According to7, identifying fragile ecological areas is imperative for ecological protection and environmental organization and management. Therefore, assessing ecological vulnerability is crucial for the study of ecosystem vulnerability45. Based on the current conditions and previous predictions, the EVI was classified from the lowest vulnerability (potential) to the highest vulnerability (high), as shown in Table 4. Overall, this study obtained three main results, which are highlighted below.
The first result concerned the spatial variation in EVI. In the composite system, the EVI (EVIPCA) varied from north to south, with Littoral being a vulnerable province and Alibori being a stable province. In the additive system, EVI (EVIad), both southern and northern Benin were identified as vulnerable, especially northern Benin, and Littoral (which was identified as vulnerable by the composite system) and central Atacora (which was identified as potentially vulnerable by the composite system), respectively, were identified as vulnerable.
The second result was the calculation of the spatial autocorrelation coefficients (Moran’s I) of each EVI, which were IPCA = 0.955256 and IAD = 0.989222 for the composite and additive systems, respectively. Both of these values are very high and are better than those reported in46. Although the spatial variations in these systems were obviously different, their Moran’s I values remained very high. However, according to Moran’s I, the spatial autocorrelation of the additive system was higher than that of the composite system. The principal component analysis approach assumes no prior relationship between the different factors and allows their relationships to develop from the statistical analysis, thus indicating the regional spatial variability of the components8. The observed discrepancies in spatial variation outcomes did not mean that there was a lack of spatial organization between the components. Therefore, graphic dissimilarities (differences in spatial distributions) do not challenge the spatial layout of the components or notably, their correlations.
The third result was from the cluster analysis, showing high-high clusters in the south for the composite system and in the north for the additive system. We deduce that regardless of the system used to calculate vulnerability, ecosystems in central Benin are still relatively stable. Central Benin has a moderate population density and moderate soil organic carbon levels. Littoral has a high population density rate, while Borgou has a high soil organic carbon level. These outcomes reveal that southern Benin is seriously threatened according to the composite system and that northern Benin is seriously threatened according to the additive system. These findings were explained and discussed with reference to available studies.
We used IDW interpolation, as opposed to41, who used kriging interpolation. We note that the indicators used in that study were slightly different from those in this study and were not classified similarly; in addition, different analysis assumptions were applied. His results show a strong positive correlation between sensitivity and the additive EVI (EVIAD), which is slightly different from the results of our study. In this study, we found a moderate correlation between these two factors. This difference in the outcomes can be attributed to the difference in the indicators and their distribution in the system. Nonetheless, that study showed that additive vulnerability is primarily influenced by adaptation, exposure and sensitivity; our study led us to put these elements in the order of adaptation, sensitivity and exposure. Both studies placed adaptation in the same position. Although the considered variables were different, we reached the same conclusion regarding adaptation, which can be considered a strength of our additive system.
Densely populated areas were determined to be very vulnerable47. High sensitivity rates were detected in southern Benin, including in Littoral, Atlantique, and Oueme. Housing and density indicators were classified as sensitivity variables, which means that density is still a threat to ecosystem stability. Littoral Province, the economic capital of Benin, which has the highest population density (more than 8000 inhabitants per square kilometer, according to the averaged raw data), and Atlantique and Oueme provinces, newly developed residential areas, were classified as extremely vulnerable. Alibori Province, the largest and least populated province, was classified as the most stable area in the composite system. We can deduce from this analysis that the population density also has a great impact on the composite system. In the additive system, Littoral remained an extremely vulnerable area, and central Atacora and Collines were the most stable areas. This outcome confirms that density in Littoral is a serious challenge to stability according to both systems.
However, the composite system than the additive system is more credible since it is based on SPSS, a statistical software, and is therefore empirical. In contrast, the additive system can be unreliable, since the indicators, as a whole, are classified according to the user. This classification method is subjective, and therefore theoretical (here, we based our indicators on expert advice and IPCC recommendations); hence, it leaves room for doubt. This study found that coastal zones, i.e., Littoral, are the most vulnerable33,34,48. This finding indicates the reality for our study. The extremely vulnerable areas identified by the composite system were high per capita density areas, which emphasized that density was a decisive indicator in our composite system. This analysis uncovered significant spatial variation in population vulnerability in southern Benin. According to the raw data we collected, the average density per capita in Borgou is 35.909%, while in Littoral, it is 8003.636%, i.e., 223 times higher than that in Borgou. Borgou is made up of several communes, while Littoral consists only of Cotonou, the economic and administrative capital of Benin, which is a highly desirable area. The demand for buildings has forced people to occupy some natural drainage channels, making this commune vulnerable to flooding. Southern Benin is less spacious but has more inhabitants than northern Benin because almost the entire administrative system of the country is located there, as well as one of the largest markets in West Africa. There is a need for an efficient decentralization process according to the determined standards. Our study revealed that regions with lower density per capita were the least vulnerable.
The additive system found that the areas with high bush fires and soil organic carbon rates were the most vulnerable. Thus, vulnerability is specific to the context34, since the factors that make a region or a community vulnerable can vary among different regions and community. The vulnerability of the northern area that was highlighted by the additive system can be explained by the practice of intensive agriculture (soil organic carbon) and the bush fires involved in these practices. Northern Benin is an agricultural area, and cotton cultivation is common; hence, there are high levels of pesticide use. Agriculture is very important for the Beninese economy and hence pesticides are used. Vulnerability in southern Benin is related to climate, flooding, and the high population density, while vulnerability in northern Benin is related to bush fires and soil organic matter levels. Although the systems and indicator groupings were different, they reached the same conclusion about Littoral Province. In the additive system, the vulnerable areas corresponded to areas with high soil organic carbon.
It is important to point out that this study suffers from certain limitations38. For example, data for all the indicators from the same time period were not always available, some required data were inaccessible and some data were gathered from the public domain. This can be interpreted as a weakness of our system. Since public-domain data are not accurate, they can result in biased outputs, which should not be ignored. The determined spatial and temporal variation, as well as the type of degradation under consideration, depends on the input data sets for the analysis and modeling39. Using automatic linear modeling model building (ALMMB), our results were improved.
The main objective of automatic linear modeling model building (ALMMB) was to improve the present study outcomes by enhancing the accuracy of the established system based on the adjusted chi-square Pearson correlation. Using automatic linear modeling regression combined with the best subsets method in SPSS 23, we tried to enhance each observed vulnerability level. Table 7 displays both the observed and enhanced rates for each EVI, and Fig. 6 displays the map of the enhanced values. We note that the potentially vulnerable areas32 increase or decrease in size less than the highly vulnerable areas.
Improving composite and additive EVI map.
Based on Table 8, in the composite system, increases in both the potentially and highly vulnerable areas were highlighted. The observed potentially vulnerable area was 48,600 km2, and the enhanced potentially vulnerable area was 60,269 km2. The observed highly vulnerable area was 3729 km2, and the enhanced highly vulnerable area was 4812 km2; the differences in these values were 11,669 km2 and 1083 km2, respectively. A decrease in the potentially vulnerable area and an increase in the highly vulnerable area were noted in the additive system. In the additive system, the observed potentially vulnerable area was 36,450 km2, and the enhanced potentially vulnerable area was 32,119 km2, for a difference of 4331 km2. The observed highly vulnerable area was 3007 km2, and the enhanced highly vulnerable area was 6977 km2, for a difference of 3970 km2, i.e., more than the double the observed value. However, according to the enhanced composite model, much attention should be paid to all southern provinces, especially Zou, Oueme and Plateau. Figure 6 displays the enhanced vulnerability mapping for a) the composite system and b) the additive system. Figure 7 summarizes the different classified areas and their differences.
Synthesis of different classified areas.
In summary, the composite system was vulnerable to climate and flooding (and to some extent to population density as well, as in Littoral), while the additive system was vulnerable to bush fires and soil organic matter. Littoral was identified as a vulnerable area in both systems. Finally, to improve the accuracy of our results, we used ALMMB. The results showed both increases and decreases in the size of vulnerable areas. The present study used a combination of GIS, PCA and ALMMB to accurately assess the vulnerability of terrestrial ecosystems in Benin.
Source: Ecology - nature.com
