Abstract
In East Africa, community-based conservation models (CBCMs) have been established to support the conservation of wildlife in fragmented landscapes like the Tarangire Ecosystem, Tanzania. To assess how different management approaches maintained large herbivore populations, we conducted line distance surveys and estimated seasonal densities of elephant, giraffe, zebra, and wildebeest in six management units, including three CBCMs, two national parks (positive controls), and one area with little conservation interventions (negative control). Using a Monte-Carlo approach to propagate uncertainties from the density estimates and trend analysis, we analyzed the resulting time series (2011–2019). Densities of the target species were consistently low in the site with little conservation interventions. In contrast, densities of zebra and wildebeest in CBCMs were similar to national parks, providing evidence that CBCMs contributed to the stabilization of these migratory populations in the central part of the ecosystem. CBCMs also supported giraffe and elephant densities similar to those found in national parks. In contrast, the functional connectivity of Lake Manyara National Park has not been augmented by CBCMs. Our analysis suggests that CBCMs can effectively conserve large herbivores, and that maintaining connectivity through CBCMs should be prioritized.
Similar content being viewed by others
Introduction
Populations of large herbivores in East African savanna and grassland ecosystems move widely through landscapes where the distribution of forage and water vary considerably across space and time, resulting in seasonal animal migrations1. These herbivore populations in East Africa have declined markedly during the last decades2 both inside and outside fully protected areas3,4,5 as a result of unsustainable legal and illegal hunting6, deterioration of rangelands7, and loss and fragmentation of habitat due to expansion of agriculture and infrastructure8,9. Acknowledging that government protected areas alone are insufficient as a single measure to halt or reverse wildlife declines and to create landscapes that support both people and wildlife10, a diverse set of community-based conservation models (CBCMs) has been implemented in several parts of East Africa. In practice, these CBCMs often augment existing protected area networks and safeguard critical habitats on community and private lands11,12. In ecosystems that still sustain long distance migrations of large herbivores, CBCMs can provide suitable and safe habitat for wildlife and can contribute to effective conservation of migratory populations1,13. CBCMs in the region often aim for sustainable coexistence between wildlife and the livestock of pastoralist communities, as well as the Indigenous communities’ rangeland management practices, which include seasonal grazing reserves and rules for pasture access14,15.
Testing the effectiveness of conservation interventions is a key topic in conservation biology16, yet frequently hampered by a lack of monitoring data in CBCMs17,18. The few site-specific assessments of the ecological effectiveness of different Tanzanian CBCMs indicate mixed results, including wildlife population declines, stable population trends, and marked population size increases19,20,21,22,23,24. Ecological assessments of CBCMs are often solely based on temporal trends of target species in one study area25 or on comparisons with density estimates in either human-dominated20,21,26 or fully protected areas22. While valuable for informing local management in the short-term, such site-specific monitoring efforts only partially inform whether CBCMs effectively contribute to conserving wildlife populations at the ecosystem scale. This is because the distribution of large mammals in heterogeneous savanna ecosystems is dynamic27,28,29 with local population sizes affected by animal movement30. To more accurately assess the ecological effectiveness of CBCMS, it is beneficial to compare wildlife densities across multiple management units, including positive reference points such as national parks (while keeping in mind that they are not entirely pristine31,32), and areas with minimal conservation efforts.
The fragmented Tarangire Ecosystem of northern Tanzania (Fig. 1) maintains one of the last remaining long-distance migrations of large herbivores in Africa1,33. Compared to historical baselines, wildlife population sizes have declined substantially inside and outside of protected areas, with particularly pronounced declines during the 1980s and 1990s31,34,35. During the last two decades, three strategically placed community-based conservation areas [Burunge and Randilen Wildlife Management Area (WMA), and Manyara Ranch] were established in parts of the ecosystem36, mostly around Tarangire National Park (NP), to counteract these declines. These community-based conservation areas protect specific habitats for wildlife while allowing limited human activity. Some CBCMs allow seasonal livestock grazing in specified areas, some allow regulated hunting, and they typically employ game scouts to enforce wildlife laws and community land-use regulations. The type and intensity of natural resource utilization and the degree of community involvement in governance differ by CBCM category (see Table 1).
Map of the Tarangire Ecosystem in northern Tanzania; the inset in the top right indicates the location of the ecosystem within Tanzania. Terrestrial line transects (black lines) were carried out in Lake Manyara (LMNP) and Tarangire National Park (TNP), Burunge (BWMA) and Randilen Wildlife Management Area (RWMA), Manyara Ranch (MR), and the Mto wa Mbu Game Controlled Area (MGCA, no boundary data available). Mkungunero Game Reserve (MGR) and Makame Wildlife Management Area (MWMA). ‘Northern Plains’ and ‘Simanjiro Plains’ denote approximate locations of the wet season ranges of zebra (Equus quagga) and wildebeest (Connochaetes taurinus). For completeness, we also mapped Mkungunero Game Reserve (MGR) and Makame Wildlife Management Area (MWMA) which are part of the TE but not sampled. Blue polygons are alkaline lakes. We created both maps (study area within Tanzania and details of our study area) using the 'ggplot2' (version 3.3.6), 'rgdal' (version 1.5–23), and 'ggsn' (version 0.5.0) packages in R (version 4.2.2)70. Maps are based on open source (area polygons, lakes) and our own (transects) shapefiles.
Here, we test how these conservation efforts affected site-specific population dynamics of four wide-ranging, abundant, and functionally important wildlife species: African savanna elephant (Loxodonta africana), Masai giraffe (Giraffa tippelskirchi), plains zebra (Equus quagga), and wildebeest (Connochaetes taurinus). Wildlife populations were estimated from line distance sampling carried out seasonally from 2011 to 2019 in Burunge WMA, Randilen WMA, and Manyara Ranch. As spatial reference points, we considered population densities and associated trends in two fully protected areas: Tarangire NP, which is buffered by the considered CBCMs and Lake Manyara NP, which is not directly bordered by CBCMs. In addition, as counterfactual, we considered wildlife population trends in an area with few restrictions on human land use, Mto wa Mbu Game Controlled Area (GCA) (Fig. 1). To estimate area-specific annual trends, we fitted generalized additive models to seasonal density data and used a two-stage Monte Carlo simulation approach which propagates uncertainties from both the distance sampling estimates and trend analysis.
Results
Area-specific trends of large herbivore populations
Based on estimates from terrestrial line distance surveys and generalized additive models, elephant densities in Burunge WMA (Fig. 2c) increased over the survey period, fluctuated widely but appeared to remain fairly constant in Tarangire NP (Fig. 2a) and Manyara Ranch (Fig. 2e), and seemed to decline in Lake Manyara NP (Fig. 2b, Table S1). During 24 seasonal surveys, we did not detect any elephant in the Mto wa Mbu GCA (Fig. 2f).
Population density estimates and associated 95% confidence intervals for elephant (Loxodonta africana). The trend lines are based on 1000 Monte Carlo replicates and modelled as season-specific (LR: long rains; Dry: dry season; SR: short rains) generalized additive models across six management units (TNP: Tarangire National Park; LMNP: Lake Manyara National Park; BWMA: Burunge Wildlife Management Area; RWMA: Randilen Wildlife Management Area; MR: Manyara Ranch; MGCA: Mto wa Mbu Game Controlled Area) of the Tarangire Ecosystem, northern Tanzania. Population density estimates are based on terrestrial line distance sampling surveys. In RWMA, elephant were not counted during the surveys.
Across the monitoring period, giraffe densities showed a slight increase in Tarangire NP (Fig. 3a) and remained relatively constant in both Burunge WMA (Fig. 3c) and Lake Manyara NP (Fig. 3b; Table S1). In Randilen WMA (Fig. 3d) and Manyara Ranch (Fig. 3e), giraffe densities showed a slight upward trend. In the Mto wa Mbu GCA, giraffe densities were much lower than in the other management units (Fig. 3f).
Population density estimates and associated 95% confidence intervals for giraffe (Giraffa tippelskirchi). The trend lines are based on 1000 Monte Carlo replicates and modelled as season-specific (LR: long rains; Dry: dry season; SR: short rains) general additive models across six management units (TNP: Tarangire National Park; LMNP: Lake Manyara National Park; BWMA: Burunge Wildlife Management Area; RWMA: Randilen Wildlife Management Area; MR: Manyara Ranch; MGCA: Mto wa Mbu Game Controlled Area) of the Tarangire Ecosystem, northern Tanzania. Population density estimates are based on terrestrial line distance sampling surveys.
Zebra densities increased markedly in Tarangire NP (Fig. 4a), and remained fairly constant in Lake Manyara NP (Fig. 4b), Burunge WMA (Fig. 4c), Randilen WMA (Fig. 4d), and Manyara Ranch (Fig. 4e; Table S1). In the Mto wa Mbu GCA (Fig. 4f), zebra densities were considerably lower compared to the other management units. Seasonality strongly affected zebra densities in Tarangire NP (Fig. 4a), reflecting their seasonal long-distance movements (i.e. concentration during the dry season inside Tarangire NP, migration to areas outside the NP during the long rains).
Population density estimates and associated 95% confidence intervals for zebra (Equus quagga). The trend lines are based on 1000 Monte Carlo replicates and modelled as season-specific (LR: long rains; Dry: dry season; SR: short rains) generalized additive models across six management units (TNP: Tarangire National Park; LMNP: Lake Manyara National Park; BWMA: Burunge Wildlife Management Area; RWMA: Randilen Wildlife Management Area; MR: Manyara Ranch; MGCA: Mto wa Mbu Game Controlled Area) of the Tarangire Ecosystem, northern Tanzania. Population density estimates are based on terrestrial line distance sampling surveys.
Across the survey period, wildebeest densities remained relatively constant in Tarangire NP (Fig. 5a), with potential increases during the dry season in this core dry season range. In Lake Manyara NP (Fig. 5b) and Manyara Ranch (Fig. 5e; Table S1), their densities did not change substantially over the study period. In Burunge WMA, however, their densities increased markedly (Fig. 5c). Similar to zebra, wildebeest densities were comparably low in Randilen WMA (Fig. 5d) and the Mto wa Mbu GCA (Fig. 5f). As with zebra, seasonality strongly affected wildebeest densities in Tarangire NP, with both species practically absent from the park during the long rains and reaching high densities during the dry season.
Population density estimates and associated 95% confidence intervals for wildebeest (Connochaetes taurinus). The trend lines are based on 1000 Monte Carlo replicates and modelled as season-specific (LR: long rains; Dry: dry season; SR: short rains) generalized additive models across six management units (TNP: Tarangire National Park; LMNP: Lake Manyara National Park; BWMA: Burunge Wildlife Management Area; RWMA: Randilen Wildlife Management Area; MR: Manyara Ranch; MGCA: Mto wa Mbu Game Controlled Area) of the Tarangire Ecosystem, northern Tanzania. Population density estimates are based on terrestrial line distance sampling surveys.
Discussion
Based on long-term, systematic wildlife monitoring data, we describe population dynamics of four wide-ranging large herbivore species for five protected areas, including three CBCMs, two national parks, and a lesser protected Game Controlled Area in the Tarangire Ecosystem of Tanzania. While our site-based monitoring highlights heterogeneity and seasonality in species-specific densities, we show that densities of the four target species in CBCMs are comparable to those in adjacent Tarangire NP22, and occasionally even higher than those in Lake Manyara NP (Fig. 6). Moreover, large herbivore densities were consistently greater than in the Game Controlled Area, which served as a negative control as there are limited conservation efforts in place. Moreover, in Burunge WMA, we detected marked increases in wildebeest and elephant densities over our study period. In Randilen WMA, giraffe densities have increased (Fig. 6). Especially for the central part of the ecosystem (Tarangire NP, Burunge WMA, and Manyara Ranch), population trends of the target species were mostly either stable or indicated population growth over time. This mirrors data from photographic mark-recapture studies of both wildebeest, showing that their population in the ecosystem has stabilized since the early 2000s34, as well as giraffe, whose populations in Manyara Ranch and much of Tarangire NP were stable from 2012–201637. Our population trend estimates also align with results of a photographic mark-recapture study conducted from 2012 to 2016, indicating a slight decrease in the giraffe population over that time frame in Burunge WMA37 (Fig. 2c); our more recent data suggest that this trajectory has since reversed (Fig. 3c). Overall, our site-based, seasonal monitoring efforts suggest that CBCMs contributed to the stabilization of large herbivore populations, with some areas experiencing increases in numbers, highlighting that strategically placed28,38 and locally supported39 conservation approaches effectively improved the resilience of wide-ranging herbivore populations in an increasingly fragmented ecosystem.
Trends in population density estimates for elephant (Loxodonta africana), giraffe (Giraffa tippelskirchi), zebra (Equus quagga) and wildebeest (Connochaetes taurinus) across six management units (TNP: Tarangire National Park; LMNP: Lake Manyara National Park; BWMA: Burunge Wildlife Management Area; RWMA: Randilen Wildlife Management Area; MR: Manyara Ranch; MGCA: Mto wa Mbu Game Controlled Area) in the Tarangire Ecosystem, northern Tanzania. The plotted values represent density estimates for each season (LR: long rains; Dry: dry season; SR: short rains) over different years, based on model predictions. Initial density estimates (lighter circles) are averages from the first two years of the time series, and final density estimates (bolder circles) are from the last two years (note varying time series lengths across areas). Density estimates were derived from terrestrial line distance sampling surveys, using Monte Carlo simulations and generalized additive models to model trends. Percent changes indicate the relative difference in average population density between initial and final periods.
Our results confirm previous observations that the oldest CBCM model in the ecosystem, Manyara Ranch (established in 2001), sustains densities of giraffe, zebra, and wildebeest that are similar to those observed in adjacent Tarangire NP19 and much higher than in the adjacent Mto wa Mbu GCA26 as well as a relatively small (compared to Tarangire NP) population of mostly male elephant40. These relatively high densities of giraffe, zebra, and wildebeest (and other species19) have been sustained over a long time span, suggesting that the concept of Manyara Ranch appears to be working for conserving the current densities of large herbivores. Likely, this is attributed to effective anti-poaching efforts and the enforcement of a limited grazing regime. The grazing strategy aims to build up sufficient grass biomass during the rainy season, allowing for dry season grazing of surrounding communities’ livestock. Data associated with these interventions are currently not accessible, yet analyzing such data would be valuable for assessing the effectiveness of these conservation interventions.
Wildlife monitoring efforts in the Burunge WMA (established in 2006) further highlight the contribution of CBCMs in conserving large herbivores in the ecosystem. Marked increases in elephant densities over time (Fig. 2c), increased density of giraffe after an initial decrease (Fig. 3c), zebra densities greater than outside the area dedicated to wildlife21, and a growing wildebeest population (Fig. 5c) that reached densities similar to those observed in neighboring, fully protected Tarangire NP22 suggest that Burunge WMA has been effective in conserving large herbivores. This is substantiated by a before-after-control-impact study which documented that adjusting management in Burunge MWA improved giraffe survival21.
In the case of the newest CBCM of the ecosystem, Randilen WMA (established in 2014), the contribution to herbivore conservation efforts is perhaps not as obvious. However, as Randilen WMA is not situated in the core migratory routes of the ecosystem28,41,42, it is not surprising that zebra and wildebeest densities were low, or occurred sporadically during the rainy season, and did not seem to have increased systematically over the relatively short monitoring period. Nevertheless, the seemingly positive trends in the giraffe population (also supported by more detailed studies within the Randilen WMA20) is encouraging and suggest that resident wildlife populations may have benefitted from the implementation of similar management activities to those instituted in Burunge WMA: the effective protection of rangelands from conversion to agriculture or settlements, locally enforced grazing regulations, and community enforcement of anti-poaching measures20,22.
During the last two decades, most conservation efforts have focused on areas outside established protected areas in the central part of the ecosystem, i.e. areas around Tarangire NP that are essential for the seasonal migration in and out of the national park. A substantial fraction of these migratory routes are now protected by CBCMs and other community land-use regulations12,43, with only a few bottlenecks remaining along the migratory routes28,42. Our ground-based monitoring indicates that the zebra population has increased (Fig. 4a) and that the wildebeest population has stabilized and possibly even increased during the dry season (Fig. 5a) in Tarangire NP from 2011 to 2019. As both species spend approximately half of the year outside Tarangire NP in village lands and CBCMs, concentrating in Tarangire NP only during the dry season, these population increases provide circumstantial evidence that conservation measures outside Tarangire NP are effective in bolstering migratory ungulate populations. Since Tarangire NP constitutes the main dry season range for wildebeest and zebra34,44, it is likely that these population size trajectories are due to intrinsic population growth and not due to immigration. On the other hand, demographic monitoring of the Tarangire elephant population documented rapid population growth over the past decades, once poaching was effectively curbed in the 1990s45. This rapid growth is not evident from our line transect monitoring for the core of Tarangire NP (Fig. 2a).
Our comparison of wildlife trends within the ecosystem (summarized in Fig. 6) supports the idea that conserving functional connectivity8,13 is key to supporting populations of large herbivores in the ecosystem. Little to no conservation efforts have been directed towards protecting the remaining connectivity between Lake Manyara NP and the wider ecosystem. While factors internal to Lake Manyara NP (especially bush encroachment)46, may have contributed to stagnant and low densities of zebra and wildebeest (which have become largely resident) and a declining population of elephant, it is plausible that the insularization of Lake Manyara NP47 contributed to this worrisome development. There remains some connectivity38, as evidenced by one documented giraffe movement between Manyara Ranch to Lake Manyara NP and back again48. Further, high levels of precipitation in 2019 and 2020 swamped the shortgrass plains habitat along Lake Manyara, and in 2021, no wildebeest were observed in Lake Manyara NP (DEL and MLB, pers. obs.). It is assumed these animals moved out of Lake Manyara NP, again suggesting some connectivity remains.
Similar to Lake Manyara, population densities of resident large herbivores in the Mto wa Mbu GCA, where there are limited anti-poaching measures, are very small (giraffe), functionally absent (elephant), and generally well below historical baselines35. Nevertheless, wildlife still occurs in this area and the landscape seems to be permeable for wildlife, especially for wildebeest and zebra which use this area for their annual migration to the Northern Plains28, even though agricultural and infrastructural development threaten this functional connectivity in several locations49.
Small, isolated populations are more vulnerable to extinction than large, connected populations because of stochastic demographic, environmental, and genetic threats50. In light of these threats, and the presence of small populations in the Tarangire Ecosystem such as in Lake Manyara NP, our results suggest that the conservation value of the CBCMs is at least twofold: (1) they effectively increase the area of suitable habitat well beyond core protected areas such as national parks; (2) they support densities of resident large herbivores comparable to those in national parks, thereby increasing effective population sizes of these species in the ecosystem. In addition, CBCMs effectively protect areas that are essential parts of migratory routes in the ecosystem and provide access to seasonally available resources28,42, and thus likely contribute to population increases of migratory ungulate populations inside Tarangire NP. Consequently, CBCMs in the Tarangire Ecosystem have been instrumental in preserving space for the annual migration of large herbivores and maintaining ecosystem functioning associated with migratory and resident wildlife populations51.
In times of a global biodiversity crisis52 and widespread wildlife declines across Africa2,53, halting these declines and demonstrating population increases in newly established CBCMs are important steps towards longer-term conservation success. However, such ‘success’ should be viewed within the context of long-term environmental processes and cognitive biases such as the shifting baseline syndrome54. We are aware that wildlife populations were historically much more numerous in the Tarangire Ecosystem31. Historical anecdotes and long-term data further affirm higher wildlife densities in the Tarangire Ecosystem in the past31,35. For instance, based on aerial ecosystem-wide surveys, the wildebeest population in the ecosystem exhibited markedly greater densities during the late 1980s and early 1990s, with an average population size from 1987 to 1994 estimated at approximately 39,000 wildebeest. In contrast, by 2011, it had dwindled to only around 12,000 wildebeest34. Because human impacts on wildlife populations started well before the first biodiversity assessments were conducted55, we may never know the “true” potential for wildlife densities in the ecosystem, which in any case may have fluctuated considerably. For instance, the first dry season counts of wildebeest and zebra in Tarangire NP during the early 1960s estimated only 1,200 and 2,500 animals, respectively, and while this was likely only a portion of the total Tarangire Ecosystem population, populations of both species had already suffered significant declines in the previous 50 years, which was attributed to a decline in dry season water sources27. Given the past and current human population growth rate in Tanzania56 and associated need for land required for infrastructure and agriculture57, it is debatable if historical wildlife population sizes (if they were ever to be known) are realistic quantitative targets for ecosystem restoration efforts. Critical wildlife habitats, such as dry season concentration areas north of Lake Manyara NP27, have been irrevocably lost for wildlife35. These substantial and likely irreversible landscape changes have likely reduced the overall carrying capacity for wildlife in the ecosystem. However, it is plausible that wildlife densities in the Tarangire Ecosystem have the potential to increase even under the current extent of core protected areas, combined with the growing extent of multi-use areas where wildlife coexists with livestock and people (e.g. Manyara Ranch, and village lands designated primarily for livestock grazing). Our findings demonstrate that such multi-use, locally managed conservation areas can contribute to the restoration of wildlife numbers in important migratory routes and seasonal ranges, which, with appropriate support, could provide the basis for further wildlife recoveries in the ecosystem over time.
In the last three decades, the main strategy for ecosystem restoration in the Tarangire Ecosystem has been to focus conservation efforts on protecting functional connectivity by creating diverse CBCMs to connect dry and wet season ranges of migratory wildlife. Overall, site-based monitoring suggests that this pragmatic conservation approach was effective. Considering that the majority of wildebeest and zebra spend approximately half of the year outside protected areas, crossing several main roads and traversing areas of human settlement during their seasonal migration, this is a remarkable example for human-wildlife coexistence in a human-dominated landscape, with important implications for conservation policy in East African savanna rangelands and potentially beyond. At the same time, we caution against excessive contentedness with the achieved outcomes. Foremost, several bottlenecks along migratory routes are still threatened by land-use change and expanding settlements; effectively securing these remaining gaps through community-supported actions should be prioritized and may yield a high return on investment for ecosystem conservation efforts49. Considering much greater densities of migratory wildlife species during the not too distant past31, we encourage conservationists to formulate and strive for bolder goals for the restoration of wildlife populations in the Tarangire Ecosystem. Such ecological restoration goals are likely best achieved if conservation measures are designed as social-ecological endeavors58, support Indigenous land and resource tenure and management systems that foster coexistence of livestock and wildlife, ensure that people benefit from increasing wildlife populations, and provide cost-effective ways to minimize costs associated with increasing wildlife populations such as crop damages, livestock depredation, threats to human wellbeing and opportunity, and transaction costs associated with preventing wildlife-related damages. Assessing the effectiveness of such restoration efforts not only requires renewed investment in long-term and ecosystem-wide wildlife monitoring efforts but also in monitoring schemes that assess indicators of social sustainability in CBCMs.
Methods
Study area
The climate of the Tarangire Ecosystem is semi-arid and characterized by a bimodal rainfall pattern: a long dry season (June to October) is followed by a short rainy season (November to December), a short dry season (January to February), and then long rains (March to May). Annual precipitation ranges between 434 and 824 mm; the vegetation is characterized by mosaics of Vachellia-Commiphora bushland and woodland, edaphic grasslands, and riverine vegetation59. The extent of the ecosystem (c. 30,000 km2) encompasses the annual movement range of migratory grazers (Fig. 1): during the dry season, zebra and wildebeest mainly concentrate in the northern part of Tarangire NP (total area c. 2650 km2) and Manyara Ranch (total area c. 182 km2). In these areas, surface water (in Tarangire NP provided by the Tarangire River, the Silale swamps, and some human-enhanced waterholes; in Manyara Ranch provided by the Makuyuni River and human-made dams) and sufficient grass biomass is available during the dry season. At the onset of the short rains, wildebeest and zebra leave Tarangire NP. About half of the wildebeest population migrates eastwards to the Simanjiro Plains and the other half migrates to the northern plains near Lake Natron44,60 where females give birth in the nutrient and mineral rich grasslands61. Around June, as the surface water on the plains dries up, they return to their dry season ranges.
The eastward migration to the Simanjiro Plains is facilitated by Certificates of Customary Rights of Occupancy (CCRO) that began as conservation easements in 2006. The CCROs secure legal communal title over lands used traditionally for seasonal livestock grazing. These rangelands are conserved by local communities to protect their livestock grazing areas; CCRO by-laws permit livestock keeping but agriculture and settlements are not allowed12. The northern migration is strengthened by Manyara Ranch (established in 2001), an unfenced area managed for coexistence between livestock and wildlife; here, pastoralists of two adjacent communities are granted grazing rights and anti-poaching and grazing laws are enforced by rangers19. Along the northern migration route, wildlife moves through the Mto wa Mbu GCA, where few restrictions on natural resource extraction are in place28,35. Land-use changes from rangeland to agriculture and settlements are constricting wildlife movements42. Giraffe62 and elephant63 do not typically move as far and as predictably as wildebeest and zebra, yet both species have large annual home ranges that exceed the boundaries of protected areas45,62. Savanna elephants are mixed feeders (grazing and browsing) and considered a facultative partially migratory species64 whereas Masai giraffes are primarily browsers and are considered a resident species with seasonal movements65. Both species occur year-round in all of our study sites.
Adjacent to Tarangire NP, several villages established two Wildlife Management Areas, i.e. community-based conservation and development areas that are spatially structured by land-use plans. Burunge WMA (c. 220 km2 delineated for wildlife conservation), officially gazetted in 2006, lies to the west of Tarangire NP and connects to Lake Manyara NP21,22 whereas Randilen WMA (c. 300 km2 delineated for wildlife conservation), established in 2014, is located northeast of Tarangire NP20. In the wildlife areas of Burunge and Randilen WMA, agriculture and permanent settlements are not allowed, and village game scouts enforce community land-use regulations and protect wildlife from illegal hunting. The ecosystem also contains the Mkungunero Game Reserve, located between Tarangire NP and the Makame WMA. Some portion of the zebra and wildebeest migrate to these areas during the wet season. However, for these areas we do not have long-term wildlife monitoring data.
Lake Manyara NP (lowland area covers c. 168 km2), located at the western edge of the ecosystem, is increasingly isolated due to human development along its northern and southern boundaries. Wildlife in this NP is mostly resident, although occasional movement does occur42,48. In Tarangire and Lake Manyara NPs, conservation authorities restrict human use to photographic tourism and research.
Area-specific density estimates
All wildlife surveys were conducted with permission from the Tanzania Commission for Science and Technology, Tanzania Wildlife Research Institute, Tanzania National Parks, Burunge and Randilen WMA, Manyara Ranch, and the villages of Losirwa, Esilalei, Mswakini, Lolkisale, and Emboret. All methods were carried out in accordance with relevant guidelines and regulations, and our observational studies did not involve experiments with animals. Starting in 2011, we established site-specific wildlife monitoring across multiple management units of the ecosystem. From the end of 2011 to the end of 2019 (see Table 1 for start and end months of the monitoring efforts), we surveyed Lake Manyara NP, Tarangire NP, Manyara Ranch, and the Mto wa Mbu GCA three times per year (24 surveys; in 2019, we omitted the dry season survey) to capture the main seasons66. In Burunge WMA, we conducted seven surveys from 2011 to 201822 and in Randilen WMA, we monitored wildlife populations during twelve seasonal surveys from 2012 to 201520. These monitoring efforts were designed as line transect surveys67, with transects mostly following roads (except for Burunge WMA, where transects were placed systematically). In Tarangire and Lake Manyara NP, Manyara Ranch, and Mto wa Mbu GCA, transect length was typically 2 km and consecutive transects were separated by 0.5 km; in Burunge WMA, transect length ranged from 0.5 to 10.3 km; in Randilen WMA, we used one single transect per survey (23–85 km, length varied due to road conditions). For most surveys, we used open-top 4 WD vehicles and slowly drove (10–20 km h-1) along transects. A minimum of two trained observers counted animals along transects. In Burunge WMA, approximately half of the transects were walked by groups of three persons each. Upon detecting an animal or animal group, we moved to a position perpendicular to the sighting, stopped the vehicle or halted the walk, counted the number of individual animals, and measured the perpendicular distance between the transect and the initial position of the animal using a laser rangefinder. If animals occurred in a group, we measured the perpendicular distance to the approximate center of the group. These data have been analyzed and published previously19,20,22,46,66, but we re-analyzed these data for consistency and comparison within the same modeling framework.
To estimate densities we used Distance 6.067. Line distance methodology accounts for imperfect detection and explicitly models the probability \(\widehat{P}\) of detecting an animal as a function of the distance from the transect. Due to the non-random placement of transects, density estimates are possibly biased46 (but see19,24). Therefore, we focus on the temporal trend of density estimates and density comparisons across sites. In cases when surveys were repeated within a single season, we summed effort and sightings68. We truncated the farthest 10% of distances, fit species- and area-specific half-normal detection functions with cosine extension69, and used the mean cluster size of each season to extrapolate from cluster to animal density. Previously, we tested if detection functions were mediated by season19,66; since including this covariate was not supported by model selection, detection functions were pooled across all seasons. Sample size exceeded the recommended threshold of 60 sightings for fitting robust detection functions68 in the majority (18/22) of species-area combinations (Table S2). Based on Kolmogorov–Smirnov goodness of fit tests (Table S2), the majority of detection models (16/22) fit the observed data well. Visual assessment of detection functions (Figs. S1–S4) suggested a relatively good fit but also indicated that target species occasionally either avoided (few detections in first distance bins) or were attracted to roads (steep peak of detections in first distance bin). Based on the derived global detection models, we used the post-stratification option in Distance 6.0 for estimating season-specific densities (Table S3). This option allowed us to generate separate density estimates for each season-year-species combination. The density estimate Dij for each area i in year-season combination j was computed as:
where \({n}_{ij}\) is the number of detections in stratum i during year-season j; \(w\) is the effective strip half-width; \({L}_{ij}\) is the total length of transects surveyed in stratum i during year-season j, and \(\widehat{P}\) is the global detection function.
Trend analysis
To assess overall temporal trends for site-specific population trends, we used generalized additive models coupled with a two-stage Monte Carlo sampling approach which enhances the robustness of our analyses by propagating uncertainties from both the distance sampling-derived density estimates and the subsequent trend analysis53. Furthermore, this methodological choice was driven by the heterogeneity in our dataset, characterized by unequal year-season combinations across areas. Consequently, we constructed area-, species- and season-specific time series. One exception was the Burunge WMA, where sampling occurred more sporadically. In this case, we aggregated all survey data, disregarding seasonal variations to maintain consistency in trend analysis.
The first stage of our Monte Carlo simulation, implemented in R 4.2.270, involved generating 1000 replicates for each density estimate. We achieved this by simulating data points within the 95% confidence intervals of the original density estimates. For zero densities, we used the normal distribution. For non-zero densities, we employed truncated normal distributions (implemented via the truncnorm package71), bounded by the lower (L) and upper (U) limits of the estimated 95% confidence intervals. For each density estimate (D), we calculated the standard deviation (SDestimated) as:
We then generated simulated densities (Dsimualted) as follows:
Subsequently, for each species-, area-, and season-specific time series, we fitted a generalized additive model using the mgcv package72. In these models, the year (Y) was treated as a smooth function, enabling us to capture non-linear temporal trends without presupposing any specific functional form. The target variable for these models was the set of 1,000 simulated data points generated in the first Monte Carlo stage. The model can be expressed as:
where \(s(Y, k=4)\) represents a smoothing spline function with a basis dimension of 4, allowing for the description of non-linear trends.
The second stage of our Monte Carlo methodology involved using the fitted generalized additive models to predict yearly density estimates. For each predicted year, we generated 1,000 simulated values. These simulations were designed to encompass the uncertainty inherent in the model predictions. We obtained the prediction (Dpredicted) and the associated standard error (SE) for each year (Yp) from each model. We then simulated new density values based on the predicted mean and standard errors:
We then used the mean values of these 1000 simulated values per year to depict trend lines for each seasonal time series. This two-stage Monte Carlo approach, coupled with the flexibility of generalized additive models, allowed us to produce a nuanced and statistically robust analysis of population dynamics across species and ecological contexts. We visualized the time series by plotting the observed density estimates and the predicted yearly trends in ggplot273.
To separate overall population trends from noise (arising from uncertainty in density and trend estimates), we condensed the key information of the time series. For each species, area, and season, we calculated the average density estimates for the initial and final years of the modelled time series. The initial and final density estimates were derived from the first and last two years of the monitoring period, respectively. For Randilen WMA, the initial years were 2012–2013 and the final years were 2014–2015. For Tarangire NP, Lake Manyara NP, Manyara Ranch, and Mto wa Mbu GCA, the initial and final years varied by season: short rains (2011–2012 to 2018–2019), long rains (2012–2013 to 2018–2019), and dry season (2012–2013 to 2017–2018; in 2019 we did not conduct dry season counts). For Burunge WMA, we aggregated data from 2011–2012 and 2017–2018 across all seasons. We then computed the percent change in density between the initial and final periods using the formula:
This approach smooths out non-linearity and provides an indication of the direction and magnitude of population changes over the study period.
References
Bolger, D. T., Newmark, W. D., Morrison, T. A. & Doak, D. F. The need for integrative approaches to understand and conserve migratory ungulates. Ecol. Lett. 11, 63–77 (2008).
Craigie, I. D. et al. Large mammal population declines in Africa’s protected areas. Biol. Conserv. 143, 2221–2228 (2010).
Western, D., Russell, S. & Cuthil, I. The status of wildlife in protected areas compared to non-protected areas of Kenya. PLoS ONE 4, e6140 (2009).
Stoner, C. et al. Assessment of effectiveness of protection strategies in Tanzania based on a decade of survey data for large herbivores. Conserv. Biol. 21, 635–646 (2007).
Ogutu, J. O. et al. Extreme wildlife declines and concurrent increase in livestock numbers in Kenya: What are the causes?. PLoS ONE 11, e0163249 (2016).
Lindsey, P. A. et al. The bushmeat trade in African savannas: Impacts, drivers, and possible solutions. Biol. Conserv. 160, 80–96 (2013).
Prins, H. H. T. The pastoral road to extinction: Competition between wildlife and traditional pastoralism in East Africa. Environ. Conserv. 19, 117–123 (1992).
Fynn, R. W. S. & Bonyongo, M. C. Functional conservation areas and the future of Africa’s wildlife. Afr. J. Ecol. 49, 175–188 (2011).
Riggio, J. & Caro, T. Structural connectivity at a national scale: Wildlife corridors in Tanzania. PLoS ONE 12, e0187407 (2017).
Kremen, C. & Merenlender, A. M. Landscapes that work for biodiversity and people. Science 362, eaau6020 (2018).
Ingram, J. C. et al. Evidence of payments for ecosystem services as a mechanism for supporting biodiversity conservation and rural livelihoods. Ecosyst. Serv. 7, 10–21 (2014).
Nelson, F. et al. Payments for ecosystem services as a framework for community-based conservation in northern Tanzania. Conserv. Biol. 24, 78–85 (2010).
Kauffman, B. M. J. et al. Mapping out a future for ungulate migrations. Science 372, 566–569 (2021).
Russell, S., Tyrrell, P. & Western, D. Seasonal interactions of pastoralists and wildlife in relation to pasture in an African savanna ecosystem. J. Arid Environ. 154, 70–81 (2018).
Nelson, F. Natural conservationists? Evaluating the impact of pastoralist land use practices on Tanzania’s wildlife economy. Pastoralism 2, 1–19 (2012).
Sutherland, W. J. et al. One hundred questions of importance to the conservation of global biological diversity. Conserv. Biol. 23, 557–567 (2009).
Newmark, W. D. & Hough, J. L. Conserving wildlife in Africa: Integrated conservation and development projects and beyond. Bioscience 50, 585–592 (2000).
Brooks, J., Waylen, K. A. & Mulder, M. B. Assessing community-based conservation projects: A systematic review and multilevel analysis of attitudinal, behavioral, ecological, and economic outcomes. Environ. Evid. 2, 2 (2013).
Kiffner, C. et al. Long-term persistence of wildlife populations in a pastoral area. Ecol. Evol. 10, 10000–10016 (2020).
Lee, D. E. & Bond, M. L. Quantifying the ecological success of a community-based wildlife conservation area in Tanzania. J. Mammal. 99, 459–464 (2018).
Lee, D. E. Evaluating conservation effectiveness in a Tanzanian community wildlife management area. J. Wildl. Manag. 82, 1767–1774 (2018).
Kiffner, C. et al. Community-based wildlife management area supports similar mammal species richness and densities compared to a national park. Ecol. Evol. 10, 480–492 (2020).
Riggio, J. et al. Sensitivity of Africa’s larger mammals to humans. J. Nat. Conserv. 43, 136–145 (2018).
Bierhoff, L., Bond, M. L., Ozgul, A. & Lee, D. E. Anthropogenic and climatic drivers of population densities in an African savanna ungulate community. Popul. Ecol. https://doi.org/10.1002/1438-390X.12182 (2024).
Ogutu, J. O. O., Kuloba, B., Piepho, H. & Kanga, E. Wildlife population dynamics in human- dominated landscapes under community-based conservation: the example of Nakuru wildlife conservancy, Kenya. PLoS ONE 12, e0169730 (2017).
Kiffner, C., Nagar, S., Kollmar, C. & Kioko, J. Wildlife species richness and densities in wildlife corridors of Northern Tanzania. J. Nat. Conserv. 31, 29–37 (2016).
Lamprey, H. F. Estimation of the large mammal densities, biomass and energy exchange in the Tarangire Game Reserve and the Masai Steppe in Tanganyika. East African Wildl. J. 2, 1–45 (1964).
Bond, M. L., Bradley, C. M., Kiffner, C., Morrison, T. A. & Lee, D. E. A multi-method approach to delineate and validate migratory corridors. Landsc. Ecol. 32, 1705–1721 (2017).
Hopcraft, J. G. C. et al. Competition, predation, and migration: Individual choice patterns of Serengeti migrants captured by hierarchical models. Ecol. Monogr. 84, 355–372 (2014).
Morrison, T. A. et al. Informing aerial total counts with demographic models: Population growth of Serengeti elephants not explained purely by demography. Conserv. Lett. 11, e12413 (2018).
Prins, H. H. T. & de Jong, J. F. The ecohistory of Tanzania’s Northern Rift Valley—can one establish an objective baseline as an endpoint for ecosystem restoration? In Tarangire: Human-Wildlife Coexistence in a Fragmented Ecosystem (eds Kiffner, C. et al.) 129–161 (Springer, 2022).
Clements, H. S. et al. The bii4africa dataset of faunal and floral population intactness estimates across Africa’s major land uses. Sci. data 11, 191 (2024).
Kiffner, C., Bond, M. L. & Lee, D. E. Human-wildlife interactions in the Tarangire Ecosystem. In Tarangire: Human-Wildlife Coexistence in a Fragmented Ecosystem (eds Kiffner, C. et al.) 3–22 (Springer, 2022). https://doi.org/10.1007/978-3-030-93604-4_1.
Morrison, T. A., Link, W. A., Newmark, W. D., Foley, C. A. H. & Bolger, D. T. Tarangire revisited: Consequences of declining connectivity in a tropical ungulate population. Biol. Conserv. 197, 53–60 (2016).
Kiffner, C., Wenner, C., LaViolet, A., Yeh, K. & Kioko, J. From savannah to farmland: effects of land-use on mammal communities in the Tarangire-Manyara ecosystem, Tanzania. Afr. J. Ecol. 53, 156–166 (2015).
Brehony, P., Morindat, A. & Sinandei, M. Land tenure, livelihoods, and conservation: Perspectives on priorities in Tanzania’s Tarangire Ecosystem. In Tarangire: Human-Wildlife Coexistence in a Fragmented Ecosystem (eds Kiffner, C. et al.) 85–108 (Springer International Publishing, 2022). https://doi.org/10.1007/978-3-030-93604-4_5.
Bond, M. L., Lee, D. E. & Paniw, M. Extinction risks and mitigation for a megaherbivore, the giraffe, in a human-influenced landscape under climate change. Glob. Chang. Biol. https://doi.org/10.1111/gcb.16970 (2023).
Riggio, J. et al. Predicting wildlife corridors for multiple species in an East African ungulate community. PLoS ONE 17, e0265136 (2022).
Raycraft, J. Community attitudes towards Randilen Wildlife Management Area. In Tarangire: Human-Wildlife Coexistence in a Fragmented Ecosystem (eds Kiffner, C. et al.) 109–125 (Springer International Publishing, 2022). https://doi.org/10.1007/978-3-030-93604-4_6.
Kioko, J., Zink, E., Sawdy, M. & Kiffner, C. Elephant (Loxodonta africana) demography and behaviour in the Tarangire-Manyara ecosystem, Tanzania. S. Afr. J. Wildl. Res. 43, 44–51 (2013).
Borner, M. The increasing isolation of Tarangire National Park. Oryx 19, 91–96 (1985).
Morrison, T. A. & Bolger, D. T. Connectivity and bottlenecks in a migratory wildebeest Connochaetes taurinus population. Oryx 48, 1–9 (2014).
Nelson, F., Muyamwa-Mupeta, P., Muyengwa, S., Sulle, E. & Kaelo, D. Progress or regression? Institutional evolutions of community-based conservation in eastern and southern Africa. Conserv. Sci. Pract. https://doi.org/10.1111/csp2.302 (2020).
Morrison, T. A. & Bolger, D. T. Wet season range fidelity in a tropical migratory ungulate. J. Anim. Ecol. 81, 543–552 (2012).
Foley, C. A. H. & Foley, L. S. The history, status, and conservation of the elephant population in the Tarangire Ecosystem. In Tarangire: Human-Wildlife Coexistence in a Fragmented Ecosystem (eds Kiffner, C. et al.) 209–232 (Springer, 2022). https://doi.org/10.1007/978-3-030-93604-4_10.
Kiffner, C. et al. Long-term population dynamics in a multi-species assemblage of large herbivores in East Africa. Ecosphere 8, e02027 (2017).
Newmark, W. D. Insularization of Tanzanian parks and the local extinction of large mammals. Conserv. Biol. 10, 1549–1556 (1996).
Lee, D. E. & Bolger, D. T. Movements and source–sink dynamics of a Masai giraffe metapopulation. Popul. Ecol. 59, 157–168 (2017).
Lohay, G. G., Riggio, J., Lobora, A. L., Kissui, B. M. & Morrison, T. A. Wildlife movements and landscape connectivity in the Tarangire Ecosystem. In Tarangire: Human-Wildlife Coexistence in a Fragmented Ecosystem (eds Kiffner, C. et al.) 255–276 (Springer, 2022). https://doi.org/10.1007/978-3-030-93604-4_12.
Wilcox, B. A. & Murphy, D. D. Conservation strategy: The effects of fragmentation on extinction. Am. Nat. 125, 879–887 (1985).
Western, D. et al. Conservation from the inside-out: Winning space and a place for wildlife in working landscapes. People Nat. 2, 279–291 (2020).
IPBES. Global Assessment Report on Biodiversity and Ecosystem Services. https://www.ipbes.net/global-assessment-report-biodiversity-ecosystem-services (Accessed 2 Oct 2019) (2019).
Chase, M. J. et al. Continent-wide survey reveals massive decline in African savannah elephants. PeerJ 4, e2354 (2016).
Pauly, D. Anecdotes and the shifting baseline syndrome of fisheries. Trends Ecol. Evol. 10, 430 (1995).
Mahoub, J.-B. et al. Setting temporal baselines for biodiversity: The limits of available monitoring data for capturing the full impact of anthropogenic pressures. Sci. Rep. 7, 41591 (2017).
National Bureau of Statistics (NBS) & Office of Chief Government Statistician Zanzibar (OCGS). 2012 Population and Housing Census: Population Distribution by Administrative Units: Key Findings. United Repub. Tanzania 1–10 (2013).
Msoffe, F. U. et al. Spatial correlates of land-use changes in the Maasai- Steppe of Tanzania: Implications for conservation and environmental planning. Int. J. Biodivers. Conserv. 3, 280–290 (2011).
Fischer, J., Riechers, M., Loos, J., Martin-Lopez, B. & Temperton, V. M. Making the UN decade on ecosystem restoration a social-ecological endeavour. Trends Ecol. Evol. 36, 20–28 (2021).
Prins, H. H. T. & Loth, P. E. Rainfall patterns as background to plant phenology in northern Tanzania. J. Biogeogr. 15, 451–463 (1988).
Kahurananga, J. & Silkiluwasha, F. The migration of zebra and wildebeest between Tarangire National Park and Simanjiro Plains, northern Tanzania, in 1972 and recent trends. Afr. J. Ecol. 35, 179–185 (1997).
Voeten, M. M., Van de Vijver, C. A. D. M., Olff, H. & Van Langevelde, F. Possible causes of decreasing migratory ungulate populations in an East African savannah after restrictions in their seasonal movements. Afr. J. Ecol. 48, 169–179 (2010).
Knüsel, M. A., Lee, D. E., König, B. & Bond, M. L. Correlates of home range sizes of giraffes, Giraffa camelopardalis. Anim. Behav. 149, 143–151 (2019).
Galanti, V., Tosi, G., Rossi, R. & Foley, C. The use of GPS radio-collars to track elephants (Loxodonta africana) in the Tarangire National Park (Tanzania). Hystrix 11, 27–37 (2000).
Purdon, A., Mole, M. A., Chase, M. J. & Van Aarde, R. J. Partial migration in savanna elephant populations distributed across southern Africa. Sci. Rep. https://doi.org/10.1038/s41598-018-29724-9 (2018).
Pellew, R. A. The feeding ecology of a selective browser, the giraffe (Giraffa camelopardalis tippelskirchi). J. Zool. 202, 57–81 (1984).
Kiffner, C. et al. Wildlife population trends as indicators of protected area effectiveness in northern Tanzania. Ecol. Indic. 110, 105903 (2020).
Thomas, L. et al. Distance software: design and analysis of distance sampling surveys for estimating population size. J. Appl. Ecol. 47, 5–14 (2010).
Buckland, S. T., Anderson, D. R., Burnham, K. P., Laake, J. L. & Borchers, D. L. Introduction to Distance Sampling. Estimating Abundance of Biological Populations (Oxford University Press, 2001).
Prieto Gonzalez, R., Thomas, L. & Marques, T. A. Estimation bias under model selection for distance sampling detection functions. Environ. Ecol. Stat. 24, 399–414 (2017).
R Core Team. R: A language and environment for statistical computing. http://www.r-project.org/ (2021).
Mersmann, O., Trautmann, H., Steuer, D. & Bornkamp, B. Package ‘truncnorm’. 1–4 (2023). https://github.com/olafmersmann/truncnorm (Accessed 11 Jan 2024).
Wood, S. N. Package ‘mgcv’. (2016). https://cran.r-project.org/web/packages/mgcv/mgcv.pdf (Accessed 2 Jan 2017).
Wickham, H. ggplot2: Elegant Graphics for Data Analysis. https://doi.org/10.1007/978-0-387-98141-3 (2016).
Acknowledgements
We thank the following institutions for funding the wildlife surveys: African Wildlife Foundation, Columbus Zoo, Chem Chem Associations, IGF Foundation, PAMS Foundation, Rufford Foundation, Sacramento Zoo, The School for Field Studies, Tulsa Zoo. We sincerely thank all rangers, SFS staff and students who participated in the surveys.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Contributions
C.K. designed the study, carried out fieldwork in Lake Manyara NP, Tarangire NP, Manyara Ranch, Burunge WMA and the Mto wa Mbu GCA, analyzed the data, created the figures and wrote the first draft of the manuscript. C.A.H.F. designed the study and contributed to the writing of the manuscript. D.E.L. designed the study, carried out fieldwork in Randilen WMA, and contributed to the writing of the manuscript. M.L.B. carried out fieldwork in Randilen WMA and critically reviewed and edited the manuscript. J.K. carried out fieldwork in Lake Manyara NP, Tarangire NP, Manyara Ranch and the Mto wa Mbu GCA, and critically reviewed and edited the manuscript. B.M.K. carried out fieldwork in Burunge WMA, and critically reviewed and edited the manuscript. A.L.L. critically reviewed and edited the manuscript. L.S.F. critically reviewed and edited the manuscript. F.N. designed the study, provided funding, and contributed to the writing of the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Kiffner, C., Foley, C.A.H., Lee, D.E. et al. The contribution of community-based conservation models to conserving large herbivore populations. Sci Rep 14, 16221 (2024). https://doi.org/10.1038/s41598-024-66517-9
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-024-66517-9
- Springer Nature Limited