It might be argued that pygmy smelt arose independently through parallel evolution in the 

 four lakes where it is known. For the characters to have evolved in the same direction by 

 chance in the four lakes is quite improbable. The results of the multivariate analyses by 

 Copeman (1977) and the present study make it extremely unlikely that the pygmy smelt 

 populations arose independently, in spite of the non-concordance of one character, the 

 visual pigment. Additionally an hypothesis of independent origin of each pygmy smelt 

 population would have to cope with similarities in spawning times of at least three pygmy 

 smelt populations, with slower growth rates and the earlier maturity of the two pygmy smelt 

 populations studied. These data cause no problem for the common origin hypothesis. We 

 therefore reject the hypothesis of independent origin of each pygmy smelt population. 



The lacustrine non-migratory pygmy smelt may suggest similarity to the kokanee , a 

 lacustrine non-migratory form of the anadromous sockeye salmon, Oncorhynchus nerka . 

 However, Nelson (1968a) found geographic and geological evidence that at least some of the 

 populations of kokanee had independent origins through the development of falls, or the 

 blockage of rivers by lava or glaciers. And Nelson (1968b) showed that some kokanee had 

 higher gill raker counts than sympatric sockeye salmon (Stewart and Anderson lakes), while 

 the Takla Lake the kokanee had fewer gill rakers than the sockeye. Although gill raker 

 counts were significantly different between sympatric kokanee and sockeye, none permitted 

 even subspecific levels of separation. Additionally, Nelson (1968b) found the range of 

 means of gill rakers in 8 allopatric kokanee populations and 11 kokanee populations 

 sympatric with sockeye, encompassed the range of means known for sockeye populations. We 

 concur with Nelson (1968a) that it is highly likely that kokanee had multiple origins and 

 that both forms should be designated as Oncorhynchus nerka . 



We bring the kokanee example forward to contrast the example of a form with a multiple 

 origin with the condition in the pygmy smelt where gill raker counts, body size and growth, 

 and spawning times all differ in a consistent pattern from sympatric rainbow smelt. Further 

 we have shown that gill raker counts permit high levels of separation ( 90%) of the pygmy 

 and rainbow smelts, whether sympatric or allopatric. All specimens of pygmy and rainbow 

 smelts can be distinguished using multivariate analysis. Pygmy and rainbow smelt are 

 reproductively isolated. 



There is one race of a dwarf (not the pygmy smelt) smelt in eastern North America that 

 probably did have an independent origin. In Black River Pond, Avalon Peninsula, 

 Newfoundland, there is a dwarf form of smelt with mean size of two-year-olds of 98 mm fork 

 length and a sympatric large smelt averaging 152 mm at the same age. Interestingly, here 

 the small and large smelt did not differ significantly in gill raker number which was 

 maximally 32, and the dwarf feeds mainly on benthic insects. There is an overlap of one day 

 in their spawning times. The preceding information was drawn from Bruce MS (1975). Insular 

 Newfoundland was apparently completely glaciated. If the lacustrine dwarf evolved In 

 insular Newfoundland, which seems possible, then it can be no older than 15,000 years BP. 

 It seems likely from deglaciation and gill raker evidence and the depauperate freshwater 

 fauna of the island that the Newfoundland dwarf evolved separately from the mainland pygmy 

 smelt. Mann and McCart (1981) found that the dwarf Cisco, Coregonus sardinella, had 



22 



