Diversity of psammophyte communities on sand dunes and sandy soils of the northern Sahara desert

2.1 Study area

The study area is located in the northern part of the Algerian Sahara in the regions of Ghardaïa and Ouargla (Fig. 1). Plant communities in this area are well adapted to ecological conditions of the Saharan habitats: Hamada, Regs, sandy soils, salty soils, depressions, Wadi beds. The natural and wild-grown vegetation offers the main fodder resources for the dromedary farmed in extensive husbandry and free grazing system (Chehma et al., 2008a; b). The climate is hot hyperarid ‘desertic’ with a large inter- and intra-annual variability of precipitation (occurring exclusively as rainfall), high temperatures, large temperature range, very intense evapotranspiration, severe and long drought episodes, and extreme weather events such as heat waves and extreme wind events (Quézel, 1965; Ozenda, 2004; Bradai et al., 2015b). According to Köppen climate classification, the region had an arid climate of hot desert “BWh”. De Demartone aridity index varied between 1 and 2, indicating hyperarid climates with precipitation deficit ranged between 85 and 98% (Table S1). According to ombrothermic diagram of Bagnouls and Gaussen, the drought period extended over 12 months a year (Fig. 1). Rainfall was very low (<36–76 mm/year), with <12 days of effective rainy days and the potential evapotranspiration very significantly exceeded precipitations all year round. July was the hottest month with the average of maximum temperature exceeded 41 °C, whereas the minimum temperatures in January did not drop below 3.2–4.4 °C. Monthly sunshine frequency was higher than 70% (Table S2).

Close

Fig. 1

Geographic map showing the location of sampling sites (numbered red pin icons) in sandy habitats of the Algerian Sahara Desert. Lower plots are ombrothermic diagrams of Bagnouls and Gaussen applied to meteorological data of Ouargla, Touggourt, Hassi Messouad, and Ghardaia.

Export to PPT

2.2 Choice of sampling sites in sand habitats

The study area was visited several times prior choosing the sampling sites to study the diversity of spontaneous plant communities in sandy habitats. Seven sites (Fig. 1) were selected based on landscape geomorphology, shape of sand accumulation, and homogeneity of ecological conditions. Site 1 (altitude = 359.64 m above sea level) represents a micro-dune established on Wadi beds where Stipagrostis pungens dominates, site 2 (altitude = 354.62 m a.s.l.) is a sand-filled rocky-bottom Wadi bed with S. pungens as characteristic species of the vegetation, site 3 (altitude = 145 m a.s.l.) is a silted Reg with the dominance of Stipagrostis acutiflora, site 4 (altitude = 160 m) is a silted Reg dominated by S. pungens, site 5 (altitude = 132 m) is a silted Reg characterized by S. acutiflora, site 6 (altitude = 121 m a.s.l.) is a silted Reg dominated by S. acutiflora, and site 7 (altitude = 106 m a.s.l.) is a micro-dune where S. pungens is abundant.

2.3 Plant sampling and data collection

Assuming the homogeneity of environmental conditions within each site, ten sample plots were placed randomly in each of the selected sampling sites. Floristic surveys were conducted during the optimal period of vegetation development that coincides with favorable temperature and seasonal rainfall in the Sahara Desert (February-April 2015). The size of every plot of the ten samples was 100 m² (10 m × 10 m), which is a sufficient sampling area given the scatterness of vegetation in hot desert environments, and sand habitats in particular. This sampling area is also adopted in most studies of Saharan flora and vegetation, which makes comparison with previous studies more reliable (Benhouhou et al., 2003, Bouallala, 2013). In each plot, the density of each species was determined. Plant species were identified using identification books of Algerian flora (Quézel and Santa, 1962; 1963; Ozenda, 2004).

2.4 Analysis of plant diversity

The floristic richness and diversity was assessed for each sampling site using several descriptors of alpha diversity (species richness, Shannon diversity index and evenness) and beta diversity (Jaccard index and Bray-Curtis distance). The data from the sample plots were treated as separate data entities when computing alpha diversity whereas they were lumped together for computing beta diversity indices using sums of species abundances from sample plots. In addition, the species richness recorded at the sampling sites was compared using species richness scale adopted in the Saharan regions (Daget and Poissonet, 1991). Accordingly, for each site, the status of the flora was determined following the number of species (S) as: very species-poor (S < 11), species-poor (S = 11–20), moderate (S = 21–30), fairy species-rich (S = 31–40), species-rich (S = 41–50) and very species-rich flora (S > 50).

Plant alpha diversity was measured using Shannon diversity index (H'), $H^{\text{'}}=-{\sum }_{i=1}^S{(\frac{\mathit{ni}}{N}\times \log }_2\frac{\mathit{ni}}{N})$ , with S: total number of plant species in the sampling site, n_i: number of individuals of the species i in the site sampled, N = total number of individuals of S species. The evenness index “E” was computed as the ratio between Shannon diversity index and the theoretical maximum diversity ‘H_max = log₂S’. Evenness values vary between 0 and 1, with maximum evenness recorded (i.e. ∼ 1) when all plant species have very similar densities. The assessment of evenness is useful for detecting the structure of a community as it helps effectively in detecting anthropogenic changes (Neffar et al., 2013, 2016).

Using the ratios of the generic or family richness to species richness (G/S or F/S), the phylogenetic diversity was analyzed through the taxonomic structures for each sampling site. Since the taxonomic structure of communities varies following habitat environmental conditions, the relationships between vegetation cover measured at each sampling plot and the number of families, genera and species were explored using Pearson correlation tests. We assumed that with the more favorable ecological conditions in sandy habitats, vegetation cover and accordingly the richness of hierarchical taxonomic categories (family, genus and species) would increase. Thus we predicted significant positive correlations between vegetation cover and the number of taxa at different taxonomic levels. The species-higher taxon relationships, i.e. relationships between species richness and generic or family richness, were investigated using power models linearized after natural logarithm (ln) transformations (Fan et al., 2017). We applied these models: ln(G) = a × ln(S) and ln(F) = a × ln(S) for the species-genus and species-family relationships, respectively; where G represents the number of genera, F the number of families and S the number of species. One-way ANOVA was used to test the variation of G/S and F/S ratios among sampling sites.

Spatial similarity (beta diversity) of plant diversity between sampling sites was assessed qualitatively using the Jaccard index using species presence/absence data per site, and quantitatively by Bray-Curtis distance using data of average species densities per site. A clustering analysis (Ward’s method) was applied to the results of the similarity analysis in order to distinguish homogeneous groups of sampling sites “sandy habitats” according to their composition and abundance of psammophyte species.

2.5 Species morpho-funtional traits and chorological features of plant communities

The psammophytic and other plant communities in each sampling site were characterized using several aspects related to the biology, ecology and biogeography of plant species. Using various sources as detailed here below, Raunkiaer lifeform, morphological, dispersal, Noy-Meir, and chorological types were assigned and defined for each plant species.

Plant life forms were first defined by Raunkiaer (1904), then the concept of biological spectrum was introduced by Raunkiaer (1907). The latter helps to determine the proportions of each life forms in the flora of a given site. Thus, to establish the biological spectra by site, life forms of plant communities in sandy habitats were determined based on the work of Quézel (1965). The determination of the morphological types, i.e. annual vs. perennial species, was based on the work of Quézel and Santa (1962; 1963) and Ozenda (2004).

The plants present several dispersal strategies following the types of diaspores. According to the agent involved in the dissemination of diaspores van der Pijl (1982) classified plants into five main categories namely anemochore, autochore, barochore, hydrochore and zoochore. Plants of sandy habitats were classified following this classification where the types of plant dispersal were determined based on Quézel and Santa (1962; 1963), Jauffret (2001) and Bouallala et al. (2020).

Based on plant adaptation strategy against drought, Evenari et al. (1975) and Noy-Meir (1973) classified plants into two categories: arido-passive plants which are defined as species which do not exhibit any photosynthetic activity during the dry period, and arido-active plants which exhibit photosynthetic activity, even reduced, during the dry period. For the classification of plants in sandy habitats of the Sahara Desert according to these categories, previous studies were used to determine the category of each species (Quézel and Santa, 1962; 1963; Ozenda, 2004; Jauffret, 2001; Bradai et al., 2015a; Bouallala et al., 2020).

According to the phytogeographic classification of Quézel (1978), the study area belongs to the Holarctic realm, Mesogean sub-realm, Saharo-Arabian region, Saharan sub-region, and the northern Sahara domain. The determination of the chorological types of plant species encountered in sandy habitats was based on the study of Quézel (1965).

The appreciation of the abundance for plant species identified in sandy soils and dunes of the Sahara Desert was defined based on the works of Quézel and Santa (1962; 1963) on the Algerian flora. Consequently, each plant species was classified according to its degree of rarity in Algeria, which includes these categories (from widespread to extremely rare): particularly widespread (CCC), very common (CC), common (C), fairly common (AC), quite rare (AR), rare (R), very rare (RR), and extremely rare (RRR).

3.1 Plant species composition, richness and structure

In general, in the sampled sandy habitats from the Northern Sahara Desert of Algeria we registered 29 plant species belonging to 27 genera and 18 families with a dominance of Brassicaceae (13.79%) and Poaceae (13.79%) (Table 1). During sampling, a total of 1244 individuals/100sqm were identified with a maximum density observed in site 3 (321 individuals/100sqm) followed by site 1 (262 individuals/100sqm) and a minimum in site 7 (80 individuals/100sqm). Among the species encountered, Stipagrostis pungens and Stipagrostis acutiflora showed the highest densities with a total of 426 and 402 individuals/100sqm, respectively. Stipagrostis pungens was dominant in sites 1 and 2 with mean densities of 15.1 ± 6.6 and 14.9 ± 22.5 individuals/relevé, respectively; the minimum was noted in site 3 with (0.4 ± 0.7 individuals/relevé). Stipagrostis acutiflora dominated at sites 3, 5 and 6 with 18.7 ± 7.3, 10.4 ± 6.7 and 11.1 ± 10.7 individuals/relevé, respectively. The singletons i.e. plants recorded with one individual in the entire study were Stipagrostis obtusa, Zilla macroptera, Pituranthos scoparius, Colocynthis vulgaris, Plantago ciliata and Fagonia glutinosa; most of these species occurred mainly in site 1 (Table 1).

3.2 Taxonomic structures

Overall, the ratio of generic richness to species richness (G/S) in the sandy habitats of the Sahara Desert of Algeria was 0.94 ± 0.13, whereas the ratio of family richness to species richness (F/S) was 0.91 ± 0.15 (Fig. 2). Values of both ratios were high in all sampling sites; it totaled 1.00 in sites 2, 4 and 7, while the lowest value was recorded in site 6 (F/S = 0.77 ± 0.13). According to one-way ANOVA, the values of both ratios varied significantly among the sampling sites, where the F/S ratio (F_{(6, 63)} = 6.86, P < 0.001) revealed slightly higher variability compared to the G/S ratio (F_{(6, 63)} = 6.20, P < 0.001).

Close

Fig. 2

Ratios of generic richness to species richness (G/S) and of family richness to species richness (F/S) for different sampling sites in sandy habitats of the Algerian Sahara Desert. The same letters associated with the means (solid colored circles) of F/S (lowercase letters) and G/S (uppercase letters) ratios are significantly not different (P > 0.05) among sampling sites following the Tukey’s HSD post hoc test. Vertical error bars are standard deviations.

Export to PPT

The linearized power functions assessing the relationships between species richness and richness of higher taxa (i.e. genus and family) are expressed in Fig. 3. The exponents of a in the power model “ln(G or F) = a × ln(S)” exhibited little variation in both genus-species and family-species relationships. According to scores of the goodness-of-fit of the models expressed as R², the linearized model of the family-species relationships displayed lower stability in comparison to the genus-species relationship. In all sampling sites, both family-species and genus-species relationships were positive and statistically significant (linear regressions: P < 0.01).

Close

Fig. 3

Relationships between species richness (S) and generic (G)/family (F) richness for different sampling sites in sandy habitats of the Algerian Sahara Desert. The color of points is mapped to sampling sites, and size values are set to F/S (top plot) and G/S (bottom plot) taxonomic ratios. The solid black lines represent a simple linear regression with 95% confidence regions in light gray. The R-squared is a goodness-of-fit measure for the linear regression model.

Export to PPT

3.3 Relationships between taxonomic structures and vegetation cover

Pearson’s correlation tests showed that the variations in the number of families, genera and species were positively associated to vegetation cover measured at the level of sampling plots (Fig. 4). While the number of all taxonomic levels was deemed to be positively correlated to plant cover, the number of species (r = 0.51, P < 0.001) then genera (r = 0.49, P < 0.001) showed slightly higher increase with the increase in vegetation cover in comparison to the number of families (r = 0.43, P < 0.001).

Close

Fig. 4

Relationships vegetation cover and the richness of species, genera and families of psammophytic communities sampled on sand dunes and sandy soils of the northern Sahara Desert. The solid black lines represent a linear regression with 95% confidence regions in light gray.

Export to PPT

3.4 Plant diversity

The species richness in sampling sites fluctuated between 5 and 17 species; it varied according to the landscapes. Overall, the floristic diversity of sandy habitats was very low. Species richness ranged between poor in site 1 (S = 17 species) or very poor (<10 species) in the rest of the sampling sites where species richness varied between 5 and 8 (Table 2). The biodiversity measurement revealed that site 1 had the highest value in the Shannon index (H = 1.49) and that site 2 had the lowest value (H = 0.86). In sites 3–7, Shannon index values varied between 1.12 and 1.30. The most balanced plant communities were noted in sites 4 and 6 (E = 0.72), followed by sites 3 and 7 (E = 0.63–0.64). The lowest value of evenness was recorded in site 2. Due to the presence of many rare species, site 1 showed relatively low evenness although it had the highest values of species richness and Shannon index.

3.5 Analysis of similarity between sites

The Venn diagram shown in Fig. 5 represents the distribution of species between the study sites. The diagram showed that there is a single species (Stipagrostis pungens) omnipresent and common among all the sites studied. Most of shared species were present in site 1. While the sites 4–6 had no exclusive species (i.e. all their species were recorded in other sites), eleven species occurred exclusively in site 1, site 3 had three, site 2 contained two and site 7 had one exclusive species (Fig. 5). These species were Colocynthis vulgaris, Pituranthos scoparius, Stipagrostis obtuse, Zilla macroptera, Pulicaria crispa, Danthonia forskalii, Antirrhinum ramosissimum, Monsonia nivea, Linaria aegyptiaca, Helianthemum lippii, and Rhantherium suaveolens for site 1; Erodium glaucophyllum, Malcolmia aegyptiaca, and Savignya longistyla for site 3; Fagonia glutinosa, and Plantago ciliata for site 2 and Cyperus conglomeratus for site 7 (Table 1).

Close

Fig. 5

Seven-set Venn diagram displaying plant species richness (S) recorded at various sites of sand dunes at the Sahara Desert of Algeria. Values of S reported between square brackets represent the total number of species in each sampling site, whereas numbers within the diagram are species shared exclusively between the corresponding sites.

Export to PPT

The qualitative analysis of plant community similarity between sites using Jaccard index showed the highest similarity scores between sites 5 and 6 (similarity = 75%) and sites 4 and 5 (similarity = 62.5%). These three sites along with sites 3 and 7 were clustered as a homogenous group which had high similarities of plant composition (Fig. 6). Besides, sites 1 and 2 were clustered a unique group (similarity = 26.3%). These two sites exhibited low similarities (<18%) with sites composing the first cluster. For the quantitative similarity computed using Bray-Curtis distance based on plant abundance data, sites 5–6 (similarity = 85%), sites 1–2 (similarity = 64.4%) and sites 4–7 (similarity = 58.2%) represented the highest similarity values. All other similarities were <50%, except for the similarity between sites 3–5 (similarity = 53.6%). The clustering analysis revealed that sites 1, 2, 4 and 7 had quantitatively similar plant composition, whereas sites 3, 5 and 6 were also homogenous. Intergroup similarities between sites composing these two groups were low.

Close

Fig. 6

Heatmaps, with dendrograms (side colored plots) of hierarchical clustering analysis (Ward’s method), displaying the matrix of plant similarities between sampling sites of sandy habitats in the Sahara Desert of Algeria. Similarity was displayed in color intensity proportionally to similarity scores: Jaccard’ coefficient (upped plot) and Bray and Curtis distance (bottom plot). Yellow rectangles within the heatmaps matrices represent homogeneous clusters of sites based on plant composition.

Export to PPT

3.6 Plant functional traits

With the exception of sites 3 and 7, chamaephytic species had the highest values in the biological spectra of lifeforms in sites 1, 2, 4, 5 and 6 with 10, 3, 3, 5 and 4 species, respectively (Table 3, Fig. 7). Hemicryptophytes, with 3 species, had slightly higher values than chamaephytes in sites 3 and 7. Therophytes were only observed in sites 1, 2 and 3 with respectively 2, 2 and 3 species. Phanerophytes were represented by one species (Retama retam), which occurred at sites 1, 2 and 7. As morphological types were concerned, the flora was composed of 7 annuals and 22 perennial plants. While the annual species were completely absent in sites 4–7, perennials were numerous and well represented in sites 1–3, with a higher value at site 1 (15 species). Basically in almost all sampling sites, the dominant plant dispersal types were anemochores then barochores. The zoochore type was weakly represented, where only three species were recorded at site 1. Regarding the Noy-Meir's classification, arido-active plants dominated in all sampling sites, except site 3, with a good representation of species in site 1 (12 species) and sites 5 and 6 with six species for each (Table 3, Fig. 7).

Close

Fig. 7

Distribution of ecological and functional traits (life forms, morphological types, dispersal types, Noy-Meir’s types, chorological types, and abundance appraisal categories) for different sampling sites of sandy soils and dunes in the Sahara Desert of Algeria. Figures within histograms refer to the number of plant species.

Export to PPT

3.7 Chorological types

The analysis of phytogeographic (chorological) characteristics of the flora of sandy habitats of the northern Sahara Desert revealed a strong presence of the Saharo-Arabian chorological element in all sampling sites, with the highest level (10 species) recorded in site 1 and lowest level (3 species) noted in site 4 (Table 3, Fig. 7). The Mediterranean-Saharan-Arabian chorological element was observed only at sites 1–3 with 6, 1 and 1 species, respectively. As for the endemism, we noted the presence of the Saharan endemic element at sites 1, 3, 4, 5 and 7 represented by two species - Oudneya africana and Euphorbia guyoniana.

3.8 Appreciation of species abundance

Plant species of psammophytic communities inhabiting sandy soils and dunes of the Sahara Desert of Algeria included six categories of abundance for the flora of Algeria (AC: fairly common, C: common, CC: very common, CCC: particularly widespread, AR: fairy rare, and R: rare) (Table 3, Fig. 7). The particularly widespread category was recorded at site 1 with a single species (Helianthemum lippii). The common and very common categories were observed at all sites with varying rates. The fairly common category was represented by two species (Cornulaca monacantha and Pulicaria crispa), which were identified at sites 1, 5 and 6. The fairly rare category was represented by a single species (Oudneya Africana) recorded at sites 3–5. Whereas the category of rare species was represented by three species (Stipagrostis acutiflora, Rhantherium suaveolens and Randonia Africana), which occurred in sites 1, 2, 3, 5 and 6.