4.1. Land Use Change Scenarios
This study focuses on changes from all LULC classes except built-up areas to either cropland or cropland-rice, although local studies in the catchment indicate the problem of deforestation as well [
28]. However, deforestation is not as pronounced for the entire catchment (
Figure 4) [
1] and therefore only changes into agricultural LULC classes were explicitly modeled. Nevertheless, difficulties in the classification scheme among the different forest classes and open forest areas or savanna were apparent [
2]. Natural classes in particular were prone to errors due to gradual differences in reflectance characteristics, although post-classification comparisons were mostly consistent and conform to historical maps. Yet, conversion into cropland was the focal LULCC and less prone to errors due to strong spectral changes [
2]. Skill measures (
Table 6) for both transitions to cropland or rice were satisfactory, nevertheless the exact distribution of LULC pixels in the 2030 scenarios should be interpreted carefully for several reasons. Firstly, the computed rate of change from 2004 to 2014 was transferred linearly until 2030. Secondly, this analyzed pattern is based on explanatory spatial factors like the altitude above sea level. Therefore, a saturation effect might occur due to limited space, e.g., in the wetland fringes surrounding the floodplain. The wetland fringes are nearly completely used as agricultural land in the setups of 2014 and 2030 (
Figure 4). A growing demand for agricultural land in this area is a source of uncertainty, because the interaction with the enclosing landscape, and therefore agriculturally less suitable areas like upland forests or flood-prone areas, is a different process compared to the observed LULCC in the wetland fringes. Thirdly, alterations in demographic growth including natural birth rates and immigration are not included in this linear approach. However, the impacts of demographic growth on LULCC are indirectly integrated due to the transfer of observed changes from 2004 to 2014 to the year 2030. The demographic growth accelerated in the 90s and after 2000 due to the migration of mainly pastoralists into the valley [
23] and correlates with the growing share of cropland in the valley, which was increasingly converted from grassland and savanna into cropland to feed the growing population. We use these conversions into cropland as a proxy for demographic growth, due to the stagnating trends in rice yields in the area [
73,
74], although conversions into cropland might also be affected by investors from outside the valley and other factors. Lastly, the influence of politics and the economy is not included, but might change the LULC drastically by setting incentives for agricultural activities e.g., the SAGCOT initiative [
22], changing the allocation or status of conservation areas or by developing the infrastructure. Furthermore, the spatial structure of the SWAT model and its HRU approach, which summarizes results for HRUs and neglects interactions among neighboring grid-cells within a subcatchment [
75], has structural limitations compared to a fully distributed grid-based solution. However, SWAT is a well proven tool to determine impacts on water resources due to LULCC [
76,
77,
78]. Analysis of impacts on water resources on grid-cell scale is not the goal of this study, but rather to identify general trends of LULCC and their impact on specific areas prone to these LULCC in order to assist the local water resource management authorities to enable a sustainable use of the available water resources. Hence, a business as usual scenario until 2030 was developed using the LCM and all analyses with regard to water resources were performed from catchment to the subcatchment scale.
The general distribution and spread of the modeled cropland/rice production area is reasonable. The hot spot of change for both scenarios is the fringe of the wetland. However, the center of the wetland is not transformed to agricultural fields on both setups, which is also reasonable due to the extended flooding and the threat to lose the harvest [
79]. Other areas of agricultural expansion are the western parts and the central northern parts, near the cities of Makambako and Njombe, and the main roads A104 and B4 (
Figure 1). Although some rice is grown in the Njombe region, it is mainly an important production region for maize, Irish potato, tea, and flowers and therefore it is rather unlikely to dispense the income within these agriculturally suitable value chains for less suitable large scale rice production in this region. The southern part, which expanded from 2014 to the 2030 setups, was already confirmed by local experts in a participatory mapping exercise as a recent rice growing area in the framework of a stakeholder workshop in February 2019. The transformation of cropland to cropland-rice in the very northeastern parts is unlikely due to the existing and growing sugarcane fields of the Kilombero Sugar Company.
4.2. Land Use/Cover and Climate Change Impact Assessment on Water Resources
The impact of LULCC on average stream discharge seems to be negligible at the first glance (
Figure 5). This is also in line with a former study on historical LULCC on cropland [
2], and was also observed in another catchment in Tanzania [
80] as well as in small catchments in West Africa with conversion of savanna into rice [
81]. One important factor for the low impact at the main outlet is also the stable share and distribution of forest classes in the upland of the catchment (
Figure 2,
Figure 4 and
Figure 5 and reference [
2]). Yet, LULCC are still seen as the main driver for decreasing streamflow in Eastern and Southern Africa [
82]. These minor changes in streamflow at the main outlet due to LULCC detected in this study can be attributed to concealing effects for large catchments [
83] like the Kilombero Catchment. Therefore, it is important to analyze the water balance on several spatio-temporal scales like the subcatchment scale or monthly averages and also analyze changes in low- or high-flow patterns.
The Q90 as representative index for the low flows decreases for both scenarios—the cropland and the cropland-rice scenarios—by 6% or 8%, respectively, from 1970s to 2030. An environmental flow assessment found several parts of the catchment to be differently vulnerable to decreases in mean annual flows [
36]. The upstream margin of the floodplain with a monthly recommended flow of 82.3% of the mean annual flow was defined as highly vulnerable concerning environmental flows [
36]. Therefore, these decreasing trends of 6% and 8% of the Q90 at the outlet should be considered carefully for further analyses. Stakeholder interviews and discussions with local farmers revealed, that perennial tributaries of the Kilombero in the northeastern part of the catchment turned into seasonal tributaries in the last decades. This change is attributed to deforestation activities and expansion of cropland, and therefore needs to be taken seriously to maintain the socio-ecological system that depends on continuous availability of water resources and the transported sediments and attached nutrients [
28,
35].
High flows are more distinct in the rice scenarios (
Figure 6), although they decrease with an increasing share of rice (
Figure 7b), especially in the months of April and May (
Figure 9g,h). The general differences among the cropland and cropland-rice scenarios arise from their different shares of all LULC classes (
Figure 5). The rice setups have a lower share of forest classes for example and therefore a comparison that aims to determine the impact of a growing agricultural share should be done separately within the cropland-rice or cropland setups. Though, the decreasing high flows within the rice scenarios (
Figure 6) can be attributed to the high water requirements of the rice plants [
29].
The cropland scenario for 2030 (
Figure 7b) displays a strong increase in the discharge amount of Q10, which is distributed to the months of March to May (
Figure 9d). This might lead to aggravated flooding events, which could either endanger the farmer’s harvest [
73,
74] their lives, critical infrastructure and their livelihood [
84]. Especially newly promoted, high yielding, but low growing improved varieties such as like SARO5 (TXD 306) might be negatively affected by these changes. These strong increases of water yield are accompanied by slightly decreasing evapotranspiration throughout the year (
Figure 9d). These patterns with regard to LULCC and Q10 are aggravated by the effects of climate change. The combined effect of climate change and LULCC inherits an increase of 84% between the two scenarios comparing the lowest (LULC 1994, RCP4.5, dry model) and the highest (LULC 2030, RCP8.5, wet model) value for Q10 shown in
Figure 11b. The effect of climate change outperforms the impact of LULCC, yet the contribution of LULCC to changes in Q10 is still substantial (
Figure 7b). It is necessary to add that changes in management practices are not included in these LULCC, but several practices, like the establishment of year-round irrigation schedules, will further affect water resources [
85,
86]. Furthermore, the uncertainty of the climate change signal, represented by the huge span in the GCM-RCM model runs [
33] (
Figure 11), is much higher than in LULCC scenarios. While climate change models show diverging trends of more dry or more wet conditions and changes in the seasonality, the impact of conversion from natural LULC into agricultural utilized fields is more explicit [
3,
4,
81], although it still depends on the specific crops grown. Nevertheless, intensification of precipitation might foster groundwater recharge and therefore access to renewable water resources in the Kilombero Catchment as well as already described in other catchments in SSA [
87]. This indicates a particular resilience to climate change and intensification of precipitation events. However, more observation-driven research is needed on the relation of surface water and groundwater resources on this topic [
87]. Moreover, data availability on the hydrogeology of the Kilombero Catchment is still poor to be modeled precisely, although some data and a local conceptual model exists [
88,
89]. Furthermore, the groundwater routines of the SWAT model are not sufficient for adequately modeling groundwater processes, because distributed parameters like the hydraulic conductivity and storage coefficients are disregarded in the linear reservoir approximations [
90].
Overall, analyses on subcatchment scale (
Figure 10 and
Figure 12) show that the conversion into cropland leads to increasing surface runoff and overall water yield (
Figure 10a,b), whereas a more diverse picture is shown for the rice setups (
Figure 10d,e), due to the differences in LULC in the setups (
Figure 5) and the aforementioned water demand of rice plants [
29]. Average annual evapotranspiration is decreasing in both agricultural setups in most of the subcatchments, especially where natural systems are converted into agricultural production zones, which is in line with other studies from the tropics [
3,
6,
7]. Still, there are studies that report increasing evapotranspiration due to conversion of forests to cropland [
91]. However, farming in the Kilombero Catchment is mainly done by low input rain-fed rice cultivation and only a few rice irrigation schemes do exist [
92]. Therefore, the rice setups were established using the default management plan from SWAT. A high input management setup would change the plant growth and consequently the evaporation of the plants [
81].
The scale dependency of hydrological processes and the spatio-temporal heterogeneity of water movement within the catchment are apparent by comparing the different characteristics of selected water balance components in the subcatchments (
Figure 10 and
Figure 12) and their monthly deviations for the entire catchment (
Figure 9 and reference [
2]). The deviations that include the effects of climate change (
Figure 12) are substantial, even though they compare the extreme situations concerning LULCC and climate change scenarios.
The manifold scenarios, inherited uncertainties and their implications on water resource management reveal the difficulties for local authorities and the need for further research in the area. The population in the catchment districts Kilombero and Ulanga has been and is currently increasing [
47] and road infrastructure as well as the Stiegler’s Gorge power station are being constructed. This will lead to further LULCC, and locally rapid deforestation has already been reported [
28,
93], consequently affecting the water balance [
94]. Forest protection against unregulated degradation is still problematic in Tanzania. There is a need to understand the social-ecological system to strengthen strategies, that ensure socio-economic benefits of local people, while preventing ecosystem degradation to allow a sustainable utilization and protection of the resource base [
5]. The local scale and the understanding of the local communities that depend on the wetland resources and their adjacent mountain forests and savannas could be the key for the development of management policies in the Kilombero Catchment [
95]. These could be for example the promotion of environmentally friendly sources of livelihood such as beekeeping, a sustainable forestry system accompanied by education on the socio-ecological system and improvements in the agricultural practices [
23]. Still, migration into the valley and population growth are critical factors for the pressure on the ecological system [
23,
95,
96]. Further information on the flooding extent, timing and duration using a hydraulic model with regard to the LULCC and climate change scenarios should support to manage the floodplain under future conditions. Beyond that, there is still not sufficient data on water quality, especially with regard to the emerging use of fertilizers, herbicides, and pesticides [
35,
95].