Introduction
Phylogenetic resources are in perpetual evolution and transformation in the face of constraints and changes (e.g., climate, predation, parasites, resources) affecting the environment (Vaughan et al. Reference Vaughan, Brydges, Fenech and Lumb2001, Mortimore Reference Mortimore, Behnke and Mortimore2016). In the Sahel, this evolution could be occurring more rapidly due to the great instability of the climate (Hulme Reference Hulme2001, Ali et al. Reference Ali, Lebel and Amani2008). Repeated droughts have been observed, the most recent being those of the 1910s, 1940s and 1970s–1980s (Ozer et al. Reference Ozer, Hountondji, Niang, Karimoune, Manzo and Salmon2010). The 1970s–1980s droughts, in terms of duration and severity, had devastating environmental consequences (Lebel & Ali Reference Lebel and Ali2009, Ozer et al. Reference Ozer, Hountondji, Niang, Karimoune, Manzo and Salmon2010, Steinig et al. Reference Steinig, Harlaß, Park and Latif2018), including significant declines in woody plant cover (Brandt et al. Reference Brandt, Hiernaux, Rasmussen, Mbow, Kergoat and Tagesson2016). The causes are both climatic, with water deficit over several successive years and increases in temperature, and anthropogenic, including expansion of cultivated areas, increased demand for firewood and overgrazing (Kandji et al. Reference Kandji, Verchot and Jens2006, Vincke et al. Reference Vincke, Dhiou and Grouzis2009, Sissoko et al. Reference Sissoko, Keulen, Verhagen, Tekken and Battaglini2011, D’Odorico et al. Reference D’Odorico, Bhattachan, Davis, Ravi and Runyan2013).
Since the 1970s–1980s droughts, environmental monitoring has shown a great improvement in plant cover in the Sahel (Sendzimir et al. Reference Sendzimir, Reij and Magnuszewski2011, Brandt et al. Reference Brandt, Verger, Diouf, Baret and Samimi2014b, Epule et al. Reference Epule, Peng, Lepage and Chen2014, Zida et al. Reference Zida, Traoré, Bationo and Waaub2019b). However, detailed analyses of the nature of the woody plant species that now colonize the re-greened Sahelian agrosystems are rare. These have been largely confined to surveys of local people’s perceptions of changes in woody plant species since the 1970s–1980s droughts (Sop & Oldeland, Reference Sop and Oldeland2013, Hänke et al., Reference Hänke, Börjeson, Hylander and Enfors-Kautsky2016) and plant inventories at the scale of village landscapes (Hänke et al., Reference Hänke, Börjeson, Hylander and Enfors-Kautsky2016, Savadogo et al., Reference Savadogo, Ouattara, Pare, Ouedraogo, Sawadogo-Kaboré, Barron and Zombre2016). Comprehensive data over a larger scale are urgently needed.
Sustainable management of forest resources is an urgent global challenge, especially in arid areas, which account for 41% of the world’s land area and contain more than 2 billion, people living mainly in rural areas (EEM 2005, Dijk & Bose Reference Dijk, Bose, Bose and Dijk2016). The survival of the rural population in these areas depends largely on trees and dry forests (EEM 2005, Sattout & Caligari Reference Sattout and Caligari2011, Dijk & Bose Reference Dijk, Bose, Bose and Dijk2016). Burkina Faso, with a population of c. 21 million (INSD, 2009), does not escape this reality. The landscape, consisting mainly of agroforestry parks and rangelands, provides for a human population made up mainly of agropastoralists, their subsistence and income depending in part on the exploitation of both wood and non-wood tree products. Understanding these agrosystems (ecosystems under agricultural management) with respect to the dynamics of habitats and associated plant species is an essential step in the development of biodiversity management policies (Antos & Parish Reference Antos and Parish2002, Sattout & Caligari Reference Sattout and Caligari2011).
The aim of our study is to assess the woody floristic composition of re-greened agrosystems following the 1970s–1980s droughts in the Northern Region of Burkina Faso in order to provide greater ecological detail on re-greening areas at a much larger scale, with a focus on the analysis of plant communities by habitat. Has the re-greening been accompanied by a change in floristic composition? Sahelian woody plants play important roles in the sustainability of production systems and the improvement of the livelihoods of human populations (Breman & Kessler Reference Breman and Kessler1997, Harris et al. Reference Harris, Hobbs, Higgs and Aronson2006, Jose Reference Jose2009, Brandt et al. Reference Brandt, Hiernaux, Rasmussen, Mbow, Kergoat and Tagesson2016, Sidibé et al. Reference Sidibé, Sanou, Bayala and Teklehaimanot2017, Dollinger & Shibu Reference Dollinger and Shibu2018). We combine remote sensing to identify areas with improving woody plant cover with vegetation surveys to characterize the woody plant species involved.
Methodology
Study area
The study was conducted in the Sudano-Sahelian climate area of the Northern Region of Burkina Faso (Fig. 1) between latitudes 12°38’ and 14°02’ north and longitudes 1°33’ and 2°55’ west; this is an area of 13 950 km2. The rainfall pattern in the area is undergoing rapid variability, with an average annual rainfall varying by c. 700 mm/year (ANAM-BF 2017). Administratively, the area is composed of the Provinces of Loroum, Passoré, Yatenga and Zondoma.
Fig. 1. Location of study area, areas with an increase in enhanced vegetation index (EVI) and the spatial distribution of the vegetation survey plots.
Location of areas with increasing woody plant cover
The woody plant cover was monitored using satellite imagery at 30 m resolution from 1986, 1999 and 2015 from the Landsat 5 TM, Landsat 7 ETM+ and Landsat 8 OLI/TIRS sensors (USGS 2017). Images at the beginning of the dry season (October, November) were analysed in order to discriminate the herbaceous plants and crops. This period marks the end of the vegetative stage of herbaceous plants and crops and the persistence of the vegetative stage in almost all of the Sahelian woody species (Hiernaux et al. Reference Hiernaux, Cissé, Diarra and Leeuw1994, Brandt et al. Reference Brandt, Hiernaux, Rasmussen, Mbow, Kergoat and Tagesson2016). The atmospheric and terrain effects on the images were corrected using the ATCOR Ground Reflectance algorithm of PCI Geomatica 2017. The enhanced vegetation index (EVI; Equation 1), used to monitor changes in vegetation cover (detailed description in Zida et al. Reference Zida, Traoré, Bationo and Waaub2019b), has the advantage of integrating wavelength in the blue range, a soil reflectance correction factor and aerosol scattering correction coefficients to correct for the variation of the solar angle of incidence and to reduce atmospheric effects, as well as the signals emitted by the soil below the vegetation (Huete Reference Huete1988, Huete et al. Reference Huete, Didan, Miura, Rodriguez, Gao and Ferreira2002). This improves the accuracy of multi-date image comparisons taken at different times of the day under different soil and atmospheric conditions (Huete Reference Huete1988, Huete et al. Reference Huete, Didan, Miura, Rodriguez, Gao and Ferreira2002).
$${\rm{EVI = G\; \times \;}}{{\left( {{{\rm{\rho }}_{{\rm{nir}}}}\,{\rm{-}}\,{{\rm{\rho }}_{\rm{r}}}} \right)} \over {\left( {{{\rm{\rho }}_{{\rm{nir}}}}{\rm{\; + \;C1\; \times \;}}{{\rm{\rho }}_{\rm{r}}}\,{\rm{-\;C2\; \times \;}}{{\rm{\rho }}_{\rm{b}}}{\rm{\; + \;L}}} \right)}}$$
In Equation 1, ρnir refers to pixel values of the near-infrared band, ρr refers to pixel values of the red band, ρb refers to pixel values of the blue band, G is the gain factor (G = 2.5), L is the ground reflectance correction factor (L = 0.5) and C1 and C2 are correction coefficients of the aerosol diffusions (C1 = 6, C2 = 7.5).
The differencing method of EVI, using a pixel-by-pixel comparison, was used to detect changes in the vegetation index (Gandhi et al. Reference Gandhi, Parthiban, Thummalu and Christy2015, Jamali et al. Reference Jamali, Jönsson, Eklundh, Ardö and Seaquist2015). Areas with an increase in EVI between two dates were subjected to overlay analysis in order to establish the change sequences of the vegetation cover. This made it possible to identify and account for the areas where the EVI increased during 1986–2015 and those areas where it increased from 1999 (1999–2015). Figure 1 shows the different sequences of EVI enhancement obtained by image analysis.
Sampling
The mapping of the sequences that the EVI enhancement produced was used to define the number of plots to be inventoried. For each sequence, this was calculated using the normal approximation of the binomial distribution (Equation 2) (Dagnelie Reference Dagnelie1998).
$${\rm{n = }}{{{\rm{U}}_{{\rm{1-\alpha /2}}}^{\rm{2}}{\rm{\; \times \;p}}\left( {{\rm{1-p}}} \right)} \over {{{\rm{e}}^{\rm{2}}}}}$$
In Equation 2, n is the number of vegetation survey plots, p is the proportion of the considered sequence of woody plant cover improvement, e is the margin of error (a value of 8% is assumed) and U1–α/2 is the value defined by the normal law according to the desired confidence level (95% confidence level for a value of U1–α/2 = 1.96 is assumed).
A total of 118 vegetation survey plots were inventoried in areas with an increase in EVI (Supplementary Table S1, available online). The spatial distribution of the plots was randomly selected on a set of systematic grid points (400 m × 400 m) covering the entire area with an increase in EVI (Fig. 1). A number of the plots corresponding to the number defined in Table S1 was assigned to each sequence of EVI enhancement. A total of 61 of the randomly distributed plots corresponded with agroforestry parks (cultivated fields in association with woody plants including fallow land) and 57 with unprotected areas (bushland dominated by savanna and steppe mainly used for grazing).
In addition to the areas with an increase in EVI, 30 vegetation survey plots distributed equally over 3 protected areas (classified forests and locally protected patches of dense woody vegetation used for diverse sociocultural and religious ceremonies, commonly named ‘sacred forests’) dating from the period before the droughts of the 1970s–1980s were also conducted. These protected areas were composed of classified forest, sacred forest and community forest and were selected on the recommendation of the forest services of the Ministry of Environment of Burkina Faso because of their relatively good state of conservation, although these areas have also been subject to droughts and anthropogenic activities, such as wood harvesting, grazing and agricultural expansion.
Circular vegetation survey plots of 25 m radius (area 1963.5 m2) were used for the inventory of woody flora. All trees and shrubs of diameter at breast height (dbh) ≥5 cm were systematically inventoried and identified according Thiombiano et al. (Reference Thiombiano, Schmidt, Dressler, Ouédraogo, Hahn and Zizka2012). Plants were identified by their species, genus and family name. The number of plants of each species (n), the dbh of the plants and the habitat types (agroforestry park, unprotected area or protected area) of the vegetation survey site were also recorded.
Data analysis
The data analysis consisted of describing the floristic composition and characterizing the ecological status of the species based on the relative abundance of the species and the conservation status of the species according to the International Union for Conservation of Nature (IUCN 2019).
The floristic diversity of survey plots and habitats was assessed by calculating the species richness (S), Simpson’s diversity index (D’), the Shannon–Weaver diversity index (H’) and Pielou’s evenness index (J′) (Kindt & Coe Reference Kindt and Coe2005, Magurran Reference Magurran2005, Kempton Reference Kempton2006, Marcon Reference Marcon2017):
$${\rm{S = n}}$$
$${{\rm{D}}^{\rm{'}}}{\rm{\ = 1\;-}}\sum {\rm{p}}_{\rm{i}}^{\rm{2}}$$
$${\rm{H' = -}}\mathop \sum \nolimits_{{\rm{i\; = \;1}}}^{\rm{n}} {{\rm{p}}_{\rm{i}}}^{\rm{*}}{{\rm{log}}_{\rm{2}}}{{\rm{p}}_{\rm{i}}}$$
$${{\rm{J}}^{\rm{'}}}{\rm{\; = \;}}{{\rm{H}}^{\rm{'}}}{\rm{/lo}}{{\rm{g}}_{\rm{2}}}{\rm{S}}$$
where n is the number of species present in the vegetation sample plot and pi is the proportion of the sample belonging to the ith species.
The species richness is the total number of species present on a site, while Simpson’s diversity index measures the probability that two individuals randomly selected from a community are of the same species; it ranges from 0 to 1, with 1 being the maximum probability that individuals are of different species, reflecting a high diversity. The Shannon–Weaver diversity index describes the structure of a community by considering both species richness and the percentage of importance of individuals of each species in relation to all individuals of all species present: it usually ranges from 0 to 5, and the higher it is, the more diverse the community is. Pielou’s evenness index measures the distribution of individuals within species in a given community independently of the species richness; it ranges from 0 to 1, with 0 expressing the abundance of one of the species and 1 corresponding to an equitable distribution of individuals in the species. The species richness and diversity values of habitats were compared using descriptive statistics and Student’s t-tests.
The homogeneity between habitats and elemental survey plots was measured using Bray–Curtis dissimilarity according to the following formula (Bray & Curtis Reference Bray and Curtis1957, Ricotta & Podani Reference Ricotta and Podani2017):
$${\rm{B}}{{\rm{C}}_{{\rm{UV}}}}{\rm{\; = \;}}{{\mathop \sum \nolimits_{{\rm{j\; = \;1}}}^{\rm{s}} \left| {{{\rm{x}}_{{\rm{Uj\;}}}}{\rm{-\;}}{{\rm{x}}_{{\rm{Vj}}}}} \right|} \over {\mathop \sum \nolimits_{{\rm{j\; = \;1}}}^{\rm{s}} \left( {{{\rm{x}}_{{\rm{Uj}}}}{\rm{\; + \;}}{{\rm{x}}_{{\rm{Vj}}}}} \right)}}$$
where xUj and xVj are the abundance values of species j in plots U and V, respectively, and s is the total number of species recorded in these two plots.
Differences in species populations between two habitats are based on floristic composition (presence/absence of species) and species abundances in them (Chao et al. Reference Chao, Chazdon, Colwell and Shen2006). Abundance data were normalized to the total number of individuals in the habitat to minimize the effect of differential sampling size between habitats and elevated to log(x + 1) to decrease the influence of dominant species (Kindt & Coe Reference Kindt and Coe2005, Magurran Reference Magurran2005). The Bray–Curtis dissimilarity index ranges from 0 to 1, where 0 means that the two habitats share all of the same species and 1 means that they have no common species. Hierarchical clustering and ordination analyses based on Bray–Curtis dissimilarity were used to determine the existence of a type of plant community specific to each habitat and the species distribution around habitats (Kindt & Coe Reference Kindt and Coe2005).
The analyses were performed using JMP Pro 14 software and the R Vegan package.
Results
Species composition and ecological status
A total of 71 species in 51 genera and 23 families were identified (complete floristic composition data in Table S2). A total of 53 species in 39 genera and 21 families were identified in the protected areas, and 52 species in 37 genera and 19 families were in unprotected areas. In the agroforestry parks, 46 species in 34 genera and 17 families were recorded. The most diverse families in terms of species and genera were the Mimosaceae, Combretaceae and Caesalpiniaceae (Fig. S1).
The abundant species, regardless of their habitat, were mainly Combretaceae: Combretum micranthum, Combretum glutinosum, Combretum nigricans, Guiera senegalensis and Anogeissus leiocarpa (Table S2). There were also Acacia erythrocalyx and Acacia ataxacantha of the Mimosaceae and Saba senegalensis (Apocynaceae) in the protected areas; Cassia sieberiana (Caesalpiniaceae) in the unprotected areas; and Vitellaria paradoxa (Sapotaceae), Piliostigma reticulatum (Caesalpiniaceae) and Balanites aegyptiaca (Balanitaceae) in the agroforestry parks. In low-abundance or rare species, there were socioeconomically important species such as Bombax costatum, Tamarindus indica, Adansonia digitata, Parkia biglobosa and Pterocarpus erinaceus (Table S2). A total of 13 species were exclusively recorded in the protected areas, 6 in the unprotected areas and 4 in the agroforestry parks (Table S2). Two exotic species were identified: Azadirachta indica in unprotected areas and agroforestry parks and Mangifera indica in agroforestry parks (Table S2). Four species were threatened and at risk of extinction according to the IUCN criteria: Afzelia africana, Khaya senegalensis and V. paradoxa (vulnerable) and P. erinaceus (endangered; Table S2).
Density and structure of woody plants
The average density of woody plants was significantly different (p = 0.0420 and p < 0.0001) between habitats and increased in the following order: agroforestry parks (52 ha–1), unprotected areas (121 ha–1) and protected areas (443 ha–1) (Table 1).
Table 1. Mean density (calculated using value per plot) and structural characteristics (per plant) of woody plants for each habitat.
* Statistical significance (p < 0.05).
dbh = diameter at breast height; PA = protected area; UA = unprotected area; AP = agroforestry park.
However, the average size of woody plants was greater in the agroforestry parks, with an average dbh per plant of 20.91 cm, greater (p < 0.0001) than in protected areas (10.30 cm) and unprotected areas (10.71 cm) (Table 1).
Species diversity of habitats
The greatest species richness (S) values were observed in protected areas (53 species) and unprotected areas (52 species; Table S3), with 46 species identified in agroforestry parks. In contrast, agroforestry parks had the highest Simpson’s, Shannon–Weaver and Pielou’s diversity values, while the lowest values of these indices were observed in protected areas (Table S3).
Homogeneity of habitats
The Bray–Curtis dissimilarity index showed greater homogeneity of species in unprotected areas and agroforestry parks (dissimilarity 0.41; Table 2), while the mean values of the Simpson’s, Shannon–Weaver and Pielou’s diversity indices were also non-significant (Table 3).
Table 2. Bray–Curtis dissimilarity: total number of species in each habitat (bold), number of species in common between habitats (top right) and Bray–Curtis dissimilarity between habitats (bottom left).
Table 3. Comparisons of mean species richness (S) and diversity values (Shannon–Weaver diversity index (H′), Pielou’s evenness index (J′) and Simpson’s diversity index (D′)) of sampling plots by Student’s t-test.
* Statistical significance (p < 0.05).
PA = protected area; UA = unprotected area; AP = agroforestry parks.
Most distinct were protected areas and agroforestry parks with a Bray–Curtis dissimilarity of 0.66 (Table 2) and significantly different mean values of diversity indices (p = 0.0004 to <0.0001; Table 3). Protected areas had the highest mean diversity index values and agroforestry parks had the lowest, except for Pielou’s evenness index, which was higher in the latter (Table 3).
Analysis of plant communities
The hierarchical clustering analysis shows that the basic vegetation survey data were not distributed by habitat; the different habitats thus consisted of the same plant communities (Fig. 2). However, groupings at an aggregation level corresponding to a lower Bray–Curtis dissimilarity (below 0.7) show small groups specific to protected areas, unprotected areas and agroforestry parks (Fig. 2).
Fig. 2. Clustering dendrogram of the vegetation survey plots carried out in the different habitats based on Bray–Curtis dissimilarity. Bray–Curtis dissimilarity < 0.7: groups of at least three vegetation survey plots specific to protected areas (PA) = black; unprotected areas (UA) = dark grey; and agroforestry parks (AP) = light grey.
The ordination highlights that species were more dependent on habitat type (Fig. 3). Thus, species such as A. ataxacantha, A. erythrocalyx, A. gourmaensis, A. africana,Ancylobotrys amoena, Baissea multiflora, Commiphora africana, Detarium microcarpum, Grewia flavescens and S. senegalensis were more prevalent in protected areas. Boscia senegalensis, C. glutinosum, Lannea velutina, Pterocarpus lucens and Sterculia setigera were more frequent in unprotected areas. Acacia nilotica, A. indica, Dichrostachys cinerea, Faidherbia albida, Flueggea virosa, Leptadenia hastata, Gardenia ternifolia, P. reticulatum, B. costatum, M. indica, T. indica and V. paradoxa were much more common in the agroforestry parks.
Fig. 3. Distribution of vegetation surveys plots (grey) and species (black) by non-parametric multidimensional scaling (NMDS) based on Bray–Curtis dissimilarity. PA = protected area; UA = unprotected area; AP = agroforestry park. Acronyms of species are listed in Supplementary Table 2.
Discussion
This study greatly advances understanding of the post-drought (1970s–1980s) woody floristic composition of re-greening Sahelian agrosystems. Protected areas are made up of the same plant communities as unprotected areas and agroforestry parks; however, they remain the most diverse (species richness) and the most dense, probably because of their low anthropogenic pressure (Savadogo et al. Reference Savadogo, Tigabu, Sawadogo and Odén2007, Bognounou et al. Reference Bognounou, Thiombiano, Savadogo, Boussim, Oden and Guinko2009, Ceperley et al. Reference Ceperley, Montagnini and Natta2010, Sambare et al. Reference Sambare, Ouedraogo, Wittig and Thiombiano2011).
The diversity of woody flora in the present study (71 species from all habitats) is greater than that of Bognounou et al. (Reference Bognounou, Thiombiano, Savadogo, Boussim, Oden and Guinko2009), who reported 36 species in 27 genera and families and 57 species in 41 genera and 23 families, respectively, in the southern Sahelian and northern Sudanian regions of Burkina Faso. It is, however, below the 90 woody species of socioeconomic importance in 32 families reported by Sop and Oldeland (Reference Sop and Oldeland2013) from the Sudano-Sahelian region of Burkina Faso based on interviews with local people. These semi-structured interviews were with 87 groups of informants from 20 villages belonging to three ethnic groups (Mossi, Fulani and Samo), where each group was asked to list all of the woody plants they used. It is worth recognizing that the woody flora of the present study is only a small part of the total national woody flora of at least 376 species (Lebrun et al. Reference Lebrun, Toutain, Gaston and Boudet1991, Thiombiano et al. Reference Thiombiano, Schmidt, Dressler, Ouédraogo, Hahn and Zizka2012). The low floristic richness is partly attributable to the difficult climatic conditions of the study area, which does not favour the development of many species, the distributions of which are more in the south of the country, where the climate is wetter (Bognounou et al. Reference Bognounou, Thiombiano, Savadogo, Boussim, Oden and Guinko2009).
Compared to the pre-drought flora, the study area is experiencing a substantial decline in the diversity of woody species. The dominance of Combretaceae (C. micranthum, C. glutinosum, C. nigricans, G. senegalensis and A. leiocarpa) highlighted in this study, and confirmed by post-drought floristic surveys in the Sudano-Sahelian region (Savadogo et al. Reference Savadogo, Tigabu, Sawadogo and Odén2007, Bognounou et al. Reference Bognounou, Thiombiano, Savadogo, Boussim, Oden and Guinko2009, Tindano et al. Reference Tindano, Ganaba, Sambare and Thiombiano2015, Kusserow Reference Kusserow2017), differs from the 1940s–1950s observations of Lavauden (Reference Lavauden1941) and Roberty (Reference Roberty1946, Reference Roberty1954), which, in addition to the Combretaceae, noted the following as common species: A. ataxacantha, B. senegalensis, Diospyros mespiliformis, B. aegyptiaca, B. costatum, K. senegalensis, P. lucens, Sclerocarya birrea, T. indica and Terminalia avicennioides. The agroforestry parks were then almost all made up of trees in mixed stands dominated by V. paradoxa and P. biglobosa (Roberty Reference Roberty1946, Reference Roberty1954). Species of great socioeconomic importance once fairly common or common (Roberty Reference Roberty1946), such as K. senegalensis, P. biglobosa, P. erinaceus, T. indica and V. paradoxa, are now rare and threatened with extinction (Sop & Oldeland Reference Sop and Oldeland2013, IUCN 2019).
These findings corroborate previous work demonstrating a significant decrease in diversity and density of post-drought woody species compared to the pre-drought situation (Gonzalez et al. Reference Gonzalez, Tucker and Sy2012, Sop & Oldeland Reference Sop and Oldeland2013, Brandt et al. Reference Brandt, Hiernaux, Rasmussen, Mbow, Kergoat and Tagesson2016, Kusserow Reference Kusserow2017). Botanical inventories in 1985, 1991 and 1992 by Kusserow (Reference Kusserow2017) in the Canal du Sahel area in Mali showed a decline and significant changes in species composition towards more robust and arid-tolerant species since the work of Roberty (Reference Roberty1946). Gonzalez et al. (Reference Gonzalez, Tucker and Sy2012) detected a significant decline in woody tree density of 18 ± 14% between 1954 and 2002 in the Western Sahel and a significant decrease of 21 ± 11% in species richness in the Sahelian, Sudanian and Guinean areas. Interviews regarding perceptions of woody species dynamics with local people in the Sudano-Sahelian area of Burkina Faso show that more than 80% of 90 listed species are in decline, while species such as B. aegyptiaca, A. nilotica, Acacia seyal, Eucalyptus camaldulensis and Moringa oleifera have increased in terms of numbers of individuals (Sop & Oldeland Reference Sop and Oldeland2013). The last two species are of exotic origin. Inventories of woody plants in re-greening and control areas using high-resolution imagery and population surveys in central Senegal between 1983 and 2010 point to a decline in tree population sizes, a strong increase in shrub density and a decline in species diversity, interpreted as a shift towards more arid-tolerant species (Herrmann & Tappan Reference Herrmann and Tappan2013). Inventories of woody plants in the North and Northern Centre Regions of Burkina Faso showed higher densities, numbers of species, height classes, diameter classes and Simpson and Shannon indices in re-greening areas than in degraded areas; re-greening was more perceptible in agroforestry parks with greater densities and heights of plants with lower diameters (Savadogo et al. Reference Savadogo, Ouattara, Pare, Ouedraogo, Sawadogo-Kaboré, Barron and Zombre2016). Inventory and interview data also reveal that the species composition has changed substantially towards the dominance of drought-resistant and exotic species in the same northern region (Ouahigouya) of Burkina Faso (Hänke et al. Reference Hänke, Börjeson, Hylander and Enfors-Kautsky2016).
Post-drought floristic composition change can be explained by natural selection, which here is in favour of species that are more tolerant to drought, and a selective decline of the species that do not thrive in extreme water stress (Hiernaux et al. Reference Hiernaux, Diarra, Trichon, Mougin, Soumaguel and Baup2009, Kusserow Reference Kusserow2017). The dominance of Combretaceae and thorny Acacia and Balanites species, which are tolerant to drought and shallow and skeletal soils (Lebrun et al. Reference Lebrun, Toutain, Gaston and Boudet1991, Gonzalez et al. Reference Gonzalez, Tucker and Sy2012, Sop & Oldeland Reference Sop and Oldeland2013, Brandt et al. Reference Brandt, Verger, Diouf, Baret and Samimi2014b, Schmidt & Zizka Reference Schmidt and Zizka2014), attests to this vegetation shift. The increased presence of food-producing species in agroforestry parks such as V. paradoxa, B. aegyptiaca, Lannea microcarpa and S. birrea also reflects a shift towards the needs of the human populations. These species contribute to the improvement of livelihoods, especially of rural populations, and the diversification of the sources of production in the event of the loss of the agricultural production season (Jose Reference Jose2009, Sop & Oldeland Reference Sop and Oldeland2013, Brandt et al. Reference Brandt, Hiernaux, Rasmussen, Mbow, Kergoat and Tagesson2016, Ouédraogo et al. Reference Ouédraogo, Bationo, Sanou, Traoré, Barry and Dayamba2017, Sidibé et al. Reference Sidibé, Sanou, Bayala and Teklehaimanot2017, Dollinger & Shibu Reference Dollinger and Shibu2018). Their low numbers in unprotected areas could be due to climatic reasons, but also could be due to the sustained exploitation of the local population; however, they are also protected, maintained and even sometimes planted on farms, on which farmers have property rights (Reij et al. Reference Reij, Tappan and Smale2009).
Some diversification in the floristic composition since the mid-1990s, which could bring back the pre-drought flora in the long term, might yet testify to the resilience of the Sahelian vegetation (Hiernaux et al. Reference Hiernaux, Diarra, Trichon, Mougin, Soumaguel and Baup2009). However, plant succession studies show that natural and anthropogenic disturbances influence the mechanism of community assembly (Walker & Moral Reference Walker and Moral2003, Chang & Turner Reference Chang and Turner2019) and make it unlikely that the present vegetation will return to that of the past (Walker & Moral Reference Walker and Moral2003). Although precipitation tended to increase in the Sahel from the 1990s (Anyamba & Tucker Reference Anyamba and Tucker2005, Nicholson Reference Nicholson2005, Bamba et al. Reference Bamba, Dieppois, Konaré, Pellarin, Balogun and Dessay2015), temperatures continue to rise, affecting soil water availability for plants (increased evapotranspiration) and likely damaging the reproduction and development of many species (Hedhly et al. Reference Hedhly, Hormaza and Herrero2008, Hatfield & Prueger Reference Hatfield and Prueger2015). Added to this is the increasing ecological footprint due to human population growth (Thiombiano et al. Reference Thiombiano, Schmidt, Dressler, Ouédraogo, Hahn and Zizka2012, Brandt et al. Reference Brandt, Romankiewicz, Spiekermann and Samimi2014a) and the development of biophysical and social adaptation practices (Zida et al. Reference Zida, Bationo and Waaub2019a), which also affect evolution and floristic composition. Ecosystems will therefore evolve in the Sahelian context towards new stable states in harmony with the combined effects of disturbances and abiotic factors (Lockwood Reference Lockwood1997, Suding et al. Reference Suding, Gross and Houseman2004, Couwenberghe Reference Couwenberghe2011, Erktan Reference Erktan2013, Hänke et al. Reference Hänke, Börjeson, Hylander and Enfors-Kautsky2016). The species that demonstrate ecological plasticity and/or that contribute to the needs of the human populations will be those that will tend to prevail.
Conclusion
Since the 1970s–1980s droughts, the woody vegetation of Sahelian agrosystems has shown improving woody plant cover, but this has become dominated by species of Combretaceae and thorny Acacia and Balanites, which are more tolerant to drought; this post-drought woody vegetation is more resistant to water stress. Several species common before these droughts are now rare and threatened with extinction. The continued degradation of woody resources has important socioeconomic consequences and requires conservation action. The preservation of these habitats through conservation efforts can be achieved through the strengthening of community initiatives to preserve existing forest resources, although the promotion of agroforestry cannot be neglected.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S037689292000017X.
Acknowledgements
We thank Emmanuel Amoah Boakye, Michel Ouédraogo, the editors and the reviewers for their very insightful comments and suggestions.
Financial support
We are grateful to the Canadian Francophonie Scholarship Program (CFSP) and the Conflict and Cooperation over Natural Resources in Developing Countries (CoCooN) Program of Dutch Cooperation for funding this research.
Conflict of interest
None.
Ethical standards
None.
