Distribution
The Serval occurs widely through sub-Saharan Africa, with the exception of tropical rainforests and deserts (Nowell & Jackson 1996). Within recent years there are new records of Servals, implying an expanding population that is recolonising areas, such as Gabon, eastern Central African Republic, southwestern Uganda and central Namibia (Thiel 2015), and within the assessment region, the provinces of the Free State, North West and Northern Cape have been recolonised (Hermann et al. 2008; Power 2014). It is common within wetland habitats of the Drakensberg Midlands, and rare in the lowland wetlands of South Africa (Ramesh & Downs 2013; Ramesh et al. 2016).
In the latter half of the 20th century, the Serval was considered extinct, or near extinct, in many areas of its historical range (Stuart 1985; Smithers 1986; Skinner & Chimimba 2005; Hermann et al. 2008; Thiel 2015), primarily due to accidental persecution as a damage-causing animal and competitive exclusion by other carnivore species (Stuart 1985). For example, it is thought the Serval occurred historically in the eastern Free State (Hunter & Bowland 2013) and as far west as the Cape Peninsula (Skead 2011). Skead (2007) reported that Servals historically occurred along the entire coastal and sub-coastal belt of the Eastern Cape, and that they were nearly extinct in that province in 1987. In the early 2000s, Servals were reintroduced into Shamwari and Kwanwde Private Game Reserves in the Eastern Cape (Hayward et al. 2007). Although they are rare, they are also currently present in the following properties in the Eastern Cape (D. Peinke pers. comm. 2015): Amakhala, Lalibella, Hopewell, Kariega Park, Samara, and Mount Camdeboo. They have also recently been recorded from the Western Cape (Figure 1). With no known reintroductions having taken place, these observations either represent greater sampling effort that revealed an already existing low-density subpopulation or a range expansion from the Eastern Cape or undetected sources in the southern Northern Cape. Such hypotheses need to be researched. It is not impossible too that undisclosed introductions took place from captive facilities.
In the previous assessment (Friedmann & Daly 2004), it was thought that Servals were restricted to the wetter parts of the country, including the provinces of KwaZulu-Natal, Mpumalanga, Limpopo and the northeastern parts of the North West. However, a number of older records from central South Africa were not included in the speciesâ distribution range (Friedmann & Daly 2004; Skinner & Chimimba 2005), as they were considered to be very rare in this part of the country. These records include four pre-1980 specimens from the eastern Free State (Lynch 1983), four sight records from the northern parts of the Eastern Cape (Lynch 1989), as well as a report that this species occurred within what is now the western part of the North West (Stuart 1981). However, more recent records from the Free State provincial conservation authorities, as well as personal observations, suggest that Servals have re-colonised and bred successfully in the western Free State and the eastern boundary of the Northern Cape (Hermann et al. 2008; Figure 1). These sightings are from vleis or riverine vegetation and suggest that Servals use the drainage lines as corridors for movement through the drylands of the Free State. As such, this is likely to be a genuine range expansion as there is no evidence of any ad hoc reintroductions. This expansion was most likely facilitated by the increase in man-made habitats, such as impoundments, weirs and dams, that have enhanced vlei-type habitats favourable to Servals (such as reeds and other dense vegetation that support their main prey species). Reduced threat levels from larger predators or domestic dogs and the adoption of holistic control methods for damage-causing animals or increased landowner tolerance for this species (Hermann et al. 2008) may also have facilitated range expansion. If true, then similar to Greater Cane Rats (Thryonomys swinderianus) (van der Merwe & Avenant 2004), Servals may continue to expand their range into other areas of South Africa. Encouragingly, most specimens were collected in the vicinity of perennial and non-perennial rivers or dam swinderianus) (van der Merwe & Avenant 2004), Servals may continue to expand their range into other areas of South Africa. Encouragingly, most specimens were collected in the vicinity of perennial and non-perennial rivers or dams, which are the preferred habitats of their main prey species, in landscapes that would otherwise be considered unsuitable for Servals (Hermann et al. 2008).
Similarly, within the North West Province, they have expanded their range westwards by at least double since 1983 (Thorn et al. 2011; Power 2014), but have not been officially confirmed further west than the 25° meridian (Power 2014). In the protected areas, they were most common at Vaalkop Dam Nature Reserve, and, interestingly, the species did not even appear on any earlier checklists at Pilanesberg or Borakalalo (for example, Newbery 1995), although they are present in these areas (Power 2014). Reintroductions to these parks, as well as Madikwe Game Reserve and Kgaswane Mountain Reserve, took place in the 1980sâ1990s (van Aarde & Skinner 1986). It is hypothesised that the above average rainfall years that have been experienced since 2000 have led to favourable habitat throughout (for example, dense grass, rejuvenated marshes) (Power 2014). This subsequent increase in both extent of occurrence (EOO) and area of occupancy (AOO), probably due to the creation of artificial water-bodies and a subsequent increase in prey, may indicate continuing recovery from historical persecution and habitat loss (Thorn et al. 2011).
Within KwaZulu-Natal, the population appears stable, especially as densities are comparable across a range of farming intensities (Ramesh & Downs 2013), as long as there are sufficient wetlands with natural vegetation evenly spaced within the landscape (Ramesh et al. 2015a). Although no previous records exist, an individual was seen at the Rooikop waterhole in the dry bed of the Nossob River in the Kgalagadi Transfrontier Park (T. Jackson pers. obs. 1990). They do not occur in the arid parts of the Nama or Succulent Karoo Biomes (Skinner & Chimimba 2005). The species also occurs throughout Swaziland (Monadjem 1998; Skinner & Chimimba 2005), and, although listed as a species predicted to occur in Lesotho lowlands, was not recorded by Lynch (1994). However, they have recently been caught on camera traps in the Lesotho Highlands (A. Jones unpubl. data).
At larger spatial scales, mesocarnivores like Serval are considered generalists, but at finer scales, Servals are habitat specialists in fragmented landscapes (Ramesh et al. 2015a). So they may be locally restricted to smaller areas within their broad distribution range, and conservation planners should integrate these scale considerations into regional management plans. To incorporate the fine-scale habitat selection and wetland spatial configuration into our estimate of AOO, we used the wetland cluster layer from Driver et al. (2012) to calculate a minimum area of occupancy for Serval across its range. The wetland cluster layer represents clusters of at least three wetlands (within 1 km of each other) embedded within relatively natural areas (50% natural area or more â compared to artificial wetlands). We used maximum Serval home range size (60 km2; Ramesh et al. 2015a) to buffer current (post-2000) distribution points using the equation:
where r is the radius distance and A is the home range area, to generate a buffer radius of 4.4 km. This buffer layer was then overlaid onto the wetland clusters (buffered by 500 m) and any clusters containing Serval home ranges were summed. This yielded 10,426â11,144 km2 of wetland cluster areas overlapping with current Serval records. This is likely to be a significant underestimate in minimum AOO for viable Serval subpopulations as there are false negatives in the Serval distribution data and many occurrence records fall outside of the cluster areas, which indicates Servals can make use of smaller wetland areas. Thus, within the total EOO (9,762,333 km2), there is an estimated 18,789 km2 wetland cluster area in total (not clipped to current Serval sightings). Furthermore, summing all buffered wetlands (small waterbodies and artificial waterbodies, not just wetland clusters) within the EOO yields a total AOO of 209,738 km2 within the Servalâs range. These AOO estimates represent a potentially useful baseline of core habitat occupancy for this species. Further research is required to refine these. For example, Ramesh et al. (2015a) use a 20 m buffer from the boundary line of each wetland to define core areas.
Population trend
Trend
The Serval is commonly recorded from most major national parks and reserves within the assessment region. From elsewhere in Africa, the minimum density of Servals in optimal habitat in Ngorongoro Crater was 0.42 animal / km² and 0.1 animal / km² in Luambe National Park in Zambia (Thiel 2015). Their status outside protected areas is uncertain, but they are inconspicuous and may be common in suitable habitat as they are tolerant of farming practices, provided there is cover and food available (Bowland 1990; Hunter & Bowland 2013; Ramesh & Downs 2013; Thiel 2015). Indeed, Bowland (1990) estimated 8 individuals / 100 km² in the Drakensberg foothills, while recently in the same area, a similar density of 6.2 ± 1.9 to 7.7 ± 1.6 individuals / 100 km² was recorded on farmland (Ramesh & Downs 2013), which perhaps indicates a stable subpopulation over the past 20 years. Additionally, ongoing work at the Sasol refinery site in Secunda (c. 3000 ha), a site with very disturbed artificial wetlands, but affording a good source of rodents, shelter, and protection (from the industry itself as well as persecution, and other predators), shows that there are 45 individuals thus far (based on markârecapture camera trap survey), which equates to 1.5 individuals / km2 (Matthews et al. 2014, 2015). This is an unusually high density for a medium-sized felid. There are also indications that the Sasol coal mining site, Syferfontien, (approx. 30 km away from the Secunda site) also has a healthy Serval subpopulation (W. Matthews unpubl. data). This very high density may not be the norm, but does indicate that, where Servals are protected, they can thrive.
However, this high density is not the norm for many game reserves, farms and agricultural areas. The occurrence of Servals in farmlands is probably enhanced by the relatively low density of large or competing carnivores such as Caracal and African Wildcat (Felis silvestris). Finally, the new records from the western Free State and eastern Northern Cape represent the most western records of Servals in recent years and may represent recolonisation of grassland habitats over the past 20 years in central South Africa (Hermann et al. 2008). Thus, the Serval population may have remained stable over the past two decades (regionally) and may be increasing along the edge of its range (Hermann et al. 2008; Thorn et al. 2011; Power 2014). However, it is difficult to infer a population increase as Servals could simply have been undetected in these areas or have always existed at low density in sub-optimal habitat. Additionally, the recent surge in the use of camera traps in ecological studies may partially explain the increasing number of Serval observations.
Using the minimum and maximum density estimates from the KwaZulu-Natal Midlands (4.3â9.3 individuals / 100 km2) and the AOO estimates, potential population size ranges from 448â1,747 individuals (using wetland clusters) and 9,018â19,506 individuals (using all buffered wetlands, to account for false negatives in the dataset). This corresponds to a range of 240â1,223 mature individuals using a 50â70% mature population structure for the wetland cluster AOO and 4,509â13,654 for all wetlands within the range. An alternative method of estimating population size, by adjusting the density estimate per vegetation type (sensu Mucina & Rutherford 2006) and estimating population size using the proportion of untransformed land across the Servalâs range, yields 10,264 ± 812 individuals in total (R.J. Power unpubl. data). This corresponds to 5,538â6,614 mature individuals. In 2004, the mature population size was estimated at around 2,500 individuals using a 50% mature population structure (Friedmann & Daly 2004). Although the current population estimate is far higher, it is not necessarily reflective of population increase but rather better data (and improved methods of data collection) on Serval distribution and density, especially in habitats once presumed to be unsuitable for the species.
This species can subsist away from wetlands, as long as adequate grass cover and/or ecotonal tree cover is available, and artificial wetlands in dam outflows are utilised. These habitats are being increasingly created on low density eco-estates, wildlife ranches, mine buffer zones and agricultural systems. However, where subdivision of farms is taking place, due to inheritance and deeds transfers, habitat quality is often reduced through alien plant invasion and increases in traffic. Nevertheless, the Serval remains a wetland (and moist grassland) specialist and these habitats are the most threatened ecosystem in South Africa (Driver et al. 2012) with an ongoing loss in pristine and functional wetland areas. The extent to which this is compensated for by increasing artificial wetlands remains to be seen. Although Servals make use of such areas, these may not represent resilient and viable long-term habitats, especially as they are vulnerable to changes in management or land-use. Thus, ongoing wetland habitat loss and degradation, combined with a continuing loss of mature individuals to persecution, trade, road collisions or failed reintroductions, indicates that a decline in numbers or loss of core subpopulations cannot be ruled out.
Generation length has been estimated as 8.3 years (Pacifici et al. 2013), which makes the three-generation window 24.8 years.