Density responses of lesser-studied carnivores to habitat and management strategies in southern Tanzania's Ruaha-Rungwa landscape.

Compared to emblematic large carnivores, most species of the order Carnivora receive little conservation attention despite increasing anthropogenic pressure and poor understanding of their status across much of their range. We employed systematic camera trapping and spatially explicit capture-recapture modelling to estimate variation in population density of serval, striped hyaena and aardwolf across the mixed-use Ruaha-Rungwa landscape in southern Tanzania. We selected three sites representative of different habitat types, management strategies, and levels of anthropogenic pressure: Ruaha National Park's core tourist area, dominated by Acacia-Commiphora bushlands and thickets; the Park's miombo woodland; and the neighbouring community-run MBOMIPA Wildlife Management Area, also covered in Acacia-Commiphora. The Park's miombo woodlands supported a higher serval density (5.56 [Standard Error = ±2.45] individuals per 100 km2) than either the core tourist area (3.45 [±1.04] individuals per 100 km2) or the Wildlife Management Area (2.08 [±0.74] individuals per 100 km2). Taken together, precipitation, the abundance of apex predators, and the level of anthropogenic pressure likely drive such variation. Striped hyaena were detected only in the Wildlife Management Area and at low density (1.36 [±0.50] individuals per 100 km2), potentially due to the location of the surveyed sites at the edge of the species' global range, high densities of sympatric competitors, and anthropogenic edge effects. Finally, aardwolf were captured in both the Park's core tourist area and the Wildlife Management Area, with a higher density in the Wildlife Management Area (13.25 [±2.48] versus 9.19 [±1.66] individuals per 100 km2), possibly as a result of lower intraguild predation and late fire outbreaks in the area surveyed. By shedding light on three understudied African carnivore species, this study highlights the importance of miombo woodland conservation and community-managed conservation, as well as the value of by-catch camera trap data to improve ecological knowledge of lesser-studied carnivores.

Introduction

Emblematic large carnivores often benefit from significant conservation investment and are prioritised in terms of resource allocation over more threatened species [1]. Such prioritisation stems in part from their essential role in shaping ecosystems [2], but also their charisma and cultural significance, which are used to highlight the problem of biodiversity loss to the public [3, 4]. In comparison, carnivores with smaller body size or less aesthetic appeal tend not to spark as much conservation interest and remain understudied globally [1, 5]. The African conservation landscape conspicuously illustrates these disparities, with iconic apex predators such as lion (Panthera leo), leopard (Panthera pardus) and cheetah (Acinonyx jubatus) drawing substantial research attention [6]. Overshadowed by these flagship species, many lesser-studied carnivores nonetheless face declining trends in number and range [7], with anthropogenic threats including habitat loss/fragmentation [8], and persecution [9] jeopardising their survival alongside a growing human population [10].

This lack of research hinders conservation status assessments for a number of African carnivores, thus precluding effective conservation planning [11], notwithstanding the ecological importance and vulnerability to extinction of many of these species. For instance, striped hyaena (Hyaena hyaena) have received very little conservation attention across their African range, although they provide essential ecosystem services in consuming carcasses and dispersing their nutrients [12]. The species is thought to be declining globally, mainly caused by poisoning and decreasing sources of carrion, leading to its classification as “Near Threatened” [13]. Moreover, little information exists on the population status of many mesocarnivores, despite their key contribution to regulating lower trophic levels and maintaining biodiversity [14]. Classified as “Least Concern”, serval (Leptailurus serval) and aardwolf (Proteles cristata) nevertheless face increasing pressure due to continued urbanisation and agricultural expansion [15, 16]. Serval are particularly vulnerable to the degradation of wetland and grassland habitat, as the species prefers well-watered environments which attract higher densities of small mammals [17]. The insectivorous diet of the aardwolf exposes the species to insecticide poisoning, an ever-increasing threat with the intensification of agriculture [15, 18].

The IUCN recommends further research to bridge knowledge gaps for these lesser-studied carnivore species’ spatial ecology and population status [13, 15, 16]. The distribution of serval, striped hyaena and aardwolf, as well as population abundances and densities across their range, need to be investigated and accurately estimated to inform effective management strategies and conservation planning [19]. The cryptic nature of these species often impedes direct observation, and camera trapping offers an efficient alternative for collecting data [20]. Over the past decade, spatial capture-recapture [21], a well-established method for estimating the density of species with individual markings such as patterned coats, has been successfully used for several populations of these three carnivore species. Most of the published estimates for striped hyaena, however, concern Asiatic populations [22-24], with a single study of an African population from Kenya [25]. Spatial capture-recapture studies of aardwolf density are similarly limited, with only two published estimates from Kenya and Botswana [25, 26]. A few more surveys have been carried out for serval in Southern Africa [26-29], and Western and Central Africa [30, 31], revealing large variations in density across the species’ range. The lack of estimates and the disparity in published results suggest the need to investigate additional populations of serval, striped hyaena and aardwolf in areas where robust estimates do not exist. Such information is essential to better understand how and why population status varies across the species’ geographic range [32, 33].

The second largest National Park in East Africa at over 20,000 km2, Tanzania’s Ruaha National Park (RNP) belongs to the greater Ruaha-Rungwa landscape, which harbours important wildlife biodiversity, partly due to the convergence of Acacia-Commiphora and miombo woodland (Brachystegia-Jubelnardia) ecotypes [34]. The Ruaha-Rungwa landscape also encompasses a range of fully and partially protected areas, including Game Reserves, Game Controlled Areas, Open Areas, and community-managed Wildlife Management Areas (WMA). Established on communal lands, WMAs are one of the most novel features in Tanzania’s conservation landscape, which intend to promote a sustainable scheme involving local communities in wildlife protection and its economic benefits [35]. Various sources and levels of anthropogenic pressure add to the vulnerability of carnivores in the Ruaha-Rungwa landscape, including: habitat conversion for agriculture, prey base depletion due to bushmeat poaching, and direct persecution by humans in retaliation to livestock predation [36, 37]. RNP also lies at the southern range limit for both the global striped hyaena population and the East & Northeast African aardwolf population [13, 15]. This ecosystem is therefore of particular conservation interest for the species, since peripheral populations often differ in density response and appear more sensitive to habitat change than more centrally located populations [38].

In addition to providing baseline status estimates for three African carnivore species for which insufficient data currently exist, this study intended to investigate local variations in density across different habitats, management strategies, and levels of protection and anthropogenic pressure. To this end, we used camera trapping and spatially explicit capture-recapture (SECR) modelling [39] to estimate population densities of serval, striped hyaena and aardwolf at three sites in Ruaha-Rungwa, representative of different ecosystem types and levels of protection and anthropogenic pressure: the Acacia-Commiphora core tourist area of RNP; the miombo woodland of western RNP; and the community-run Acacia-Commiphora MBOMIPA WMA, which adjoins Ruaha to the east and acts as a buffer between the Park and unprotected village lands.

Materials and methods

Ethics statement

Data collection consisted of camera trapping, a non-invasive method which avoids contact with the study species and minimises interference with their natural behaviour. Fieldwork was carried out under research permits 2018-368-NA-2018-107, 2019-96-ER-97-20 and 2019-424-NA-2018-184, granted by the Tanzania Commission for Science and Technology (COSTECH; Dar es Salaam, Tanzania; rclearance@costech.or.tz) and Tanzania Wildlife Research Institute (TAWIRI; Arusha, Tanzania; researchclearance@tawiri.or.tz).

Study area

The Ruaha-Rungwa ecosystem in southern Tanzania extends over 45,000 km2 across three ecoregions: Central Zambezian miombo woodlands, Eastern miombo woodlands, and Eastern African acacia savannas [40] (Fig 1A and 1B). RNP lies at the centre of this ecosystem and was the largest National Park in Tanzania until the formation of Nyerere National Park in 2019 [41]. With an altitude ranging from 716 m to 1,888 m [42], the terrain features low rolling hills overlooked by an escarpment to the north, and the Great Ruaha River runs along the RNP’s eastern border [43]. RNP experiences a semi-arid to arid climate and a unimodal rainfall regime, from December to April [44]. Acacia-Commiphora deciduous bushlands and thickets predominate in about two-thirds of the park, while miombo woodlands cover its western part [45] (Fig 1B). A number of Game Reserves, Game Controlled Areas, and Open Areas complete the Ruaha-Rungwa landscape to the north and the west of RNP, while the Matumizi Bora ya Malihai Idodi na Pawaga (MBOMIPA) and Waga WMAs stretch along its eastern border. MBOMIPA WMA was established in 2007 based on the principles of community-based natural resource management [35] and encounters greater levels of anthropogenic impacts than the core tourist area of RNP [46]. The landscape is unfenced, which facilitates free movement of wildlife.

Fig 1

Ruaha-Rungwa landscape and spatial distribution of camera trap stations.

(A) Location of the Ruaha-Rungwa landscape in Tanzania (made with Natural Earth). (B) Ruaha-Rungwa landscape’s ecotypes [40] and land uses. The map depicts, but does not explicitly name, boundaries of additional protected areas, and only shows villages and towns near protected areas. (C) Core RNP Acacia-Commiphora grid. (D) RNP miombo grid. (E) MBOMIPA WMA Acacia-Commiphora grid.

Camera trap surveys

The study used a photographic dataset collected between June and November 2018 in RNP and MBOMIPA WMA to assess the status of leopard [46]. We surveyed two sites within RNP, and one in MBOMIPA WMA (Fig 1C–1E). The first camera trap grid covered an area of 223 km2 in highly-productive riverine Acacia-Commiphora habitat [47], at the core of the park’s tourist area, with 45 camera stations set out for 83 days. The second grid, located in miombo woodland in the western part of RNP, consisted of 26 camera stations over an area of 152 km2 for 90 days. We deployed the third grid in highly-productive riverine Acacia-Commiphora habitat in MBOMIPA WMA, with 40 stations spanning a total area of 270 km2 for 69 days.

The duration of each survey was kept below 90 days both to ensure sufficient captures/recaptures and to approximate closed populations, as per previous studies [20, 21, 28, 48–52]. The design of the camera trap layout sought to optimise capture rates of large carnivores, facilitate individual identification, and ensure no gaps in the trapping array. Camera placement prioritised roads and junctions, or game trails, as large carnivores preferentially travel along roads [53]. With a single road crossing the study area within the park’s miombo woodland, half of the cameras in the RNP miombo grid were set on game trails along semi-open mbugas (drainage lines). The mean distance between two cameras ranged from 1.88 km in RNP miombo to 1.96 km and 2.08 km in Core RNP Acacia-Commiphora and WMA Acacia-Commiphora, respectively. Given the published home range estimates for the target species [54, 55], this spacing enabled to recapture individuals at multiple camera trap stations [21, 56], in compliance with SECR requirements for modelling space use.

All but one of the stations consisted of paired cameras facing one another on either side of a road or trail, to increase the likelihood of capturing both flanks of passing animals [48]. Cameras were mounted on trees at approximately 40 cm height, with the surrounding vegetation regularly cleared to prevent any obstruction of the lens or sensor and reduce the risk of bush fire damaging the camera. The survey deployed several motion-activated camera models: Cuddeback Professional Color Model 1347 and Cuddeback X-Change Color Model 1279, Non Typical Inc., Wisconsin, USA; HC500 HyperFire, Reconyx, Wisconsin, USA. The majority of cameras featured xenon flashes, which produce higher definition images of animal markings than infrared LED flashes [56].

Density estimation

We estimated the population density (defined as the number of individuals per 100 km2) of serval, striped hyaena, and aardwolf at each survey site via Maximum Likelihood Estimation SECR analysis, with the package secr v3.2.0 [57] in R version 3.6.3 [58]. SECR combines two sub-models representing the spatial distribution of a population and the detection process respectively [59], and therefore does not require ad hoc estimation of the effective trapping area as per conventional capture-recapture methods [60].

Data inputs consisted of detection histories detailing sampling occasions and locations of captures for each individual throughout the survey, and a trap layout listing the coordinates of every camera trap station and their activity periods (S2 and S3 Appendices). We visually identified individuals in camera trap images through their unique pelage markings [48], focusing on flanks for serval and fore-quarters and hind-quarters for striped hyaena and aardwolf, as detailed in S1 Appendix. We sexed individuals based on the unobstructed view of external genitalia, late pregnancy signs such as weight gain and enlarged abdomen (see S1 Appendix), or the presence of cubs. Individuals whose sex could not be confidently distinguished were classified as “unknown sex” (coded NA in the detection histories). We selected the flank with the greatest number of capture events for each species and grid to produce detection histories, using the following framework: (i) sampling occasions spanned 24-hours from midday to midday, to account for the nocturnal nature of the studied species; (ii) individuals could be recorded at different locations during a sampling occasion, but only once per given location (as per standard practice [61]).

The SECR observation sub-model describes the individual detection probability as a monotonous function of the distance to home range centre, characterised by g0, the detection probability at the range centre, and σ, a spatial scale related to home range width [62]. We tested three standard functions (half-normal, negative exponential and hazard rate) to model the decrease of the detection function with the distance to the home range centre. To minimise any bias in estimated densities, we selected a buffer width at which density estimates reached a plateau for each analysis.

We tested for variation in the baseline encounter probability g0 through the use of embedded predictors and covariates hypothesised to influence detection probability [63]. Specifically, we investigated the behavioural response of individuals to capture events by fitting the global trap response model (b predictor) and the local trap response model (bk predictor), and assessed the influence of the type of flash at each station (xenon versus LED flash), and station location (on-road versus off-road) on detection. Finally, we examined the effect of sex on detection function parameters g0 and σ by fitting a hybrid mixture model, which accommodates individuals of unknown sex [64]. Model selection was carried out by ranking models based on their Akaike Information Criterion score, corrected for small sample size (AICc) [65]. When more than one candidate model had substantial empirical support (ΔAICc < 2) [65], a final density estimate was derived by applying model averaging based on AICc weights [21].

Results and discussion

Results

Survey effort

The sampling effort across the 111 stations deployed at the three study sites totalled 8,477 camera trap nights and yielded 226 images of serval, 125 images of striped hyaena and 1,119 images of aardwolf (see S4 Appendix). We could identify individuals in 86.3% of serval photographs, 72.8% of striped hyaena photographs, and 81.1% of aardwolf photographs. Detection histories for serval featured 38 independent capture events for 13 identified individuals in Core RNP Acacia-Commiphora, 31 events for 10 identified individuals in WMA Acacia-Commiphora, and 23 events for 12 identified individuals in RNP miombo. Striped hyaena were captured in WMA Acacia-Commiphora only, with 42 independent capture events and 12 identified individuals. Aardwolf were only captured at camera stations in Core RNP Acacia-Commiphora and WMA Acacia-Commiphora, with 36 and 37 individuals identified, accounting for a total of 240 and 185 independent capture events, respectively.

Aardwolf displayed the highest recapture rate, i.e. the percentage of individuals captured more than once, across all grids (84.9%), followed by striped hyaena (58.3%) and serval (57.1%). All captured individuals were adults or subadults, with the exception of one female serval and one aardwolf accompanied by offspring. The target species showed predominantly nocturnal activity, with 86% of identified serval captures, 98% of striped hyaena captures, and 99.4% of aardwolf captures occurring between 7 pm and 6 am. The poor visibility of genitalia and lack of clear secondary sexual traits limited sex determination to 2.9% of the identified servals and 16.7% of identified striped hyaenas, preventing the modelling of sex differences in the density estimation process. Aardwolf sexing proved more successful, with sex assigned to 35.6% of identified individuals, which allowed us to model sex differences in detection function parameters.

Population density

Of the three detection functions tested, the negative exponential function best fitted the datasets for all target species at the three sites. We found no evidence of learned behavioural response or influence of covariates on detection for serval in Core RNP Acacia-Commiphora and WMA Acacia-Commiphora. However, the model in which g0 varied with station location was supported in RNP miombo. S5 Appendix details the ranking results based on AICc score. RNP miombo yielded the highest serval density, with 5.56 [Standard Error = ±2.45] individuals per 100 km2, followed by 3.45 [±1.04] individuals per 100 km2 in Core RNP Acacia-Commiphora, and 2.08 [±0.74] individuals per 100 km2 in WMA Acacia-Commiphora (Fig 2, Table 1).

Fig 2

Population density estimates for the target species at the three survey sites.

Bars represent standard errors.

Table 1

Population density and parameter estimates, with standard errors/95% confidence intervals for the target species at the three survey sites.

			Estimate ± SE		Estimate ± SE 95% CI
Species	Site	Predictor	g0	σ(m)	Density (ind/100 km2)
Serval	Core RNP Acacia-Commiphora		0.041 ± 0.014	1062 ± 171	3.45 ± 1.04
					1.93–6.16
	RNP miombo	On road	0.033 ± 0.017	1249 ± 361	5.56 ± 2.45
		Off road	0.006 ± 0.004		2.44–12.69
	WMA Acacia-Commiphora		0.034 ± 0.013	1566 ± 299	2.08 ± 0.74
					1.05–4.09
Striped hyaena	WMA Acacia-Commiphora	bk = 0	0.032 ± 0.016	2625 ± 521	1.36 ± 0.50
		bk = 1	0.078 ± 0.029		0.68–2.72
Aardwolf	Core RNP Acacia-Commiphora	bk = 0	0.030 ± 0.009	1220 ± 131	9.19 ± 1.66
		bk = 1	0.263 ± 0.038		6.46–13.06
	WMA Acacia-Commiphora	Female	0.135 ± 0.085	476 ± 111	13.25 ± 2.48
		Male	0.323 ± 0.075	553 ± 42	9.22–19.05

Predictor bk- = 1 for site-specific step change after the first capture.

For striped hyaena in WMA Acacia-Commiphora, two models received strong support (ΔAICc < 2, [65]); the model with constant g0 and sigma, and the local trap response model (bk). Model averaging returned an estimated density of 1.36 [±0.50] individuals per 100 km2.

For aardwolf in Core RNP Acacia-Commiphora, learned behavioural response influenced capture probability, with the local trap response model (bk) ranking highest, while the models in which g0 or sigma varied with sex ranked highest in WMA Acacia-Commiphora. Aardwolf density was estimated at 9.19 [±1.66] individuals per 100 km2 in Core RNP Acacia-Commiphora, and 13.25 [±2.48] individuals per 100 km2 in WMA Acacia-Commiphora.

Discussion

Density comparison

This study produced the first population densities estimated through spatial capture-recapture for three lesser-studied carnivores, serval, striped hyaena and aardwolf, in Tanzania. Using camera trap data from the Ruaha-Rungwa landscape, we shed light on the population density responses of these increasingly threatened species to habitat and land management strategies. Our results provide an important baseline and reference estimate that can be employed for future ecological monitoring of striped hyaena, aardwolf and serval.

Serval densities were higher within the RNP grids than in WMA Acacia-Commiphora, and in particular within the miombo woodland. This density distribution follows the precipitation pattern over the Ruaha-Rungwa landscape, as the miombo woodlands in the south-western part of RNP receive the greatest amount of rainfall on average and WMA Acacia-Commiphora is the driest of the three surveyed areas [44]. The miombo woodlands feature open, grassy mbuga (“dambo”) drainage lines. Seasonal rains flood mbugas and water remains present year-round, usually in distinct springs or pools during the dry season. This result aligns with the preference of servals for well-watered environments such as wetlands and riparian habitat [17]. The difference between RNP miombo and Core RNP Acacia-Commiphora could also stem from the lower abundance of lion, leopard and spotted hyaena (Crocuta crocuta) in the miombo woodlands [46], as these apex predator species occasionally kill serval [66]. The lower serval population density in WMA Acacia-Commiphora likely results from the WMA’s proximity to unprotected land, inducing greater edge effects [67], and a more pronounced level of anthropogenic pressure. For instance, the camera stations in the WMA recorded nine illegal incursions, two of them with evidence of bushmeat hunting, as well as two spotted hyaenas and two striped hyaenas bearing marks of snare captures. The common use of dogs for bushmeat hunting may also negatively impact resident populations of serval, as occasional killings by domestic dogs have been observed elsewhere in Tanzania [68]. In contrast, the survey only recorded one illegal incursion in the RNP grids. Moreover, the lower abundance of small ungulates in the WMA driven by anthropogenic impacts and the lack of habitat immediately outside the WMA may intensify competition between serval and leopard [47]. In areas with fewer impala or other medium-sized antelopes, leopard may shift their diet towards smaller prey species, such as large rodents and birds, and thus increase their dietary overlap with serval [69].

Serval density estimates using spatial capture-recapture vary greatly across Africa, from 0.63 and 1.28 individuals per 100 km2 in northern Namibia [28] to 101.21 in Mpumalanga, South Africa [29]. The latter estimate comes from a heavily modified habitat where the absence of persecution and interspecific competition, combined with high rodent biomass, enables serval to attain such high numbers. In comparison, our results lie within the lower range of published densities and are close to estimates for the Niokolo Koba National Park, Senegal (3.49–4.08 individuals per 100 km2) [31] and the Drakensberg Midlands, South Africa (6.0–8.3 individuals per 100 km2) [27]. The only other density estimate available for the species in Tanzania exceeds our results, with 10.9 [SE = 3.17] individuals per 100 km2 in Tarangire National Park [70]; however, the study employed conventional capture-recapture, which has been shown to overestimate density compared to spatial capture-recapture [71].

We could only estimate striped hyaena population density in WMA Acacia-Commiphora, as the camera trap survey did not detect any individuals in Core RNP Acacia-Commiphora or RNP miombo. The Ruaha-Rungwa landscape lies at the southern limit of the species’ global range [13]; deeper within the species’ range, in Laikipia County, Kenya, population density estimated through SECR was around six times higher than the density estimated in WMA Acacia-Commiphora [25]. Such a pattern is consistent with widespread biogeographic trends, typically showing diminished abundance towards species range edges [72]. Furthermore, the species is closely associated with Acacia-Commiphora bushland within Africa [68], and the fact that our study landscape traverses a transition from this habitat to miombo woodland may point to a mechanism delimiting the range edge in this area. The low density observed in WMA Acacia-Commiphora may also ensue from high edge effects due to the area’s proximity to human-dominated areas [67], as highlighted by the record of two individuals showing evidence of wire snare capture during the study period.

In contrast to serval, we estimated a higher aardwolf density in WMA Acacia-Commiphora than in Core RNP Acacia-Commiphora. The difference probably results from milder intraguild predation in the WMA, which supports lower densities of apex predators than the core tourist area of RNP [46]. Moreover, as aardwolf primarily feed on harvester termites, depletion of ungulates would not trigger dietary overlap with larger carnivores [18]. In addition, aardwolf may benefit from accidental fire outbreaks which often occur later in the dry season in the WMA; the reduced grass cover caused by this burning may force harvester termites to range more widely in search of food, thus lengthening their exposure time to predation on the ground [73, 74]. The reverse trend in aardwolf density compared to serval could indicate different sensitivities to anthropogenic pressures between the two species, although poaching activities in the area do not specifically target either species. Aardwolf may also be more prone to predation by apex predators, with greater stealth and agility giving serval an edge on the slower-moving aardwolf when threatened [17]. The absence of records of aardwolf in miombo woodland reflects the species’ known preference for Acacia–Commiphora bushland [68]. Our results align with aardwolf population densities estimated through conventional capture-recapture and spatial capture-recapture in Tarangire National Park and Kenya’s Laikipia District, of 9.04 and 11.63 individuals per 100 km2, respectively [25, 70].

Camera trap location impacted detection probability for serval in RNP miombo. Rather than denoting serval’s preference for larger roads, the higher detection probability at stations located on roads probably results from the particular road network in this portion of RNP, consisting of a single road complemented by two infrequently-used vehicle tracks and numerous subsidiary game trails. Though we tried to select the most heavily used trails, their large number offers wildlife multiple equivalent options compared to the stations set on the most readily accessible road. For aardwolf in Core RNP Acacia-Commiphora and striped hyaena, the probability of detection at a camera trap station was higher for individuals that had already been captured at the same site. This suggests that individuals favour the core area of their home range more often than the periphery and are not deterred by camera trap flashlights (xenon or LED flash). Sex appeared to have a mild impact on aardwolf capture probability in WMA Acacia-Commiphora, with males having a higher probability of capture than females.

Conservation implications and recommendations

This study made use of photographic by-catch data from a research project targeting leopard. Considering the logistical and budget requirements of camera trapping, this study illustrates the value of by-catch data in terms of resource optimisation, provided that the survey design also complies with the ecology of the non-target species [75]. Our trap layout met spatial sampling recommendations for the species we considered: the array coverage areas were larger than single home range areas of all species [54, 55, 76, 77], and spacing was sufficient to yield recaptures of individuals across multiple stations in all cases [21, 76, 78–80]. However, the trap layout configuration was tailored to maximise leopard capture rate, with stations preferentially deployed on roads and junctions [53], which may not be optimum for serval, striped hyaena and aardwolf. For instance, striped hyaena have been recorded to move mostly cross-country rather than along roads in the Serengeti [12]. Similarly, mesocarnivores such as serval and aardwolf may avoid areas commonly used by apex predators, which would lower their capture rate along roads [81]. Although the analysis did not find any evidence of road avoidance behaviour, we suggest that future work targeting these species might improve detectability by positioning camera stations around sightings of the animals or their signs.

The higher population density estimated for serval in miombo woodland indicates that this is an important habitat for the species in the region. One of the most extensive ecosystems in Sub-Saharan Africa, miombo woodlands cover a substantial portion of Tanzania’s protected and unprotected land [82]. Miombo plays a key role in the livelihood of rural communities but faces unsustainable resource exploitation [83, 84]. As such, increased woodland degradation and deforestation would negatively impact serval as well as a range of other mammal species. As miombo woodlands have received little conservation attention compared to Acacia habitats in East Africa [82], we recommend additional survey work across their extent to improve the understanding of these systems and protect their unique biodiversity.

Additionally, our results highlight the importance of the community-managed MBOMIPA WMA for the striped hyaena and aardwolf in the Ruaha-Rungwa landscape. Along with the adjacent RNP, MBOMIPA WMA provides the two hyaenids with suitable habitat at the edge of their global range. More generally, MBOMIPA WMA acts as a buffer zone between the highly protected RNP and the surrounding villages, and our results demonstrate the potential of community conservation in complementing more traditional conservation strategies, particularly in areas with lower densities of vulnerable species such as striped hyaena. Nevertheless, the camera trap survey found evidence of bushmeat hunting in the WMA; although pictures show that hyaenas can break free from snares and survive with amputated limbs, snaring and other non-selective poaching methods generally induce an increase in anthropogenic mortality of non-target species in savannah ecosystems [85, 86]. Considering the species’ low density in the Ruaha-Rungwa landscape, a conservation priority for striped hyaena should aim to reduce the impact of snaring on resident populations and assess the corresponding risk to their viability.

In conclusion, our study illustrates how data from camera trap surveys targeting charismatic large carnivores can shed light on populations of lesser-studied carnivores, broadening the scope of capture-recapture studies beyond focal species. While adjustments to sample design can help to extend the range of accessible species in surveys [26], in this case, we were able to generate useful data without such adjustments. Similar efforts, employing by-catch data, could be easily replicated elsewhere to provide information on the spatial and population ecology of a range of species. Finally, we recommend complementing the baseline estimates provided for serval, striped hyaena and aardwolf in the Ruaha-Rungwa landscape with investigations into spatial variation in site use across the landscape to determine the environmental and anthropogenic factors driving their habitat selection.

Individual identification and sexing.

(PDF)

Click here for additional data file.

Capture histories.

(PDF)

Click here for additional data file.

Trap layouts.

(PDF)

Click here for additional data file.

Survey grid summary information.

(PDF)

Click here for additional data file.

SECR model ranking.

(PDF)

Click here for additional data file.

26 Nov 2020

PONE-D-20-33931

Density responses of lesser-studied carnivores to habitat and management strategies in southern Tanzania’s Ruaha-Rungwa landscape

PLOS ONE

Dear Dr. Hardouin,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Jan 10 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Bi-Song Yue, Ph.D

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. In your Methods section, please provide additional information regarding the permits you obtained for the work. Please ensure you have included the full name of the authority that approved the field site access and, if no permits were required, a brief statement explaining why.

3. Thank you for stating the following in the Acknowledgments Section of your manuscript:

"We would like to thank the Government of Tanzania, Tanzania Wildlife Research

Institute (TAWIRI), Commission for Science and Technology (COSTECH), Tanzania National

Parks Authority (TANAPA), Tanzania Wildlife Management Authority (TAWA), and Idodi

Pawaga MBOMIPA WMA for their support of this research."

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

"Scholarship funding for CS and PS is provided by the University of Oxford NERC Environmental Research DTP (https://www.environmental-research.ox.ac.uk). AD is funded by a Recanati-Kaplan Fellowship (https://www.wildcru.org). Additional funding was awarded to CS for this research from National Geographic Society Early Career Grants (https://www.nationalgeographic.org/funding-opportunities/grants), Cleveland Metroparks Zoo Africa Seed Grants (https://www.clevelandmetroparks.com/zoo), Chicago Zoological Society Chicago Board of Trade (CBOT) Endangered Species Fund (https://www.czs.org/Chicago-Zoological-Society/Conservation-Leadership/CBOT-Endangered-Species-Fund), and Pittsburgh Zoo & PPG Aquarium Conservation & Sustainability Fund (https://www.pittsburghzoo.org/conservation/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

4. We note that Figure 1 in your submission contain map images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

4.1.    You may seek permission from the original copyright holder of Figure 1 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

4.2.    If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Hardouin et al.

This paper is better than it might appear. As a comparison of three sites, with no replicates, one might dismiss it as a pedestrian. As an ecological study, it is. What makes it important is its methods. The authors use bycatch — and I just love their use of that term — the incidental records of species in a study of leopards. There are now surely thousands of camera trap studies aimed are finding large mammals that, often enough, have other species walk by. These other species are interesting, important, and sometimes poorly known. They are grist to this paper’s mill and good for the authors.

Second, the methods of setting up the trapping grid and identifying the individuals also provide a model for other studies. I see a lot of camera trap studies to review: this is a good one.

Oh, but the English is awful. I suspect the senior author is not a native speaker, and so has an excuse. The other authors need to be ashamed. I have rarely seen so many sentences written in the passive voice.

Overall, a basic, unpretentious paper that is useful contribution to a genre that will assume importance with the expanding number of papers about camera trapping.

Reviewer #2: Specific Comments:

Lines 21-22: Is ‘less charismatic’ necessary? Also, the use of body size adjectives is relative. What is large and what is smaller? The species listed could easily be viewed as large carnivores in many systems.

Lines 31-33: It isn't clear here what the 'respectively' is referencing. Is it the 3 sites listed above in the order they are listed?

Line 46: The use of camera traps is not novel. There are many, many published studies that have used this technique. I wouldn’t suggest it as a rationale or novelty of the study here.

Lines 85-91: Is this difference in geography important? How so? Is the distinction of just these 2 countries important? Do you expect differences compared to other parts of the geographic range for this species? Justification for study seems tenuous and not convincing given how it is presented.

Lines 109-110: The technique is not novel. Based on the rationale here and the issue of geographically incomplete data sets, the need for study is questionable, as presented.

Lines 155-156: So no movement in or out over 90 days. This doesn't seem realistic.

Lines 161-163: What is the significance here on distance? Was independence of camera trap being sought?

Lines 180-189: So the entire 90 days was extent of the capture history matrix? How wouldn't there be deaths, immigration and emigration occurring across such a long time period? Thus, if this population is actually open, why not use a Jolly-Seber model to assess population numbers? How certain was the individual identification? Any estimate of your error rate?

Lines 205-206: Any concern with count data and overdispersion of data?

Lines 228-232: I don't recall any mention of diagnostics used to identify gender in the Methods. Any biases and how were they minimized?

Lines 243-245: AIC weights range from 0 to 1 but the values that are presented in S5 Appendix range from 0 to >71. The delta AICs are not calculated correctly either, at least given the AICc values presented. Did the authors interpret results based on the values presented? If so, I would question the results and interpretations. Also, if models were similar in terms of the AIC weight, were the parameters and their SE evaluated to see if a model had uninformative variables (sensu Arnold 2010. Journal of Wildlife Management)?

Lines 343: I don't understand the point of this first sentence.

Lines 345-348: How so and what does this mean to be 'compatible with the home range'?

Lines 354-356: Does this conclusion follow based directly on the design of the study and analyses? I didn't see this as an objective in the Introduction.

Reviewer #3: This paper used photos of three mesocarnivores, aardwolf, serval, and striped hyena, from a previous camera trap study of leopards along with spatial capture-recapture modeling methods to estimate their population abundance/density in Tanzania. The authors set up the need for the study well by explaining the conservation challenges for these medium-sized carnivores in that they are all facing a variety of threats but do not receive the same conservation attention as charismatic megafauna (like lions). Because the three species involved have patterned coats with unique markings among individuals, the authors developed capture histories for use in the modeling (included in the supporting information). Using SECR, they estimated population density for each of the three species while addressing model assumptions and testing the potential influence of the camera flash and station location on detection. Their results provide density estimates for the three species and are given in context to habitat, previous density estimates, and conservation concerns. Overall, I thought the paper was really well written and the spatial capture-recapture modeling was thorough and addressed the important assumptions involved in using this approach. The density estimates all appear realistic given the species and habitat and were effectively compared to other published results. I also appreciated the discussion of the data in the context of conservation.

There are two points for clarification that would strengthen the paper. First, it would be helpful to add details to the methods for identification of individual animals using the pelage patterns. Correctly identifying unique individuals is essential for their use of capture-recapture analysis. When describing their identification methods, the authors cite a paper from 1995 that manually identified 10 individual tigers so I assume that is the method they also used but it is unclear. The supplemental materials include images of each species with polygons around markings but there’s no caption to explain whether this was done by an observer or software like Hotspotter. It would be helpful to include more of a description about how the individuals were identified and ID was confirmed.

Secondly, the authors note that that large carnivores preferentially use roads for movement and the camera photos used were originally collected for a study of leopards. However, there is no mention of the potential intraspecific interactions between the three focal species in this study and leopard (or even interactions among the three species). Was this considered as a covariate in the model and if so, why was it not included? Adding addition information and discussion about the intraspecific responses would be helpful.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

10 Mar 2021

Journal requirements:

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

• https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

• https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Response:

We renamed the supporting information files in accordance with PLOS ONE guidelines, i.e. “S1_File.pdf”, and we uploaded them individually.

2. In your Methods section, please provide additional information regarding the permits you obtained for the work. Please ensure you have included the full name of the authority that approved the field site access and, if no permits were required, a brief statement explaining why.

Response:

Tanzania Commission for Science and Technology (COSTECH) and Tanzania Wildlife Research Institute (TAWIRI) are the full names of the institutions which granted research permits and approved field site access. We updated the paragraph with their headquarters location and research clearance email address, as well as an additional permit number, as follows (lines 126-129): “Fieldwork was carried out under research permits 2018-368-NA-2018-107, 2019-96-ER-97-20 and 2019-424-NA-2018-184, granted by the Tanzania Commission for Science and Technology (COSTECH; Dar es Salaam, Tanzania; rclearance@costech.or.tz) and Tanzania Wildlife Research Institute (TAWIRI; Arusha, Tanzania; researchclearance@tawiri.or.tz).”

3. Thank you for stating the following in the Acknowledgments Section of your manuscript: "We would like to thank the Government of Tanzania, Tanzania Wildlife Research Institute (TAWIRI), Commission for Science and Technology (COSTECH), Tanzania National Parks Authority (TANAPA), Tanzania Wildlife Management Authority (TAWA), and Idodi Pawaga MBOMIPA WMA for their support of this research."

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows: "Scholarship funding for CS and PS is provided by the University of Oxford NERC Environmental Research DTP (https://www.environmental-research.ox.ac.uk). AD is funded by a Recanati-Kaplan Fellowship (https://www.wildcru.org). Additional funding was awarded to CS for this research from National Geographic Society Early Career Grants (https://www.nationalgeographic.org/funding-opportunities/grants), Cleveland Metroparks Zoo Africa Seed Grants (https://www.clevelandmetroparks.com/zoo), Chicago Zoological Society Chicago Board of Trade (CBOT) Endangered Species Fund (https://www.czs.org/Chicago-Zoological-Society/Conservation-Leadership/CBOT-Endangered-Species-Fund), and Pittsburgh Zoo & PPG Aquarium Conservation & Sustainability Fund (https://www.pittsburghzoo.org/conservation/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

Response:

The institutions mentioned in the Acknowledgments Section did not provide funding for this research but administrative, logistics, and intellectual support. Therefore, our Funding Statement does not need to include them. We amended the Acknowledgments Section to avoid potential confusion, as follows (lines 409-412): “We would like to thank the Government of Tanzania, Tanzania Wildlife Research Institute (TAWIRI), Tanzania Commission for Science and Technology (COSTECH), Tanzania National Parks Authority (TANAPA), Tanzania Wildlife Management Authority (TAWA), and Idodi-Pawaga MBOMIPA WMA."

4. We note that Figure 1 in your submission contain map images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

4.1. You may seek permission from the original copyright holder of Figure 1 to publish the content specifically under the CC BY 4.0 license. We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text: “I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission. In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

4.2. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

• USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

• The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

• Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

• NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

• Landsat: http://landsat.visibleearth.nasa.gov/

• USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

• Natural Earth (public domain): http://www.naturalearthdata.com/

Response:

In Figure 1A, we used for the shaded relief a shapefile from the Natural Earth website (https://www.naturalearthdata.com). As declared on the Natural Earth website, “All versions of Natural Earth raster + vector map data found on this website are in the public domain”. Ecoregions in Figures 1B-E originate from the WWF website (https://www.worldwildlife.org/publications/terrestrial-ecoregions-of-the-world), which did not mention any copyright issues. Other elements of the maps such as country’s borders, protected areas delineation, rivers, roads, or human settlements also originate from shapefiles which are not copyrighted. We mentioned the Natural Earth website and the reference for ecoregions in Figure 1 caption, as follows (lines 147-152):

“Fig 1. Ruaha-Rungwa landscape and spatial distribution of camera trap stations. (A) Location of the Ruaha-Rungwa landscape in Tanzania (made with Natural Earth). (B) Ruaha-Rungwa landscape’s ecotypes [40] and land uses. The map depicts, but does not explicitly name, boundaries of additional protected areas, and only shows villages and towns near protected areas. (C) Core RNP Acacia-Commiphora grid. (D) RNP miombo grid. (E) MBOMIPA WMA Acacia-Commiphora grid.”

Review Comments to the Author

Reviewer #1:

Hardouin et al.

This paper is better than it might appear. As a comparison of three sites, with no replicates, one might dismiss it as a pedestrian. As an ecological study, it is. What makes it important is its methods. The authors use bycatch — and I just love their use of that term — the incidental records of species in a study of leopards. There are now surely thousands of camera trap studies aimed are finding large mammals that, often enough, have other species walk by. These other species are interesting, important, and sometimes poorly known. They are grist to this paper’s mill and good for the authors.

Second, the methods of setting up the trapping grid and identifying the individuals also provide a model for other studies. I see a lot of camera trap studies to review: this is a good one.

Oh, but the English is awful. I suspect the senior author is not a native speaker, and so has an excuse. The other authors need to be ashamed. I have rarely seen so many sentences written in the passive voice.

Overall, a basic, unpretentious paper that is useful contribution to a genre that will assume importance with the expanding number of papers about camera trapping.

Response:

The manuscript was edited and proofread by native English speakers to improve the language standard and to reduce the use of passive voice.

Reviewer #2:

Specific Comments:

Lines 21-22: Is ‘less charismatic’ necessary? Also, the use of body size adjectives is relative. What is large and what is smaller? The species listed could easily be viewed as large carnivores in many systems.

Response:

We deleted the references to species’ charisma and body size and amended the manuscript as follows (lines 21-23): “Compared to emblematic large carnivores, most species of the order Carnivora receive little conservation attention despite increasing anthropogenic pressure and poor understanding of their status across much of their range.”

Lines 31-33: It isn't clear here what the 'respectively' is referencing. Is it the 3 sites listed above in the order they are listed?

Response:

We rephrased the sentence for clarification, as follows (lines 31-33): “The Park’s miombo woodlands supported a higher serval density (5.56 [Standard Error = ±2.45] individuals per 100 km2) than either the core tourist area (3.45 [±1.04] individuals per 100 km2) or the Wildlife Management Area (2.08 [±0.74] individuals per 100 km2).”

Line 46: The use of camera traps is not novel. There are many, many published studies that have used this technique. I wouldn’t suggest it as a rationale or novelty of the study here.

Response:

The manuscript was amended as follows (lines 44-47): “By shedding light on three understudied African carnivore species, this study highlights the importance of miombo woodland conservation and community-managed conservation, as well as the value of by-catch camera trap data to improve ecological knowledge of lesser-studied carnivores.”

Lines 85-91: Is this difference in geography important? How so? Is the distinction of just these 2 countries important? Do you expect differences compared to other parts of the geographic range for this species? Justification for study seems tenuous and not convincing given how it is presented.

Lines 109-110: The technique is not novel. Based on the rationale here and the issue of geographically incomplete data sets, the need for study is questionable, as presented.

Response:

We have replied to these two comments simultaneously as they both relate to the justification of the study. Population densities vary across species’ geographical range in response to environmental and anthropogenic factors [1]. For instance, published estimates for serval density using spatial capture-recapture show large variations depending on the ecosystem considered, from 0.63 individuals per 100 km2 in northern Namibia [2] to 101.21 in Mpumalanga, South Africa [3]. In light of such variability, large-scale extrapolations would be ill-advised, leading us to state the need to investigate populations in areas where robust estimates do not exist, such as East Africa. More generally, assessing density across different parts of a species’ range, particularly across different habitats and land use types, and surveying different components of a landscape is important to get a better understanding of how population status varies across the species’ geographic extent [4,5].

In addition, the Ruaha-Rungwa landscape presents interesting characteristics for the study of our target species. First, it lies at the southern limit of the striped hyaena geographic range and the East & Northeast African aardwolf population range. As a result of the biogeographic pattern of density decline toward the boundaries of a species range, peripheral populations are expected to be less abundant and more sensitive to habitat change than more centrally located populations [6]. Therefore, we hypothesised that striped hyaena and aardwolf densities in the Rungwa-Ruaha landscape would differ from the estimates in Laikipia County, Kenya, to which we refer in the manuscript [7]. Second, the Ruaha-Rungwa landscape’s variety of ecotypes and management strategies described in the introduction allows to study local variations in density and assess the species’ response against environmental and anthropogenic factors. We amended the introduction (lines 88-94 and 106-115) to include this additional information.

Lines 161-163: What is the significance here on distance? Was independence of camera trap being sought?

Response:

Spatial Capture-Recapture (SCR) uses location-specific individual encounter histories to construct a spatial model of the detection process, where the individual detection probability is characterised by the detection probability at the range centre, and a spatial scale related to home range width [8]. To estimate this spatial scale parameter, SCR models need some individuals be re-encountered at several camera trap sites, which constitutes an additional level of information compared to conventional capture–recapture models. Therefore, the distance between camera stations should allow the observation of individuals at multiple camera stations [9-11]. According to published home range estimates [12,13], the station spacing adopted in the survey design was compatible with spatial recaptures for serval, striped hyaena and aardwolf, despite the survey primarily targeting leopard, a species associated with larger home ranges. We added a sentence at the end of the paragraph to clarify this point (lines 171-173).

Lines 155-156: So no movement in or out over 90 days. This doesn't seem realistic.

Lines 180-189: So the entire 90 days was extent of the capture history matrix? How wouldn't there be deaths, immigration and emigration occurring across such a long time period? Thus, if this population is actually open, why not use a Jolly-Seber model to assess population numbers? How certain was the individual identification? Any estimate of your error rate?

Response:

We acknowledge that the sampling periods used in our study, i.e. 70, 83 and 90 days, increase the risk of violating the population closure assumption and introducing a bias in density estimations. However, we also note that shorter sampling periods cannot guarantee closed populations and also produce a bias [9]. Considering the low detectability of the studied species, the lengthening of the sampling period allowed for a larger number of individuals and recapture events, resulting in more accurate density estimations [9,14]. Even so, the number of serval and striped hyaena individuals in our study remained less than 15, as shown in S4 Appendix. Such bias/precision trade-off frequently occurs for elusive species, with sampling periods of 60-90 days widely adopted as a compromise between ensuring sufficient data and approximating closed populations [2,9,14,15,16,17]. Moreover, a study of the bias-precision trade-off showed a positive impact of longer sampling periods for slow-life history species and intermediate-life history species (up to 3 months) [15]. Lines 162-164 were amended to introduce this notion of compromise when choosing sampling durations.

The lead author performed individual identification twice and identifications were verified by a different observer to minimise mismatches. Any photographs with uncertain identification were excluded from the analysis. However, based on the verification performed by the second observer, we estimate that the average error rates for serval and hyaena aligned with the results of a recent snow leopard study [18] i.e. circa 10%. For aardwolf we anticipate a slightly higher error rate, between 15-20%, as aardwolf markings are less numerous and distinctive than the other study species. We added more details about individual identification into S1 Appendix.

Lines 205-206: Any concern with count data and overdispersion of data?

Response:

Overdispersion in SCR models most likely arises from non-independent spatial configurations between individuals such as clustering into groups [19]. Amongst the studied species, serval are solitary and the two hyaenids display solitary foraging behaviour despite forming monogamous pairs for aardwolf or resting in small groups for striped hyaena [20]. The only case where more than one individual featured in a camera trap photograph corresponded to females with dependant young, and we did not include accompanying offspring in detection histories. We tested the goodness of fit of the supported models with a Monte-Carlo test using the residual deviance divided by the residual degrees of freedom method [21], and the results did not show any evidence of a lack of fit.

Lines 228-232: I don't recall any mention of diagnostics used to identify gender in the Methods. Any biases and how were they minimized?

Response:

Sexing was based on the unobstructed view of external genitalia, late pregnancy signs such as weight gain and enlarged abdomen, or the presence of cubs. Individuals whose sex could not be confidently distinguished were classified as “unknown sex” (coded NA in the detection histories). We added a sentence in the manuscript (lines 195-198) and some pictures in S1 Appendix to clarify this point.

Lines 243-245: AIC weights range from 0 to 1 but the values that are presented in S5 Appendix range from 0 to >71. The delta AICs are not calculated correctly either, at least given the AICc values presented. Did the authors interpret results based on the values presented? If so, I would question the results and interpretations. Also, if models were similar in terms of the AIC weight, were the parameters and their SE evaluated to see if a model had uninformative variables (sensu Arnold 2010. Journal of Wildlife Management)?

Response:

We made a mistake when reporting data in S5 Appendix, with the last three columns of the table corresponding to {AIC, AICc, �AICc} instead of {AICc, �AICc, AICcwt}. We apologise for this error and amended S5 Appendix with the correct values. The case of a model falling within 2 AIC units (�AIC < 2) of the top-ranked model occurred twice, for striped hyaena and aardwolf in MBOMIPA WMA Acacia-Commiphora. Both aardwolf models (g0[sex] σ[.] and g0[.] σ[sex]) have 5 parameters, whereas the second-ranked model for striped hyaena (g0[.] σ[.]) has one parameter less than the top model (g0[bk] σ[.]). Therefore, �AICc between the top and second ranked models mainly result from differences in the maximised log-likelihood of the model and not from the addition of an uninformative parameter [22,23].

Lines 343: I don't understand the point of this first sentence.

Response:

We amended this sentence in the manuscript to link it more clearly to the following one about the value of by-catch data, as follows (lines 359-360): “This study made use of photographic by-catch data from a research project targeting leopard.”

Lines 345-348: How so and what does this mean to be 'compatible with the home range'?

Response:

Simulation studies investigating the impact of spatial design on SCR parameter estimates have shown that trap arrays should cover at least one home range for SCR models to perform well [24,25]. Based on the maximum distances between spatial recaptures (see S4 Appendix) and the published home range estimates for our study species [12,13], our survey design complied with this requirement. Simulations have also demonstrated the impact of trap spacing on model precision, with the existence of an optimal spacing range between 1–3 σ under the half-normal encounter probability model with spatial scale parameter σ [9,24,26,27,28]. Our survey design also met this requirement, with σ exceeding half the distance between traps for all species and sites except aardwolf in WMA Acacia-Commiphora. In any case, the design allowed recaptures of individuals across multiple stations We added a sentence in the manuscript to detail these two points (lines 362-366).

Lines 354-356: Does this conclusion follow based directly on the design of the study and analyses? I didn't see this as an objective in the Introduction.

Response:

Based on observations of the species’ behaviour in published studies, we highlight in this paragraph a potential limitation of the survey design which may have reduced capture rates and make recommendations for future work to prevent it. Our analysis tested the influence of station location on detection probability and did not find any evidence of road avoidance behaviour. However, the positioning of off-road camera stations prioritised areas used by leopards, which might have also resulted in a lower capture rate for mesocarnivores. Therefore, further information about the species’ movement in the Ruaha-Rungwa ecosystem would be needed to state confidently on this topic and future work should ideally investigate the species’ preferences prior to camera trapping. We rephrase the sentence in the manuscript for more clarity (lines 371-374).

Reviewer #3:

This paper used photos of three mesocarnivores, aardwolf, serval, and striped hyena, from a previous camera trap study of leopards along with spatial capture-recapture modeling methods to estimate their population abundance/density in Tanzania. The authors set up the need for the study well by explaining the conservation challenges for these medium-sized carnivores in that they are all facing a variety of threats but do not receive the same conservation attention as charismatic megafauna (like lions). Because the three species involved have patterned coats with unique markings among individuals, the authors developed capture histories for use in the modeling (included in the supporting information). Using SECR, they estimated population density for each of the three species while addressing model assumptions and testing the potential influence of the camera flash and station location on detection. Their results provide density estimates for the three species and are given in context to habitat, previous density estimates, and conservation concerns. Overall, I thought the paper was really well written and the spatial capture-recapture modeling was thorough and addressed the important assumptions involved in using this approach. The density estimates all appear realistic given the species and habitat and were effectively compared to other published results. I also appreciated the discussion of the data in the context of conservation.

There are two points for clarification that would strengthen the paper. First, it would be helpful to add details to the methods for identification of individual animals using the pelage patterns. Correctly identifying unique individuals is essential for their use of capture-recapture analysis. When describing their identification methods, the authors cite a paper from 1995 that manually identified 10 individual tigers so I assume that is the method they also used but it is unclear. The supplemental materials include images of each species with polygons around markings but there’s no caption to explain whether this was done by an observer or software like Hotspotter. It would be helpful to include more of a description about how the individuals were identified and ID was confirmed.

Response:

Individual identification was performed by visually inspecting coat markings. The lead author examined and named each record once, then carried out a second run to check each of them. A different observer subsequently verified identifications to minimise mismatches, and any photographs with uncertain identification were excluded from the analysis. We also tested the I3S pattern software on a sample batch of serval pictures, but visual identification proved more effective. We started with the species presenting the most defined markings, i.e. serval, and compared the position of spots on individuals’ flanks [10]. The method consisted of processing all the pictures coming from one camera station, in chronological order, and then going to the next camera station. Each identified individual was characterised by an ID number, its sex (if determinable), and a distinctive spot arrangement (indicated in red in S1 Appendix pictures). The comparison of the coat markings of individuals featuring in subsequent records with identified individuals extended over the whole flank, with the key marker acting as a starting point to ease the process. We repeated the operation with striped hyaena and aardwolf, but this time focusing on fore-quarters and hind-quarters and selecting several key markers. We added a paragraph and some pictures in S1 Appendix to clarify the identification process.

Secondly, the authors note that that large carnivores preferentially use roads for movement and the camera photos used were originally collected for a study of leopards. However, there is no mention of the potential intraspecific interactions between the three focal species in this study and leopard (or even interactions among the three species). Was this considered as a covariate in the model and if so, why was it not included? Adding addition information and discussion about the intraspecific responses would be helpful.

Response:

This is an excellent suggestion, which we have in fact addressed, but the resulting set of analyses was too extensive to contain easily within a single paper. We therefore plan to submit a separate paper that looks at spatial correlates of distribution, including inter-specific interactions.

References:

1. Brown JH. On the relationship between abundance and distribution of species. Am Nat. 1984; 124(2): 255-279.

2. Edwards S, Portas R, Hanssen L, Beytell P, Melzheimer J, Stratford K. The spotted ghost: density and distribution of serval Leptailurus serval in Namibia. Afr J Ecol. 2018; 56: 831–840. doi: 10.1111/aje.12540.

3. Loock DJE, Williams ST, Emslie KW, Matthews WS, Swanepoel LH. High carnivore population density highlights the conservation value of industrialised sites. Sci Rep. 2018; 8: 16575. doi: 10.1038/s41598-018-34936-0.

4. Foster RJ, Harmsen BJ. A critique of density estimation from camera-trap data. J Wildl Manage. 2012; 76(2): 224–236. doi: 10.1002/jwmg.275.

5. Suryawanshi KR, Khanyari M, Sharma K, Lkhagvajav P, Mishra C. Sampling bias in snow leopard population estimation studies. Popul Ecol. 2019; 61(3): 268–276. doi: 10.1002/1438-390X.1027.

6. Doherty P, Boulinier T, Nichols J. Local extinction and turnover rates at the edge and interior of species' ranges. Ann Zool Fennici. 2003; 40(2): 145-153.

7. O'Brien TG, Kinnaird MF. Density estimation of sympatric carnivores using spatially explicit capture-recapture methods and standard trapping grid. Ecol Appl. 2011; 21: 2908-2916. doi:10.1890/10-2284.1.

8. Efford MG. Estimation of population density by spatially explicit capture–recapture analysis of data from area searches. Ecology. 2011; 92: 2202-2207. doi:10.1890/11-0332.1.

9. Royle JA, Chandler RB, Sollmann R, Gardner B. Spatial Capture-recapture. 1st Edition. San Diego: Academic Press; 2013. 612 p.

10. Rovero F, Zimmerman F. Camera trapping for wildlife research. Exeter: Pelagic publishing; 2016. Chapter 7, Capture-recapture methods for density estimation; p. 95-142.

11. Sollmann R. A gentle introduction to camera‐trap data analysis. Afr J Ecol. 2018; 56(4): 740– 749. doi: 10.1111/aje.12557.

12. Mills G, Hofer H. Hyaenas: status survey and conservation action plan. IUCN/SSC Hyaena Specialist Group, IUCN, Gland, Switzerland. 1998. Available from: https://portals.iucn.org/library/node/7402.

13. Ramesh T, Kalle R, Downs CT. Spatiotemporal variation in resource selection of servals: Insights from a landscape under heavy land‐use transformation. J Mammal. 2016; 97(2): 554–567. doi: 10.1093/jmammal/gyv201.

14. Harmsen BJ, Foster RJ, Quigley H. Spatially explicit capture recapture density estimates: Robustness, accuracy and precision in a long-term study of jaguars (Panthera onca). PLoS One. 2020; 15(6): e0227468. doi: 10.1371/journal.pone.0227468.

15. Dupont P, Milleret C, Gimenez O, Bischof R. Population closure and the bias-precision trade-off in spatial capture-recapture. Methods Ecol Evol. 2019; 10: 661–672. doi: 10.1111/2041-210X.1315.

16. Boron V, Tzanopoulos J, Gallo J, Barragan J, Jaimes-Rodriguez L, et al. Jaguar Densities across Human-Dominated Landscapes in Colombia: The Contribution of Unprotected Areas to Long Term Conservation. PLoS One. 2016; 11(5): e0153973. doi: 10.1371/journal.pone.0153973.

17. Satter CB, Augustine BC, Harmsen BJ, Foster RJ, Sanchez EE, Wultsch C, et al. Long‐term monitoring of ocelot densities in Belize. J Wildl Manage. 2019; 83: 283-294. doi: 10.1002/jwmg.21598.

18. Johansson O, Samelius G, Wikberg E, Chapron G, Mishra C, Low M. Identification errors in camera-trap studies result in systematic population overestimation. Sci Rep. 2020; 10: 6393. doi: 10.1038/s41598-020-63367-z.

19. Bischof R, Dupont P, Milleret C, Chipperfield J, Royle, JA. Consequences of ignoring group association in spatial capture–recapture analysis. Wildlife Biol. 2020. doi: 10.2981/wlb.00649.

20. Foley C, Foley L, Lobora A, De Luca D, Msuha M, Davenport TRB et al. A Field Guide to the Larger Mammals of Tanzania. Princeton: Princeton University Press; 2014. 320 p.

21. Efford MG. Secr 3.2 - spatially explicit capture–recapture in R. 2019 [cited 2021 February 15]. 20 p. Available from: https://cran.r-project.org/src/contrib/Archive/secr/.

22. Burnham KP, Anderson DR. Multimodel inference: understanding AIC and BIC in model selection. Sociol Methods Res. 2004; 33(2): 261–304. doi: 10.1177/0049124104268644.

23. Arnold TW. Uninformative Parameters and Model Selection Using Akaike's Information Criterion. The Journal of Wildlife Management. 2010; 74: 1175-1178. doi:10.1111/j.1937-2817.2010.tb01236.x.

24. Sollmann R, Gardner B, Belant JL. How Does Spatial Study Design Influence Density Estimates from Spatial Capture-Recapture Models? PLoS ONE. 2012; 7(4): e34575. doi: 10.1371/journal.pone.0034575.

25. Tobler MW, Powell GVN, Estimating jaguar densities with camera traps: problems with current designs and recommendations for future studies. Biol Conserv. 2013; 159: 109-118. doi: 10.1016/j.biocon.2012.12.009.

26. Sun CC, Fuller AK, Royle JA. Trap Configuration and Spacing Influences Parameter Estimates in Spatial Capture-Recapture Models. PLoS ONE. 2014; 9(2): e88025. doi: 10.1371/journal.pone.0088025.

27. Clark JD. Comparing clustered sampling designs for spatially explicit estimation of population density. Popul Ecol. 2019; 61: 93–101. doi: 10.1002/1438-390X.1011.

28. Efford, MG, Boulanger, J. Fast evaluation of study designs for spatially explicit capture–recapture. Methods Ecol Evol. 2019; 10: 1529–1535. doi: 10.1111/2041-210X.13239.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

17 Mar 2021

Density responses of lesser-studied carnivores to habitat and management strategies in southern Tanzania’s Ruaha-Rungwa landscape

PONE-D-20-33931R1

Dear Dr. Hardouin,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Bi-Song Yue, Ph.D

Academic Editor

PLOS ONE

19 Mar 2021

PONE-D-20-33931R1

Density responses of lesser-studied carnivores to habitat and management strategies in southern Tanzania’s Ruaha-Rungwa landscape

Dear Dr. Hardouin:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Bi-Song Yue

Academic Editor

PLOS ONE