It has been previously noted that elephant have been reintroduced into several protected areas where vultures nest and that there may be a risk to the persistence of vulture populations if elephant impact gets to the point where the availability of suitable nesting trees may be reduced (e.g. Rushworth 2008). The objectives of this paper are to: (1) assess the potential vulnerability of tree-nesting vulture populations, as resident breeding species in KZN, to elephant impact by quantifying the extent of overlap of elephant populations and vulture nesting sites, (2) evaluate the ability to manage elephant populations by assessing current elephant densities of all KZN elephant populations against preferred densities as defined in the various elephant management plans, thereby indirectly assessing the effectiveness of the National Norms and Standards for the Management of Elephant in South Africa to protect other elements of concern where there are elephant, and (3) to provide management recommendations to reduce the vulnerability of tree-nesting vultures to elephant.

Systematic aerial surveys of tree-nesting vultures in Zululand have taken place biannually since 2004 (Howells & Goodman 2013a, 2013b, 2013c; Howells et al. 2010; Rushworth 2017b). While these surveys have focused on protected areas and known nesting areas, they have targeted other suitable habitat outside protected areas, and have been supplemented by ground-based observations, including inter alia (1) reporting of nests by patrolling rangers, (2) landowner reporting of vulture activities and nests (there have been significant vulture awareness activities conducted with landowners in Zululand) and (3) reporting of nests by avifaunal experts during Environmental Impact Assessment activities. The pilot of the Ezemvelo aircraft routinely surveys for vulture nests during other monitoring or transport flights. Each nest was mapped using a GPS while flying and so is accurate to within 100 m of the true ground position. While it is undoubtable that some nests have not been accounted for, there is a high degree of confidence that more than 95% of tree-nesting vulture nests in the province are known. It is reasonably assumed that the map of nesting sites represents an unbiased assessment of the current distribution of vulture nests in KZN.

All elephant populations in KZN are known and all properties have, or are in the final stages of completing, elephant management plans in terms of the requirements of the National Norms and Standards for the Management of Elephant. Ezemvelo KZN Wildlife (Ezemvelo), as the biodiversity regulatory authority for KZN, approves all elephant management plans for non-state areas and maintains a database of all properties with elephant. Numbers are updated annually based on censuses in Ezemvelo protected areas and reports received by Ezemvelo from other properties. The boundaries of all elephant properties as demarcated in the elephant management plans are kept in GIS format by Ezemvelo. There are some challenges in providing the exact areas available to elephant because the fenced and legal boundaries are not always completely the same; however, the most accurate fenced boundaries available were used in this assessment, captured at a 1:50 000 scale or better. In some cases, natural features make up part of the boundary (e.g. the eastern boundary of the Phongolo Nature Reserve and the Pongola Game Reserve East Association, referred to as the Phongolo/Pongola Complex in this paper, is the Pongolapoort Dam) and these were digitised off aerial photographs or satellite images.

Each property with elephant is required to specify the preferred management density of elephant. In most cases, these have been approved as part of the approval of the elephant management plans; in some cases, these remain as recommendations, while the plan goes through the approval process. In some cases, this is expressed as a density (elephants per km²) and in other cases as a number of elephant, in which case these were converted to densities based on the property size. Where the preferred number was provided as a range, the upper limit of the range was used to calculate preferred density.

We used GIS to compare vulture nest distribution from the 2016–2017 surveys, with boundaries of properties with elephant to quantify the proportion of nesting sites occurring within areas occupied by elephant. In addition, we quantified the proportion of properties currently (2016) stocked above the ‘preferred density’ as defined in the respective elephant management plans. Where properties were still stocked below the preferred density, we calculated the time (years) to reach the preferred density based on the growth rate for each specific population as reported or calculated from annual reports provided to Ezemvelo.

Given that (1) other extraneous factors such as poisoning events influence the number of pairs and/or nesting attempts in any particular year and subsequent years, and (2) there is a lag effect between current elephant density and vegetation response, it is not possible to use this data to determine the relationship between elephant density and vulture nest density or change in density.

Not all areas with elephant have vultures, but the majority of vulture nests occur in areas with elephant. A total of 623 active vulture nests were recorded in KZN in 2016/2017. Only 36 of these nests occurred in areas where no elephant were present, whereas 587 (94.2%) of all vulture nests occur in areas where elephant occur (Figure 1, Table 1). When analysed by species, 100% of Lappet-faced (n = 17), 100% of White-headed (n = 2) and 94.0% of African White-backed Vulture (n = 568) nests co-occur with elephant. However, the imminent dropping of fences between the ‘Big 5’ properties (which have 12 pairs of African White-backed Vultures and no elephant) and the neighbouring Hluhluwe–iMfolozi Park will result in the immigration of elephants, thereby increasing the overall proportion of vulture nests co-occurring with elephant to 96.1%. If a recent proposal for the development of a large open cast coal mine goes ahead in the Thula Thula area, 58 African White-backed Vulture breeding pairs would be displaced, resulting in 99.5% of remaining nests co-occurring with elephant.

The three most important properties for tree-nesting vultures (in terms of total number of vulture nests of all species combined) are the Hluhluwe–iMfolozi Park (73.2% of all vulture nests in KZN), the Phongolo/Pongola Complex (8.8%) and Thula Thula (5.9%). It should be noted that the current number of nests in uMkhuze Game Reserve is significantly reduced compared to the recent past owing to ongoing poisoning events; otherwise, uMkhuze would have been in the top three most important sites and should be regarded as one of the most important sites in the province in terms of efforts to recover vulture populations as per the targets in the provincial vulture conservation strategy (Rushworth 2008). Both the Phongolo/Pongola Complex (second most important) and Thula Thula (third most important) are under threat of land-use change, further emphasising the importance of Hluhluwe–iMfolozi Park and uMkhuze.

The majority of areas, including half of those with vultures, are stocked at densities above the preferred density recorded in the elephant management plans. Fourteen out of 21 properties (66.7%) were already over the preferred or recommended stocking density by 2016 (Figure 2, Table 2), with one property being stocked at almost double the preferred stocking density. Of the remaining seven properties currently below the preferred stocking density, three (Mawana, uMkhuze Game Reserve, KwaZulu Private) will be over the preferred stocking density within 5 years, one in 5–10 years (Hluhluwe–iMfolozi Park), one in 10–15 years (Somkhanda) and one (Thanda/Mduna Royal) will take over 15 years (based on current growth rates, incorporating effects of contraception, and assuming no further introductions) (Table 2). The last (Zulu Nyala) is a female-only population. Five of the eight properties where elephant and vulture nests co-occur are currently stocked above the preferred density specified in the elephant management plans; three are marginally (less than 10%) above, while two (representing 14.7% of all vulture nests) are stocked significantly above the preferred density (Figure 2).

Given that 94.2%, and potentially up to 99.5%, of nests of tree-nesting vultures in KZN occur in areas now occupied by elephant, and the potential for elephant to impact on large trees in general and specifically those selected by vultures, we propose that the way in which elephant and elephant impacts are managed will be an important factor affecting vultures in the medium to long term. While there is no evidence at this stage to suggest that suitable nesting trees are becoming limiting, there is evidence of a reduction in density, and further evidence of a developing recruitment bottleneck, of preferred nesting trees in the most important areas for vultures in KZN. Although it is complex to fully understand the interacting drivers of change, there is evidence that elephant are at least a contributory factor in these vegetation changes. We predict that if an appropriate balance between elephant numbers and vegetation is not achieved in existing areas with elephant, this could conceivably lead, in the long term, to the complete or almost complete loss of tree-nesting vultures as breeding residents from KZN.

Given that many elephant populations are still growing (i.e. the impacts are increasing over time at a range of scales), and that there are significant lag effects in vegetation response to herbivory, there is a need for a concerted effort to model or predict future scenarios of vegetation composition and structure so that management interventions can take place early enough. In particular, planning for elephant management should not consider average conditions, but should explicitly incorporate the impact of elephant during prolonged periods of drought. While several authors have advocated implementing measures to protect individual trees, we believe that this approach is not practical in protected areas and largely fails to recognise that the key issue is allowing for ongoing recruitment into the adult size classes as individual protected trees will eventually senesce. This interventionist approach is also largely incompatible with the process-based management philosophy adopted by conservation agencies and that is entrenched in management plans. A risk-averse approach in relation to setting preferred management densities for elephant, or other thresholds of potential concern, should be adopted in the Hluhluwe–iMfolozi Park given that 73.2% of vulture nests occur in this area.

Two-thirds of properties with elephant, including two of the three most important vulture nesting areas, were stocked above their preferred densities by 2016. Preferred densities of elephant in these confined areas were set in relation to multiple factors, one of which may have been concern over maintaining nesting sites or opportunities for vultures. It is not the intention here to review the logic for the setting of preferred densities or to question whether they were set appropriately for vultures where vultures were listed as an element of concern. However, the key thesis is the seeming inability to manage according to the preferred densities, whether these were set correctly or not. What this could mean for vultures is that even if threshold densities were set specifically and appropriately to allow for the long-term persistence of suitable nesting trees for vultures, based on a good scientific understanding of vegetation dynamics and ecological processes, the socio-political factors related to elephant population reduction may still mean that these thresholds are exceeded. For example, the management authority of Ithala Game Reserve has been trying without success for more than 6 years to find homes for excess elephant, all the time being over the density approved in the elephant management plan (currently stocked at nearly 170% of the preferred density in the elephant management plan).

The failure to manage at preferred densities is probably the result of a combination of factors, notably limited opportunities to translocate live elephants, including the limitations imposed by the Norms and Standards regarding translocation of previously translocated elephants, coupled with a reluctance to invoke the culling option given the continued opposition from some quarters. Interacting with the aforementioned reasons, there is still a level of uncertainty regarding the ecological interactions between browsing, vegetation and other drivers, creating both uncertainty in setting the preferred density and possibly a reluctance to face scrutiny of these estimates in the face of challenges to overturn decisions². The Norms and Standards and elephant management plans have clearly been successful in securing ethical and humane treatment of elephants, but seemingly there is a significant risk in failing to achieve other equally important key purposes of the Norms and Standards which are to ensure that elephant are managed in a way that ‘promotes broader biodiversity and socio-economic goals that are ecologically, socially and economically sustainable’ and ‘does not disrupt the ecological integrity of the ecosystems in which elephants occur’. There is seemingly a risk that short-term elephant welfare considerations may take precedence over long-term ecosystem and biodiversity management considerations, particularly for smaller properties.

The Norms and Standards require a rigorous process of evaluating all possible management options to manage elephant density or impact. Only if all other management options have been attempted and densities or impact are still too high will the culling option be considered. In this case, in addition to the management plan, a separate culling plan has to be prepared, submitted and approved prior to permits being issued. All of these processes can introduce significant lag effects in managing elephant numbers and complicate the adaptive management approach advocated in the same document. The Norms and Standards also prohibit selective culling from within family groups. For small properties that can only hold one family group, and where contraception has not been used or has not contained the population size, this may mean an ‘all or nothing’ approach where the entire family group has to be removed and a new smaller group introduced. Properties in this category are at present choosing to allow the population density to exceed the preferred density.

The National Norms and Standards for the Management of Elephant refer to an adaptive management approach, defined as ‘integrated research, planning and monitoring in repeated cycles of learning in order to better define and achieve objectives, and is built on the assumption that natural extensive wildlife systems are complex, our knowledge is imperfect but we can learn from purposeful, documented objectives and actions’. They go on to state that ‘measures to manage elephants must be informed by the best available scientific information, and where the available scientific information is insufficient, adaptive management forms the cornerstone of the management of elephants and adaptive decision-making tools must be adopted’. What this actually means in practice is unclear. However, in the case of managing for vulture nesting opportunities, we believe that adaptive management should be interpreted as setting relatively conservative elephant stocking densities, and then monitoring the impact of that density on the recruitment and availability of suitable large trees over an appropriate ecologically meaningful time frame. If the impacts are within acceptable limits, the elephant density could be allowed to increase. The process of monitoring and re-assessment is then repeated until an acceptable stocking density that is compatible with all objectives is reached. It may take several decades to then arrive at a final agreed preferred density using this approach. Advances in immunocontraception would allow this approach to be implemented without the need for, or at least significantly reducing the need for, removals. The likelihood of exceeding the acceptable density would be diminished by adopting this approach. It would however require specific emphasis on, and resources allocated to, monitoring. The tendency at present, by and large, seems to be to try and estimate the maximum density of elephant that an area may support and manage towards that target. This creates a significant risk that if the target was set incorrectly, then incompatible changes in habitat, or other impacts on objectives, may have been initiated or may take place prior to the preferred density being reached.

The vulnerability of vultures to elephant impact should be seen in the context of both rapid habitat loss outside protected areas as well as the unsuitability of otherwise suitable habitat owing to anthropogenic disturbance effects. Natural vegetation in KZN is decreasing at 1.2% per annum (Jewitt et al. 2015), while human population densities are increasing at 1.61% per annum (Statistics South Africa 2017). In communal areas, which cover a significant portion of the savanna biome, there is an ongoing loss of larger trees (e.g. see Higgins, Shackleton & Robinson 1999). If vultures are lost from protected areas and other extensive wildlife systems, then there are extremely limited opportunities for them elsewhere. The opportunities to expand or increase the number of protected areas suitable for vulture nesting are also decreasing rapidly, and there are no targets set for protected area expansion in the province. There is a strong push to grow the ‘wildlife economy’ in KZN, and this is likely to translate in the medium term to some increase in extensive wildlife systems, but at the same time also likely to see introduction of elephant to more areas. There is also a growing tendency to subdivide extensive ranches into smaller, intensively managed camps to breed high-value game species (Taylor, Lindsey, Davies-Mostert & Goodman 2015), and this may further reduce the extent of suitable nesting habitat for vultures.

We have shown for KZN that (1) a high proportion of vulture nests occur in areas with elephant, (2) there is evidence of elephant selection for tree species preferred by vultures for nesting, as well as of actual declines in density and recruitment of preferred species, (3) there are challenges in managing elephant populations according to preferred densities and (4) there are limited and reducing opportunities for vultures to use other areas for breeding. We conclude that retaining adequate habitat in protected areas and extensive wildlife systems with the appropriate structural attributes for vulture nesting and foraging is essential. We also conclude that retaining vultures as breeding residents in KZN would be dependent on appropriate ecological management of these areas, including managing the relationship between elephant and their habitat. Loss of vultures from the areas where they currently breed could result in the loss of vultures as breeding residents from KZN, even if other anthropogenic factors influencing vulture mortality could be managed. While many of the anthropogenic threats to vultures are difficult to manage and may largely be outside of the control of conservation programmes, managing elephant impact on vegetation structural attributes and hence the availability of vulture breeding habitat is directly within the control of land managers. It is important to note that the recruitment of trees into adult size classes is affected by the presence of multiple browsers and fire (O’Kane et al. 2011; Staver et al. 2009), and can be slow especially where both processes act together. For example, there has been essentially no regeneration of riverine forest, once the preferred nesting habitat of African White-backed Vultures in the Hluhluwe–iMfolozi Park, more than three decades after Tropical Cyclone Domoina (Staver, Beckett & Graf 2017). While it can be debated whether elephant and vegetation may reach a new stable state or that vegetation may eventually recover in a cyclical process (Scholes & Mennell 2008), in the time taken for this recovery to take place vultures could end up with significantly reduced or no suitable breeding habitat and could largely or completely disappear as breeding residents, with a low probability of eventual population recovery.

An indirect impact of elephant on vultures may be mediated through the competitive release of small shrubs and trees in response to reduced competition from large trees. A reduction in large trees has been associated with an increase in smaller size classes in several areas containing elephant (e.g. Druce et al. 2017; Kalwij et al. 2010). The increasing density of shrubs may reduce foraging ability of vultures, both in terms of carcass detectability and ability to take off after feeding (Bamford, Monadjem & Hardy 2009b; Brown 1985). While vultures can and do forage over large areas, loss of foraging habitat in protected areas and other extensive wildlife systems would result in vultures spending more time foraging in areas where they are more susceptible to mortality from poisons, power lines and direct human persecution.

The recent addition of all three tree-nesting vulture species occurring in KZN to Appendix 1 under the Convention on Migratory Species in 2017 makes the management of vultures, including the management of breeding habitat, not merely a provincial or national issue, but an international obligation.

We propose that the most important management actions to reduce vulnerability of vulture populations are:

Based on what we have observed in KZN, we hypothesise that tree-nesting vultures in the rest of South Africa could be vulnerable to elephant impact in the medium to long term. Loss of suitable breeding habitat in protected areas and extensive wildlife systems would result in significant negative effects on vulture populations. Unlike vegetation that can potentially recover in the medium to long term following perturbations, vulture populations would not be able to persist for the length of time that this recovery would take.