OUP user menu

To Twin or Not to Twin? Trade-Offs in Litter Size and Fawn Survival in Mule Deer

Tamara L. Johnstone-Yellin, Lisa A. Shipley, Woodrow L. Myers, Hugh S. Robinson
DOI: http://dx.doi.org/10.1644/08-MAMM-A-030.1 453-460 First published online: 14 April 2009


Parents often face a trade-off between allocating resources to many young with lower survival, or fewer young with higher survival. To examine trade-offs in litter size and juvenile survival in mule deer, and how survival was influenced by the nutritional condition of the female, we compared the survival of 30 twin and singleton mule deer (Odocoileus hemionus) fawns during their 1st summer in eastern Washington. Overall fawn mortality was high (51.6%), especially during the 1st half of the summer (61.3%). Coyote predation caused 58.3% of all identified fawn mortalities. Twin fawns had a risk of dying 2.6 times higher than single fawns, and the difference between twin and singleton survival was most drastic during the first 1.5 months of life. Body fat of females during their last trimester of pregnancy predicted the number of fetuses they were carrying and whether they had at least 1 fawn surviving until the fall, but not the number of fawns surviving. Under these conditions, a litter size of 2 would be considered optimal because mothers giving birth to twins produced an average of 0.92 fawns by fall, whereas mothers producing singles ended up with only 0.75 fawns. However, our model suggested that a population producing only twins would be expected to increase only 4% faster than one producing only singletons.

Key words
  • litter size
  • maternal condition
  • mule deer
  • Odocoileus hemionus
  • survival
  • twin
  • Washington

Because resources are limited, wild animals face trade-offs between allocating resources to survival and reproduction. One such trade-off occurs when parents produce either few young with higher survival or many young with lower survival (Case 1978). This optimal clutch size for birds was 1st modeled by Lack (1947) as the clutch size that produces the most fledglings. This optimal clutch depends on the influence of clutch size on survival and future reproductive performance of juveniles, parents, and their subsequent offspring (Case 1978; Stearns 1992). Larger broods tend to suffer higher mortality, and, to a lesser degree, larger broods reduce parental survival and the future reproductive performance of parents and daughters (Case 1978). Because the intensity of these trade-offs depends on quality and quantity of resources available, the optimal or “Lack” clutch may vary among years and locations (Lack 1947). For example, when resources were limited in 9th century rural Finland, women who produced a single child achieved a higher lifetime reproductive success than those who produced twins (Haukioja et al. 1989). However, in modern society where resources are more abundant, women producing twins have a higher lifetime reproductive success, despite a slightly higher mortality of twins.

According to resource allocation theory (Williams 1966), trade-offs in litter size are achieved through balancing levels of prenatal and postnatal parental investment. Maternal care can be defined as the amount of resources invested in an offspring multiplied by the number offspring (Andersen et al. 2000). As the number of offspring increases, the amount of parental care per individual decreases but total care increases (Lloyd 1987; Read and Harvey 1989). As a group, ungulates are large, long-lived, and iteroparous, with small litters, and allocate high levels of energy in each breeding attempt (Pelabon et al. 1995). Some species (e.g., elk [Cervus elaphus], American bison [Bison bison], and caribou [Rangifer tarandus]) typically produce singletons, whereas others (e.g., mule deer [Odocoileus hemionus], roe deer [Capreolus capreolus], moose [Alces americanus], and pronghorn Antilocapra americand) commonly produce litters of 2 or 3. Because relative size at birth and early growth of offspring do not differ between monotocous and polytocous ungulates, polytocous ungulates allocate more in maternal care than monotocous ungulates (Andersen et al. 2000). For example, it costs 1.6 times more in lactation energy to raise a set of twin mule deer or white-tailed deer (Odocoileus virginianus) fawns than a singleton (Carl and Robbins 1988; Mauget et al. 1999; Pekins et al. 1998; Oftedal 1985; Sadleir 1980). Although ungulates are generally considered capital breeders that rely on stored energy for reproduction, Andersen et al. (2000) showed that small roe deer fall more toward the income breeder side of the continuum because they use energy acquired during the reproductive period to support maternal care (Stearns 1992).

Despite the wealth of literature on ungulates, surprisingly little is known about trade-offs in litter size and survival of young in this group. Therefore, we sought to examine the reproductive trade-offs faced by mule deer, by monitoring the survival of mule deer fawns from near birth to 3 months of age. We tested the hypothesis that because of trade-offs associated with maternal condition and maternal care, twin fawns would have higher mortality than singleton fawns, and that both twin and singleton fawns of females with higher maternal condition would have higher survival rates. However, we expected that because twinning is the norm in mule deer (Anderson and Wallmo 1984), that on average, females that produced twins would end the critical lactation period (first 3 months of life) with more fawns than those that produced singles. In addition, we used a simple population model to demonstrate the potential effects of a twinning tactic compared to a singleton tactic on population growth in mule deer.

Materials and Methods

Study area.—We examined survival of free-ranging mule deer fawns in 2 sites within the channeled scablands of eastern Washington. The Washington Department of Fish and Wildlife-managed Revere Wildlife Area and Bureau of Land Management-managed Escure Ranch (henceforth referred to as RWA), located in Whitman and Adams counties, contained 3-tipped sage (Artemisia tripartita)–Idaho fescue (Festuca idahoensis) communities, interspersed with winter wheat cultivated by local farmers. Our 2nd site was the Bureau of Land Management-managed Coffeepot Lake-Pacific Lakes Wildlife Areas (henceforth referred to as CPWA), located in Lincoln County (approximately 64 km from RWA). This site contained 3-tipped sage-Idaho fescue and big sage (Artemisia tridentata)–Idaho fescue zones. Adjacent areas were farmed for winter wheat, lentils, and alfalfa. These areas supported resident (overlapping summer and winter range) or semi-resident (partially overlapping summer and winter range) mule deer. Fawn-doe ratios at RWA as estimated by routine late-November aerial surveys after hunting season were 40–50 fawns per 100 females (Washington Department of Fish and Wildlife 2003). Average maximum daily temperatures in the study area in 2003 were 21.1°C in May, 26.9°C in June, 32.2°C in July, and 30.7°C in August. Average minimum temperatures were 7.0°C in May, 9.6°C in June, 9.2°C in July, and 10.1°C in August. Precipitation averaged 3.6 cm in May, 0 cm in June, 0.10 cm in July, and 0.80 cm in August.

Capturing and monitoring female mule deer and, their fawns.—To monitor survival of wild fawns from shortly after birth to 3 months and to obtain information about fetal rates and maternal body condition, a Washington Department of Fish and Wildlife crew captured 17 females in RWA and 13 females in CPWA in March 2003 using a net gun. Each female had an existing radiocollar or was fitted for a new one (Advanced Telemetry Systems, Inc., Isanti, Minnesota). We scored the body condition of the rump, the thickness of the longissimus dorsi muscle, and maximum thickness of subcutaneous rump fat via ultrasonography (Sonovet 600; Universal Medical, New Bedford, New York; Cook et al. 2001), and estimated total body fat from equations described in Cook et al. (2007). We also used transabdominal ultrasonography to determine the expected number of fawns. One female died before parturition.

To assist in capturing fawns, in each female we inserted a 15-g vaginal-implant transmitter (M3940; Advanced Telemetry Systems, Inc.) with a temperature-sensitive sensor that emitted a postpartum signal when expelled immediately before fawns were born, and a precise-event transmitter that recorded time at which the vaginal-implant transmitter was expelled. We prepared and inserted the vaginal-implant transmitters according to procedures described in Bishop et al. (2002) and Johnstone-Yellin et al. (2006).

To capture and radiomark wild fawns, we checked the signals of each vaginal-implant transmitter every 1–2 days from 27 May through 5 June using either ground or aerial telemetry. Ground crews investigated postpartum signals from vaginal-implant transmitters, searching for fawns at the expulsion site. Because no fawns were found near the expulsion sites (Johnstone-Yellin et al. 2006), females were radiolocated by ground crews or helicopter and watched until fawns were spotted. Crews wearing surgical gloves to minimize transfer of human scent captured and collared each fawn. However, we were still only able to find 13 fawns belonging to 9 of 29 remaining radiocollared adult females for which body condition was known. To increase the sample size of fawns for survival analysis, crews captured and collared 17 additional fawns of unmarked females, using a helicopter to locate females and fawns visually. Ground crews weighed each fawn using a spring-loaded scale, determined its sex, and measured the length of the hind leg and the distance from crown to rump. Although we were not sure if all vaginal-implant transmitters were expelled at birth, we used the precise-event transmitter as the most objective estimate of the age of fawns (Johnstone-Yellin et al. 2006). We had no objective estimate of age for fawns from unmarked females. To each fawn captured, crews attached an expandable radibcollar (M4210; Advanced Telemetry Systems, Inc.), weighing approximately 120 g. The collars were programmed with a mortality sensor that switched from a pulse of 60 ppm to 120 ppm if the fawn remained motionless for 4 or more hours. All methods met guidelines approved by the American Society of Mammalogist (Gannon et al. 2007) and were approved by the Institutional Animal Care and Use Committee of Washington State University, Pullman.

At RWA and CPWA, we listened for mortality signals every other day from the time of fawn capture until 12 September 2003. Upon hearing a mortality signal, we located the carcass within 4–24 h to assess the cause of death via a postmortem examination (Wade and Bowns 1982) and used clues from the mortality or cache site and any remains of the carcass and collars to determine the cause of death. If the carcass was intact, we submitted it to Washington Disease Diagnostics Laboratory (Washington State University, Pullman, Washington) for necropsy.

Because we were able to capture and monitor the survival of fawns from only one-half of the radiocollared adult females, we radiolocated 27 of the 29 remaining original females that had vaginal-implant transmitters using a helicopter or on foot and checked if they had fawns at heel in early fall, although the survey took a month to conduct, we pooled all data and considered it “survival to fall.” Fawns we did not see during fawn captures (May–June), but saw in September, we assumed were with the female in May. However, if ultrasound from March 2003 indicated twins and we only saw 1 fawn in September, we did not assume she had twins in May. Thus, we removed the family from the analyses involving twinning. If we did not see fawns with a female in May or September but ultrasonography in March confirmed she was pregnant, we included the female-fawn pair in the survival-to-fall analysis, but not survival to parturition, because we were unable to confirm if fawns were present in May. If a female not observed during captures was seen with a single fawn in the fall (which corresponded with fetus count), we assumed that fawn was present in May; we also assumed there were no twins even though we could not confirm the absence of a 2nd fawn.

Data analysis and hypothesis testing.—We estimated survival rate (Ŝi ± SE) of 30 radiocollared mule deer fawns (16 singletons and 14 twins, 13 males and 17 females, and 20 RWA and 10 CPWA), over the summer using the Kaplan Meier (1958) method. To examine the effects of categorical variables (sex, site, litter size, and season) on survival of radiocollared fawns we used the Cox proportional hazards model (Cox 1972), comparing competing models using Akaike's information criterion correct for small sample sizes (AICC; PROC PHREG, version 8.1; SAS Institute Inc., Cary, North Carolina). To calculate the relative importance of explanatory variables in the model given the data and the pool of candidate models, we calculated Akaike weights (wis) and summed them over the subset of models that contained each variable (Burnham and Anderson 2002). To examine the effect of twinning on fawn survival to fall further, we performed a chi-square analysis for comparing proportions on a larger data set of 48 fawns (24 singles and 24 twins, and 30 RWA and 18 CPWA). These included our 30 radiocollared fawns combined with 18 additional fawns from radiocollared females for which we knew their twinning status near birth and for which we were only able to make a one-time observation of their unmarked fawns in the fall. We used the same data set of 36 families to determine whether the fates of twins were independent, and the chi-square method of Gaillard et al. (1998), in which expected numbers of family types (i.e., families with twins in which none, 1, or both survive, and families with singles in which the fawn did or did not survive) were compared to observed number of family types.

For 27 radiocollared pregnant females that we confirmed with live fawns, we used logistic regression (PROC LOGISTIC, version 8.1; SAS Institute Inc.) to determine whether maternal condition (i.e., body mass, body length, thickness of longissimus dorsi muscle, and body fat composition) in March predicted the number of fetuses, whether a female had at least 1 fawn (yes/no), and the number of fawns (0, 1, or 2) surviving to the fall. We determined the correlation among model variables to reduce the candidate pool of models, and then used AICC to determine the most-parsimonious model predicting fetal rates and fawn survival.

To examine the value of twinning to population dynamics, we built deterministic population models using a Leslie matrix (Leslie 1945) in program RAMAS GIS (Akcakaya et al. 1999) and compared predicted growth of a population bearing just twins with a population bearing just singletons. We assumed a population structure of fawn, yearling, and adult (3–12 years) with survival of yearlings and adults constant within models and survival of fawns constant between models. A total of 7 population models was produced with survival of adults and yearlings ranging from 0.70 to 1.0 at 0.05 increments. A survival rate of 0.853 for adult females may be necessary to maintain a stable population (Unsworth et al. 1999). We set survival at stage 12 to 0, assuming mortality of all deer on their 13th birthday. Both populations were given an initial abundance of 100 animals set to a stable age distribution. We also assumed adult fecundity was the product of female survival and fetal rates (2 for females bearing twins and 1 for females bearing singletons). Yearling fecundity was set at one-half of adult fecundity. From 0 to 6 months of age, we assigned twins the survivorship obtained from our wild twins and singletons the survival rate obtained from our wild singletons from birth to fall. Because we did not measure overwinter survival of fawns, we assumed survival of twins and singletons was equal and we used the mean overwinter survival (6–12 months) for mule deer found in Colorado and Idaho (0.389— Bartmann et al. 1992; Bishop et al. 2005; Unsworth et al. 1999; White et al. 1987).


Of the 30 free-ranging fawns collared, 16 died by 12 September 2003. Mortalities peaked between 15 and 25 days after birth, and 12 of the 16 mortalities occurred during the first 54 days of the summer interval, between 27 May and 19 July 2003. Predation was the most common cause of fawn mortality. During the first 54 days, 7 fawns were depredated by coyotes, 1 fawn died from an intestinal infection, and 1 was hit by a tractor. Three others died of unknown causes. During the last 54 days, 2 of the remaining 18 fawns were depredated by coyotes, 1 died of unknown causes, and 1 dropped its collar. Because we did not know the fate of the fawn that dropped its collar, we censored it from that point.

Survival of 30 radiocollared fawns during the 108-day summer interval was 0.516 ± 0.090, 0.613 ± 0.088 during the first 54 days and 0.790 ± 0.094 (n = 18) during the last 54 days of the interval. The most-parsimonious model for fawn survival included litter size, which alone accounted for 23% of the weight of evidence for fawn survival given the data and the other models (Table 1). Together, all models that included litter size had a relative weight of 57%. Singletons had a higher survival rate (0.625 ± 0.12) and mean survival time (69.3 ± 6.4 days) than twins (0.400 ±0.13, 16.3 ± days, hazard ratio = 2.63). Similarly, in a larger group of 48 fawns, a higher proportion of single fawns (0.75, 18 of 24) survived the interval from capture until 12 September than did twin fawns (0.46, 11 of 24, Z = 2.07, P = 0.04, d = 3.04) and the survival of twin fawns was independent of each other (i.e., no family effects, χ2 = 1.55, d.f. = 1, P = 0.34). Therefore, the group of 12 females producing twins ended the summer with 11 fawns, or 0.91 fawns per female, whereas the group of 24 mothers that produced singles ended the summer with 18 fawns, or 0.75 fawns per female.

View this table:
Table 1

Weight of evidence for general linear models predicting survival rates of 30 neonatal mule deer (Odocoileus hemionus) fawns from birth until 12 September 2003 in 2 study sites in the channeled scablands of eastern Washington based on whether they were singletons or twins (litter size) and their sex. Test statistics include Akaike's information criterion corrected for small sample size (AICC), AICC difference (Δc), and Akaike weight (wi).

Litter size81.300.23
Litter size × sex820.70.16
Litter size × site83.041.740.1
Litter size × site × sex83.492.190.08
Site × sex83.842.540.06
Study site84.172.870.05

The effect of a fawn's sex on its survival through the summer also showed moderate support in the data; the model with sex alone had a 17% probability of being the best model given the data and the other models included in the analysis. Together, all models that included the variable sex had a combined relative weight of 47%. Males tended to have a lower survival rate (0.385 ± 0.11) than females (0.695 ± 0.13) across the summer interval. However, we cannot completely rule out the possibility that survival was independent of the variables considered because the intercept-only model had a 14% probability of being the best model analyzed.

Body mass of 27 pregnant mule deer in March was positively correlated with the thickness of the longissimus dorsi (r = 0.44, P = 0.02) and body length (r = 0.63, P = 0.004), so only body mass, body fat, and location were included in the pool of candidate models for predicting fetal rates and fawn survival. The model containing all 3 variables had the greatest relative weight of evidence in the data (wi = 0.85) for predicting whether pregnant females had single or twin fetuses (Table 2). Twin fetuses were more common in CPWA than RWA, and the probability of having twins increased with body fat, but decreased with body mass. Similarly, the model containing only body fat of females had the most support in the data (wi = 0.26) for predicting whether females had at least 1 fawn surviving until fall, but the intercept-only model contained nearly the same relative weight of evidence (wi = 0.19, Δc = 0.53). The intercept-only model had the greatest relative weight of evidence of the candidate models for predicting the number of fawns a female had surviving in the fall (Table 2).

View this table:
Table 2

Weight of evidence for general linear models predicting fetal rates, and whether 27 female mule deer (Odocoileus hemionus) had fawns and number of fawns surviving from near birth until fall in 2 study sites in the channeled scablands of eastern Washington, 2003, based on their body mass (kg) and body fat (%). Test statistics include Akaike's information criterion corrected for small sample sizes (AICC), AICCdifference (Δc), and Akaike weight (wi).

Twin fetusesFawn(s) surviving to fallNumber of fawns in fall
Site × mass × fat23.770.000.8640.293.230.0555.003.090.05
Site × mass28.014.240.1040.703.640.0454.792.880.05
Study site31.878.100.0239.031.970.1052.080.170.20
Body mass (kg)33.229.450.0138.861.800.1153.531.620.10
Site × fat33.539.760.0138.331.270.1453.061.150.12
Body fat (%)34.6610.890.0037.060.000.2652.260.350.18
Mass × fat34.6810.910.0038.671.610.1254.092.180.07

Across the 7 levels of adult and yearling survival explored, our population model indicated that a wild population in which only twins were born had an average λ = 1.000, whereas a population having only singletons had an average λ = 0.9660. Therefore, all else being equal, twinning would be expected to increase the population 4% faster than producing singletons.


Nearly one-half of the newborn mule deer fawns captured in the channeled scablands of eastern Washington died within their first 1.5 months of life, primarily from predation by coyotes. However, if a fawn survived this early period (61.3% survival), it had a greater chance (79.0%) of surviving the rest of the summer. Survival rates of neonatal mule deer fawns in other locations have shown similar survival rates over the 1st summer (Table 3). Coyote predation accounted for more deaths (58.3%) than any other single cause of mortality of mule deer fawns identified in our study area, intermediate to the rates observed in Montana (90%) and Colorado (22%) during summer (Table 3). We were unable to determine the cause of mortality for 25% of dead fawns because of inconclusive clues at the mortality site, thus we may have underestimated the extent of coyote predation.

View this table:
Table 3

Finite survival rate during the age interval, mortality rate (mortality %, no. dead/total no. at beginning of interval × 100), and percent of deaths attributed to coyotes (% coyote mortality) and all predators (% predator mortality) in mule deer fawns (Odocoileus hemionus) in the western United States.

Age (months)LocationnNo. yearsYearsSurvival rateMortality %% coyote mortality% predator mortalityReference
0–1.5Arizona7171Fox and Krausman 1994
Colorado23041999–200142Pojar and Bowden 2004
Colorado10621991–19920.6613279Whittaker and Lindzey 1999
Oregon27891971–1979273858Trainer et al. 1981
Washington30120030.613385858This study
1.5–3.5Oregon18481971–19796Trainer et al. 1981
Washington18120030.792255050This study
0–3.5Montana9161976–19812290Hamlin et al. 1984
Washington30120030.516525656This study
Utah101951–196024Robinette et al. 1977
Colorado14132002–20040.52860–70Bishop 2007
Colorado11932002–20050.42860–70Bishop 2007

Although we did not detect starvation as a proximate cause of death for any of our fawns, we cannot rule out the possibility that inadequate nutrition was ultimately responsible for some unknown deaths or deaths from predation. In a much larger study, Bishop (2007) reported that 20–30% of neonatal fawns died from starvation or disease. Inadequate nutrition of fawns or females may compromise the ability of fawns to escape predators. For example, fawns receiving inadequate nutrition may be weaker (Bartmann et al. 1992; Boyce et al. 1999; Tollefson 2007; Tveraa et al. 2003), receive less care from their mothers (Landete-Castillejos et al. 2001; Tollefson 2007), or may be more conspicuous as they search for and call out to their mothers (Carl and Robbins 1988; T. L. Johnstone-Yellin, pers. obs.).

Births of fawns peaked on 2 June, similar to fawns in Hanford, Washington (29 May ± 6 days—Steigers and Flinders 1980) and Richland, Washington (30 May ± 9.1 days—Hedlund and Rogers 1976). Mortalities of fawns in our study area subsequently peaked at the end of June, a month after birth, whereas mortalities of fawns in Colorado (Bishop 2007; Pojar and Bowden 2004), Oregon (Trainer 1975), Montana (Hamlin et al. 1984), and south-central Washington (Steigers and Flinders 1980) were concentrated toward the end of the first 2 months of life. Mule deer are 1 of the 4 temperate cervid species with a relatively long neonatal hiding phase (Linnell and Andersen 1998). Fawns may be more vulnerable at about 30–45 days old because they are old enough to flush from hiding, but not yet strong enough to outrun predators (Lingle 2000). As fawns age, the average distance they move increases (Riley and Dood 1984), following dams more closely (Jackson et al. 1972; Riley and Dood 1984; Steigers and Flinders 1980) or spending time foraging (Sadleir 1980). With increased mobility comes a higher risk of predators locating and preying upon fawns. Deaths of fawns also may coincide with maturation of juvenile coyotes and the formation of familial hunting (Steigers and Flinders 1980). Newborn fawns create an irruption of vulnerable prey, and coyotes seem to require time each fawning season to develop a search image and hunting technique that allows them to recognize quickly and exploit fawns as a resource (Lingle 2000; Testa 2002; Whittaker and Lindzey 1999). In addition, male fawns in our study had a mortality rate almost twice that of female fawns. In many populations, male cervids grow faster and have a greater nutritional requirement than females (Clutton-Brock et al. 1985; Tollefson 2007). Therefore, male fawns may become more active and visible to predators sooner than females.

Twin mule deer fawns had a risk of dying 2.6 times higher, especially during the first 1.5 months of life, than singletons, and single fawns had a mean survival time 3 times longer than twin fawns. Likewise, Robinette et al. (1957) found increased mortality of mule deer triplets and twins than singletons from birth to fall. In contrast, singleton and twin moose calves tend to have equal chances of surviving (Bertram and Vivion 2002; Osborne et al. 1991; Testa et al. 2000). Unlike moose, however, mule deer rely on a hiding phase for protection from predators. Twin mule deer fawns may suffer higher mortality from predation because they tend to be more ostentatious than singletons, thus increasing their visibility to predators (Linnell and Andersen 1998; Riley and Dood 1984). In large, multiyear studies with neonatal moose (19 pairs—Testa et al. 2000) and roe deer (79 sets of twins and 40 sets of triplets—Gaillard et al. 1998), siblings tended to survive or die together more than expected by chance, especially those in poorer habitats and during the first 15 days after birth. However, we were unable to detect family effects in our smaller sample of 36 deer families. Likewise, in a much larger data set of 270 neonates and 178 families, Bishop (2007) found only a slight interdependence of fates of siblings (overdispersion = 1.25). Because fawns normally bed apart from each other, they may be protected from being killed by the same predator.

Many studies have reported links between nutrition and reproduction in wild ungulates (Cook et al. 2004; Côté and Festa-Bianchet 2001; Julander et al. 1961; Tveraa et al. 2003).

Our study with mule deer provides further support for these links. Body fat of the female mule deer in this study measured in March before the 3rd trimester of pregnancy was included in the most-parsimonious models that predicted whether a female had twin fetuses and whether they had at least 1 fawn surviving until fall. However, our Δc values (Table 2) suggest that the relationship between body fat of females and survival of fawns in our data is relatively weak. We may have lacked adequate statistical power in our sample size of 27 adult females that we were able to monitor until fall, which also were uniformly low in body condition (i.e., 23 of 27 had between 6% and 7% body fat). On the other hand, nutritional condition of females before the 3rd trimester may not be the most sensitive or critical point at which survival of neonates is influenced or predicted. Mule deer are small and polytocous like roe deer (Andersen et al. 2000), and thus may rely more on energy consumed during late pregnancy and lactation than stored fat for financing reproduction, falling more toward the income side of the capital-income breeder continuum (Stearns 1992) than do larger monotocous ungulates. Likewise, Bishop (2007) found that survival of fawns was not greatly improved when free-ranging female mule deer were given a nutritional supplement during early-mid-pregnancy in winter.

Instead, nutrition of females during the last trimester, when the fetus is growing most rapidly, or during lactation may have a greater influence on fawn survival than condition earlier in pregnancy (Armstrong 1950; Pekins et al. 1998). For example, Mander et al. (1961) found more mule deer twins in summer habitats containing more-nutritious forage. In addition, female reindeer (Tveraa et al. 2003) and female mountain goats (Côté and Festa-Bianchet 2001) that were heavier near parturition gave birth to heavier neonates. Following parturition, growth of fawns depends on the amount and quality of milk obtained by fawns (Cook et al. 2004; Robbins and Robbins 1979; Tollefson 2007), which may depend on the quality and quantity of forage available during lactation (Loudon and Kay 1984) more than the body condition of females during pregnancy. Fawns that grow poorly may be more susceptible to predation and have less chance of surviving the harsh effects of winter (Cook et al. 2004; Gerhart et al. 1996; Mackie et al. 1998). Clearly, monitoring condition and intake of females and fawns during the last trimester and throughout the summer is critical to establishing the role of condition of females on survival of neonates.

Despite the fact that fawns born as twins were 2–3 times more likely to die than fawns born as singletons, a litter size of 2 maximized the reproductive success of females in our population. Our results agree with those of Gill et al. (2001), who found that twin mule deer had a higher mortality rate than singletons, but that recruitment rate increased with litter size. Although females invest 1.6 times more energy to produce twins than singletons (Carl and Robbins 1988; Mauget et al. 1999; Oftedal 1985; Pekins et al. 1998; Sadleir 1980), our model showed that the lower survival of twins in our study produced a growth rate 4% higher than populations that produced all singletons. Elasticities (the proportional change in lambda resulting from a proportional change in the population parameter) can be added together to predict the joint effect of changes in multiple population parameters (Caswell 2001; Mills 2007). Although the summed elasticities of adult survival are greater than those of fawn survival, no single parameter within our population models shows a higher elasticity than survival from fawn to yearling. By combining elasticity and annual variation of vital rates, recent studies have shown the importance of the youngest age classes in ungulate population growth (Gaillard et al. 2000; Gaillard and Yoccoz 2003; Raithel et al. 2007). However, if raising twin fawns reduces survival or productivity of females in the subsequent year, the benefits of producing twins would be reduced. Likewise, if predation is additive with starvation or disease in neonatal mule deer, then populations in areas with high predation rates on fawns might find smaller litter sizes adaptive. However, if resources are abundant and predation is minimal, the advantage of producing twins or even triplets would be greater.


We thank N. Thines, S. McCusker, D. Williams, J. Kujala, J. McCanna, G. Paulsen, and volunteers from the Inland Northwest Wildlife Council of Spokane, Washington, for help capturing fawns. D. Parker, of Northern Air, Inc., Bonners Ferry, Idaho, was instrumental in telemetry flights for fawn captures. J. Gaillard provided helpful comments on an earlier draft of this manuscript. This study was funded by Bonneville Power Administration and Washington Department of Fish and Wildlife.


  • Associate Editor was Martin B. Main.

Literature Cited

View Abstract