BULLETIN FLORIDA STATE MUSEUM


 A number of species that are completely or largely restricted to
the Roanoke River system, which drains into the Atlantic Ocean, have
affinities not with other Atlantic coastal forms, but with species in
the Ohio Valley. The headwaters of the Kanawha River are narrowly
separated from those of the Roanoke over a fairly extensive area in
western Virginia. Much of the present Roanoke fauna, including
Notropis cerasinus and, possibly, N. albeolus, almost certainly reached
this drainage by headwater transfer from the old Teays system dur-
ing preglacial time (Wright, 1934: 65), though whether the invasions
by these species occurred simultaneously cannot be determined.
 Evolution of the common N. cornutus-N. chrysocephalus stock
probably took place in the Mississippi Valley. This hypothesis is sup-
ported by the distribution pattern of these two species, which cen-
ters in this region. Some of the common cornutus-chrysocephalus
stock is believed to have become separated from the main popula-
tion in late Pliocene in one of the following areas: the Atlantic coast,
the upper Mississippi Valley, the Great Plains, or the Great Lakes.
The relative distributions of the two species suggest this segment
evolved into N. cornutus, while that part of the population to the
south became N. chrysocephalus.
 There is little evidence that N. cornutus evolved in either the Mis-
 sissippi Valley or on the Atlantic coast, except for its presence there
 today. Geological evidence (Flint, 1947: 163-167; and 1957: 168-170)
 indicates that before the Pleistocene the middle and upper parts of
 the Missouri did not flow into the Mississippi River as they do today,
 but instead "flowed north into the Souris-Assiniboine system, perhaps
 discharging via the Lake Winnipeg depression, the Nelson, and the
 Hudson Bay region. The advent of the ice sheet flowing from the
 northeast blocked all this drainage and detoured the Missouri along
 the ice margin." Contact was then established with the Kansas River
 below Kansas City, thus effecting a new outlet for the upper Missouri
 River. The upper Missouri therefore provided an isolated region
 where evolution of Notropis cornutus could have occurred, but its
 apparent absence from the upper Missouri drainage today argues
 strongly against this area as a place of origin.
 Radforth (1944: 10-11) describes and pictures a hypothetical course
 for the Laurentian River, which drained the present Great Lakes re-
 gion (with the possible exception of Lake Superior) in pre-Pleistocene
 time. Possibly a segment of the cornutus-chrysocephalus ancestral
 stock entered this river system during late Pliocene and there evolved
 into N. cornutus. This Laurentian center of origin for N. cornutus


Vol. 8