Senegal is a country in West Africa located between 12˚8'N-16˚41'N and 11˚21'W-17˚32'W. It has a surface area of 196,722 km² and an estimated population of 17,215,433 in 2021 [22] . This Sudanese-Sahelian country has few rivers and regular rainfall deficits. It also has a rainy season (from June to October) and a dry season.

The rainy season peaks in August-September and varies with latitude. This rainy season corresponds to the monsoon period in the Sahel. Rainfall is lower in the north than in the south. The northern region of Senegal registers an average annual-rainfall of about 400 mm, whereas in the south it reaches 1000 mm. The Senegalese growing vegetative season extends from July to November. In the north, this season starts generally in August and ends in October [17] .

Figure 1 shows the Great African Green Wall (GGW) crossing northern Senegal. It is 15 km wide and about 545 km long. It represents 7% of the total length of the GGW. This line covers a surface of 817500 hectares. It passes through three regions: Louga, Matam and Tambacounda (Table 1). For this study, we selected at least one site in each region that has been subjected to various GGW reforestation

surveys since 2006 (See Tables 1-3 and Figure 1).

The main soil types encountered in Senegal are, in order of importance: non-leached and leached tropical ferruginous soils (34.40%), lithosols (21.38%), hydromorphic soils (10.93%), regosols (8.16%), less developed soils (7.74%), red-brown soils (6.15%), ferrallitic soils (5.78%), halomorphic soils (2.90%), vertisols (1.65%), sub-arid brown soils (0.64%) and crude mineral soils (0.27%)

[23] . On the other hand, the main soil types crossed by GGW Senegalese, from west to east, are dry sandy soils, ferruginous soils, red-brown soils, lithosols and sub-arid brown soils.

The land cover dataset is from the European Space Agency Climate Change Initiative Land Cover (ESA-CCI LC) and Copernicus Climate Change. This dataset provides global maps describing the land surface into 22 classes, which were defined using the United Nations (UN) Food and Agriculture Organization (FAO) Land Cover Classification System ( [24] [25] [26] [27] ). This database has a spatial resolution of 300 meters and available from 1992 to the present, with one year delay (1992-2020). The ESA-CCI LC is a combined product of global surface reflectance from different satellites missions (ENVISAT, MERIS, SPOT 4, SPOT 5, Proba-V, NOAA-15 (AVHRR)). They are available on http://maps.elie.ucl.ac.be/CCI/viewer/download.php and https://cds.climate.copernicus.eu/cdsapp#!/dataset/satellite-land-cover/.

Details of the in-depth validation of the land use maps are available in the product quality assessment report (https://datastore.copernicus-climate.eu/documents/satellite-land-cover/D5.2.3_PQAR_ICDR_LC_v2.1.x_PRODUCTS_v1.1.pdf; [28] ).

Thanks to long-term data consistency, annual updates and a high level of thematic detail based on global observational data, the ESA-CCI LC map series served as an input for various applications such as the impact of land cover change (LCC) on climate ( [29] ), long-term historical reconstructions for LC climate modeling and biodiversity accounting ( [30] [31] [32] [33] ), as well as forest [34] [35] [36] ) and desertification monitoring in scientific research ( [37] [38] ), carbon and climate change models ( [36] [39] ), policy-making and in the business sector ( [40] [41] [42] [43] ).

As part of the 2018 reporting to the UNCCD (United Nations Convention to Combat Desertification) and assessment of these indicators (Land Cover Trends; Land Productivity Trends; Organic Carbon Stock Trends), LC ESA-CCI land cover data has been advised and provided to member countries as default level 1 data ( [40] ).

The land use database derived from LC ESA-CCI allowed the establishment of the land cover maps of Syer, Tessekere, Loughere-Thioly and Ballou from 1992 to 2020.

1) Syer (Louga)

The analysis of land use between 1992 and 2020 shows 14 land cover classes in Syer area (Figure 2; Table 5 and Table 6). Figure 2, Figure 3 and Table 5 illustrate the extend of changes in land use during this period. There is an increase in the area of tree and shrub savannas (+8.5%), rainfed crops (0.95%), irrigated crops (+0.68%), mosaic crops (1.55%), tree cover flooded saline (+0.38%). Table 5 shows that there is a decrease in area of steppe/grassland (−9.8%),

2) Tessekere (Louga)

Figure 4 represents the spatiotemporal distribution of the nine land cover classes over Tessekere site from 1992 to 2020. The main land cover classes are tree savanna and steppe. The rates of evolution and change in area are recorded in Table 6.

There is a decline area for steppe/grassland, rainfed crops, mosaic agriculture, natural mosaic vegetation, shrub steppe and bare soils during 1992-2020 varying from 38.7% to 95.74% (Table 6). On the other hand, there is a spatial extension

of savanna tree and savanna shrub of respectively 78.43% to 89.82% from 1992 to 2020.

3) Loughere-Thioly (Matam)

Seven land cover classes were identified in Loughere-Thioly according to the 1992 land cover map (Figure 5 and Table 6). The spatial extension of these classes from 1992 to 2020 shows that there is an increase of the area corresponding to mosaic agriculture of 1.85%, natural mosaic vegetation (+1.91%), tree savanna (0.76%). The rate of steppe/grassland area declines from 13.30% to 9.0%. The other classes remained stable.

4) Ballou (Tambacounda)

The Ballou land cover data processing provided 11 classes (Figure 6). The main land use classes are rainfed crops (47.33%) and shrub steppe (33.66%). There is a reduction in the surface area of woodland/open forest of 14.73% (Table 6). In the other hand we can observe an increase in the surface area associated to urban and shrub savanna classes of respectively 125% and 40.5%. The other classes areas remained stable from 1992 to 2020. So, we can notice a great

spatial extension and development of urban areas in Ballou.

The temporal evolution of the land-use maps from 1992-2000 show globally a degradation of the vegetation cover. There is a reduction in the surface area of tree savanna, rainfed and mosaic crops, woodland/open forest and water classes. This situation results from the drought recorded particularly in the 1970s, 1980s and 1990s in the Sahel region ( [4] [5] [17] [69] [70] [71] [72] [73] ). The overexploitation of land combined with irregulat rainfall have caused the reduction of agricultural, tree savanna and forest areas during dryness episodes. The rainfall deficit on the study areas recorded varies from 11% to 20% and the lost of vegetative activities through the vegetation index is between 4% and 8.5% [17] . These results are in agreement with the works of ( [6] [9] [11] [15] [74] ) in the Ferlo region (north of Senegal).

A study based on 40 years satellite data carried out by the CILSS (Comité Inter-états de Lutte contre la Sécheresse dans le Sahel “CILSS” [9] ) during 1975 and 2013 has shown that 26% of the land in Senegal are overexploited, included the agricultural areas causing savanna and open forest fragmentation. According to [15] ), the spontaneous vegetation has been highly degraded (−13.4%) during 1974-2013 in the Ferlo region (Tessekere). [12] and [75] have shown that the losses in surface of protected areas are low for the steppe and high for the wooded and tree savanna. The land cover degradadtion results from the human activities and the lack of precipitations ( [5] [70] [71] [72] [76] ).

A regreening of the different localities and a positive dynamic are observed with the return of a rainfall values higher than the those recorded in 1992-2000. Gains in surface of tree savanna and cultivation areas instead of bare soils are observed. There is an improvement of the surface areas of tree and shrub savanna of 11.40% (Tessekere), 8.25% (Syer) and 2.70% (Loughere-Thioly). In spite of the gradual improvement of the rainfall in Ballou after the year 2000, the vegetation cover was continuously in degradation. 37 km² of deciduous forest areas was lost after the year 2000s in benefit to shrub savanna and urbanization area.

With regrad to the non-uniform distribution of vegetation cover in the Sahel, particularly in our study area, the return of rainfall could not be the only factor in the vegetation regreening ( [77] ).

Agroforestry practices, reforestation, good management of natural resources should be taken into account in the success of the positive vegetation dynamic ( [6] [9] [12] [18] [72] [74] [77] [78] [79] . According to [11] [12] , the conversion with improvement of vegetation cover has affected 4000 ha in the Tessekere area. The regressing of bare soil surface areas was also due to the implementation of the Senegalese-German GTZ project since 1986. The progressing of cultivation areas was favorited by the planting of trees in the Ferlo region since 2007 (Senegalese Great Green Wall). With the collaboration of the local population, the GGW project has managed to plant 13,000 ha (case of Koyli Alpha Park), 52 plots of forest plantations and agroforestry (18,599 ha), about ten multi-purpose village gardens at Syer, Tessekere, Labgar, Loughere-Thioly and Sakal ( [80] ) and more than 27000 ha of land restored ( [81] ).

1) Analysis of estimated carbon storage on Senegalese GMV sites

The results of the evolution of land use carried out in the first part of this work were used as tools for quantifying and evaluating the dynamics of carbon stocks on our different study sites.

The method uses the different types of land cover and land use over each site and their average carbon stock values. In this way, we are able to estimate the total quantity of carbon and the average stock of plant cover from 1992 to 2020 (Figure 7), and carry out a comparative analysis of the evolution of carbon between these periods.

Figure 7 corresponds to the estimate of average carbon stocks in the different study sites. The carbon values per hectare show two dynamics. Indeed, carbon sequestration increases in the soils of Syer, Tessekere, Loughere-Thioly from 1992 to 2020. On the other hand, Ballou recorded a drop in carbon stock over the same period. Note that the average carbon stock values recorded in our study

sites vary between 11.18 ± 0.001 t.C/ha and 27.33 ± 1.34 t.C/ha. With the highest value observed in the locality of Tessekere (29.80 t.C/ha in 2020) and the lowest in Ballou (11.185 t.C/ha in 2020). For the different localities, the average annual values of organic carbon stocks from 1992 to 2020 range from 4,177,726 ± 12,539 t.C (Syer); 5,740,287 ± 23,306 t.C (Tessekere); 3,257,418 ± 7087 t.C (Loughere-Thioly) and 1,353,630 ± 2779 t.C (Ballou). The relative uncertainties of the study sites are in the range 0.2% to 0.4% of the different annual storages.

The results obtained on the quantities of average carbon stocks sequestered in the GGW zones of Senegal present large differences from one site to another in terms of carbon storage potential. This disparity is linked both to climatic conditions ( [56] [82] [83] ) but it is reinforced by the areas of trees in the different plant covers. Tree savannas, for example, represent on average 81.4% in Tessekere, 45% in Syer and 28% in Loughere-Thioly of the land use of these localities.

These averages of carbon storage are of the same order of magnitude as those reported by [52] [61] [84] [85] [86] [87] .

In 1992, the quantities of sequestered carbon were estimated at 4,027,459 ± 12,360 t.C (Syer), 5,550,062 ± 22,785 t.C (Tessekere), 3,197,238 ± 7173 t.C (Loughere-Thioly) and 1,353,884 ± 2778 t.C (Ballou). The main land use classes in these localities’ contribution vary from 35% to 96% of carbon storage. Tree savanna allows to sequester 2,850,821 ± 17,305 t.C (Syer); 5,339,604 ± 23683 t.C (Tessekere) and 1,636,259 ± 13,110 t.C (Loughere-Thioly) (Table 7).

In 2010, the annual carbon sequestration increased from 0.3% to 2.7% compared to that of 1992 on the Syer, Tessekere and Loughere-Thioly sites. On the other hand, sequestration is down by −0.02% in Ballou site (Table 8) due to the high urbanization.

In 2020, the total carbon storage in the soils of Syer, Tessekere and Loughere-Thioly is estimated between 3.5% and 12.7% compared to 1992. Compared to 2010, carbon sequestration increased from 1% to 12.4% (9.74% in Syer; 12.44% in Tessekere and 0.93% in Loughere-Thioly). In Ballou soils, the storage loss is −0.04% and −0.02% compared respectively to 1992 and 2010.

The improvement in the areas of land use classes of tree savannas, shrub savannas and crop areas to the detriment of bare soils, steppes and grassland generated a gain of 510,756 ± 26,490 t.C. in Syer, 705,834 ± 47,560 t.C in Tessekere and 110,784 ± 14,357 t.C. in Loughere-Thioly between 1992 and 2020. The spatial extension of the urban zone and shrub savanna classes at the expense of woodland/open deciduous forest, generated a loss in carbon stock in Ballou site of −505 ± 5557 t.C.

2) Analysis of carbon storage dynamics from 2000-2020

Figure 8 which represents the evolution of standardized carbon sequestration anomalies in the soil from 2000 to 2020 exhibits two dynamics.

The localities of Syer, Tessekere and Loughere-Thioly present a period with a negative anomaly followed by a second phase of positive anomaly. The positive periods began in 2007, 2013 and 2014 respectively for Loughere-Thioly, Tessekere and Syer. The positive sequestration anomalies became significant (greater than 1) in 2017 (Tessekere), 2018 (Syer) and 2019 (Loughere-Thioly). On the other hand, the dynamic in the locality of Ballou is reversed. This evolution of carbon storage is decreasing. Indeed, a negative anomaly is observed since 2010 in this site. The significance of soil carbon storage in reforested sites occurs 8 to 10 years after the start of the various tree plantations of the Senegalese GGW.

For Ballou, the loss of capacity to store carbon in the soil is observed and became significant since 2019.

The results of the break tests applied to the carbon sequestration data are presented in the Table 9 and Figure 9. The Buishand test, which shows the main break in the time series, reveals very significant breaks in the different localities. These breaks occurred in 2012 (Tessekere) 2013 (Syer), i.e., 3 to 4 years after the beginning of reforestation on these sites, and in 2006 at Loughere-Thioly. The

results of the BEAST test, which detect several breaks, show a change break in the trend in 2006, 2007, 2008 respectively in Loughere-Thioly, Tessekere and Syer. These years correspond to the beginning of tree planting on these sites. Then other breaks are located in the trends of the carbon storage series in 2016 (Tessekere) and 2017 (Syer and Loughere-Thioly). This second break corresponds to the year when carbon sequestration became significant in these localities. This result is in agreement with the representation of the standardized anomalies. The Bayesian test confirms the break change in 2012 at Tessekere. In the non-reforested locality (case of Ballou), the breaks are observed in 2004, 2009 and 2016. That of 2009 corresponds to the beginning of loss of open forest area.

One can note also that the average carbon stocks (t.C/ha) are still increasing and the slopes of the trends remain positive after the breaks in the reforested localities. On the other hand, in the locality not yet reforested, the average carbon stock is decreasing and the slopes are negative.

3) Quantity of carbon sequestered on GGW sites since reforestation.

One of the objectives of the Great Green Wall initiative is to sequester 250 million tons of carbon in the soil by 2030. Thus, this part of the work is dedicated to the assessment of the quantity of carbon stored on each site after the reforestation campaigns of the GGW initiative in Senegal.

In the commune of Syer, the quantity sequestered after 2010 is estimated at 1,679,931 ± 127,085 t.C. which represents an annual gain of 119,406 ± 9550 t.C/year. In Tessekere the carbon stock for the period 2009-2020 represents 3,729,318 ± 283,822 t.C and corresponds to 248,848 ± 17,420 t.C/year. In Loughere-Thioly after the various reforestation campaigns (started in 2009), the sequestration of organic carbon amounts to 362,902 ± 29,032 t.C, or 30,495 ± 1525 t.C/year. For all of the reforested sites in our study, the gain in carbon sequestration amounts to 5,772,151 ± 439,939 t.C or 398,749 ± 27,910 t.C/year.

On the other hand, on the non-reforested site of Ballou, the loss is estimated at −3055 ± 153 t.C, or -204 ± 10 t.C/year (see Figure 10).

The results of the organic carbon storage estimation in GGW Senegalese soils show an upward dynamic in carbon sequestration for the reforested sites of Syer, Tessekere (Louga), Loughere-Thioly (Matam) and a loss capacity to store carbon on non-reforested sites (case of Ballou in Tambacounda region) over the period 1992-2020. Other works ( [17] [50] ) had shown a positive dynamic of reforested sites based on vegetation indices (NDVI and VHI). The results of these studies reinforce ours on the positive dynamics of carbon sequestration observed in the different reforested soils of the GGW. Carbon storage and emission could be considered as good integrators of the state of plant cover health and anthropogenic pressures [88] .

In the Senegal Sudanian zone, land cover and land use have a strong impact on the storage of organic carbon [84] . This study shows that the conversion of forests to peanut fields causes a transformation of soil texture and a loss of organic carbon. Indeed, the carbon storage of dominant crop soils, in this case

peanut fields, is 27% to 35% lower than in semi-natural savanna soils. Organic carbon stock total is lower for sandy soils than in sandy clay soils. Furthermore, the study of [52] which was carried out on carbon stock measurements of 15 sites ranging from the North (Dagana department) At the South-East (Tambacounda region) in Senegal made a comparison between soil carbon stocks under the canopy and outside the canopy of trees. The study reveals that under the crown the carbon stock is estimated to 22.1 t.C/ha compared to 17.6 t.C/ha of outside the crown.

In the sub-region particularly in the agroforestry park of Saria in Burkina Faso, it is shown that the first layers of the soil contained 91% of the total carbon stock and these quantities varied between 8.74 ± 6.05 t.C/ha and 26.59 ± 7.94 t.C/ha ( [85] ). In the semi-arid landscape of Dano (Burkina Faso), the average stock is estimated at 24 t.C/ha. This study mentioned that the gallery forest soils stored more carbon (30.2 ± 15.6 t.C/ha) than those of savanna (22.1 ± 6.1 t.C/ha), forests (22 ± 8.2 t.C/ha), tree savanna (21.4 ± 7.4 t.C /ha), and cultivated land (14.9 ± 5.7 t.C/ha) ( [86] ). For these authors, these dry zone forest management systems play an important ecological role by contributing significantly to the fight against climate change through the strong potential for carbon sequestration. Trees and shrubs used in agroforestry systems increase carbon storage thanks to the addition of aerial and root biomass ( [56] [85] [87] [89] ). In Mali, the quantity of carbon sequestered varies depending on the agroforestry system and the maintenance techniques of these systems [89] . The rate of sequestered carbon differs from one system to another. It is estimated at 26 t.C/ha in the Fallow system followed by the Savanna orchard system of 19 t.C/ha and 8 t.C/ha for the Plantation system.

The dynamics of the average carbon stock of the different reforested sites is increasing. In municipalities not yet reforested, the temporal evolution of organic carbon storage is decreasing. This is the case of areas in full urbanization (Ballou).

In Senegal, the estimation of the temporal evolution of carbon storage in soils is often based on modeling approaches ( [55] [56] [84] [90] ). The Century and RothC models were the most frequently used to predict soil carbon dynamics under different agronomic and soil scenarios ( [56] [90] ). [90] applied the Century model to Senegalese agrosystems. The different modes of agricultural land use lead to a reduction in soil organic carbon stocks between 2002 and 2050. Continuous crop rotation (groundnut-millet) without external carbon input leads to the greatest reduction in carbon stocks (−1.4 to −3.9 t.C/ha). On the other hand, agroforestry based on Faidherbia albida plantations generates the greatest increase in soil organic carbon stocks of +11 t.C/ha between 2002 and 2050 or 0.23 t.C/ha/year. The application of the RothC model in the same region shows that crop rotation (groundnut-millet) leads to the most significant reduction in organic carbon stocks from −1.8 to −5.1 t.C/ha ( [55] [56] ). Organic carbon stocks (from layer 0 - 25 cm) went from 8.1 t.C/ha in 2009 to 3.3 t.C/ha in 2050, a loss of 4.8 t.C/ha. By integrating an agroforestry scenario based on Faidherbia Albida, the increase in carbon stock is estimated at +12 t.C/ha between 2009 and 2050, i.e., an annual gain of 0.29 t.C/ha/year. In northern Senegal, [91] quantified carbon stock dynamics with the GEMS model. In 2000, the average carbon stock, taking into account land use (cultivated plots, fallow plots, forest areas, savannas and reforestation), was estimated at 28 t.C/ha in the 0 - 40 cm layer. According to [91] , carbon stocks derived from bare soils vary between 5 t.C/ha and 15 t.C/ha for soils on irrigated cropping systems [91] .

In the semi-arid region of Dano in Burkina Faso, the results of the HadGEM2-ES and MPI-ESM-MR models by 2070 showed that climate change could affect the carbon storage potential of woody species in different land use and land cover (LULC). These models estimate the reduction in carbon storage capacity of 90% for the HadGEM2-ES model and 89.4% for the MPI-ESM-MR model ( [86] ).

The results of the temporal evolution of carbon sequestration from these different simulations corroborate the dynamics observed on our study sites. On the other hand, a difference is noted between the carbon stock values obtained by the models and that derived from land use and land Cover maps. This difference could be explained by the input parameters used in the models (choice of planted species, soil texture, crop rotation, plantation ages) and the land cover data only taken into account in our study. Indeed, taking into account the species of reforested trees and the age of the plantations could improve the estimate of carbon storage ( [86] [92] ). The study carried out by [92] in Sudan, showed that aboveground biomass in Acacia-Senegal plantations (Senegalia Senegal) is more important in 45-year-old plantations than in plantations of 16-year-old trees. Then, the average stock of sequestered carbon is 11.9 t.C/ha for 35-year-old plantations and 1.2 t.C/ha for 16-year-old plantations.

This study, which was carried out in certain localities of the GGW in Senegal shows the impact of reforestation (started in 2006 on certain sites) on the capacity of trees to capture atmospheric carbon dioxide and bury it in the soil. The results of our study estimated the average total sequestration has 4,177,726 ± 12,539 t.C to Syer; 5,740,287 ± 23,306 t.C in Tessekere; 3,257,418 ± 7087 t.C in Loughere-Thioly and 1,353,630 ± 2779 t.C for Ballou. After reforestation, the storage of carbon in the soil and the reduction of CO₂ in the atmosphere of these different sites increased by +3.5% at Loughere-Thioly and by +12.7% at Syer and Tessekere.

These results are consistent with the work of [93] [94] [95] . This work shows that vegetation restoration is an effective tool for increasing plant biomass and organic carbon content. And, Soil organic carbon content is significantly higher in natural vegetation restoration than in managed vegetation and plowed land.

In Senegal, the implementation and inventories of forest areas as part of the PROGEDE 1 and 2 projects (Sustainable and Participatory Management of Traditional and Alternative Energy Project; [54] [96] ) in seven regions of Senegal since 1998, have made it possible to quantify 15.37 million tons of sequestered carbon and 56.41 million tons of CO₂. The highest values of stored carbon were recorded in the forests of eastern and southern Senegal. The carbon quantities of the Koar, Guimara and Kandiator sites were 3,488,107 t.C respectively; 2,444,917 t.C and 2,422,389 t.C.

In China, the implementation of Grain to Green Program (GTGP) and eco-environmental emigration in the rocky Karstic desert region (Southwest China) have made it possible to sequester carbon and produce oxygen. Between 2000 and 2010, [97] showed the impact of ecological rehabilitation initiatives started in 1999 in this region on soil carbon storage and oxygen production. It notes a significant annual increase of 20.94% in carbon sequestration and oxygen production. The total increase in carbon and oxygen production in different counties in this region was estimated to be 7.66 million tons and 3.51 million tons respectively in Baise and Huanjiang.