Mate choice, reproductive success and inbreeding in white rhinoceros: New insights for conservation management

Abstract Improving our sparse knowledge of the mating and reproductive behaviour of white rhinoceros (Ceratotherium simum Burchell, 1817) is essential for the effective conservation of this iconic species. By combining morphological, physiological and habitat data with paternity assignments of 104 known mother–offspring pairs collected over a period of 13 years, we provide the most comprehensive analysis of the mating system in this species. We show that while the overall mating system was promiscuous, and both males and females produced more offspring when mating with several partners, half of all females with multiple offspring were monogamous. Additionally, we find that mating and reproductive success varied significantly among territorial males in two independent sets of males. In females, however, variation in the mating and the reproductive success was not larger than expected by random demographic fluctuations. Horn size, testosterone metabolite concentration, territory size, habitat openness and the volume of preferred food within the territory did not seem to influence male mating or reproductive success. Moreover, there was no sign of inbreeding avoidance: females tended to mate more frequently with closely related males, and one daughter produced a progeny with her father. The lack of inbreeding avoidance, in combination with the skew in male reproductive success, the partial monogamy in females and the territorial‐based mating system, jeopardizes the already low genetic variation in the species. Considering that the majority of populations are restricted to fenced reserves and private farms, we recommend taking preventive measures that aim to reduce inbreeding in white rhinoceros. A video abstract can be viewed here.


Supplement 1
Transect points: Transect points were located within a square grid of 0.8 minutes length, distributed over the entire study area. The points were located using a GPS, giving a ± 100 m deviation in distance between points. The survey was done in October 1998 and lasted three weeks. During this period rainfall was little (6.9 mm) and no fresh grass was observed.

Supplement 2 Food selection and volume of selected food:
For the food selection study we compared the number of times an individual grass species was available along a feeding path with the number of times this particular grass species has been eaten by a rhinoceros. This study is a preliminary analysis for the study on differences in habitat characteristics and was conducted on females only. For this, we followed the footprints of foraging females (n = 6) with the help of a local game tracker. Once a feeding event has been identified, such as by the number of tracks that were scattered within circles, the location of a feeding patch could be recognized by the characteristic fresh green bite marks on the top of a grass species. We placed a 1m x 1m plot on each feeding patch and identified each individual grass species therein. Once a certain grass species has been eaten within a feeding plot, it was counted as a single feeding event for this grass species.
Additionally, we established the availability of grass species along the feeding path by placing 1m x 1m plot along the feeding path approximately 100m apart from each other. Again all grass species within this plot were identified and the presence of a single species within each plot was counted as a single event. We carefully balanced the number of feeding and non-feeding plots within one tracking session in order to minimize autocorrelation. Data collection was terminated if no feeding occurred within a distance of 1 km, which resulted in a total of 91 feeding plots and 59 non-feeding plots.
To establish whether a particular grass species has been selected during feeding, all feeding events for this grass species were added and divided by the sum of all non-feeding events for this particular grass species. Finally, the grass species with the highest frequency has been used for the comparison of habitat characteristics within male territories.
We estimated the volume of selected grass species within male territories using four 1m x 1m plots placed in the four directions of the compass, ten meters apart from the main transect point. We herewith accounted for heterogeneous vegetation and included a larger area in the transect analysis. Within each plot, we estimated the presence of individual grass species, their cover (Greig-Smith 1957) and average height. We then calculated the mean volume of all plots and all transect points.

Supplement 3
Microsatellite loci: Each locus possessed 2-4 alleles with frequency skewed towards one allele. Only SR63 had equal weighting to the two alleles present at that locus. There was a presence of unique alleles at six of the 11 loci (BR06, DB44, WR32A, WR32F, WR32A, SR54). Four individuals in the parent pool were not typed at SR63 (1 breeding female, 3 males), one male at WR32F and one male at SR262, resulting in a white rhinoceros being genotyped at an average of 98.01% of the 22 potential alleles. We found no evidence of null alleles or stuttering for the final 10 loci used in this study. Mean probability of non-exclusion of the true father given the known mother was 0.333 for the 11 loci, and 0.389 when WR32F was excluded. The highest non-exclusion probability occurred at locus SR262 and the lowest at locus SR281. S1: Summary statistics for the microsatellite loci used in this study. Primer sequence, publication source, the number of alleles per locus (k), the number of samples (N), the observed (H obs ) and expected heterozygosity (H exp ), the polymorphism information content (PIC), non-exclusion probability of the father given the known mother (NE-1P) and the individual genotyping error for 11 microsatellite loci.