Dynamics of the Inland Valley Occupation in the Oti Watershed in Benin: Diagnosis and Prospective Analysis

The use of inland valley has highly increased in number and area over the last three decades because of their soils fertility and their hydromorphic character. The present research analyzes the dynamics of the occupation and use of inland valley in the south of the Oti watershed in Benin and their future by the years 2030 and 2050. It is based on a combinative approach which integrates spatial Remote Sensing, Geographical Information Systems and Land Use Change Model incorporated in Idrisi Selva software. Landsat TM images of 1988, ETM of 2000 and OLI TIRS of 2015 were used for land cover information extraction. The projections’ period was made based on the observed socioeconomic conditions in the study area. The 1988 to 2015 change detection indicates that plant formation (woodlands and savanna) characterized by the inventoried inland valley evolved toward the anthropogenic formations (mosaic of crops, fallows and bare soils). The annual rate of change is estimated at -0,83 %) for the first period and -1,6 % for the second period. The anthropogenic procedure was strongly observed from 1988 to 2015 with a global annual regression of 0,38 % which leads to an extension of the agricultural areas. Considering the three topographic sectors between the period of 1988 and 2015, about 22,01 % of the Atacora massif inland valley, 44,88 % of the wavy interfluves and 30,41 % of the low altitude sector are occupied by crops. This evolutionary trend of agricultural areas comes from the prominent role of the agriculture in the economy of the Oti watershed. By 2030 and 2050, we will probably witness a significant extension of agricultural areas which could respectively go from 105 461,63 hectare in 2015 to 115987 hectare in 2030 and 144182 hectare in 2050.


Introduction
Climate change has contributed to the greatest famines in sub-Saharan countries Since the 1970s (Diello, 2007). They are in combination with anthropogenic actions which caused poor agricultural performance and degradation of natural resources (Grouziz, 1986 ;Ouédraogo, 1998 ;Kabré, 2008).
In West Africa, the changes observed have had repercussions on water systems and huge socioeconomic impacts which may constitute a constraint for the integrated management of available water resources. (Vissin, 2007 ;Totin, 2010). The progressive decrease in surface water resources in West Africa at the level of the main basins has reached 40 to 60% and is greater than that of rainfall (15% to 30%) depending on the area. (Afouda et al., 2007). The Oti watershed in northwest of Benin is not on the sidelines of these new hydro-climatic data with all their consequences on the livelihoods and on the land use model. (Idiéti, 2012). To this, must be added the rapid growth of the population which has lead to a great pressure on resources because of the positive interaction between this demographic growth and the demand for seeded areas. (Soubérou, 2013). The environmental interest that followed is reflected in a regressive evolution of forest areas and an uncontrolled extension of agricultural areas especially the hydromorphic zones, and particularly the inland valley. (Djihinto, 1997 ;FAO, 2007 ;Ouorou Barré, 2014). Temporarily flooded areas, these inland valley have particular hydrological and pedological characteristics that make them areas with high agricultural potential (Assigbé and Mama, 1993). These characteristics make these inland valley receivers of strategic agricultural lands that help the population to solve the issue of agricultural land's decrease upstream of the inland valley and the timespace irregularities of precipitation (Mahaman and Windmeijer, 1995).
Thus, inland valley development has accelerated in recent years because of farmers who have oriented agricultural activities more towards hydromorphic zones (Houndagba et al., 2007, Oloukoi andMama, 2009) which show obvious socio-economic opportunities (rice production and other vegetable products) (Oloukoi et al., 2013). This new agricultural dynamics is reflected in a significant regression of natural plant formations in favor of anthropogenic formations. However, it is important to remember that no work has oriented the analysis towards the prospective dynamics of occupation and land use in the inland valley. And this is at least at the scale of a watershed in northwest of Benin despite the modern tools offered by Remote Sensing and the Geographic Information System (GIS).
For a better understanding of current and future interest of inland valley exploitation, it is important to identify the sites affected by the changes observed. Then, calculate their evolution rates and simulate the possible changes in the future, based on the results obtained and the socio-economic and biophysical information in south of the Oti watershed.

Geographical scope of the study
The study area is a portion of the Volta river basin (BHV) that extends over the territory of the Republic of Benin. This portion is precisely located in the Oti watershed (WHYCOS, 2006). Located in the northwest of Benin, it straddles the Atacora and Donga departments and occupies 47.20% of its total area of the Oti watershed in Benin (Figure 1) The climate is of Sudanese and Guinean-Sudanese type, which integrates with that of West Africa (Adjanohoun et al., 1989). It is characterized by two great seasons: a rainy season (May to October) and a dry season (November to April). The annual rainfall is between 967 and 1255 mm. The number of rainy days in the north-west varies from 60 to 70 days (Ouorou Barré, 2014

Data processing and results analysis
The downloaded Landsat images (TM, ETM +, OLI TIRS) were processed (geometric and radiometric image correction, color composition and visual interpretation) and then classified into the Idrisi software environment to retrieve information on the occupation of the inland valley. After the process of image supervised classification, the results were improved with fieldwork and the analysis of the dynamics of the inland valley occupation was based on the change detection approach to compare the different land-use classes from 1988 to 2000 and 2000 to 2015. Detecting changes involves identifying differences in the states of an object or phenomenon through observation at different time periods (Hountondji, 2008). Oloukoi, 2012, Tohozin, 2016. She uses the GIS approach through the determination of the transition matrix and visual interpretation to understand the different types of conversion between these occupation units. Thus, changes in land use were assessed on the basis of the rate of the annual change of the inland valley occupation which shows the proportion of each land-use unit that annually changes. This is the annual average rate of spatial LCM is a probabilistic model with a discrete event that is one that does not evolve continuously over time but with discrete time steps (Coquillard and Hill, 1997). As a prospective model, it shows the landscape and its organization in a future temporal dimension (projected time t). Thus, two maps of the land use and use of the inland valley (LULC) at known times t 0 (previous) and t 1 (current) were highlighted for a comparative analysis in order to determine the transitions, to quantify the concerned area for each type of transition and locate the changes. On the basis of the transitions matrix of obtained, the future potential changes are calculated at a time t 1 + ... n. The model was calibrated by factors that explain the changes that have occurred (socio-economic and biophysical factors that have contributed to the evolution of the land use inland valley units until 2015). Each type of transition can be explained with known quantifiable and mappable factors (maps of slopes, expansion of the agricultural areas, population progress, etc.). The model will then combine the different factors by attributing to them more or less strong weights to explain the transitions it has detected between the two maps provided at the beginning ( Figure 2). Thanks to the linking and the combination of the explanatory factors and the changes that occurred between the two original land use maps, the model will first carry out via a statistical method (logistic regression or neural network) probability maps to changes (susceptibility maps to changes). Then, the prospective land-use map at a given date t 1 + ... n will be used based on the previously identified changes to simulate the next one. Finally, the last step is the evaluation of the simulation. According to Vigneau (2013), this is an optional step but it helps to give more credibility to the reliability of the simulations.   From the analysis of the information in Table I (Table  II). The analysis in Table II shows a gradual evolution of the fields and fallows mosaics of (3.48%) between 1988 and 2000, (3.92%) between 2000 and 2015 as shown in Picture 3. On the other hand, the other units area is reduced mostly that of the tree and shrub savannas with respectively a change rate of (-1,57%) and (1,12). The average of the annual change rate is estimated at (-0.83%) for the first period and (-1.6) for the second period.

Picture 3: Occupation and use of inland valley in the south of the Oti watershed in 1988, 2000 and 2015
The examination of Figure 3 shows an extension of cultivation and fallows areas and an increase in the area of tree and shrub savannas at the expense of gallery forests, woodlands and wooded savannas in both periods. From these results, the continuity of the processes of savanization and anthropisation in the units of occupation emerges. In sum, from 1988 to 2015, the dynamics of the inland valley units of occupation in the south of the Oti watershed were made with an overall annual regression of (-0.38%) and led to an expansion of wooded and shrub savannas, especially mosaics of crops and fallows and the bare soils.

Evolution of inland valley occupation units based on the morphological structure
The analysis of the land use units' small scale helped to assess the spatial dynamics trends of inland valley' occupation in the three predefined areas (Atacora Massif, Peneplain and Plain). At the level of the three sectors, there was an extension of the mosaic unit of crops and fallows (increase in areas) during the two periods (Table III) and a reduction in the area of forest formations. This analysis confirms the processes of savanization and anthropisation that appear through the conversion of the land use classes.  (12) The R indicator shows a phenomenon of anthropisation over the whole area (Table III) The Gourma Plain sector is increasingly experiencing a mosaic of crops and fallows during both study periods, while the other two were stabilizing between 2000 and 2015. Consequently, anthropization would migrate more towards the low altitude sector with more hydromorphic lands, after taking place mainly in medium altitude sector (peneplain) and high altitude sector (Atacora massif). This evolutionary trend of agricultural land (Picture 4) in the Oti watershed comes from the prominent role of agriculture in the economy of the townships which constitute the study environment, It also comes from the depletion of farmland in high and medium altitudes and fallows land cultivation over a long period (0 to 5 years).

Projection of the inland valley occupation from 2013 to 2030 and 2050
The projection based on previous trends reveals that forest formations are declining in favor of mosaic crops and fallows in 2030 and 2050 (Table V). The savannas will experience a reduction in their areas (about 4.63% in 2030 and 15.39% in 2050) while the areas of crops and fallows will continue to expand. In 2013, the mosaics of crops and fallows which accounted for 29.31%, will increase by 2.93% in 2030 and 7.83% in 2050. On the other hand, the other units will be reduced gradually. The gallery forests will increase from 1.72% in 2015 to 1.63% in 2030 then to 1.55% in 2050. The woodlands formations and wooded savannas will go from 3. The sectors (1, 2, 3) will experience a stable area of crops and fallows mosaics from 2030 to 2050 (6). The regressive trend of plant formations in sector 2 and 3 noted between 2015 and 2050 will help decision-makers to take actions in order to manage and preserve natural resources so as to anticipate the irreversible degradation and the maintenance of the regulatory role played by this basin in the hydrological cycle. The agricultural activities movement from sectors 2 and 3 to sector 1 is therefore to be planned for a rational and sustainable use. It is important to note that the resolution of the image used did not help to have full access to the land use details which implies that for safety measures we explore for the future studies more improved resolution images than those of the Landsat series. This is a match with Oloukoi's (2012) analyses which show that the classification of Landsats data used in his research presents some uncertainties in terms of planimetric accuracy. Considering the three sectors (low, medium and high altitudes) of the inland valley' characterization, the same evolutionary tendency of the crops and fallows' areas has been noted with the depletion of arable lands at high and medium altitudes and the orientation of crops towards the low altitude sector. The results support those of Albergel et al. (1993) who finds a migration of farmers from areas subject to degradation of the natural environment and drought to more fertile and wetter areas. The use of the Land Change Modeler (LCM) model for simulating the future occupation of the inland valley through the previous observed dynamics helped to predict the future of the Oti watershed part. Indeed, this scenario provides for an extension of agricultural areas in 2030 and 2050 and has shown that the next conversions will be more in the plain of Gourma.

Analysis of the inland valley use dynamics
This upward trend in crops and fallows mosaics areas is mainly due to the agricultural potential (availability of water, presence of permanent moisture, soil fertility, etc.) that the inland valley have for high agricultural production (Thenkabail, 2013). It is therefore important to understand that the decrease in It was noted that a gradual evolution of field and fallows mosaics respectively 3.48% between 1988 and 2000, (3.92%) between 2000 and 2015 and a reduction in the area of the other units especially tree and shrub savanna with a respective rate of change of -1.57% and 1.12%. The average annual rate of change is estimated at -0.83% for the first period and -1.6% for the second period. The anthropisation process was strongly noticed from 1988 to 2015 with an overall annual regression of -0.38% which led to an extension of the wooded and shrub savannahs and especially the mosaics of crops and fallows, agricultural areas and consequently the inland valley. It is important to remember that the inland valley have been exploited in recent decades in a very active way.
Simulations made on the basis of this trend in order to assess the behavior of land occupation units in the future revealed that forest formations are still in decline in favor of mosaics of fallows and crops in 2030 and 2050. Consequently, there's an increase in the farming areas in the inland valley. Considering the average annual growth of the cultivated areas over the three decades being raised to 2.30%, we will witness a remarkable extension of agricultural areas the next thirty-five years. The regressive trend of plant formations between 2015 and 2050 will help the decision-makers to take actions in order to manage and conserve natural resources so as to anticipate the irreversible degradation and the maintenance of the regulatory role played by this basin in the hydrological cycle.