Population Ecology

, Volume 52, Issue 1, pp 89–102

Rainfall extremes explain interannual shifts in timing and synchrony of calving in topi and warthog

Authors

    • International Livestock Research Institute (ILRI)
    • Institut fuer Pflanzenbau und GruenlandUniversitaet Hohenheim
  • Hans-Peter Piepho
    • Institut fuer Pflanzenbau und GruenlandUniversitaet Hohenheim
  • Holly T. Dublin
    • Centre for Biodiversity Conservation, c/o South African National Biodiversity InstituteSpecies Survival Commission (SSC), The World Conservation Union (IUCN)
  • Nina Bhola
    • University of Groningen
  • Robin S. Reid
    • International Livestock Research Institute (ILRI)
Original Article

DOI: 10.1007/s10144-009-0163-3

Cite this article as:
Ogutu, J.O., Piepho, H., Dublin, H.T. et al. Popul Ecol (2010) 52: 89. doi:10.1007/s10144-009-0163-3

Abstract

We tested the hypothesis that ungulates time and synchronize births to match gestation and lactation with peak food availability and quality in seasonal environments, using ground counts of topi and warthog conducted over 174 months (July 1989–December 2003) in the Mara–Serengeti ecosystem. During this 15-year period, 2,725 newborn and 45,574 adult female topi and 933 newborn and 7,831 adult warthogs were recorded. Births were distinctly synchronized in both species but far less so than in ungulates in temperate regions. Extreme droughts delayed onset and reduced synchrony of calving and natality rates but high rainfall advanced onset and increased synchrony of calving and natality rates in both species, supporting the seasonality hypothesis. Annual shifts in birth peaks were significantly negatively correlated with the preceding wet season rainfall. Varying the timing and synchrony of births and natality rates are widespread but little understood adaptations of ungulates to climatic extremes. Climate change heightens the need for advancing this understanding because increasing frequency and severity of droughts is likely to decouple phenology of breeding in seasonally breeding ungulates from that in their food plants. Similar studies of African ungulates are either extremely rare or non-existent. New approaches to estimating the time of peak births and its confidence limits and the degree of synchrony of breeding are also presented.

Keywords

BreedingDroughtsFloodsMara–Serengeti ecosystemPhenologyUngulates

Introduction

Timing and synchrony of births are important components of fitness for animals inhabiting seasonally variable environments. In mammals, the timing of reproduction is determined primarily by resource availability during gestation and lactation (Millar 1977; Oftedal 1984), whereas breeding synchrony, especially among gregarious, placental mammals is by adjustment of oestrous and not births (Berger 1992). Two hypotheses have been proposed to explain birth synchrony in ungulates. The first states that birth synchrony in ungulates is an adaptation to seasonality in weather and resource supplies (Ims 1990a) and predicts that ungulates match their breeding with the period of peak food availability and plant quality to meet the high energy demands of gestation and lactation (Clutton-Brock et al. 1989). The second hypothesis states that birth synchrony is an adaptation to predation on neonates (Estes 1976; Rutberg 1984; Testa 2002) and predicts that ungulates synchronize birth peaks to reduce predation on newborns through predator glutting (satiation) and confusion (Estes 1976). These hypotheses are not necessarily mutually exclusive since seasonality in resources and predation on neonates may interactively determine timing and synchrony of births in ungulates (Sinclair et al. 2000; Moe et al. 2007).

The role of seasonality and predation in shaping the timing and synchrony of births has been extensively investigated in north temperate ungulates but relatively few similar studies have thus far been undertaken in African savannas. Studies of northern ungulates suggest that seasonal variation in weather or productivity provides a better explanation of timing and synchrony of parturition than the predation hypothesis in Dall’s sheep (Ovis dalli) (Rutberg 1984; Rachlow and Bowyer 1991), bighorn sheep (Ovis canadensis) (Festa-Bianchet 1988), bison (Bison bison) (Green and Rothstein 1993), roe deer (Capreolus capreolus) (Linnell and Andersen 1998), and caribou (Rangifer tarandus) (Post et al. 2003). Although weather, nutritional condition, and social environment determine the onset of mating, body condition is the primary determinant of the onset of the rut in ungulates (Sadleir 1969; Estes 1976). Nutritional restriction apparently delays oestrus in sheep and cattle (Verme 1965; Sadleir 1969), oestrus and ovulation in elk (Cervus elaphus nelsoni) (Cook et al. 2001) and red deer (Cervus elaphus) (Langvatn et al. 2004), and increases early embryonic mortality in white-tailed deer (Odocoileus virginianus, Teer et al. 1965), moose (Alces alces, Testa and Adams 1998), caribou (Russell et al. 1998), and domestic livestock (Ayalon 1978).

Ungulates exhibit substantial plasticity in gestation length, which is ultimately constrained physiologically, implying that timing of mating results from selection acting primarily upon timing of birth (Kiltie 1982). Ungulates can vary the mean gestation time depending on their species, nutritional status, or age. Poor nutrition lengthens mean gestation time in white-tailed deer (Verme 1965), horses (Howell and Rollins 1951), cattle (Hutchinson and MacFarlane 1958), and Dall’s sheep (Rachlow and Bowyer 1991), but shortens it in sheep (Thomson and Thomson 1949; Alexander 1956). Alternatively, ungulates (e.g., bison) in good condition can shorten gestation length whereas those in poor condition cannot (Berger 1992). Also, ungulates such as pronghorn antelope (Antilocapra americana) can prolong gestation depending on age (Byers and Hogg 1995). Gestation may be shortened to trade diminished neonatal growth for improved survivorship by minimizing predation (Berger 1992). However, shortened gestation length results in lighter neonates than those gestated to full term, amplifying the risk of mortality under stressful ecological conditions, e.g., in bison (Berger 1992) and reindeer (Rangifer tarandus; Holland et al. 2006).

In the tropics and subtropics, birth peaks are often timed to coincide with seasonal rainfall and the appearance of green foliage (Talbot and Talbot 1963; Sinclair et al. 2000). In African savannas, seasonality in food availability and quality for grazing ungulates derives from a tight relationship between vegetation production and quality and a markedly seasonal rainfall (Desmukh 1984). Unlike grasses, savanna trees and shrubs exhibit less pronounced seasonality in the availability and quality of leaves and twigs (Rutherford 1980). It follows that timing and synchrony of births should differ among dietary guilds of tropical ungulates. Indeed, patterns of breeding among African ungulates form a continuum ranging from highly seasonal and synchronized to seasonal and uncoordinated breeding, with mixed grazer-browsers and browsers hardly showing seasonal and synchronized breeding patterns (Leuthold and Leuthold 1975; Sinclair et al. 2000). Suggested mechanisms determining the location of specific ungulate species on this continuum include predation on neonates (Estes 1976) and predation plus seasonality in food availability and quality (Sinclair et al. 2000; Moe et al. 2007).

If birth synchrony is a response to seasonal availability of nutrients, then the degree of synchrony should vary from year to year in response to unpredictable patterns of forage growth (Bunnell 1982). Further, if food determines breeding synchrony in ungulates, leading to distributions of births matching the seasonal distribution of food (Rutberg 1987; Sinclair et al. 2000), then extreme climatic events, such as extreme droughts and floods, should significantly shift the timing and synchrony of parturition in seasonally breeding ungulates. Alternatively, if synchrony in births is primarily an antipredator adaptation by ungulates (Rutberg 1987; Ims 1990a), then episodic and unpredictable droughts and floods should not significantly shift the timing and synchrony of births in seasonally breeding ungulates. Here, we examine how droughts and floods shift the timing and synchrony of births in topi, Damaliscus korrigum (Ogilby) and warthog, Pharcocoerus aethiopicus (Pallas), two grazing ungulates with synchronized breeding cycles in the Mara–Serengeti ecosystem of Kenya and Tanzania. In particular, we examine interannual variation in timing, synchrony, and peak calf: female ratios using 15 years of monthly monitoring data.

Materials and methods

Study area

The Masai Mara National Reserve (MMNR; 1°13′–1°45′S, 34°45′-35°25′E) covers 1,530 km2 in southwestern Kenya. It is the northernmost limit of the Mara–Serengeti ecosystem (25,000 km2). It is bounded by the Serengeti National Park to the south, the Siria escarpment to the west and Maasai pastoral ranches to the north and east. The Sand, Talek, and Mara are the three major rivers draining the reserve. Enormous herds of migratory wildebeest (Connochaetes taurinus Burchell), zebra (Equus burchelli Gray) and Thomson’s gazelle (Gazella thomsoni Günther) migrate into and occupy the reserve from July to October (or later) each year.

Animal counts

Monthly changes in herd size, abundance, age and sex composition of seven ungulate species including topi and warthog were monitored in the MMNR by the Maasai Mara Ecological Monitoring Programme (MMEMP) from July 1989 to December 2003 as part of a broader ecological monitoring programme.

The counts were conducted from a vehicle along three strip-transects selected to enable representative coverage and year-round accessibility by car out to a distance of 1 km either side of the transect centerline. The three transects measured 47.4, 144.0, and 71.0 km and sampled 450.4 km2 of the 1,530-km2 reserve. The transects were fixed, except for minor changes in 1996 described in Ogutu et al. (2008), and the same counting procedure used throughout to obtain information on long-term population trends. The surveys were conducted from 0700 to 1500 hours and took four consecutive days to complete. Surveys were not conducted in 17 of the 174 months due to logistical difficulties.

The observers recorded complete groups by date, transect, species, and number of individuals in each age-group by sex in each count. Animals were counted, aged and sexed using binoculars. The vehicle was stopped or driven-off the transect path to obtain accurate counts, age and sex of animals. Combinations of body size, horn length, shape and size, and coat color were used to assign juveniles to age categories, but ages were not assigned to adult animals (Sinclair 1995, pp. 194–219; Ogutu et al. 2008). Individual topi and warthog were assigned to age and sex (topi only) classes as follows. Topi were classified as newborn (<1 month), quarter (<6 months), half (<10 months), yearling (11–18 months) and older (>18 months) animals whereas warthog as newborn (<1 month), quarter (<4 months), half (4–8 months) and older (>8 months) individuals. Newborn topi calves have the umbilical chord still attached but no horns. Three-quarter topi and warthog have their back lower than that of an adult but are nearly full-grown and so were lumped together with adults. Yearling-to-adult topi were sexed but all warthogs were not because sexing warthog proved unreliable under field conditions. Male topi have a larger boss or thicker horns than females.

Rainfall and ENSO

Rainfall was monitored monthly from July 1989 to December 2003 using a network of 15 gauges distributed throughout the study area and averaged over all gauges to capture spatial variability. Rainfall in the Mara–Serengeti ecosystem has two major modes in the wet season and a minor one in the dry season and displays a 5-year oscillation (Ogutu et al. 2007). The early wet season rains (“short rains”) fall during November–December and the late wet season rains (“long rains”) during March–June, and the dry season rains during July–October. January–February, although part of the wet season (November–June), is often dry (Norton-Griffiths et al. 1975; Ogutu et al. 2007).

Fluctuations in the hemispheric El Niño and the Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD; Saji et al. 1999; Webster et al. 1999) influence interannual fluctuations in regional rainfall in equatorial East Africa (Nicholson and Entekhabi 1986) and hence the quantity and quality of food available for ungulates in Mara–Serengeti (Ogutu et al. 2007). The ENSO time series for 1951–2003 downloaded from the website of the Climate Prediction Centre of America (http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml). Analysis of the rainfall time series (Ogutu et al. 2007) revealed mild droughts in 1991 and 1994 and severe droughts in 1993, 1997 and 1999–2000, and an extreme flood in 1997–1998 (Fig. 1). A severe drought also occurred in 2005 due to failure of the short rains (Hastenrath et al. 2007). These extremes in rainfall coincided with the strongest (1997–1998) and longest (1990–1995) ENSO events on instrumental record. The 1997–1998 floods were also associated with an abnormally high activity in the IOD (Saji et al. 1999; Webster et al. 1999). Similar to 1997–1998, a severe drought in 1960 and early 1961 was also followed by the greatest floods since 1914 in late 1961 and early 1962 (Ogutu et al. 2007). The 1961–1962, 1967 and 1994 floods were also associated with unusual activity in the IOD during October–December (Saji et al. 1999; Webster et al. 1999). Ogutu et al. (2007) describe fluctuations in wet and dry season rainfall in the reserve during 1965–2003 (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs10144-009-0163-3/MediaObjects/10144_2009_163_Fig1_HTML.gif
Fig. 1

Standardized wet (solid line and needle) and dry (dashed line and needle and triangle) season rainfall (a) and long rains (March–April–May, solid line and needle) and short rains (October–November–December, dashed line and needle and triangle) (b) in the Maasai Mara National Reserve from 1965 to 2004

Statistical modeling and data analyses

We first interpolated the 17 missing counts of newborn calves and breeding females using a generalized state space mixed model with a seasonal trend component and assuming a Poisson distribution for counts of calves and females (Piepho and Ogutu 2007). We then regressed monthly count totals of newborn calves on quadratic functions of birth month, assuming a Poisson error distribution and a log link function and using the logarithm of the number of breeding females as an offset variable to obtain estimates of the expected ratio of calves to breeding females. Since warthog were not sexed, the number of calves was divided by the count total of all individuals of both sexes in the breeding age class. The regression models were fitted in the SAS procedure GLIMMIX (SAS Institute 2005). Separate models were fitted to counts of newborns for (1) each breeding cycle (July to June), and (2) all breeding cycles combined (1989–2003) to get the expected distribution of birth rates within a year. The time of peak births (timing) and the spread of births around the peak (synchrony) and its standard error were estimated using the method described in Appendix A in the electronic supplementary material (ESM), and the confidence intervals around the time of peak births using the modification of Fieller’s method (Kendall and Stuart 1979, §20.34) described in ESM, Appendix B. Birth rate at the time of peak births was calculated by substituting the estimated time of peak births into the quadratic equation. Shifts in the timing and synchrony of births were related to rainfall by regressing the deviation in time of birth peaks from its long-term mean against the standardized deviation in wet and dry season rainfall, and in early and late wet season rainfall from their respective long-term means. Similar regressions were performed for the degree of synchrony and peak birth rates.

Results

A total of 91,582 topi, including 2,725 newborns and 45,574 adult females and 12,214 warthog, comprising 933 piglets and 7,831 adults of both sexes, were assigned to age classes during 1989–2003.

Phenology

Here, we examine year-to-year variation in the onset of births and birth season length as well as the within-year distribution of births in relation to rainfall. The distribution of births within the year showed peaks in November for both species (Table 1). Topi births started, on average, in mid-August and ended in March, lasting an average of 7.43 months, spanning July–January, whereas 90% of topi calves were born within 5.07 months of the onset of births (Table 2). Similarly, warthog farrowing began, on average, in mid-August but ended 1–2 months earlier spanning January–February and lasted an average of 5.5 months. Also, 90% of warthog piglets were born within 4.3 months of the onset of births. Topi births started as early as July in rainy years, such as 1995–1996, but were delayed in years preceded by droughts. Thus, for example, topi births started 3 months later in November 1994 and 2000 following extreme droughts in 1993 and 1999–2000, respectively (Table 2). The onsets of births were also delayed for warthog but, unlike topi, the delays (lasting 2 months) occurred during and not after severe drought years in 1993, 1999, and 2000. Effective birthing period varied by as much as 3 months for topi and 2 months for warthog, with the longest period for topi recorded during the severe drought of 1993 and that for warthog during the droughts of 1993 and 1997. The monthly distributions of births in each breeding cycle are shown in Fig. 2 for topi and Fig. 3 for warthog.
Table 1

The number of all newborn and adult female topi and all adult warthog recorded in each month in the Maasai Mara National Reserve, summed over 1989–2003

Month

Topi

Warthog

Newborn

Adult females

Newborn per adult female (%)

Newborn

Adults

Newborn per adult (%)

Jul

3

3,263

0.1

3

518

0.6

Aug

12

4,026

0.3

31

584

5.3

Sep

200

2,894

6.9

57

602

9.5

Oct

655

3,213

20.4

313

830

37.7

Nov

906

3,083

29.4

342

843

40.6

Dec

467

2,392

19.5

118

698

16.9

Jan

357

3,757

9.5

43

776

5.5

Feb

92

3,612

2.5

15

656

2.3

Mar

27

4,218

0.6

11

762

1.4

Apr

2

4,547

0.0

0

575

0.0

May

4

5,312

0.1

0

482

0.0

June

0

5,257

0.0

0

507

0.0

Total

2,725

45,574

6.0

933

7,831

11.9

Table 2

The onset, end and length of birth periods for topi and warthog in the Mara reserve during 1989–2003

Topi births

Warthog births

Onset

End

Length

Onset

End

Length

Observed

Effective

Observed

Effective

Aug 1989

Mar 1990

8

4

Sep 1989

Feb 1990

6

4

Aug 1990

May 1991

10

4

Aug 1990

Feb 1991

7

5

Sep 1991

Jan 1992

5

4

Sep 1991

Feb 1992

6

4

Sep 1992

Mar 1993

7

5

Aug 1992

Feb 1993

7

6

Aug 1993

Mar 1994

8

7

Oct 1993

Mar 1994

6

5

Nov 1994

May 1995

7

4

Sep 1994

Jan 1995

5

4

Jul 1995

Feb 1996

8

6

Sep 1995

Jan 1996

5

4

Jul 1996

Feb 1997

8

6

Sep 1996

Mar 1997

7

3

Aug 1997

Feb 1998

7

6

Jul 1997

Jan 1998

7

6

Aug 1998

Mar 1999

8

5

Aug 1998

Nov 1998

4

4

Sep 1999

Mar 2000

7

5

Oct 1999

Dec 1999

3

3

Nov 2000

Feb 2001

4

4

Oct 2000

Jan 2001

4

3

Aug 2001

Mar 2002

8

6

Sep 2001

Jan 2002

5

4

Sep 2002

May 2003

9

5

Aug 2002

Dec 2002

5

5

Observed length is the number of consecutive months over which newborn calves were recorded, whereas effective length is the number of consecutive months over which 90% of calves were born, starting from the month in which the first newborn calf was recorded

https://static-content.springer.com/image/art%3A10.1007%2Fs10144-009-0163-3/MediaObjects/10144_2009_163_Fig2_HTML.gif
Fig. 2

The distribution of the number of newborn calves per 100 female topi of breeding age and the fitted Poisson regression model and the associated 95% confidence bounds. Month 1 is July and 12 is June

https://static-content.springer.com/image/art%3A10.1007%2Fs10144-009-0163-3/MediaObjects/10144_2009_163_Fig3_HTML.gif
Fig. 3

The distribution of the number of newborn piglets per 100 warthogs of breeding age and the fitted Poisson regression model and its 95% confidence limits. Month 1 is July and 12 is June

We classified years into groups showing similar patterns of deviation in timing of births to infer how droughts and floods influenced the timing and synchrony of birthing in topi and warthogs, and then assessed the extent and nature of deviation of birth rates from expectation based on distribution of births over all the 15 years. The timing of birth peaks for both topi and warthog advanced by up to 1.5 months during wet years but were delayed by up to 1.5 months during severe drought years. Topi birth peaks advanced by up to 1.5 months in years of high rainfall, with the earliest peak occurring in 1998 when the greatest ENSO floods occurred. Other notable early birth peaks for topi occurred in years with good rainfall including 1990–1991, 1995–1996, and 2003 (Table 3; Fig. 4a). Topi births peaks were also delayed by up to 1.5 months in severe drought years, including 1993, 1997, and 1999–2000, or in years preceded by severe drought years, such as 1994. Likewise, warthog birth peaks advanced by up to 1.43 months in wet years, for instance in 1998, but were delayed by as much as 1 month in extreme drought years, including 1993, 1997, and 1999–2000, or in years following severe drought years, e.g., 1994 (Table 3; Fig. 4a). Topi and warthog showed similar patterns (Table 3; Fig. 4a–d). For both species, the number of calves per female during birth peaks fell far below average during the severe droughts of 1993 and 1997, when the dry season rainfall failed, and during the record-breaking El Niño floods of 1997–1998. The number of calves per female at the peak birth period was, however, far higher than average in 1990–1992, 1995, 2001, and 2003 (Fig. 4c), when both the dry and wet season rainfall components were average or higher (Ogutu et al. 2007).
Table 3

The estimated linear slope \( (\hat{\beta }), \) quadratic slope \( (\hat{\gamma }), \) month of peak births \( (\hat{\phi }) \) from 1 July, its standard error \( ( {{\text{SE}}(\hat{\phi })}), \) lower \(( {\hat{\phi }_{\text{LCL}} }) \) and upper \(( {\hat{\phi }_{\text{UCL}} }) \) 95% confidence limits and the number of calves per 100 females at the time of peak births (Max) for topi and warthog

Species

Birth year

\( \hat{\beta } \)

\( \hat{\gamma } \)

\( \hat{\phi } \)

\( {\text{SE}}(\hat{\phi }) \)

\( \hat{\phi }_{\text{LCL}} \)

\( \hat{\phi }_{\text{UCL}} \)

\( \hat{\sigma } \)

Max

Topi

1989

4.4

−0.5

4.6

0.5

4.5

4.8

1.0

19.7

1990

5.3

−0.6

4.5

0.5

4.4

4.6

0.9

47.2

1991

6.3

−0.6

5.1

0.7

5.0

5.2

0.9

65.3

1992

6.7

−0.6

5.9

0.5

5.8

6.0

0.9

73.3

1993

2.0

−0.2

5.6

0.6

5.3

5.8

1.7

20.6

1994

4.9

−0.4

6.0

0.6

5.9

6.2

1.1

20.4

1995

5.4

−0.6

4.8

0.5

4.7

4.9

0.9

44.0

1996

6.6

−0.7

5.1

0.8

5.0

5.2

0.9

26.8

1997

5.5

−0.5

5.8

0.9

5.6

6.0

1.0

25.7

1998

2.4

−0.3

3.9

0.5

3.6

4.2

1.3

19.0

1999

4.9

−0.4

6.0

0.9

5.8

6.3

1.1

27.0

2000

8.4

−0.6

6.6

1.2

6.4

6.7

0.9

34.9

2001

4.9

−0.5

5.4

0.8

5.2

5.6

1.1

48.7

2002

2.9

−0.3

4.7

0.4

4.5

4.8

1.3

32.4

All years

3.2

−0.3

5.2

0.1

5.1

5.2

1.3

28.2

Warthog

1989

2.8

−0.3

4.8

0.7

4.5

5.0

1.3

69.2

1990

3.7

−0.4

4.9

0.7

4.7

5.1

1.1

113.8

1991

1.9

−0.3

3.6

0.6

3.3

3.9

1.4

85.8

1992

4.7

−0.5

5.1

1.0

4.9

5.3

1.0

54.1

1993

2.2

−0.2

4.9

1.0

4.3

5.3

1.5

32.0

1994

2.5

−0.2

5.5

0.9

5.1

5.9

1.5

54.2

1995

5.4

−0.6

4.7

1.3

4.4

4.9

0.9

48.9

1996

3.4

−0.4

4.4

1.0

4.0

4.7

1.1

36.3

1997

1.8

−0.2

4.7

1.1

4.0

5.4

1.6

37.6

1998

4.6

−0.6

3.8

1.9

3.3

4.4

0.9

56.3

1999

9.6

−1.0

4.9

4.1

4.6

5.1

0.7

72.6

2000

6.5

−0.6

5.1

1.9

4.8

5.5

0.9

89.8

2001

4.1

−0.4

5.0

1.5

4.4

5.4

1.1

63.0

2002

3.3

−0.4

4.2

1.0

3.7

4.5

1.1

47.8

All years

2.7

−0.3

4.7

0.2

4.6

4.8

1.3

55.9

The larger the absolute value of gamma or the smaller the value of \( \hat{\sigma } \) , the stronger the degree of synchrony of births

https://static-content.springer.com/image/art%3A10.1007%2Fs10144-009-0163-3/MediaObjects/10144_2009_163_Fig4_HTML.gif
Fig. 4

Deviations in the timing of birth peaks for each year from the peak for 1989–2003 and the mean and 95% confidence limits (solid horizontal lines) (a), degree of synchrony for each year divided by the degree of synchrony for 1989–2003 (b), deviation in peak calving rate for each year from the peak rate for 1989–2003 (c), and the relationship between the deviations in the timing of birth peaks and the wet season rainfall normalized by subtracting the mean for 1965–2003 and then dividing the difference by the standard deviation for 1965–2003 (d). For warthog, the deviation for 1991 was outlying relative to the others and thus was deleted before fitting the quadratic model in d

Shifts in birth peaks were clearly associated with amount of the wet season rainfall in the preceding year. For both topi and warthog, the wet season rainfall component was the best predictor of the variation in the timing of birth peaks. Hence deviation in the birth peak from its long-term mean \( (\tilde{\phi }), \) was significantly negatively correlated with the standardized deviation in the wet season rainfall from its long-term mean (wetstd; F1,13 = 23.76, P = 0.0003): \( \tilde{\phi } = 0.027( \pm 0.12) - 0.53( \pm 0.11) \) × wetstd (Fig. 4d). For warthog, the wet season rainfall was also best correlated with deviation in peak births, following the quadratic regression relation (F1,11 = 3.71, P = 0.081, linear coefficient; F1,11 = 13.53, P = 0.0036, quadratic coefficient): \( \tilde{\phi } = 0.38( \pm 0.11) - 0.13( \pm 0.07) \) × wetstd −0.23(±0.062) × wetstd × wetstd (Fig. 4d).

Three other observations of topi in the Mara also noted late onset of births following droughts and early onset of births following high rainfall, implying that plasticity in the timing of births in topi represent an important adaptation to extreme fluctuations in rainfall. Specifically, topi calved well but much later than usual in the Mara during a severe drought in 1953 (GDAR 1954, p. 29). In addition, most topi calves were born a month earlier in September–October instead of the usual time in October–November in the Mara in 1962 following extreme floods in 1961–1962 (GDAR 1962, p. 23). Also, during 2005–2006, we recorded 116 newborn and 346 adult female topi in the Mara during September–October 2005, but only 1 newborn and 377 adult female topi during September–October 2006 due to a severe drought in September–December 2005. The delay in the onset of births in the 2006–2007 calving season was similar to those observed in 1994–1995 and 2000–2001, which were also preceded by droughts.

Synchrony

The distribution of births for topi and warthog were far less synchronized than expected during the severe droughts of 1993–1994 and 1997 (warthog), and the severe floods of 1998 (topi), caused by failure of the dry season rainfall, but far more coordinated than expected during the extreme droughts of 1999 and 2000, caused by failure of the wet season rainfall (Tables 1 and 3; Fig. 4). Synchrony was also high for topi in 1996 with relatively low wet season rainfall. Births were much more synchronized than expected in all years but 1998 and 2002, when the degree of synchrony was average for topi, and in 1993, which is noteworthy as the only year in which births were much less synchronized than expected for both species. Linear regressions of degree of synchrony on the wet, dry and annual rainfall components revealed the strongest relationships with the dry season rainfall that was significant for topi [−0.52 (0.032)−0.088 (0.039) × drystd, F1,13 = 5.70, P = 0.0401] but insignificant for warthog (F1,13 = 0.21, P = 0.656). Hence, low dry season rainfall was associated with decreased degree of coordination of births. Only for topi was the effective length of the breeding period (see Table 1) significantly correlated with the total rainfall during long rains in March–May [7.85 (1.10)–0.008 (0.0032) × MAM]. Timing and synchrony of births were not significantly correlated for either topi (Spearman r = −0.035, 95% LCL = −0.555 and 95% UCL = 0.505, P = 0.904, n = 14 years) or warthog (r = 0.0044, 95% LCL = −0.527 and 95% UCL = 0.534, P = 0.988, n = 14 years).

Discussion

Timing of births

Timing of calving is one of the principal means of modifying calf survival and female fitness in ungulates (Bunnell 1982) and varies geographically with the degree of seasonality in weather and resources. Accordingly, calving in topi in the Mara and Serengeti (Sinclair et al. 2000) is synchronized, but less so than in its congenerics in Southern Africa, with one rainy season (Fairall 1968; Du Plessis 1972; David 1975), but is bimodal in some equatorial populations (Estes and Estes 1979). Timing of breeding in warthog also varies considerably geographically with distinct farrowing seasons reported for parts of Eastern (Child et al. 1968; Leuthold and Leuthold 1975; Boshe 1981; Rodgers 1984; Sinclair et al. 2000) and Southern (Mason 1986) Africa, and bimodal or continuous breeding for Zaire and Congo Braziville, with high rainfall year-round (Brown 1936).

If breeding in ungulates is determined by the quantity and quality of forage (Sadleir 1969), then onset of births should be delayed in dry and advanced in wet relative to an average year. Topi and warthog conformed to this expectation by interanually varying the onset of calving by up to 3–4 months during 1989–2003. This remarkable plasticity in the onset and duration of birthing in both species implies that droughts delay whereas floods advance calving. The annual changes in peak calving dates were correlated with the preceding wet season rainfall, implying that the effect of food on maternal condition is probably the proximal determinant of timing of births.

Delaying births has significant fitness consequences because female ungulates calving late must produce milk from forage of later phenology with higher fibre content and lower digestibility, and because late-born calves have less resources and time for growth (Bunnell 1982) and are therefore more likely to die (Festa-Bianchet 1988). Yet shifting annual onset and duration of calving is widespread among ungulates, implying adaptive benefits. Delaying births during droughts can reduce mortality of expectant females and calves born before sufficient new forage is available. It is, however, impossible to ascertain whether individual topi or warthog delayed or advanced births after conception or simply varied the timing of conception, since our data only show how the whole populations of both species shifted the annual onset and duration of births. For wildebeest in Ngorongoro Crater in Tanzania, the rut of 1965 occurred a month later than in 1964, coincident with a similar delay in the onset of the long rains (Estes 1976). Similarly, poor rains during the La Niña drought of 2000 suppressed conceptions in elephants (Loxodonta africana) in Samburu Game Reserve in Kenya and modified the synchronization of their reproductive activity (Wittemyer et al. 2007). Thus, variation of the timing of conception is a widely used mechanism for coping with climatic variability among native tropical ungulates. Even though it is more obvious why topi and warthog should delay the onset of birthing during drought years, early calving during flood years may not always be an optimal decision, since floods may decrease neonatal survival through increased cost of movement and susceptibility to predation, etc.

Delayed onset and poor calving during severe droughts, due to poor condition of breeding adults in the normal period of rut and pregnancy, and early onset and prolific calving in wet years, due to abundant forage produced by very heavy rains, has also been observed in wildebeest in Nairobi Park of Kenya (Talbot and Talbot 1963) and in Ngorongoro Crater and elsewhere in northern Tanzania following a severe drought in 1960–1961 and subsequent extreme floods in 1961–1962 (Estes 1976). In both cases, the onset of calving was delayed by 1–3 months during the drought and advanced by a similar period following the floods, with the breeding cycle gradually returning to the normal time in the subsequent few years.

Delaying calving also has associated costs. These include the cost of gestation as delayed calving also relates to the condition of breeding females in the subsequent rutting season. As a result, females may trade investing in their current offspring (gestation and/or lactation) with investing in their own maintenance to attain above-average condition during the next rutting season. So, delaying calving too much may actually mean “accepting” to be in below-average condition during the next rutting season. The fact that many females bred unusually earlier in good years (warthog) or in good years following drought years (topi) suggests that some females invested in their own maintenance and were in above-average condition during the subsequent rutting season. If topi and warthog require high quality resources in the wet season before ovulation and forgo reproduction during severe droughts or floods, then there may be a physiological threshold below which conceptions do not occur in both species.

In Southern Africa, impala (Aepyceros melampus (Lichtenstein)) have highly synchronized births, and birth peaks and onset of calving occur earlier and lamb:ewe ratio is high after good rainy seasons, but the lambing period is significantly delayed in extremely dry years (Moe et al. 2007). The onset of lambing and peak lambing time of impala was also significantly negatively correlated with rainfall in the preceding wet season (Moe et al. 2007). Ryan et al. (2007) also noted early onset of calving in South African buffalo, following unusually high rainfall. Scottish red deer (Mitchell and Lincoln 1973) and Alaskan caribou (Adams and Dale 1998) also breed early in years when they are in good condition. Lastly, reproductive phenology in Kenyan elephants was also correlated with resource availability during the season of conception (Wittemyer et al. 2007).

Timing of conception in African savanna ungulates appears determined by rainfall (Child et al. 1968; Boshe 1981) acting through maternal body condition, because high rainfall improves forage quantity and quality (Desmukh 1984). Post et al. (2003) also reported clear synchronization of the timing of parturition by caribou to plant phenology, regardless of predation pressure, suggesting that food sufficiency also governs the timing and spread of births in northern ungulates. Ungulates may vary the onset of births by varying timing of the rut. For example, topi in the Mara–Serengeti rut for about 1.5 months during the long rains in March–May (Bro-Jørgensen 2003) when more than 90% of topi calves are conceived (Bro-Jørgensen 2002). The maximum shift we observed in the timing of peak births of 1.5 months for topi during 1989–2003 matches the period of the rut reported by Bro-Jørgensen (2003) well, indicating rainfall extremes in the year of the “rutting season” preceding the calving year probably shift timing and synchrony of births by shifting timing of the rut and ovulation. Scottish (Mitchell and Lincoln 1973) and Norwegian (Langvatn et al. 2004) red deer hinds in poor condition also tend to conceive late or not at all, implying that poor condition may cause failure to ovulate or conceive. Ungulates can also delay births after conception, but the extent of this delay is constrained by gestation length, which is controlled physiologically (Kiltie 1982), and is typically much less than the interannual variation in the onset of birthing of 3–4 months observed for topi and warthog in Mara–Serengeti.

Topi and warthog are seasonal breeders but calved almost year-round, indicating considerable plasticity in calving time. Such high plasticity may be achieved through manipulation of gestation length or varying the timing of conception. Stalling the labor process to delay parturition by 3–4 months seems rather extreme. Topi have a gestation period of 7.5–8 months, produce one young per birth and wean their young after 4 months. Warthog have a gestation period of 5.7–5.8 months and produce a typical litter of three to four piglets. Conception in both species was dependent on wet season rainfall in the season of conception but timed so that peak calving occurred at the onset of the next wet season rainfall in October–December, 5–8 months later. Consequently, timing of conception in both species likely involves integrating information on experienced current and expected future conditions when calving (Wittemyer et al. 2007).

It is likely that some calves died after birth but before being observed. This is likely in an ecosystem with such a high density of large predators. Severe droughts and floods could also increase calf mortality, so that only a few newborn calves are actually seen during the surveys. Since weather-related survival is age dependent, newborns and juveniles would be more severely affected (Gaillard et al. 1998, 2000). This, however, seems less likely since greater predation on neonates would reduce their ratio relative to breeding females, but not instigate the shifts observed in the onset of births nor the interannual variation in synchrony of births. Furthermore, for herbivores of >50 kg such as topi (weight 75–160 kg) and warthog (weight 50–150 kg), fecundity of young females rather than juvenile survivorship may be the fitness component most sensitive to environmental perturbations (Gaillard et al. 1998, 2000).

Synchrony of births

If gestation plasticity also occurs in topi and warthog then breeding synchrony should be higher following good rainy seasons. As expected, the degree of synchrony of births in topi and warthog was higher in years preceded by good rains. Even so, synchrony of births for topi and warthog was remarkably lower than those reported for some tropical and subtropical ungulates able to produce 74–90% of their calves within 14–60 days, including wildebeest in Mara–Serengeti (Watson 1969; Estes 1976), impala in Botswana (Moe et al. 2007), bontebok (Damaliscus dorcas dorcas) (David 1975) and blesbok (D. d. phillipsi) (Du Plessis 1972) in South Africa. Furthermore, breeding in topi and warthog in Mara–Serengeti was even far less synchronized than in temperate ungulates inhabiting highly seasonal environments, including the Norwegian roe deer (Linnell and Andersen 1998), Scottish red deer (Clutton-Brock et al. 1982), Alaskan moose (Bowyer et al. 1998), caribou (Adams and Dale 1998), American mountain goat (Oreamnos americanus) (Côté and Festa-Bianchet 2001) and pronghorn antelope (Kitchen 1974) that produce 70–90% of their calves within 10–26 days.

Birth synchrony can reduce predation in some mammal species (Estes 1976) but not others (Bowyer et al. 1998; Linnel and Andersen 1998). Estes (1976) showed that wildebeest calves deviating from the birth peak experience greater predation. However, the observed responses to rainfall extremes of topi, warthog, wildebeest (Talbot and Talbot 1963; Estes 1976), buffalo (Ryan et al. 2007), impala (Moe et al. 2007), and elephant (Wittemyer et al. 2007) imply that even though predation may contribute to shaping the timing and synchrony of births in African ungulates, it plays a less influential role than seasonality in resources. Our data thus clearly support the hypothesis that ungulates time their births to match seasonality in resources.

Nevertheless, topi and warthog calve in the dry season when the migratory wildebeest, zebra and Thomson's gazelle occupy the Mara and predation pressure on neonates is lowest, implying that shifts in birth dates during droughts and floods, leading to more births when migrants are absent from the Mara, could heighten neonatal predation. If only specialist predators are likely to shape reproductive synchrony in topi and warthog (Ims 1990b), then, even though they have a large population of large predators in Mara–Serengeti, a predator swamping strategy alone is insufficient to explain their reproductive synchrony because these predators are generalist predators dependent on a variety of other prey species. Neonatal predation may nonetheless cause substantial mortality in both species especially during rainfall extremes when animals are greatly weakened. Moreover, calves born in very dry years are weakened by starvation and become more susceptible to predation. Thus, if changing climate modifies seasonality in rainfall by altering its regularity or timing, it could alter the synchronization between births and plant phenology and amplify neonatal predation.

It is plausible that timing and synchrony of calving may be distinct processes subject to different selective pressures. As suggested by Post (2004), timing of births in ungulates is a trait that is under intense selection, but synchrony of births or its absence may result from selection on timing. This important distinction relates to the fact that timing is an individual trait amenable to selection, unlike synchrony which is a population-level phenomenon and hence harder to envision as under selective forces. Consequently, the magnitude of synchrony in any particular year may be determined by the timing of births in that year, which in turn is influenced by environmental conditions. Nonetheless, the correlation between timing and synchrony of births was insignificant for both topi and warthog.

Birth rates

The high peak birth rates observed for topi and warthog in rainy years and lower rates in drought years indicate that extreme droughts reduce calving rates and/or increase calf and adult mortality. Similar observations have also been reported for wildebeest in Nairobi Park (Talbot and Talbot 1963; Hillman and Hillman 1977) and Mara–Serengeti (Mduma et al. 1999), buffalo in South Africa (Ryan et al. 2007) and Western Uganda (Grimsdell 1973), and elephants in Kenya (Wittemyer et al. 2007).

Shifts in the timing and synchrony of breeding in African ungulates will thus likely be accentuated by the increasing frequency and severity of droughts predicted for continental Africa (Hulme et al. 2001) and already underway in Mara–Serengeti (Ogutu et al. 2007). Such droughts can have profound and enduring legacies on ungulate population dynamics by severely reducing births or survivorship (Talbot and Talbot 1963; Estes 1976; Hillman and Hillman 1977; Mduma et al. 1999).

Rainfall also influences the success of calving through its effect on vegetation. Female fecundity in ungulates, such as wildebeest (Talbot and Talbot 1963), deer (Taber and Dassmann 1958) and domestic livestock (Hart and Guilbert 1933; Miller et al. 1942) appears to be determined primarily by the nutritional status of the animals before and during the rut. Ovulation may be delayed or prevented by a low level of nutrition, particularly by a deficiency of vitamin A and protein (Hart and Guilbert 1933; Miller et al. 1942). Such nutritional deficiency can cause prenatal mortality and the birth of weakened young which may die soon after parturition. Since the only natural source of vitamin A is green vegetation (Buechner 1960, p. 117), which is also high in protein, sufficient rainfall is necessary to assure adequate food before and during the rut. Indeed, even bulls that are nutritionally stressed show damped propensity to mate during climatic extremes. Nutritionally stressed female ungulates tend to breed at a later age than they do under better conditions. This has been demonstrated for black-tailed deer (Odocoileus hemionus) (Taber and Dassmann 1958) and suggested for wildebeest (Talbot and Talbot 1963). Not surprisingly, breeding in ungulates, such as wildebeest, follows the months of most reliable and heavy rainfall, varying interannually in response to rainfall fluctuations (Talbot and Talbot 1963; Watson 1969).

If both the timing and synchrony of birth in ungulates are adaptations to long-term patterns of climate that provide the most hospitable conditions to bear and rear young (Bowyer et al. 1998), then ungulates that exhibit highly synchronized births are likely to be more susceptible to climate change than asynchronously breeding ungulates that are better adapted to high climatic variability. Similarly, ungulates inhabiting highly seasonal environments are likely to be affected more by climatic extremes than those inhabiting less seasonal environments with more flexible onset and duration of calving.

The significant relationship between wet season rainfall and timing and synchrony of calving suggests that births would be delayed, and would involve fewer females and be less synchronized during the decreasing phase and vice versa during the increasing phase of the 10-year cycle evident in equatorial Eastern African rainfall (McHugh 2006), including in the Mara–Serengeti (Ogutu et al. 2007). The dry and wet season rainfall components varied independently in the Mara during 1965–2003 (Spearman: r = 0.0298, P = 0.5420, n = 38 years). Hence, bimodality in rainfall may constrain the ability of seasonally breeding ungulates, such as topi and warthog, to adjust the timing and synchrony of their births to match opposing changes in the wet and dry season or the long and short rainfall components.

Acknowledgments

The Maasai Mara Ecological Monitoring Program was designed and supervised by Dr. Holly T. Dublin and run by Paul Chara (July 1989–1992), John Naiyoma (1989–1993), Alex Obara (1995–1997), and Charles Matankory (1991–2003), and was funded by the World Wide Fund for Nature and Friends of Conservation. This study was also supported by the National Science Foundation (NSF) Grant Nos: BCS 0709671 and DEB-0342820 and a grant from the Belgian government (DGIC BEL011) to ILRI. We thank the Narok County Council for permission to conduct this study and for providing an office space and accommodation for MMEMP staff. BaseCamp Explorer Ltd also provided an office space during the later stages of the monitoring program. The International Livestock Research Institute (ILRI) and the Alexander von Humboldt Foundation (AvH) supported J.O. during preparation of this paper. The Kenya Meteorological Department and Dr. Kay Holekamp provided additional rainfall data for the Mara Reserve.

Supplementary material

10144_2009_163_MOESM1_ESM.pdf (73 kb)
Appendix A (PDF 72 kb)
10144_2009_163_MOESM2_ESM.pdf (25 kb)
Appendix B (PDF 25 kb)

Copyright information

© The Society of Population Ecology and Springer 2009