters off the southern 
regions of South America (approximately 55°S) are 
too cold for Atlantic Ocean and Pacific Ocean fish to 
migrate around Cape Horn. Furthermore, direct con- 
nections between the tropical Atlantic Ocean and the 
eastern Pacific Ocean were severed after the Isth- 
mus of Panama closed about 3.5 million years ago 
(e.g. Keigwin, 1982; Coates et al., 1992), a closure 
likely to have predated the origin of yellowfin tuna 
(estimated by Elliott and Ward (1995) to have oc- 
curred within the last two million years). Thus Pa- 
cific Ocean and Atlantic Ocean fish could not mix. In 
contrast, Atlantic Ocean and Indian Ocean fish could 
mix (through southern Africa waters), as could In- 
dian Ocean and Pacific Ocean fish (through Indone- 
sian waters), but tagging experiments indicate that 
most yellowfin tuna move on a scale of hundreds 
rather than thousands of kilometers (Joseph et al., 
1964; Bayliff, 1979; Hunter et al., 1986; Lewis, 1992). 
The extent of migration between ocean basins is 
therefore likely to be low, with intraoceanic recruit- 
ment predominating. Nonetheless, interoceanic 
movements are possible and could account for the 
low degree of genetic differentiation among areas. 
Further discussion of the genetic and other biologi- 
cal data with respect to Pacific Ocean fish is given in 
Ward et al. (1994). 
At present, these suggestions on the global stock 
structure of yellowfin tuna are essentially based on 
gene frequencies at a single polymorphic allozyme 
locus, GPI-A*, because no significant genetic hetero- 
geneity was detected for three other polymorphic 
allozymes and the mitochondrial DNA variants 
showed little interpopulation differentiation. It may 
well be that the stock structure of yellowfin tuna, in 
management terms, is more complex than these 
present findings suggest: very limited migration be- 
tween areas can effectively homogenise gene frequen- 
cies, and thus dispersal between areas can still be 
low even between populations that cannot be geneti- 
cally discriminated. 
Future genetic work should include the examina- 
tion of more fish from the Indian Ocean because the 
identification of these fish as a separate stock is based 
primarily on the analysis of just 21 fish for a single 
allozyme locus. Further clarification of genetic stock 
structure issues in yellowfin tuna will require larger 
sample sizes, examination of more areas (especially 
from the Indian and Atlantic Oceans), and the de- 
ployment of genetic techniques, such as microsatellite 
analysis, with enhanced resolving power and less 
concern over neutrality and selection issues. 
Acknowledgments 
This work was supported by grant 91/27 from the 
Fisheries Research and Development Corporation. 
We thank the following who assisted us by collecting 
samples: Noel Barut, Barbara Block, Chris Boggs, 
Thor Carter, Tim Davis, Ed Everett, John Gunn, 
Aubrey Harris, Nishi Karunasinghe, Theresa 
O’Leary, Dennis Lee, Dave Milton, Pianet Renaud, 
and Toki Takenaka. Shirlena Soh helped with some 
of the allozyme analyses. Chris Bolch, Rene 
Vaillancourt, Vivienne Mawson, Peter Rothlisberg, 
and two anonymous referees made useful comments 
on a draft of this manuscript. 
