M. J. Lawes, School for Environmental Research, Charles Darwin University, Darwin, NT 0909, Australia. E-mail: Michael.Lawes@cdu.edu.au
Synchronous monocarpy in long-lived plants is often associated with pollination by wind, in part because infrequent mass flowering may satiate pollinators. Selfing in synchronous monocarps may provide reproductive assurance but conflict with the benefits of outcrossing, a key evolutionary driver of synchrony. We predicted that animal-pollinated species with synchronous flowering would have unspecialised flowers and attract abundant generalised pollinators, but predictions for selfing and outcrossing frequencies were not obvious. We examined the pollination biology of Isoglossa woodii (Acanthaceae), an insect-pollinated, monocarpic herb that flowers synchronously at 4–7-year intervals. The most frequent visitor to I. woodii flowers was the African honeybee, Apis mellifera adansonii. Hand-pollination failed to enhance seed production, indicating that the pollinators were not saturated. No seed was set in the absence of pollinators. Seed set was similar among selfed and outcrossed flowers, demonstrating a geitonogamous mixed-mating strategy with no direct evidence of preferential outcrossing. Flowers contained four ovules, but most fruits only developed one seed, raising the possibility that preferential outcrossing occurs by post-pollination processes. We argue that a number of the theoretical concerns about geitonogamous selfing as a form of reproductive assurance do not apply to a long-lived synchronous monocarp such as I. woodii.
Most plants flower at the same time as conspecifics, a process that facilitates outcrossing in many species (Augspurger 1981; Stephenson 1982; Burd 1994). Masting is a special case of synchronised flowering and fruiting in which synchrony occurs at supra-annual scales, one of several possible benefits of which may be an economy of scale provided by enhancement of pollination rates (Kelly & Sork 2002; Satake & Iwasa 2002). Synchronised monocarpy in long-lived plants is an even more extreme case of synchronised flowering because the interval between flowering may be exceptionally long (e.g. up to 120 years in some bamboos; Janzen 1976), because an entire generational cohort shares a single opportunity to reproduce successfully, and because there is little or no intra-individual variation in flowering intervals. Supra-annual flowering (= temporal aggregation) often also involves spatial aggregation and even monodominance (Kelly & Sork 2002), particularly in wind-pollinated plants where the benefits of cross-pollination may only be realised at short distances (Koenig & Ashley 2003; Davis et al. 2004; Ghazoul 2005).
The evolution of synchronised supra-annual flowering may be driven by a range of processes, of which facilitation of outcrossing is but one (Kelly 1994; Kelly & Sork 2002). Regardless, this life-history strategy has consequences for the pollination biology of the plants concerned. Amongst animal-pollinated species, supra-annual flowering renders the evolution of specialised pollination syndromes improbable, and the risk of pollinator satiation should favour simple floral structures to attract and exploit a range of generalist pollinators. Monocarps, especially those that are synchronous, are under even greater selection pressure to have a reproductive strategy that is robust to among-year variations in environmental and pollinator (and dispersal or establishment) conditions to avoid total reproductive failure. Trade-offs between display size and outcross rates (Albert et al. 2008) may be overridden by the imperative to maximise reproductive success in a single flowering event by the commitment of all available resources to it (Young & Augspurger 1991).
Although increased outcrossing is a potential advantage of synchronous flowering, we propose that a mixed-mating strategy may be beneficial to synchronous monocarps because inbreeding can provide reproductive assurance. Notwithstanding the abundance of mixed-mating strategies amongst plants with less specialised life histories, theoretical justification for reproductive assurance remains elusive, and pathways for evolution of the strategy unclear (Goodwillie et al. 2005; Johnston et al. 2009). Mixed mating has been demonstrated in an animal-pollinated synchronous monocarp (Sharma et al. 2008), whereas the animal-pollinated masting tree Sorbus aucuparia is an obligate outcrosser (Pías & Guitián 2006). Ghazoul & Satake (2009) argued that production of non-viable selfed seed may be beneficial to masting trees by facilitating satiation of seed predators. Synchronous monocarpy may optimise floral density within stands, a feature that may serve to increase outcrossing rates in plants with a mixed-mating strategy (Karron et al. 1995).
In this study, we investigate the reproductive biology of Isoglossa woodii, a synchronously monocarpic herb that flowers in 4–7-year cycles (Van Steenis 1978). The objectives of the study are to: (i) examine the floral morphology and flowering phenology of I. woodii as it relates to the reproductive ecology of the species; (ii) record floral visitors, observe their behaviour, and determine the mechanism for pollination in I. woodii; and (iii) to determine the breeding system in this species. We interpret our findings in the context of the consequences of synchronous monocarpy for flowering in I. woodii and the debate on reproductive assurance in plants with a mixed-mating strategy.
Materials and methods
Study species and site
Isoglossa woodii (Acanthaceae) is a robust, perennial herb growing to a height of 1–2 m. It is endemic to northeastern South Africa and southeastern Mozambique, and is the dominant understorey plant in Indian Ocean coastal dune forests (Tinley 1985). This study was undertaken at Cape Vidal in the iSimangaliso Wetland Park, KwaZulu-Natal, South Africa (28°16′ S, 32°29′ E), where I. woodii is monodominant in the forest (Griffiths et al. 2007). The mean annual rainfall of this site is approximately 900 mm, spread evenly throughout the year, with a mean temperature of 21.5 °C (Schulze et al. 1997).
The species is synchronously monocarpic, with flowering cycles of 4–7 years (Van Steenis 1978). Following flowering and subsequent death, regeneration is from seed that has no dormancy so that seedlings germinate in a single cohort (Z. Tsvuura, M. E. Griffiths & M. J. Lawes, unpublished results). At Cape Vidal, synchronous flowering of I. woodii occurred over 1000s of hectares in 2000 and again from April to September 2007. Observations and pollination experiments on I. woodii were carried out during 2007.
Isoglossa woodii plants expand laterally from adventitious root suckers (Griffiths et al. 2007), a feature that precludes ready identification of individuals. In this study, we report some results on a per stem basis, selected stems being well spaced to avoid sampling the same plant twice. Destructive sampling 2 years after mass flowering demonstrated that plants may occupy an area of up to 1.0–1.5 m2 with 1–10 stems per individual (M. J. Lawes, unpublished results).
The inflorescences of I. woodii are arranged in racemes. Individual flowers are bilabiate, with white petals marked pink on the lower lip. The two stamens each have two anthers that are adpressed against the upper labium. The single, glabrous stigma sits between the stamens, and the bilocular ovary produces two ovules per locule. Single inflorescences on 50 randomly selected I. woodii stems were monitored throughout the flowering season to identify patterns of development and the number of flowers produced. Thirty flowers were examined and measured under a dissecting microscope with a calibrated ocular micrometer to estimate which insect visitors would be capable of pollen transfer. Measurements were made of: (i) flower length and breadth; (ii) length of the floral tube, style and stamen; (iii) distance between anther tips; and (iv) the distance from the anthers to the lower labium (Fig. 1).
Pollen production was measured in one anther from each of three flowers on 10 plants. Undehisced anthers were placed singly in microcentrifuge tubes containing 100 μl HCl and crushed using a glass rod. The tubes were centrifuged at 1118 g for 10 min, after which the supernatant was removed and the pellet resuspended in 100 μl of a 3:1 solution of lactic acid and glycerol. After the solution had been vortexed for 30 s, three 1-μl subsamples were taken and the number of pollen grains counted under 10× magnification using a compound microscope.
To determine whether stigmas are receptive upon anthesis, we abscised styles (n = 30) of I. woodii at different stages of development (unopened buds and flowers open for 1–10 days), placed them in a 5% hydrogen peroxide solution and monitored the rate of bubble formation (Mattson et al. 1974; Dafni 1992).
Nectaries are present at the base of the floral tubes in I. woodii. Diurnal patterns of nectar availability were assessed as the percentage of open I. woodii flowers in an inflorescence (one inflorescence per stem, n = 30 stems) that contained nectar in each of the early morning, midday, and late afternoon.
The presence, volume and sugar concentration of nectar produced by I. woodii flowers over 24 h was assessed for 10 stems protected by netting to exclude insect visitors. At the end of the 24 h, all open flowers were scored for the presence of nectar. Nectar was extracted from, and volumes measured for, five flowers on each stem using a graduated 5-μl microcapillary tube. Because volumes were very small, for the measurement of sugar concentration, the extracted nectar was pooled to form samples from 10 flowers and concentration measured using a Brix refractometer.
Phenological patterns were monitored at two spatio-temporal scales at Cape Vidal. Twenty stems were tagged and the number of inflorescences and flowers counted on 14 occasions during the 8.5-month flowering period, including five occasions during the 2.5-month initial flowering peak. Six 150-m transects were established and evaluated on five occasions, including three sampling sessions during the initial flowering peak. At 10-m intervals, a distance likely to exceed the spread of clonal individuals, the nearest stem within 2 m was scored for the presence of flowers. On the same stem, we randomly selected three inflorescences and counted the number of flowers per inflorescence.
The number and identity of potential pollinators visiting I. woodii flowers, and the number of flowers they visited, was documented in one 10-min survey in each of 80 randomly placed 0.25 × 0.25-m plots during the period of peak flowering activity. Observations were conducted between 09:40 and 14:00 h, and temperature recorded along with the number of inflorescences.
In a separate exercise, the number of open flowers present and visited per inflorescence, and the distance travelled between inflorescences, was recorded for the primary floral visitor, Apis mellifera adansonii. For this purpose, 45 individuals were tracked for 5 min (n = 9) or until lost from sight if this occurred sooner (n = 36, 145 ± 15 s). As A. m. adansonii sometimes visited a flower more than once, the number of flowers visited sometimes exceeds the number present.
Ten A. m. adansonii were collected in sweep nets over flowering I. woodii. Pollen loads on the top of the head, upper body, lower body and in pollen sacs on the hind legs were collected by swabbing with gelatine stained with fuchsin. Pollen loads were examined at 10× magnification using a compound microscope, grains being counted and identified as I. woodii or other species.
The efficacy of self- and cross-pollination in I. woodii was investigated on plants collected at Cape Vidal and grown in a greenhouse at the University of KwaZulu-Natal in Pietermaritzburg, South Africa. To exclude insects, plants were grown under shadecloth and the surface of greenhouse benches was sprayed with an organophosphate insecticide. Three treatments were applied to flowers on each of six inflorescences on a total of seven plants: self-pollination by hand, cross-pollination by hand, and an unmanipulated control. For each flower, fruit set was assessed 3 weeks later.
The extent and rapidity of germination of pollen was compared on selfed and outcrossed flowers on two plants under the same greenhouse conditions as above. On each plant, six flowers each were selfed and six flowers were cross-pollinated by hand, and half of the styles harvested after 24 and 48 h, respectively. Styles were prepared for examination with a modification of the staining technique described by Martin (1959). Styles were fixed in a formalin, alcohol and acetone mixture (8:1:1) for 24 h, washed with tap water and placed in NaOH solution for 24 h. The NaOH was subsequently removed by rinsing in tap water and the styles transferred to a 1% solution of aniline blue for 8 h before mounting on microscope slides. Pollen tubes were examined at 10× magnification using an Olympus Provin AX70 fluorescent microscope at a wavelength of 365 nm. Pollen was considered to have germinated when the pollen tube extended down the length of the style.
A second experiment was carried out at Cape Vidal to determine whether I. woodii is pollen-limited in its natural habitat. We compared fruit and seed set in 25 pairs of plants in a randomized block design. Plants were paired in space and matched for size and flowering state, but were sufficiently separated (3–5 m) to preclude clonal connections. On one plant of each pair, all flowers on one inflorescence were cross-pollinated by hand to ensure that they received supplemental pollen. Another inflorescence on the same plant was unmanipulated and marked as a control (internal control). On the paired neighbouring plant, one or two inflorescences were selected to serve as a second control (external control) to distinguish effects of pollen limitation from effects of resource shunting between inflorescences. Flowers open on day 1, 4 and 9 of the experiment were included, and fruit set counted as they developed over the following 6 weeks. For each inflorescence, a sub-sample of eight fruits (or all fruits if <8 were available) was dissected to determine the number of seeds per fruit.
A self-compatibility index was calculated as the percentage of flowers that develop into fruits after self-pollination divided by the percentage of flowers that develop into fruits after cross-pollination (Lloyd 1968). With the aim of assessing whether synchronised flowering may satiate the available pollinators whilst controlling for the effects of temperature on pollinator activity, the value of temperature and inflorescence density as predictors of bee visitation rates (flowers visited per inflorescence over 10 min, log10[x + 1] transformed) was assessed using simple and multiple least-squares linear regressions. Data from treatments in the greenhouse pollination experiment were pooled to the level of plants; the resulting randomized block design was analysed using Friedman’s non-parametric anova with post-hoc multiple comparisons following Zar (1984) as the control group had no variance. The response variables for the pollen limitation experiment were fruit set and seed set. Fruit set is the proportion of flowers that set fruit. Seed set was estimated as the proportion of ovules that set seed in flowers that set fruit based on four ovules per flower. In external controls with two inflorescences, the response variable was the mean of the two. The data were analysed with a randomized-block anova for each response variable. Except where stated otherwise, measurements are reported as mean ± 1 SE.
Floral phenology and morphology
Flowering commenced in late April and peaked in late May 2007 (Fig. 2). On 26 May, 91.4% of stems (n = 93) along transects were flowering, with a coincident peak in open flowers per inflorescence of 2.45 ± 0.15. Isoglossa woodii produced 215 ± 24 inflorescences per stem (n = 20) over the course of the flowering season, no new inflorescences being found after 16 June, even in the minor resurgence of flowering in September to November, which involved <10% of stems.
Inflorescences comprised from 10–60 flower buds, the number developing into flowers being 25.8 ± 1.4 (n = 50). Flowering was acropetalous (flowering from the bottom of the inflorescence upward). At a given moment, an inflorescence could contain unopened flower buds, flowers in various stages of development, developing fruits and mature pods. The flowers of I. woodii are shortly tubular (mean 5.32 mm long, about half the length of the flower, Table 1). We found little variation in measurements among flowers (Table 1). Flowers produced an average of 10,404 ± 550 pollen grains (n = 30), yielding a pollen:ovule ratio of 2601 ± 138. Pollen was available immediately upon anthesis and was quickly removed by insects. When flowers first open, the style is straight and the stigma is not receptive; stigmatic activity commenced 2–4 days after the flower opened, once the stigma was recurved outwards (Fig. 1). This demonstrates that the species is protandrous, meaning that the male organs develop before the female organs.
Table 1. Measurements (in mm) of floral morphology in Isoglossa woodii. CV: coefficient of variation.
floral tube length
distance from anther to lower labium
Isoglossa woodii flowers offered a nectar reward at the base of the floral tube, although nectar was present in <6% of flowers, even after exclusion of insect visitors for 24 h (Table 2). In flowers that did contain nectar, production over 24 h was 0.54 ± 0.08 μl (n = 50), the nectar having a sugar concentration of 34.2 ± 7.9% (n = 5).
Table 2. Percentage of Isoglossa woodii flowers containing nectar (mean ± SE) at various times of the day (n = 30 stems per sampling period) and after a 24-h nectar accumulation (n = 10 stems).
% flowers with nectar
0.56 ± 0.56
0.00 ± 0.00
4.22 ± 1.82
24 h accumulation
5.89 ± 1.25
The African honeybee, Apis mellifera adansonii (Hymenoptera), was the only insect visitor to I. woodii flowers during the 80 10-min plot surveys. On average, foraging A. m. adansonii spent 2.6 s per flower and visited 1.51 ± 0.02 flowers per inflorescence (n = 1396; range 1–8). This was 73.1 ± 0.8% of the open flowers on an inflorescence [n = 1319; range 12.5–150%; four records (0.3%) exceeded 100% because individual flowers were visited more than once]. Recorded movements between inflorescences were 19.4 ± 0.9 cm (n = 1362; range 0–600). Other species of bee, butterfly, wasp and fly were noted visiting I. woodii flowers during incidental observations.
Apis m. adansonii collected at I. woodii flowers had high loads of pollen on all parts of the body and especially on the head and pollen sacs, almost all of which was I. woodii pollen (Table 3). The orientation of the stigma in I. woodii flowers ensures that pollen can be transferred from the head of A. m. adansonii to the stigmatic surface (Fig. 3).
Table 3. Mean pollen loads (± 1 SE) on Apis mellifera adansonii individuals collected from Isoglossa woodii stands (n = 10).
number Isoglossa woodii pollen grains
number foreign pollen grains
800 ± 106
0 ± 0.0
49.7 ± 7.7
0 ± 0.0
196.8 ± 65.3
0 ± 0.0
483 ± 143.8
0.2 ± 0.133
Visit rates of A. m. adansonii to plots were assessed across temperatures ranging from 15 to 23 °C, and bees were active throughout this range (Fig. 4a). Temperature accounted for 11.5% of the variance in bee visit rates, rates increasing with increasing temperature (P = 0.001). Inflorescence density accounted for only 0.04% of the variance in visit rates (Fig. 4b; P = 0.86). The combined model (temperature + inflorescence density) performed only marginally better than temperature alone, accounting for 12.6% of the variance.
In the greenhouse pollination experiment, 37% of cross-pollinated flowers and 39% of self-pollinated flowers developed fruits, but selfed plants displayed markedly more variable responses to pollination than did cross-pollinated plants (Fig. 5). There was no fruit development in control flowers (Fig. 5). Differences among treatments were highly significant (Friedman’s anovaχ2 = 9.48, n = 7, df = 2, P = 0.009); post-hoc comparisons demonstrated significant differences only between the controls and each of the treatments (P < 0.001 in both cases). The self-compatibility index was 1.4 ± 0.7 (n = 82 flowers), samples with indices >0.75 being described as self-compatible (Lloyd 1992). For both selfed and cross-pollinated flowers, growth of pollen tubes commenced between 24 and 48 h after pollination, with no obvious difference in speed of growth between selfed and cross-pollinated flowers.
In the pollen limitation experiment, there was no significant difference between the pollen-supplemented inflorescences and the two controls for either fruit set (F2,48 = 0.04, P = 0.96) or seed set (F2,48 = 0.54, P = 0.59). Using treatments × blocks as replicates (n = 75), the proportion of flowers that set fruit was 0.60 ± 0.02 and the proportion of ovules that set seed was 0.48 ± 0.01. Notwithstanding that I. woodii has four ovules per flower, 89% of fruits contained only one seed and none contained more than two seeds (mean of 0.95 ± 0.01, n = 710; 8.3% with 0 seed, 2.8% with 2 seeds).
In addition to being synchronised among years, flowering in Isoglossa woodii was synchronised among plants and among inflorescences within plants. However, flowers within inflorescences were markedly asynchronous, with typically <10% of flowers (2.45/25.8) open at any time. This contrast is likely to reflect the tension between attracting pollinators and the risk of saturating them that underlies the evolution of floral display size in animal-pollinated plants (Harder & Barrett 1995; Benítez-Vieyra et al. 2006). Long-lived synchronous monocarps that are animal-pollinated, as in I. woodii, are a special case in that the tension is acted out with unusual strength at the scale of the group as well as at that of the individual.
Notwithstanding that I. woodii flowered over 1000s of hectares, pollen supplementation in the field failed to increase the quite high proportion (60.5%) of flowers that set fruit. Thus, I. woodii was successful in attracting pollinators without saturating them. It did so by attracting an abundant generalist pollinator, the African honeybee, to the shallowly tubular flowers with a reward mainly of pollen. The rarity of pollen grains other than those of I. woodii on bees caught at I. woodii flowers suggests high fidelity of bees to this plant species. A synchronous monocarp and a masting tree that are animal-pollinated also attracted generalist insect pollinators (Pías & Guitián 2006; Sharma et al. 2008).
A mixed-mating strategy
Isoglossa woodii was self-compatible, having a mixed-mating strategy of the ‘single-flower type’ (sensuGoodwillie et al. 2005). Selfing was geitonogamous (pollination of a flower with pollen from another flower on the same plant). We detected no evidence of preferential outcrossing, though the dramatically different variability among plants in their response to pollination treatments suggests that selection pressure on self-compatibility is weak. Nevertheless, this species displayed a number of characters (acropetally, protandry, abundant production of pollen) that may promote outcrossing. The observed pollen–ovule ratio falls between the ranges for facultative and obligate outcrossing species reported by Cruden (1977) but may be elevated towards the range associated with obligate outcrossing by the provision of pollen as a reward for pollinators. As most flowers produced only one seed from four ovules, it is plausible that preferential outcrossing occurred after pollination (Teixeira et al. 2009), either by differential growth of pollen tubes or selective abortion of selfed ovules. This would greatly increase the chance that the seed will be the product of outcrossing and perhaps simultaneously increase fitness by limiting the division of resources among seeds. Selfed flowers may produce seeds of similar quantity but not quality as that of outcrossed flowers (Young 1982). Future studies on the relative performance of selfed versus outcrossed progeny, as well as the paternity of seeds resulting from mixed pollinations, would be required to confirm the hypotheses advanced.
Geitonogamous selfing may provide I. woodii with reproductive assurance in the face of risks associated with synchronous monocarpy. However, support for the concept of selfing as a form of reproductive assurance for any plant remains equivocal because of the costs involved, and few models have successfully addressed its evolution (Goodwillie et al. 2005). Geitonogamy has been regarded as an accidental cost of the mass display needed to attract pollinators (Lloyd 1992; de Jong et al. 1993; Eckert 2000; Brunet & Sweet 2006a,b), and may also incur pollen- and seed-discounting costs (Goodwillie et al. 2005). However, reservations about the costs of geitonogamy are of reduced relevance to synchronously monocarpic species such as I. woodii for a number of reasons. First, synchrony among individuals facilitates attraction at the group level, providing a potentially beneficial mass display without cost to the individual. We observed acropetally in I. woodii, which would minimise selfing. Second, one of the putative costs of geitonogamy, foregone future outcrossed reproduction (Goodwillie et al. 2005), is not incurred by monocarpic species. Third, investment in reproductive assurance is more likely in monocarpic than polycarpic species because of the enhanced benefits and not merely because of reduced costs. The effect may be further emphasized in long-lived synchronous monocarps by group-level selection (Wilson & Wilson 2007), especially where group members are relatives (Foster et al. 2006). Finally, the costs of pollen discounting may be minimised in I. woodii by the abundant production of pollen and its evident success in attracting pollinators, while the costs of seed discounting may be minimised by post-pollination processes that favour outcrossed seed.
Considerable transfer of pollen to flowers of the same individual is an almost inevitable and substantial consequence of synchronous monocarpy both because of the necessary size of individual floral displays and the requirement for a generalised and thus reliable pollination syndrome. Animal pollination may improve rates of outcross pollination compared to wind pollination – a potential compensation for the risks of pollinator satiation – but is limited by the inevitably generalist nature of the pollinators (Brunet & Sweet 2006b). Our data on the movements of bees between I. woodii flowers demonstrate this consequence and conundrum in that many movements were between flowers within inflorescences or over distances that are mostly unlikely to preclude self-pollination. A possible evolutionary response to this inevitability is to be self-compatible and make use of self-pollinations. Reproductive assurance is a simple and parsimonious explanation for selfing in synchronous monocarps, a possibility that may be enhanced by maximising rather than minimising rates of selfing, thus allowing plants to screen selfed ovules for deleterious alleles (Armbruster & Rogers 2004). An alternative is that selfed seeds may be a sacrificial component of satiation of seed predators (Ghazoul & Satake 2009). Given that the two primary hypotheses for the evolution of supra-annual flowering relate to pollination and satiation of seed predators respectively (Kelly 1994; Kelly & Sork 2002), the drivers of selfing in synchronous monocarps appear worthy of further investigation.
Monocarpy in the Acanthaceae
Monocarpy is particularly prominent among long-lived Acanthaceae, occurring in Acanthopale laxiflora, Aechmanthera spp., Brillantaisia nitens and Mimulopsis solmsii (Young & Augspurger 1991; Struhsaker 1997), in Strobilanthes spp. (Janzen 1976), in Isoglossa spp. (Van Steenis 1978), and possibly also in Stenosiphonium spp. (Carine & Scotland 2000). Many of these are synchronous monocarps. Sharma et al. (2008) reported a generalised bee pollination syndrome and high rates of fertilisation of flowers whether outcrossed or selfed in Strobilanthes kunthianus during a gregarious flowering event. Within Isoglossa, as currently defined, I. hypoestifolia is iteroparous (C. Potgieter, pers. comm.) and, in stark contrast to I. woodii, has a specialised pollination syndrome involving a long-proboscis fly (Potgieter & Edwards 2005). The contrasts available within the family may provide an exceptional opportunity for evaluation of the evolutionary causes and consequences of synchronous monocarpy as a life history strategy in long-lived plants.
We thank Sandy Steenhuisen and Adam Shuttleworth for assistance with fieldwork. We are grateful to Ezemvelo KwaZulu-Natal Wildlife and the iSimangaliso Wetland Park Authority for permission to conduct fieldwork at Cape Vidal. The study was funded by the Andrew W. Mellon Foundation and the National Research Foundation of South Africa (grant number 2053633). Any opinion, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Research Foundation. Field vehicles were provided by the Mazda Wildlife Fund.