284 TELFORD: MULTIMAMMATE RAT POPULATION BIOLOGY IN TANZANIA

December as in Morogoro, with those juveniles found from November through
February representing reproduction comparable to that following heavy short
rains in Morogoro. The differences might also indicate a species difference in
reproductive pattern. Sample sizes reported by Chapman et al. (1959) are
inadequate to draw a meaningful comparison with the Morogoro study.
 There is agreement between conclusions presented here and those of
other authors concerning the role of precipitation on population dynamics.
Coetzee (1975) thought that the cold winters in South Africa might act toward
controlling density, with the concomitant influence of sparse plant cover during
late winter providing less protection against predators. Cold is not a factor in
tropical Africa, except perhaps with montane rodent populations, but certainly
the lack of ground cover observed during the very dry period of November-
April 1981-82 contributed to greater mortality and, thus, lower density going
into the main reproductive season of 1982. And cover was distinctly better
during the following two wetter years. Delany and Neal (1969) suggested that
rainfall might be the most important factor governing rodent breeding in
Uganda, through its indirect effect upon food availability. Coetzee (1975)
linked the abundance of food to population explosions, a conclusion also
suggested by Smithers (1971), who found the number of pregnant female P.
natalensis in Botswana to be very low during the last two years of a four year
drought. Following the end of the drought, a massive explosion took place. In
Coetzee's opinion, population explosions might be linked with the abundance
of food. He listed the following factors as important regulators of P.
natalensis density: (1) a high reproductive rate due to large litter size and short
intervals between litters, (2) the age at first litter, and (3) the duration of
breeding period, which is largely controlled by the rainy season with its indirect
influence over food supply. An optimum food supply can lead to a population
explosion which would disrupt the normal social structure, thus influencing
litter size and interval, the size of neonates, and survival rate. In South Africa,
high density coincides with cold weather, scarcity of food, lack of cover and
increased predation, all of which can lead to a population crash.
 The present study reinforces Coetzee's conclusions. Although perhaps
important only in areas with Morogoro's climatic pattern, it is likely that the
influence of precipitation is most critical during the short rain period, setting
the stage, as it were, for the density threshold upon which the P. natalensis
population enters the main breeding season in April of each year. If the short
rains fall far below the average, density will decrease to minimal levels between
January and April, and population levels will increase thereafter at a rate
proportionate to subsequent precipitation levels. With a succession of
favorable short rains, within two years the population can "explode" as was
observed in 1983-84, followed by a "crash," as seen in January 1984. That this is
a recurrent phenomenon is supported by Harris (1937) in his Morogoro study:
"The most striking feature [of fluctuations in population density] was a fall