Journal list menu

Volume 58, Issue 12 p. 2892-2902
RESEARCH ARTICLE
Open Access

Public health and economic benefits of spotted hyenas Crocuta crocuta in a peri-urban system

Chinmay Sonawane

Corresponding Author

Chinmay Sonawane

Department of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA, USA

Correspondence

Chinmay Sonawane

Email: [email protected]

Search for more papers by this author
Gidey Yirga

Gidey Yirga

Department of Biology, Mekelle University, Mekelle, Ethiopia

Search for more papers by this author
Neil H. Carter

Neil H. Carter

School for Environment and Sustainability, University of Michigan, Ann Arbor, MI, USA

Search for more papers by this author
First published: 26 September 2021
Citations: 5

Handling Editor: Fernanda Michalski

Abstract

  1. Species that depend on anthropogenic waste for food can remove pathogens that pose health risks to humans and livestock, thereby saving lives and money. Quantifying these benefits is rare, yet can lead to innovative conservation solutions.
  2. To assess these benefits, we examined the feeding ecology and population size of peri-urban spotted hyenas Crocuta crocuta in Mekelle, Ethiopia. We integrated these field data into a disease transmission model to predict: (a) the number of anthrax and bovine tuberculosis (bTB) infections arising in humans and livestock from infected carcass waste and (b) the costs associated with treating these infections and losing livestock. We compared these public health and economic outcomes under two scenarios: (a) hyenas are present and (b) the counterfactual, hyenas are absent.
  3. We estimated that hyenas annually remove 4.2% (207 tonnes) of the total carcass waste disposed of by residents and businesses in Mekelle. Furthermore, the scavenging behaviour of hyenas annually prevents five infections of anthrax and bTB in humans, and 140 infections in cattle, sheep and goats. This disease control service potentially saves USD 52,165 due to the treatment costs and livestock loss avoided.
  4. Synthesis and applications. This human–hyena interaction in Ethiopia is evidence that large carnivores can contribute to human health and economy. To retain these benefits and maintain tolerance of hyenas, we recommend introducing education programmes to promote safe outdoor behaviour around hyenas, training watchdogs to alert residents of hyena presence, constructing bomas to protect livestock from hyena attacks, and preserving the hyenas' access to carcass waste to reduce their dependency on livestock predation. With humans and carnivores coming more frequently into contact, understanding and communicating how these species can benefit humanity will be critical to motivating human–carnivore coexistence worldwide.

1 INTRODUCTION

The natural world provides critical goods and services to sustain human life. Research on ecosystem goods and services has traditionally concentrated on provisioning and regulating services from particular ecosystems (Vihervaara et al., 2010). Large carnivores are often neglected in such conversations of ecosystem goods and services (Ceauşu et al., 2019; Expósito-Granados et al., 2019). Indeed, much of the research on human–carnivore interactions has focused on the detrimental role of carnivores in zoonoses transmission, livestock depredation and attacks on humans (O'Bryan et al., 2018). These negative impacts on human life contribute to the fear of carnivores among the public, thereby motivating extirpations of carnivores from human-dominated landscapes. Worldwide, human activities threaten many large carnivore species with extinction (Wolf & Ripple, 2018).

Given this enduring threat of carnivore loss, further research into the beneficial contributions of these species on human life is important for promoting human–carnivore coexistence (Carter & Linnell, 2016; Carter et al., 2012; O'Bryan et al., 2018). The very few cases that have explicitly quantified these benefits demonstrate a range of goods and services that have previously been overlooked. For example, leopards Panthera pardus in Mumbai, India have been shown to prevent 90 human cases of rabies through their predation on feral dogs (Braczkowski et al., 2018). Recolonisation of cougars Puma concolor in the eastern U.S. is predicted to reduce deer-vehicle accidents by 22%, and prevent 21,400 human injuries over 30 years (Gilbert et al., 2017). Comprehensive valuations of these benefits can potentially inform nature-based solutions for achieving the UN's Sustainable Development Goals (Cohen-Shacham et al., 2016).

We present a methodology to quantify the beneficial contributions of carnivores to human health and economy, and use spotted hyenas Crocuta crocuta in Mekelle, Ethiopia as a case study. Although often vilified, hyenas sanitise urban environments by consuming carcass waste discarded by local communities (Gade, 2006). In Mekelle, the inadequacy of waste collection infrastructure has prompted residents to dispose two-thirds of solid waste in open areas and roadsides, thereby rendering waste management as the most serious public service issue in Mekelle (Tadesse & Hadgu, 2009; Tadesse et al., 2008). Improper management of organic waste contributes to the high burden of zoonotic diseases in Ethiopia, and compromises human and livestock health (Grace et al., 2012). Zoonoses of particular concern are anthrax and bovine tuberculosis (bTB), which are collectively responsible for approximately 6,000 human deaths and 500,000 cattle deaths annually in Ethiopia (Demirdag et al., 2003; FAO, 2018; Grace et al., 2012; Hunt, 1996; Pieracci et al., 2016). These two diseases alone impose significant financial strain on the healthcare system and the livestock industry. We hypothesise that hyenas provide public health and economic benefits by nocturnally scavenging for anthropogenic waste, thereby reducing transmission of disease from waste to humans and livestock.

To assess this hypothesis, we address three aims: (a) estimate the carcass waste annually consumed by Mekelle hyenas, (b) model the number of anthrax and bTB infections in humans and livestock in the presence and absence of hyenas and (c) appraise the financial value of the disease control service provided by hyenas. We found evidence that hyenas directly provide life-sustaining services in Mekelle, and ease the financial burden of controlling disease in humans and livestock; such benefits would not exist in the absence of hyenas. Our methodology opens new research avenues for quantifying the beneficial contributions from large carnivores and scavengers, which can be used to motivate coexistence with these species.

2 MATERIALS AND METHODS

2.1 Study area

Our study area, Mekelle, Ethiopia, covers an area of 96 km2, and hosts over 310,000 residents and 120,000 livestock animals (Mekelle University, 2012). Annual rainfall in the region averages 550 mm, and mean temperatures range from 12 to 27°C. The landscape surrounding Mekelle is at an altitude of 2,300 m and severely degraded; beyond agricultural land, vegetation is primarily limited to Acacia spp. and Eucalyptus spp. The wild prey base is also chronically depleted, and wild carnivores rely on domestic animals for food (Abay et al., 2011).

We recorded our observations of hyenas scavenging in Mekelle over 40 non-consecutive nights (with a mean of 1.14 nights between observation nights, and a range of zero to seven nights) between mid-June and mid-August 2019. In the local landfill, carcass waste was reliably disposed of, and attracted hyenas nightly (Figure 1). Local residents and livestock also frequently interacted with waste at the landfill: residents salvaged recyclable items from the waste, and livestock grazed on organic matter. Due to the limited quantity of carcass waste, hyenas frequently consumed all available food within the first few hours after human activity subsided; thus, approximately 96% of our feeding observations were recorded between 19:00 and 23:00. We also opportunistically observed hyenas feeding on the streets and periphery of Mekelle, and all observations occurred between 18:00 and 05:00.

Details are in the caption following the image
Map of Mekelle city and surrounding habitat for hyenas

Our methodology for assessing the benefits provided by hyenas was broken into four steps. First, we estimated the quantity of carcass waste consumed by an individual hyena. Second, we estimated the hyena population size in Mekelle to infer how much overall waste they were consuming. Third, to estimate the proportion of waste consumed by hyenas, we calculated the total carcass waste available in Mekelle based on the existing literature to compare against the total waste removed by the hyena population. Fourth, we integrated these data into a disease transmission model to simulate the transmission of anthrax and bTB from carcass waste to humans and livestock. By comparing disease transmission in the presence and absence of hyenas, we determined the number of infections prevented by hyena scavenging of carcass waste, and the financial value of this service.

2.2 Estimating individual consumption rate

The annual consumption rate for a single hyena was calculated with Equation 1:
urn:x-wiley:00218901:media:jpe14024:jpe14024-math-0001(1)
where T is the total annual consumption (kg yr−1 hyena−1), P is the set of prey species consumed (cattle, equid, poultry, small ruminant), vx is the frequency of prey species x at the landfill and wx is weight of prey species x consumed by a single hyena per day.

We determined the frequencies of prey species that were available to the hyenas by scanning the landfill and recording the carcass waste available. On most nights (n = 27), only one species of carcass waste was available and fed upon by the hyenas. On nights with multiple species available (n = 5), one species usually dominated in terms of biomass. As hyenas do not preferentially feed on particular species, we recorded only the species with the most abundant waste to simplify the annual consumption rate calculation (Equation 1; Hayward, 2006). The limited food in Mekelle also discouraged preferential feeding, and ensured that all waste was consumed. Thus, the annual consumption rate is likely an underestimation, as we did not account for the morsels of waste consumed nightly by the hyenas.

We determined the weight of prey species x consumed per hyena by presenting the Grid clan with waste samples and quantifying the consumed weight of these samples. Each night, we randomly selected samples of carcass waste (n = 74) from the available waste, and recorded their pre-consumption and post-consumption weights using a scientific weight scale (Tefal). For samples heavier than 50 kg (e.g. entire equid carcasses), we used body measurements and standard weight prediction equations to calculate pre-consumption weights (Carroll & Huntington, 1988; Pearson & Ouassat, 2000). After the complete consumption of the samples, post-consumption weights of large carcasses were determined by estimating the weight of the remaining skeletal features (Prange et al., 1979). We considered samples to be consumed if: (a) there were no remains of the sample or (b) all hyenas ceased feeding on the sample, and the next two hyenas to show interest in the sample, through sniffing, did not feed (this indicated that there were no remains left to feed upon). The average weight consumed (in kg day−1 hyena−1) was determined as the difference between the post-consumption weight and pre-consumption weight of the sample, divided by the number of hyenas feeding on the sample. We observed the number of hyenas feeding on these samples using night-vision binoculars (NightFox). Of the 74 samples of carcass waste selected, 28 samples were used to calculate the consumption rate. The others could not be used: four were consumed by sympatric carnivores, 15 were not consumed to completion according to the aforementioned criteria and 27 were moved out of our line of sight by the hyenas.

2.3 Estimating total carcass waste consumed

To infer the total carcass waste removed by hyenas in Mekelle, we multiplied the individual consumption rate by the estimated hyena population size. Mekelle hyenas are peri-urban and inhabit the areas around Mekelle during the day to avoid anthropogenic disturbance. We, therefore, estimated the population size as the product of hyena density and area size of hyena habitat around Mekelle. We conservatively assumed that hyenas travel 5 km daily to feed in Mekelle, and 5 km to return to their diurnal resting sites (Kruuk, 1972). Thus, the 293.7 km2 area within a 5 km radius of Mekelle describes the habitat of Mekelle hyenas (Figure 1). To estimate hyena density around the perimeter of Mekelle, we performed call-in surveys (Ogutu & Dublin, 1998). Over two consecutive nights (between 19:00 and 00:00), we broadcasted hyena distress vocalisations with a 45 W megaphone (Monacor) from the top of a vehicle at five different locations around Mekelle (Figure 1). The three eastern locations were surveyed on the first night, and the two western locations were surveyed on the second night. The sharp topographic difference between the east and west presented a physical barrier for hyenas to cross, and likely minimised the potential for double-counting over the two nights. At each location, we played the same 5-min track of vocalisations four times, rotating the megaphone 90° after each playback. After these 20 min, the sound projection was paused for 10 min, before repeating this entire cycle once more. During this hour of the call-in survey, we recorded the maximum number of hyenas visible within sight using night-vision binoculars.

Local calibration is necessary for call-in surveys: not all subjects are in range to hear the playback, and not all subjects will be attracted to the broadcast (Ogutu & Dublin, 1998). Based on calibration experiments by Yirga et al. (2017), only hyenas within a 2.8 km radius of the broadcast responded to the playback; thus, all surveyed locations were separated from one another by at least 5.6 km to avoid double counting. Furthermore, the response rate to the broadcast in these calibration experiments was 75% (95% confidence interval, CI: 41.9%–100.0%). The population size of Mekelle hyenas was, therefore, estimated with Equation 2:
urn:x-wiley:00218901:media:jpe14024:jpe14024-math-0002(2)
where P is the population size, L is the set of surveyed locations and ni is the raw count of hyenas at location i.

2.4 Estimating proportion of carcass waste consumed

To estimate the proportion of carcass waste consumed by hyenas, we first calculated the total carcass waste made available to them annually by Mekelle residents and businesses. This was achieved by consulting the literature for the annual number of domestic animal deaths and the average disposed weights of these animals (Table S1). We subsequently estimated the total quantity of carcass waste that remained in Mekelle under two scenarios: (a) hyenas are present and (b) the counterfactual, hyenas are absent (Figure 2). To estimate the carcass waste remaining in Mekelle under the current scenario wherein hyenas are present, we calculated the product of the annual consumption rate for a single hyena and the population size of Mekelle hyenas, and subtracted this value from the total carcass waste disposed in Mekelle.

Details are in the caption following the image
Hypothesised impacts of the two modelled scenarios on carcass waste removal and disease transmission in humans and livestock

2.5 Projecting public health benefits

2.5.1 Disease transmission model

To project potential public health benefits from hyena scavenging of carcass waste, we compared the transmission of disease under the two aforementioned scenarios (Figure 2). We predicted the number of anthrax and bTB cases in humans and livestock that were directly transmitted from cattle and small ruminant carcass waste in Mekelle using the linear disease transmission model in Equation 3 (Figure 3):
urn:x-wiley:00218901:media:jpe14024:jpe14024-math-0003(3)
where Fi(d) is the annual agent i (human, cattle and small ruminant) deaths from disease d (anthrax and bTB), C is the cattle and small ruminant carcasses in Mekelle (a function of the consumption rate and hyena population size), πi(d) is the prevalence of disease d in meat products of type i (%), γi(d) is the contact rate between infected carcasses and agent i (contacts carcass−1 day−1) for disease d, pi(d) is the probability of infection for agent i upon contact with disease d, and mi(d) is the mortality rate for agent i infected with disease d. The parameters used for the model are summarised in Tables S2 and S3.
Details are in the caption following the image
Linear model (Equation 3) to estimate disease transmission from carcass waste to humans and livestock

2.5.2 Parameterising contact rates

The transmission dynamics of anthrax and bTB can vary from system to system as population size and social behaviours dictate the rate at which susceptible humans and livestock come into contact with infected carcass waste (Silk et al., 2019). Thus, it was necessary to parameterise specific contact rates for Mekelle. Empirically, quantifying these contact rates demands significant time and labour resources to monitor subsamples of humans and livestock, and their interactions with waste. Instead, we estimated contact rates with an agent-based model developed on NetLogo 6.1.1, which is commonly used to simulate disease transmission (Wilensky, 1999; Willem et al., 2017). At initialisation, the model produced a grid of 100 × 100 m patches, mirroring the shape and size of Mekelle. This spatial resolution was sufficiently small to capture movement patterns while optimising the model's computation performance. For simplicity, each patch spawned agents—humans, cattle and small ruminants—in linear proportion to its distance from the centre of Mekelle. We arranged the linear relationships so that the total number of agents spawned in the model was similar to the actual population numbers in Mekelle (Mekelle University, 2012). This ultimately reflected the trends in population density gradients that we observed in Mekelle (high density of humans and low densities of livestock in the centre, and vice versa at the periphery). To add stochasticity, each patch spawned agents according to a normal distribution with the mean set as the expected number of agents as per the linear proportion, and the standard deviation as one. At initialisation, random integers were also generated by a Poisson distribution—with the mean set as the maximum distance from the centre of Mekelle (i.e. 82.3 patches)—to determine the locations of the infected carcasses. Once a random integer x was generated, the model assigned an infected carcass to a randomly selected patch x patches away from the centre. In the model, infected carcasses were typically located closer to the periphery, thereby aligning with our observations of abandoned carcasses in Mekelle.

Once initialised, the model was run: agents moved from their patch to a randomly chosen adjacent patch, thereby simulating walking behaviour. In each iteration of the model, humans walked 27 patches (2.7 km) and livestock walked 50 patches (5.0 km); these lengths reflected the daily average walking distances determined by empirical studies (Althoff et al., 2017; Anteneh et al., 2010; Öberg et al., 1993). After each step, patches with infected carcasses determined if any agents were located on their patch. If agents were present, then the patch reports a ‘contact’ with a 0.005 probability. The 0.005 probability is derived from the following assumption: if an agent comes within 0.5 m of an infected carcass, the agent has come into contact with the disease via airborne pathogens or direct contact (Chakraborty et al., 2012; Gannon et al., 2007). Thus, in a 100 × 100 m patch, there are 40,000 squares sized 0.5 × 0.5 m. An agent must theoretically travel through 200 of these squares (approximately 100 m) to walk to the next patch; we therefore divided 200 by 40,000 to derive the 0.005 probability of coming into contact with an infected carcass. After agents completed walking the average daily distance, the model calculated the daily contact rate an infected carcass had with each type of agent. The model was repeated for 5,000 iterations to determine average contact rates.

2.6 Projecting economic impacts

The cost of treating an infected human was estimated as USD 202.71; this value included the cost of care-seeking, diagnosis and treatment, and was adjusted for inflation and currency exchange using the 23/01/2020 rate of 31.9922 ETB = 1 USD (Asres et al., 2018; World Bank, 2019). The cost of livestock loss was estimated as USD 3,749.56 per cattle death from anthrax, USD 1,974.43 per cattle death from bTB, and USD 93.91 per small ruminant death from both diseases (FAO, 2018). These valuations consider the farm-gate price and the loss of production. The valuations for cattle lost to anthrax and bTB vary due to differential losses of production arising from the two diseases (FAO, 2018). The valuation of small ruminant loss was based on the farm-gate price (USD 21.92), loss of dairy production (USD 70.76), loss of hide (USD 1.23), inflation and currency exchange (Baars, 2000; World Bank, 2019); we could not estimate other losses of production for small ruminants due to insufficient data.

3 RESULTS

3.1 Individual consumption rate

A Mekelle hyena scavenged 983.46 kg (95% CI: 685.96–1,397.67 kg) of carcass waste annually. Equid carcasses provided 82.2% of total annual consumed biomass, and poultry waste was the most frequently consumed carcass waste (Table 1). The Grid clan consumed no waste on 20% of nights, and recorded a maximum consumption rate of 12.33 kg day−1 hyena−1 on one night in the presence of an equid carcass.

TABLE 1. Biomass of resources scavenged by Mekelle hyenas. The frequencies describe the nights on which different types of waste was available. Consumption data show mean ± SE. Estimate of cattle consumption was based on estimate of small ruminant consumption due to lack of data; both cattle and small ruminant waste were in the form of offal, not whole carcasses
Prey species Frequency observed (%) (n)

Consumption

(kg sample−1 hyena−1)

Consumption

(kg−1 yr−1 hyena−1)

Cattle 10.0 (4) 0.70 ± 0.25 25.72 ± 9.23
Equid 25.0 (10) 8.86 ± 0.90 808.45 ± 82.28
Poultry 42.5 (17) 0.92 ± 0.16 142.85 ± 25.21
Small ruminant 2.5 (1) 0.70 ± 0.25 6.43 ± 2.31
None 20.0 (8)
Total 100.0 (40) 983.46 ± 119.02

3.2 Total carcass waste consumed

A population size of 210 hyenas (95% CI: 157–376) was estimated from the 66 hyenas that responded to the call-in surveys. Given that the surveys covered an area of 123.2 km2, hyena density was calculated to be 0.71 hyenas/km2, after adjusting for the 75% response rate. An additional 20 hyenas were also opportunistically encountered between the five locations, though these were excluded in our calculations. Finally, we estimate that the population of 210 hyenas annually remove 207 tonnes of carcass waste in Mekelle.

3.3 Proportion of carcass waste consumed

The 210 hyenas removed 4.2% (95% CI: 2.2%–10.7%) of the 4,917 tonnes of accessible (i.e. can come into contact with local human residents and livestock) carcass waste annually discarded by residents and businesses in Mekelle (Figure 4). Additionally, 10,974 tonnes of inaccessible carcass waste were generated annually by the export abattoir, which internally manages waste through rendering and composting plants (Mekelle University, 2012). Thus, a total of 15,890 tonnes of carcass waste is annually produced in Mekelle (Figure 4).

Details are in the caption following the image
Sources of carcass waste and their contribution to disease in Mekelle. All values are tonnes of waste per year. ‘Carcass waste type’ nodes quantify the carcass waste from each species. ‘Source’ nodes describe the method of death leading to the production of carcass waste. ‘Accessibility’ nodes describe if local human residents and livestock are in contact with the waste. ‘Contribution to disease’ nodes describe if the waste poses a public health threat to local human residents and livestock

3.4 Public health benefits

The disease transmission models indicated that, each year, three fewer anthrax infections and two fewer bTB infections were transmitted to humans from carcass waste in the presence of hyenas than compared to the counterfactual wherein hyenas were absent. This annually prevented one anthrax-related human death in Mekelle. In the presence of hyenas, 63 anthrax cases and 50 bTB cases were expected, with 29 and 9 of these respective cases resulting in human fatalities. In contrast, in the absence of hyenas, carcass waste was expected to annually infect 66 humans with anthrax, and 52 humans with bTB. Of these, 30 and 10 cases, respectively, were predicted to be fatal (Figure 5). Therefore, hyenas scavenging on carcass waste reduced disease transmission and mortality in humans by 4.2% in Mekelle.

Details are in the caption following the image
Cases and financial burden of anthrax and bTB infection from carcass waste. Total bar length marks the infections in Mekelle, and the coloured bars capture the mortality from these infections. Black lines mark the cost of treating all human infections in a and b, and value of livestock loss in c and d. The error bars capture the 95% CIs for each modelled output. (a) Anthrax infections in humans. (b) bTB infections in humans. (c) Anthrax infections in cattle and small ruminants. (d) bTB infections in cattle and small ruminants. For comparison, estimates of infections and deaths reported by FAO are included. FAO data in c and d reflect cattle cases only (FAO, 2018)

In livestock, 11 fewer cases of anthrax infections and 129 fewer cases of bTB infections were modelled when hyenas were present, thereby annually saving 8 and 13 livestock animals from anthrax-related and bTB-related deaths, respectively. In the presence of hyenas, carcass waste was expected to annually infect 259 livestock animals with anthrax and 2,899 livestock animals with bTB, thereby causing 186 and 303 deaths from these diseases respectively. In the absence of hyenas, anthrax and bTB infections in livestock were predicted to increase to 271 and 3,027 cases, respectively. Similarly, anthrax-related and bTB-related deaths in livestock increased to 194 and 317 animals, respectively (Figure 5). Overall, the models demonstrated a 4.2% reduction in the transmission of anthrax and a 4.3% reduction in the transmission of bTB in livestock when hyenas were scavenging on carcass waste in Mekelle.

The 95% CIs for the model outputs, in Figure 5 and Tables S4 and S5, were also calculated by using Equation 3: the lower and upper limits of the 95% CIs for parameters calculated in this study—namely, consumption rate, population size (both of which influence C) and contact rates (γi(d))—were substituted into the equation, and parameters derived from the literature—namely, πi(d), pi(d) and mi(d)—remained unchanged, given the limited information on their CIs. Admittedly, a deeper understanding of these uncertainties would likely expand the ranges of the CIs of our model outputs.

3.5 Economic benefits

These reductions in disease transmission when hyenas are present were associated with annual savings of (a) USD 564 and USD 442 due to avoided treatment costs of anthrax and bTB infections in humans, respectively and (b) USD 28,187 and USD 22,972 due to livestock loss avoided from anthrax-related and bTB-related deaths. Treatment costs for human anthrax infections were USD 12,855 in the presence of hyenas, and increased to USD 13,419 in their absence. Similarly, treatment costs for bTB infections in humans annually amounted to USD 10,046 with hyena presence, and increased to USD 10,489 without hyena presence. Livestock loss from anthrax was valued at USD 642,835 in the presence of hyenas, and at USD 671,021 in the absence of hyenas. bTB was predicted to cause an annual livestock loss of USD 517,602 when hyenas were present, and USD 540,573 when hyenas were absent (Figure 5). Therefore, the presence of hyenas annually reduced the economic burden of disease treatment costs in humans and in livestock loss by 4.2%.

4 DISCUSSION

Quantifying the services provided by large carnivores to local communities is necessary to comprehensively assess the value of sharing landscapes with these species; however, few studies have done so (O'Bryan et al., 2018). Our quantification of the waste consumption by hyenas in Mekelle indicates the provision of substantial sanitation benefits. Waste generation in Mekelle is only anticipated to increase with the exponentially growing human population, and given the financial constraints on waste collection and disposal services, sanitary conditions are expected to deteriorate further (Fenta et al., 2017; Tadesse et al., 2008). Thus, the partial removal of waste by hyenas meets an urgent need not adequately addressed by current infrastructure. Furthermore, the anthrax and bTB transmission models indicated that the scavenging behaviour of Mekelle hyenas indirectly saves two human and 22 livestock lives annually. By consuming potentially diseased carcass waste, hyenas prevented a total of five human and 140 livestock infections of anthrax and bTB. A derived benefit of hyenas preventing zoonotic infections is the sparing of USD 52,165 in treatment costs and livestock loss. These public health and economic benefits are particularly valuable in a region with limited healthcare facilities and one of the highest zoonosis burdens in the world (Fullman et al., 2018; Grace et al., 2012). This human–hyena interaction indicates that local communities can receive substantial sanitation services from scavengers, and explicit quantification of disease control using our methodology would further motivate the conservation of scavengers worldwide.

The benefits of hyena presence must also be compared against the costs. Abay et al. (2011) reported 10 non-fatal attacks on humans, and an economic loss of USD 2,928 from 33 fatal attacks on cattle and small ruminants per year in Mekelle. Such impacts foster a fear of hyenas among the public, though further research is required into the mental health effects of sharing landscapes with large carnivores (Braczkowski et al., 2018). However, the benefits described in this study also require further exploration, as only a fraction of all the contributions that hyenas provide to human well-being are captured here. Equid and poultry waste, an important food source for some Mekelle hyenas, also likely contribute to disease transmission in humans and livestock, though this contribution could not be quantified in this study due to the lack of parameters in the literature. Hyena scavenging also indirectly reduces person-to-person transmission of anthrax and bTB, and likely limits the environmental transmission of other concerning zoonoses (e.g. brucellosis, Escherichia coli and Salmonella) as well (Pieracci et al., 2016). Moreover, by feeding on waste, hyenas are also likely preventing spillovers of novel zoonotic diseases into the human reservoir.

To preserve these benefits into the future, non-lethal management actions that minimise risks and enable hyenas to coexist are necessary. Hyena attacks on humans are primarily linked to human participation in risky nocturnal activities, such as sleeping outdoors (Abay et al., 2011). With education programmes promoting safe outdoor behaviour and watchdogs alerting residents of hyena presence, these attacks can be avoided. The majority of attacks on livestock also occur overnight, and fortified bomas have been demonstrated to economically protect livestock (Abay et al., 2011; Kissui et al., 2019). In northern Ethiopia, hyenas are more likely to predate on domestic animals in the absence of carcass waste (Yirga et al., 2012). Therefore, by maintaining the hyenas' access to waste, human–hyena conflict can be mitigated in Mekelle, and the sanitation and disease control benefits provided by hyena scavenging can be preserved or even increased.

The disease transmission model we used here relied on accurate estimates of carcass consumption rates from the field. We estimated that a Mekelle hyena scavenged 2.69 kg of anthropogenic waste daily (Table 1). Previously, Abay et al. (2011) determined that scavenged waste accounted for 84.3%–89.0% of the hyenas' diets in Mekelle. Therefore, considering both hunted and scavenged resources, we estimate that Mekelle hyenas consume 3.11 kg day−1 hyena−1. Based on the relationship between body mass and consumption rates in large carnivores, this consumption rate is within expectations for a carnivore weighing 52–70 kg (Henschel & Tilson, 1988). Thus, our estimate is likely accurate. We also acknowledge that hyena diet composition varies in Mekelle: at the landfill, we observed hyenas primarily feeding on equid and poultry waste (Table 1), whereas cattle and small ruminant waste were more frequent in the diets of hyenas near the southern pastoral villages of Mekelle (Abay et al., 2011). Despite this variation, our estimate of the consumption rate likely remains typical for all Mekelle hyenas: most hyena populations across Africa consume 3–4 kg day−1 hyena−1 (Henschel & Tilson, 1988; Kruuk, 1972). Moreover, while diet composition may vary, Mekelle hyenas almost exclusively scavenge on the carcass waste of domestic animals due to a depleted wild prey base (Abay et al., 2011). For these reasons, we are confident that our estimate of hyena consumption of carcass waste used in the disease transmission models is reasonable.

The magnitude of the beneficial services provided by hyenas is proportional to the population size of Mekelle hyenas. Our estimate of the hyena density in Mekelle is similar to the results of previous studies: a density of 0.8 hyenas/km2 was recorded for the Mekelle district, and an average density of 0.87 hyenas/km2 was observed in human-inhabited areas of northern Ethiopia (Yirga et al., 2015, 2017). However, due to the difficulty in identifying hyena individuals at night, we could only record the maximum number of hyenas present at any given moment in the 1-hr call-in surveys. Counting the total number of different individuals that responded may increase our hyena density estimate, and consequently, also our valuation of the services that hyenas provide to Mekelle. Moreover, we conservatively assumed that hyenas feeding on anthropogenic waste resided within a 5 km buffer of Mekelle, and this provided a rudimentary definition of the range of Mekelle's peri-urban hyenas. Thus, research into the movement ecology of Mekelle hyenas could better inform our understanding of their population size. Nonetheless, our estimate of the number of Mekelle hyenas indicates a relatively large population size that, if maintained, could continue to provide benefits to local communities. Moreover, our model outputs of infections and deaths were within the same magnitude as those reported by the FAO, and any discrepancies between the outputs and FAO data are likely due to the underreporting of infections, and the consequent overestimation of mortality (FAO, 2018).

In the absence of hyenas, these public health and economic benefits may be lost in Mekelle, and artificial infrastructure or other vertebrate scavengers cannot feasibly be expected to compensate (Gade, 2006). Indeed, this human–hyena interaction demonstrates the viability of nature-based solutions to address societal challenges, particularly in rural and low-income areas; in Mekelle, the scavenging behaviour of hyenas is contributing to achieving three of the Sustainable Development Goals—namely, ensuring good health and well-being (3), providing clean water and sanitation (6), and promoting terrestrial biodiversity (15) (UN GA, 2015). In the face of increasing human–hyena conflict, the continued success of human-hyena coexistence depends on recognising these critical services that hyenas provide to human well-being in the urban areas of Africa (Gade, 2006). This approach of bridging the ecological functions that carnivores perform to tangible benefits for humanity will be critical for large carnivore conservation in the 21st century.

ACKNOWLEDGEMENTS

We thank Haile Gerigis, Mekelle University and the Norwegian Agency for Development Cooperation for providing field support, and the Office of Undergraduate Research and Fellowships, Harvard University for providing financial support.

    CONFLICT OF INTEREST

    The authors have declared no conflict of interest.

    AUTHORS' CONTRIBUTIONS

    C.S. and N.H.C. designed the project; G.Y. contributed to the methodology and coordinated field support; C.S. collected and analysed the data, and wrote the manuscript draft. All authors reviewed and edited the manuscript, and gave final approval for publication.

    DATA AVAILABILITY STATEMENT

    Data are available via the Dryad Digital Repository https://doi.org/10.5061/dryad.7m0cfxprc (Sonawane et al., 2020).