T HE L ENGTH OF THE B REEDING S EASON

Abundance of PL was significantly greater than that of PMG in two out of three years in this study (Chapter 2), and these differences are reflected in the number of estimated births. A bimodal distribution of births was often observed during this study (Figure 3-2); this lends support to the hypothesis that there is a mid-summer lull in breeding in these species (Wolff 1996), which may be caused in part by the loss of overwintered females from the population. Though it is likely that some pregnancies were overlooked due to inconsistencies in longitudinal trapping records, there was no difference between species in trappability (Table 4-4); thus, the number of pregnancies should be underestimated for PL and PMG in the same way. Can differences in the length of their breeding seasons explain the changes in abundance of each species observed during this study?

Photoperiod is a primary cue for the initiation of spring breeding for P. maniculatus (Price 1966; Millar and Herdman 2004) and P. leucopus (Whitaker 1940; Heideman et al. 1999). Spring breeding is stimulated by increasing photoperiod and is mediated by temperature such that breeding begins following sudden rises in temperature that coincide with increasing day length (Sadleir 1974). Similarly, the cessation of breeding in higher latitude populations is thought to be closely tied to declining temperature, perhaps due to the increased cost of maintenance at lower temperatures (Sadleir et al. 1973) and energy requirements of lactation (Harland and Millar 1980).

Individuals in syntopic populations of PL and PMG experience the same climatic conditions; thus, if they differ in breeding season length, the explanation must lie in species-specific responses to environmental cues such as temperature or photoperiod. There is some evidence that responsiveness to photoperiod is under genetic control for PL (Heideman and Bronson 1991). Myers et al. (2005) proposed that the onset of breeding of PL in northern Michigan may be influenced by gene flow from southern populations, where this species typically begins breeding in March (Brown 1964; Baker 1983). Similarly, the onset of breeding for PMG may be influenced by gene flow from northern populations, which begin breeding in April or May (Millar et al. 1979).

133 In this study, PL began breeding earlier and ceased breeding later than PMG in all years except 2011, when PMG began breeding earlier. Analysis of ground temperature and capture records suggests that differences between species in the date of onset of breeding were not due to differences in temperature. The number of captures was not correlated with spring temperature at capture sites (Figure 4-6), but temperature varied little between sites (Figure 4-5). These observations support the hypothesis that breeding season length may be a species-specific niche difference where PL and PMG co-occur.

There is phenotypic variation in the responsiveness of individuals to photoperiod (Heideman et

al. 1999), and natural populations of PL consist of a mixture of genetically determined

phenotypes that are intermediate between absolutely photoresponsive and absolutely non- responsive (Heideman and Bronson 1991). Comparisons of the length of interval between the first and last recorded births and estimates of population-wide dates of onset and cessation of breeding suggest that there were fewer PMG than PL that bred either very early or very late. A possible explanation for this is that there is less phenotypic plasticity in the responses to

environmental cues regulating breeding in PMG than there is in PL.

Peromyscus with longer breeding seasons have a greater annual reproductive growth than those

with short breeding seasons (Millar et al. 1979); it is thus possible that differences in population growth between PL and PMG are influenced by breeding season length. Over the course of this study, PL bred for 24 days longer than PMG on average. PMG experienced the shortest breeding season and lowest abundance in 2012, at the same time that PL experienced the longest breeding season and greatest abundance. Further, PL bred for nearly 50 days longer than PMG in 2012, which was the largest difference in breeding season length between species – this coincided with the largest difference in abundance observed in this study.

A longer breeding season was not always associated with greater abundance, however. PL bred for the same amount of time in 2010 and 2011 (approximately 130 days), but its abundance was greater in 2011 than 2010. PL bred for 21 days longer than PMG in 2010, but no differences in abundance were observed that year. In contrast, PL and PMG differed in breeding season length by only 5 days in 2011, yet their abundance differed strongly. The overall length of the breeding season is thus not a consistent predictor of patterns of abundance, suggesting that the effect of

134 breeding season length on population growth is mediated by other factors. What could cause a longer breeding season to result in increased population growth in some years but not others?

For populations in which YOY breed in their natal year, an early onset of the breeding season maximizes total reproduction because mature YOY tend to dominate breeding in late summer (Havelka and Millar 2004) and early-born YOY are more likely than those born later to have litters of their own (Sharpe and Millar 1991). Myers et al. (2005) observed that PL increases faster and to a greater maximum in years following short and mild winters than in years following long and harsh ones. In this study, spring and overall PL abundance were highest in 2012, which was preceded by the mildest and shortest winter experienced by mice (Chapter 2). Reproductive output of PL YOY females was the highest in 2012 (Chapter 3), perhaps the result of successful early reproduction by their overwintered parents.

Early onset of breeding, however, is not always advantageous. Newborn mice suffer decreased growth rates at low temperatures and require greater maternal care in the nest (King 1968; Harland and Millar 1980). Breeding females must balance this need with longer bouts of foraging outside the nest to meet the energetic requirements of lactation (Millar and Innes 1985; Millar et al. 1990). Spring breeding begins following sudden rises in temperature that coincide with increasing photoperiod (Jameson 1953; Sheppe 1963; Brown 1964); however, once breeding has begun, further declines in temperature in the early spring may not cause cessation of reproduction (Sadleir 1974). An increase in temperature in the early spring may thus cause breeding to begin too early, resulting in reduced maternal survival and reproductive success.

I was not able to assess failed early reproduction; however, some trends observed during this study indirectly suggest that this occurred in 2010. The breeding season of 2010 was preceded by a relatively harsh and long winter (Chapter 2), and, consistent with the predictions of Myers et

al. (2005), both spring and overall PL abundance were low. Based on the timing of the first

known birth, I estimated that PL began breeding on April 18, 2010. This was nearly identical to the onset of breeding in 2012 (Table 3-2); however, spring conditions in 2012 were mild whereas conditions in early 2010 were colder. In 2010, female PL that established residency appeared on the grid later than female PMG, but the opposite trend was found for males (Chapter 3),

135 have begun breeding in 2010 under harsher conditions, resulting in both failed reproduction and increased mortality of overwintered females. This might explain the absence of a numerical advantage for PL even though it bred for 21 days longer than PMG.