1 Introduction

The global challenge of biodiversity conservation is the conversion of natural lands, such as forests, wetlands, and grasslands, into agricultural and developed lands. According to [1], the conservation of wildlife necessitates a thorough understanding of each species' preferred habitat. Land use changes outside of protected areas (PAs) also have a substantial impact on wildlife, which needs a vast home range. For instance, land use changes in Nepal's Trans-Himalaya Region caused blue sheep (Pseudois nayaur) to move their foraging to lower elevations, luring snow leopard (Panthera uncia) from their higher-elevation habitats to lower sites, leading to severe human-wildlife conflicts [2]. Increased human-wildlife conflicts are also linked to land use changes in developed nations including China [3].

Humans and wildlife have coexisted for centuries and wildlife has played an important role in human survival and cultural development [4]. Historically, larger wild African mammals such as Hippopotamus (Hippopotamus amphibius), African Elephants (Loxodonta africana), buffaloes (Syncerus caffer) and giraffes (Giraffa camelopardalis) were considered resources for food, medicine, shelter, and ivory [5]. The increasing protection status of these wild animals and their habitat in recent years has changed the attitudes and perceptions of local people, who now consider these animals as threats rather than resources [6]. The present concept of protected areas (PAs), including national parks, evolved in the mid-1800s during the European colonial era, when different land use types, including farms and grazing lands, were converted to PAs [7]. During this period, the main focus of wildlife protection in Africa was for eventual hunting opportunities, and nature conservation was not the main goal. In the twentieth century, the main focus was changed to nature conservation and maintaining ecological processes and native species by minimizing the influence of humans on natural ecosystems [7]. About 16.8 million km2 of the world’s land surface area is protected, covering more than 100,000 PAs, of which Africa contributes about 4.2 million km2 [8]. In Tanzania, the PAs network covers about 39.6% of the total land area, including game reserves, forest reserves, game controlled areas, national parks, and the Ngorongoro Conservation Areas [9].

Effective conservation of wildlife species entails a complete knowledge of habitat preferences and requirements from within and outside protected areas (PAs) [5, 10, 11]. LUC are causing a greater challenge in the conservation and management of wildlife species, especially in Africa where most of the PAs are not fenced [11,12,13]. For instance, in the Masai Mara–Serengeti ecosystem, a 25% decline in seven wild ungulate populations was reported due to LUC dominated by human–wildlife conflicts [14]. Additionally, LUC is also reported to escalate human-wildlife conflicts in the Rombo area, located between Kilimanjaro National Park in Tanzania and Tsavo West National Park in Kenya [12]. A study adjacent to Ruaha National Park also reported a positive correlation between crop-destruction events and farm proximity to the river and to hippos' access points [15].

LUC is attributed to the increasing human population and the expansion of settlements and agricultural lands, altering the lands that were formally used by wildlife [16, 17]. Different studies also reported alteration of the land that was formally used by wildlife to bloom crop raiding, resulting in livestock kills and endangering human survival and properties, causing conflicts [18,19,20]. In Kenya and Nigeria, human population increase and settlement expansion closer to wildlife habitats are also reported to threaten wildlife survival, including hippos [11, 21]. The same has also been reported in Tanzania, especially in Bunda and Rombo districts, where encroachment of wildlife habitats has brought conflict between wildlife and people [22,23,24]. To deter wildlife incidents, local communities in conflict hotspot areas have been reported to use ditches, local fences, wall hedges, shouting, banging tins and drums, lighting fires, throwing stones, and chili fences [17, 25].

Babati District is among potential agro-ecological zones in Tanzania with fertile soil suitable for agriculture, with diverse vegetation and water catchment areas favorable for wildlife, including Hippos [26, 27]. The district also harbors Lake Babati, recognized as a valuable wildlife habitat that was set aside as a hippo reserve since the colonial era [26]. In the year 2002, the district was upgraded to form the capital of the newly gazetted Manyara Region. The situation attracted more people and amplified the significant human population growth and LUC, which potentially threatened the survival of hippos and other wildlife species [27]. This implies that more land will still be transformed into settlement areas, thus reducing hippos' grazing lands and increasing HHI, resulting in conflicts. Hippos’ are known to foraging at night, and they preference searching for food on terrestrial lands [16, 21]. This contributes to crops raiding especially maize to the villagers close to the lake resulting to conflicts [28]. In spite of this situation, studies relating the LUC to hippos and human wellbeing in this area are scarce. Therefore, the purpose of this study is to assess the status of LUC around Lake Babati in the last twenty years, the time of human settling in areas adjacent to the lake in relation to land use changes and hippos interaction, the variation of farm size with respect to distance from the lake and hippos interaction, respondents' perceptions of the trend of hippos incidents, and the mitigation measures used.

2 Methods

2.1 Study site

The study was carried out in Babati District in the Manyara region, located between latitude 3° and 4° South and longitude 35° and 36° East in Tanzania. The study included the area around Lake Babati with at least 6 km on both sides and three villages, namely Bagara ziwani, Negamsi, and Ngarenaro, around Lake Babati (Fig. 1). The three villages were selected based on the HHI cases and distance from the lake, whereby Bagaraziwani and Ngaranaro villages are located within 2 km from the lake, while Negamsi is located within 2 to 3 km from the lake. The mean annual temperature of the area is 23 °C, with the coldest month being July, with an average temperature of 15 °C, and the warmest month being December, with an average temperature of 34 °C. Rainfall is bimodal, with short rains from October to December and long rains from March to June, with an average annual rainfall of 1200 mm. This rainfall pattern also defines the two cropping seasons of the area. The human population of the area is growing faster and the growth rate was observed to increase by 3.2% in the 2002–2012 years, with 32 people per km2 ranked third nationally for the high population growth rate across districts [27]. Local communities, primarily farmers, mainly cultivate crops including bananas (Musa spp), maize (Zea mays) and common beans (Phaseolus vulgaris) [27]. The communities also engage in other social-economic activities, including tourism in Tarangire and Manyara National Parks and fishing within Lake Babati [26, 27].

Fig. 1
figure 1

Map showing the study villages around Lake Babati in Babati District Manyara region

Full size image

2.2 Satellite image acquisition, processing and analysis

Landsat 7 Enhanced Thematic Mapper (ETM) and Landsat 8 Operational Land Imager (OLI) imagery were used to analyze land cover change in the years 1999 and 2019, respectively. The satellite images with a spatial resolution of 30 m were downloaded from the US Geological Survey (USGS) Earth Explorer (http://earthexplorer.usgs.gov). The selection of images was based on visual quality, acquisition time, and crop growing calendar (Table 1).

Table 1 Satellite image descriptions (source: http://earthexplorer.usgs.gov)
Full size table

Shape files for study area boundaries were obtained through the National Bureau of Statistics (https://openmicrodata.worldpress.com). Satellite images were imported into ArcGIS version 10.8 for processing and analysis, which included image cleaning, compositing, and clipping. The World Geodetic System 1984 (WGS84) was used as the geographic coordinate system, and zone 37S of the Universal Transverse Mercator (UTM) projection was chosen for projection. Natural color bands, which are 3–2–1 color bands for Landsat 7 Enhanced Thematic Mapper (ETM) imagery and 4–3–2 color bands for Landsat 8 Operational Land Imager (OLI) data, were utilized to detect and identify specific features from the images [12]. The maximum likelihood function was used to classify Landsat images, which is recommended to be the most common decision rule in supervised classification [13]. It is considered the best because each pixel is assigned to the class to which it has the highest probability of belonging and hence gives accurate results. Visual interpretation and digital image classification were then combined using GIS functions.

Land use classes defined in the study area included: (i) Water bodies (permanent lakes and rivers); (ii) Agroforestry (mixed perennial crops such as coffee and banana, sprinkled with tall tree species including Albizia spp., Mangifera indica, Citrus spp., Azadirachta indica, and Annona squamosal, Prunus persica, and Guava guava); (iii) seasonal agricultural lands (seasonal cultivated lands, livestock grazing lands, and bare lands); and (iv) Settlement (any type of building).

Accuracy assessment was performed by using ground truthing through error matrix technique [29], whereby 80 direct points were compared with classified images and modified accordingly. Errors in LUC categorization are rather typical. In order to assess the accuracy of the mapping, the classification in this study involved the construction of a matrix error, which is a common way to show the accuracy of classification [13, 30], sampling description, and proportion of each category in the map. The overall accuracy, producer and user accuracies, and kappa coefficient were determined from the created error matrix. Overall accuracy (1), kappa coefficient (2), user’s accuracy (3), and producer’s accuracy (4) were calculated using the following equations:

$$\mathrm{Overall \,accuracy}=\frac{\mathrm{Total \,number \,of \,correctly \,classified \,pixels}}{\mathrm{Total \,number \,of \,refence \,pixels}}\times 100$$
(1)
$$\mathrm{Kappa \,coefficient}=\frac{\left(\mathrm{TS }\times \mathrm{TCCS}\right)-\sum (\mathrm{RT}\times \mathrm{CT})}{{\mathrm{TS}}^{2}-\sum (\mathrm{RT}\times \mathrm{CT})}\times 100$$
(2)

where, TS is the total number of samples (pixels), TCCS is the total number of correctly classified samples (pixels), CT is the total column and RT is the total row.

$$\mathrm{UA}=\frac{\mathrm{Number \,of \,correctly \,classified \,pixels \,in \,each \,category}}{\mathrm{Total \,number \,of \,classified \,pixels \,in \,that \,category }\left(\mathrm{Row \,total}\right)}\times 100$$
(3)
$$\mathrm{PA}=\frac{\mathrm{Number \,of \,correctly \,classified \,pixels \,in \,each \,category}}{\mathrm{Total \,number \,of \,classified \,pixels \,in \,that \,category }(\mathrm{Column \,total})}\times 100$$
(4)

where, UA stands for user’s accuracy and PA stands for producer’s accuracy.

2.3 Data collection

Semi-structured questionnaires with both open-ended and closed-ended questions were administered to 112 respondents for primary data collection. As recommended in [20], for better representation of the population that is statistically acceptable, at least 5% of the total numbers of households per village were randomly selected from the study village register book for interview. Across the study villages, 40, 32, and 40 households were selected from Ngarenaro, Negamsi, and Bagara ziwani, respectively. With the assistance of the village leaders, the selected respondents were those aged 25 and above and who had resided for at least 5 years in the area. These respondents were considered mature enough and had experience with HHI and the LUC in the study area [12]. Before administering questionnaires to respondents, the researchers were introduced by the respective village leaders and asked for their consent to be interviewed. The information captured through semi-structured questionnaires from the agreed respondents through the informed consent included: distance from the lake; time of stay in the village; farm size; trend of hippo visits and destruction to the farm for the last ten years; and mitigation measures used against HHI. From each study village, one focus group discussion (FGDs) with at least 10 participants, including fishermen, farmers, and livestock keepers, was conducted to cross-check and verify additional information, including reaching consensus on the following: the trend of hippos’ visits to the farm for the last ten years; mitigation measures used against HHI and the LUC in the study area. Participant field observation was also conducted by the two researchers during questionnaire administering time across the study villages for 12 days. A road transect walk of within 6 km from the lake was used for the observation, which took place between the hours of 8:00 am and 18:00 pm. The observed data included the various types of current land uses and cover, mitigation strategies employed, and the location and size of the farm in relation to how distant it was from the lake.

2.4 Data analysis

Analysis of the quantitative data and graphical presentations were performed by using R software version 3.6.1. A parametric test (one way ANOVA) was used to find an association between the variables. The variables included distance from the lake and time resided in the village, together with distance from the farm. A generalized linear model was also used to test different mitigation measures used against HHI across study villages. Descriptive statistics were also used to analyze the frequencies of the mentioned variables across villages, and the in-depth information obtained based on people’s perceptions and attitudes was analyzed by using the Statistical Package for Social Science (SPSS 21).

3 Results

3.1 LUC and accuracy assessment in Babati area from the year 1999 to 2019

In this work, stratified random sampling methods were used, and an error matrix was created to evaluate the accuracy of raster images from 1999 to 2019. Overall accuracy and kappa coefficient between 1999 and 2019 were 86%, 89%, and 0.89, 0.87, respectively. The obtained accuracy in all raster layers is acceptable, falling between 0.85 and 1.

During 1999, settlements covered the largest area, followed by seasonal agricultural land, agroforestry, and water (Table 2). Twenty years later (2019), settled land increased further by while other forms of land use continued to decrease. A decrease was observed under seasonal agricultural land, and water suggesting that part of the area formally used for seasonal agricultural and water was converted into settlements (Table 2). Agroforestry was observed to have relatively minor changes (Table1, Fig. 2). LUC cross-tabulation matrix from the year 1999–2019 in Babati area indicated that, the change detection between years favored settlement whereby from 1999 to 2019, 3.5 km2 (2.6%) of the other land use types were converted into agricultural lands. Out of these 0.9 km2 (0.7%), 0.1 km2 (0.1%) and 2.5 km2 (1.8%) were from Water, Agroforestry and seasonal agriculture lands respectively (Table 2). Settlements were also observed to shift from the southern part of the lake to the northern part, which was formally dominated by seasonal agriculture and agroforestry (Fig. 2).

Table 2 LUC changes in Babati area from 1999 to 2019 years
Full size table
Fig. 2
figure 2

LUC in the years 1999 and 2019 in Babati area

Full size image

3.2 Time of the settlement establishment and farming seasons across the distance from the lake

The time for settlement establishment in study villages also varied significantly at f = 9.8597(1) p = 0.00217 and f = 12.791(2) p < 0.0001 with distance from the lake and farming seasons, respectively. The respondents’ time of establishing settlement in the village also increased with the distance from the lake, where the closest ones were from Ngarenaro, followed by Bagara Ziwani, who had settled for 5 years and above, while Negamsi village had settled who had resided for 18 years and above. Moreover, the intensity of cropping varied slightly across the study villages with distance from the lake, whereby the farms closer to the lake in Bagara ziwani village were cropped throughout the year whereas in other villages cropping were intensive during heavy rain season. About 100% of responses agreed that conflicts were severe in areas closer to the lake within the distance of < 3 km. The areas closer to the lake were also reports 89% of respondents to have high incidences of conflicts due to crop raiding especially maize.

3.3 Respondents’ perceptions and mitigation measures against hippos problem in the recent ten years (N = 112)

Across the study villages, respondents (93%) reported an increase in the hippos’ visits and problems in the last ten years. This agrees with the results from the FGDs where 100% of the panel reported an increase in hippos’ visits and destruction on their farms and home grounds. Additionally, respondents reported using different mitigation measures which varied significantly across study villages at f = 21.330(2) p ≤ 0.00), f = 26.742(2) p ≤ 0.00 and f = 3.893(2) p ≤ 0.023 for fencing, digging holes, and guarding, respectively. Across study villages, fencing (57%) was the most commonly used method, followed by digging holes (40%) and guarding fields (3%) was the least and only practiced in Ngarenaro village. Additionally, respondents from Negamsi only practiced fencing at a minimal rate (12%) while 88% did not use any mitigation measure compared to other study villages (Fig. 3).

Fig. 3
figure 3

Mitigation measures used against hippos’ problem across study villages

Full size image

4 Discussion

4.1 History of land use land cover changes around Lake Babati

The study observed the newly residential houses to be dominant in Babati town which is closer to the lake compared to the periphery areas. Respondents also reported an increment of human settlement in Babati town responding to the upgrading of the district as a capital city of the Manyara district. In the year 2002, Babati Town became the capital of the Manyara region, which previously formed part of the Arusha region [27]. The event escalated the human population in Babati district by 3.2% annually between 2002 and 2012. The district was further ranked third in terms of human population growth rate across Tanzania [27]. Although the hippo population trends and the number of conflicts in Babati are were unknown, the rate conflicts especially crop raiding are reported to increase [28]. Other studies also reported encroachment on hippo grazing lands by cultivation and habitat destruction to endanger their long-term survival and the survival of other forms of wildlife [11, 31]. Historically, before 1954, the areas adjacent to Lake Babati were dominated by grasslands and were mainly used for grazing by local communities [26]. A study in 1990 found the area around Lake Babati to be transformed from grassland used for grazing to intensive farming. The farming also extended to the areas that were previously occupied by lakeside forests. This triggered the LUC adjacent to Lake Babati, known to be a hippo reserve even before Tanzania's independence to date.

4.2 Land use types and changes in Babati area from the year 1999 to 2019

This study has indicated that settlement expanded by 2.6% within the time period of 20 years while other forms of land use decreased. Bailey et al. [31] reported that even small increases in settlement areas may have significant impacts on wildlife and their environments. This is due to the fact that settlements are more permanent and resource-intensive compared to other forms of LUC [12]. In Babati District, the human population is reported to have increased by 32% within a ten-year interval [27]. Additionally, fertile soil for agriculture and fishing activities in Lake Babati draws more people to reside in the area [26]. This suggests that more land will still be converted to the settlement areas, thus reducing hippos' grazing areas. Human population growth around wildlife habitat has negative impacts on biodiversity through actions such as deforestation and habitat encroachment [10, 19]. However, the magnitude of destruction is context-specific based on the effectiveness of border protection and the nature of the landscape [18]. For instance, in Kenya's pastoral areas with no farming, wildlife populations were reported to decrease when human populations approached 8–10 people per km2 [19]. Babati District is reported to have 32 people per km2 [27], and the results further shows the shifting of human population from the southern to the northern part of the Babati area. This is due to the fact that Babati District became the capital of the Manyara region and the center of the town was located on the northern side, which attracted shifting. This might further contribute to encroachment and much loss of hippos' grazing areas.

4.3 Time of settlement establishment in relation to the distance from the lake

This study reported time of establishing settlement to vary across study villages where the villagers (Ngarenaro and Bagara ziwani) closer to the lake had residents ranging from 5 to < 30 years of stay. This shows that the area has been receiving a significant number of residents in recent years. The same trend is observed under satellite image classification, where the settlement is concentrated on the areas closer to the lake and Babati town. This is also attributed to escalated interactions and conflicts between hippos and humans in the villages closer to the lake. Contrary to the current situation, back in the 1950s, the areas closer to the lake were reported to be dominated by grasslands, which provided a conducive environment for hippos and other wildlife species, including buffaloes and water birds [26]. Additionally, in recent years, the study area has been harboring hippos and bird species to the exclusion of other wildlife species, and the adjacent areas are dominated by settlements and agricultural lands. Other studies in Kenya and Nigeria also reported human population increase and settlement expansion closer to wildlife habitats to threaten hippos' survival and their habitats [11, 21]. The same has also been reported in Tanzania, where human encroachment into areas formally used by wildlife has brought conflict between wildlife and people [13, 20, 22]. Contrary to this, Negamsi village, which is located far (3 km) from the lake, was observed to have a majority of respondents who have resided for 20 years and above, which was a relatively long period of time compared to Ngarenaro and Bagara ziwani village villages. The study suggests that a more human population is expected in Babati town and closer to the lake compared to the peripheral areas.

Moreover, the average farm owned by respondents in Bagara Ziwani village, which is closer to the lake, was observed to be cultivated across the cropping seasons than other study villages, suggesting continuous agricultural activities adjacent to the lake. This increases the burden on the survival of hippos and other aquatic birds in the area, escalating human–hippos interactions and conflicts. The study by [15] revealed a positive correlation between crop-destruction incidents by hippos and farm proximity to the river adjacent to Ruaha National Park. Other studies also reported intensive conversion of land that was formally used by wildlife to agriculture, resulting in booming crop raiding and conflicts [18, 20, 32]. The situation calls for immediate management interventions to balance conservation of the remaining hippos’ population and development of the livelihoods in the area.

4.4 Respondents’ perceptions and mitigation measures against hippos’ problem in the recent 10 years

The findings from this study reveal an increase in hippo incidents in the last 10 years. It is attributed to the significant increase in the human population in Babati, changing land use cover types from grassland areas to agricultural lands and settlements [26, 27]. This stimulates an overlap of resource needs between humans and hippos. The study also reported different mitigation measures used by respondents to vary across the study villages, whereby Ngarenaro and Bagara ziwani closer to the lake used a combination of methods. This is because they are prone to hippos' damage compared to the Negamsi village. Other studies in Kenya [10, 19] and Tanzania [15, 25] also sited animals like wild pigs, elephants, buffaloes, hippos, and monkeys as problem animals where they live in close proximity to agricultural lands. Although the local measures used were reported to be not very effective, the fencing method was commonly used across villages because it was cheap in terms of labor and financial resources compared to digging holes and guarding fields. The materials for making local fencing were reported to be locally available, which included sisal (Agave sisalana) and Acacia nilotica, which were the main materials used in local fencing across the study villages. Hence, the situation is still challenging. In most African countries, including Tanzania, many protected areas are not fenced [33]. The situation has led local communities to use various deterrent techniques such as ditches, local fences, wall hedges, shouting, banging tins and drums, lighting fires, throwing stones and chili fences [17, 34, 35] to control wildlife incidents. In spite of all these efforts, interactions and conflicts between wildlife and people are still a problem [33]. There is a need for involvement of local communities in getting the sustainable measures for conflict mitigation to be adopted in specific localities.

5 Conclusion

Human–hippo interaction is a great challenge to the communities adjacent to Lake Babati, which is known to be a hippo reserve. The study observed an increase in LUC in the study area whereby most of the land has been converted to settlement and areas closer to the lake had relatively short lived settlement than areas farther from the lake. Across study villages, the conflicts were highly reported closer to the lake where fencing was found to be the dominant short-term mitigation measure used for combatting hippo problems. Proper land use planning and the use of alternative crops like fruit trees that are not preferred by hippos are suggested for attaining the long-term mitigation measures. The efforts will reduce burdens on hippos’ habitat and ensure sustainable conservation while attaining the development of communities closer to Lake Babati and adjacent areas.