The present study was carried on in the region of Benslimane (western Morocco), located on the Atlantic coast between Rabat and Casablanca. The bioclimate is Mediterranean, semi-arid (annual rainfall: 450 mm) with mild winters (mean daily minimum temperature: 7.5 °C) [28]. Hydromorphic soils are developed on a quartzitic sandstone substrate. The landscape is composed of forests of Quercus suber and of Tetraclinis articulata mixed with cultivated fields and restraint plantations of Pinus spp. and Eucalyptus spp. [28]. This province is characterized by a great abundance of temporary pools, which cover 2% of its total surface area, with a wide range of size, shape, depth, use and location [24,29].

In the Benslimane cork oak forests, 16 temporary pools have been selected (Table 1), 8 of them located within two ungrazed hunting reserves (thereafter named “Ungrazed” pools), and the other 8 in public forests, which are extensively grazed by cattle and sheep (thereafter named “Grazed” pools). In these 2 hunting reserves, grazing has been excluded since 1975. The vegetation of these 16 pools was studied in 2009 at two sampling dates (February and May) and at 2 locations within the pool (the Centre and the Edge), since the species composition varies with hydromorphy [23,25]. At each date and location, phyto-ecological surveys were carried out on homogenous vegetation units (46 m², corresponding to the minimal area according to Gounot 1969 [30]). These surveys consisted in a vegetation relevé, estimating the cover in 6 classes of each species according to Braun-Blanquet method [31]. The annual or perennial trait was attributed for each species, following the flora of North Africa [32] and Morocco [33]. The water depth was measured in each survey plot at each date.

The vegetation of each pool was characterized by the number of cumulated species recorded at the 2 dates of sampling for the following categories: the “Total species richness”, measured as the total number of species recorded, “Rare species” calculated as the number of species with patrimonial interest (as defined by [34]), “Pool species” calculated as the number of aquatic and amphibious species that are considered as characteristic of wetland habitat according to North African floras [22,35,36] and to previous works on Moroccan pools [24] and “Opportunistic species” calculated as the number of terrestrial species originating from surrounding forests and agricultural landscapes [32,33] that can establish in the temporary pools during the dry phases. The abundance of Pool species, Opportunistic species and Rare species were calculated for each pool as the cumulative sum of the abundances of the species in each respective group.

In order to assess the impact of grazing on the structure of the vegetation, 6 temporary pools (3 Grazed and 3 Ungrazed) were selected among the 16 pools previously studied and were monitored for three years (2007, 2008, 2009). The vegetation was measured 3 times in 2007 (March, May and June) and 4 times in 2008 and in 2009 (January, March, May and June). The vegetation of the 6 pools was studied in quadrats (0.3 × 0.3 m, divided into 9 squares of 0.1 × 0.1 m) which were placed each 2 m along 2 permanent orthogonal transects crossing in the deepest part of the pools. Within each quadrat, the abundance of each species was measured as the number of squares in which it was present (0 to 9). The water depth was measured in each quadrat of vegetation survey.

Within each pool, three plant communities common to all pools were distinguished according to their dominant species, and corresponding to different locations along the topographic gradient (Edge, Intermediate and Centre belts). The communities were respectively dominated by Ranunculus baudotii and Glyceria fluitans in the Centre, by Isoetes velata and Bolboschoenus maritimus in the Intermediate, and by Leontodon saxatilis and Narcissus viridiflorus in the Edge belt. In each of these communities, 5 successive quadrats were selected along the permanent transects. The community structure was then studied within 15 quadrats (0.3 × 0.3 m) for each pool (5 quadrats in each of the 3 topographical levels or 90 quadrats in total over the 6 pools). The number of Annuals, of Perennials and the Total number of species per quadrat (hereafter referred respectively as “Annuals”, “Perennials” and “Total” species richness) were calculated separately for each year cumulating the successive measurements made during the year.

The relationships between species richness and biomass were studied using biomass measurements made in the 6 pools (3 Grazed and 3 Ungrazed) in February 2008, at the expected peak of biomass on Benslimane vernal pools [26]. Biomass was measured in 3 plots (1 × 1 m²) in each community (Centre, Intermediate and Edge; 9 plots in each pool and 54 plots in total) randomly distributed at close vicinity of the permanent transects settled for the study of species composition (see above). The collected biomass from each plot was oven-dried (60 °C) until a constant weight. The species richness (Total and Annuals) of the vegetation nearby biomass plots was calculated using the 3 corresponding (closest) vegetation plots (transects). The species richness was calculated as the cumulated number of species encountered in February 2008 in the 3 quadrats measured.

Within each of the 16 pools, two soil samples were collected (10 first cm of soil) respectively at the Edge and the Centre near the vegetation relevés. On these samples, were measured pH, particle size (silt, clay, sand), content in organic matter, total nitrogen, total salinity and total phosphorus. The grain size analysis followed the “Bouyoucos” method [37]. The pH of the soil was measured using a pH-metre in a suspension of fine soil (soil/water in a 1/2.5 proportion). The total salinity of the sediment was measured by electric conductivity (Philipps PR 9801) of an aqueous solution of soil/bi-distilled water (in a 1/5 proportion), previously shaken for 1 hour (150 rpm) [37]. The protocol described by [38] was followed to determine the organic carbon and the total nitrogen (Kjeldahl).

2.6.1 Study at the regional scale

2.6.2 Study at the local scale

Over the 16 pools, a total of 155 species (Appendix A), have been encountered (75% annuals, 25% perennials) including 8 rare species for Morocco: Apium inundatum, Exaculum pusillum, Isoete velata, Myriophyllum alterniflorum, Pilularia minuta, Elatine brochonii, Elatine alsinastrum, and Lythrum thymifolia. Among this total, 23 species were exclusive to Grazed pools, 47 species were exclusive to Ungrazed pools and 85 species were common to both.

The most important environmental characteristics for vegetation (determined by forward selection RDA) are grazing regime, phosphorus content of soil, soil pH, and maximum water depth (Fig. 1). These variables explain 44% of the total variance of the vegetation composition (F = 2.15; p = 0.001). Grazing explains 14.1% of the total variance (F = 2.29; p = 0.001), while the cumulative soil variables (pH and phosphorus) explain 19.6% (F = 1.92; p = 0.003), and maximum water depth explains 9.9% (F = 1.94; p = 0.007). The common variance explained by these variables is weak and not significant (p > 0.05).

A group of exclusive or more abundant species in Grazed pools (Plantago coronopus, Scilla autumnalis, Aira cupaniana, Rumex pulcher…) is identified on the biplot ½ (Fig. 1). This group is opposed to a group of species most abundant or solely present in Ungrazed pools (Elatine alsinastrum, Apium inundatum, Myosotis sicula, Phalaris caerulescens, Rumex crispus, Echium plantagineum…). Most of the species are more abundant in Ungrazed pools (hunting reserves; Fig. 1). The plan ½ (Fig. 1) also separates Pool (aquatic and amphibious) species, which are abundant in deeper pools (Elatine alsinastrum, Apium inundatum, Myosotis sicula …) from Opportunistic species that are more abundant in shallow pools with soils richer in phosphorus (Phalaris caerulescens, Rumex crispus, Echium plantagineum, Linum usitatissimum).

The Total number of species per pool, the number of Pool species and the number of Rare species do not vary significantly between the two grazing regimes (p > 0.05). However, the total abundance of Pool species is significantly lower in Grazed than in Ungrazed pools (F = 11.32; p < 0.01). The abundance of Rare and Opportunistic species is not significantly different between the two pool types (p > 0.05). The Total number of species within each pool, the number of Pool species and the number of Rare species are not correlated with maximum water depth or with the size of pools (p > 0.05).

In 2008, the number of species per quadrat (respectively Total, Annuals and Perennials) decreased significantly with the maximum water depth (Table 3) but did not show significant differences between the grazing regimes. In 2009, the water depth had a similar significant effect on the Total species richness and on the number of Annual species, but was not correlated to the number of Perennial species per quadrat. The number of species per quadrat was significantly higher in Ungrazed than in Grazed pools (Table 3).

The Total number of species and the number of Annual species per quadrat (MANOVA, Table 4) differed significantly between Grazed and Ungrazed pools, along the hydromorphic gradient and between years. The number of Perennial species per quadrat was significantly different between belts (topographic gradient) and between years, with a significant effect of the interaction between belts and pool types. Similarly, the vegetation abundance in the quadrats (Total and Annuals) was significantly different between the pool types, between belts and between years (Table 4). The abundance of Perennial species per quadrat was significantly different between belts and between pool types, with a significant interaction between pool types and years. The abundance of Perennial species was significantly higher the 2nd year in the Central and Intermediate belts in Ungrazed pools. Conversely, it was significantly lower in all belts in Grazed pools. The pool type had a significant effect on the abundance of all groups of species, with a significant effect of belt and time (except for Perennial species) (Table 4).

In 2008, the richness in Annuals and the Total species richness per quadrat were positively correlated with the abundance of Perennial species per quadrat in Grazed pools (r² = 0.28; p < 0.05; n = 45 and r² = 0.44; p < 0.05; n = 45, respectively for the Annuals and the Total species richness) and in Ungrazed pools (r² = 0.23; p < 0.05; n = 45 and r² = 0.35; p < 0.05; n = 45 respectively for the number of Annual species and Total species richness). In 2009, the Annual species richness per quadrat was not significantly correlated to the abundance of Perennial species in both Grazed and Ungrazed pools (p > 0.05). The Total richness was weakly correlated to the abundance of Perennial species in the Grazed pools (r² = 0.14; p < 0.05; n = 45) and in the Ungrazed pools (r² = 0.22; p < 0.05; n = 45).

The biomass in February 2008 was significantly higher in Ungrazed (123.15 ± 25.81 g/m²) than in Grazed pools (42.26 ± 23.28 g/m²). The Total species richness and the number of Annual species per quadrat were weakly correlated to the total biomass of vegetation (Fig. 2).

The regional analysis shows differences in the species composition of the vegetation in the 16 studied pools. These differences are explained by grazing (14.1%), hydrology (9.9%) and soil characteristics (pH and phosphorus: 19.6%). Almost 30% of the species found in the 16 pools were exclusive to Ungrazed pools, while 15% were exclusive to Grazed ones and 55% common to both.

Grazing affects differentially the species and their abundance in the pools (Fig. 1) and most of species were consistently less abundant in Grazed pools than in Ungrazed ones due to the direct impact of herbivores (grazing and trampling). This notably concerns the species less tolerant to disturbance and the most palatable to livestock [26], which were more abundant in Ungrazed pools, i.e. located within hunting reserves (e.g., Juncus bufonius, Poa annua, Myosotis sicula, Phalaris caerulescens, Anthoxantum odoratum; Fig. 1). Conversely, the species that were more abundant in Grazed pools (Fig. 1) possess specific traits such as thorns (Carlina racemosa), hairiness (Hypericum tomentosum), repellent (i.e. aromatic) molecules (Glycyrrhiza foetida) or prostrate growth form (Plantago coronopus) [12,41] which make them little or ungrazed. Because these species are known to escape grazing, their abundance is used for assessment of grazing pressure [12,42].

The analysis of the impact of grazing on Pool species and Opportunistic species showed different responses of these groups to grazing. Pool species were favoured by the absence of grazing in contrast to Opportunistic and Rare species. The mechanisms involved are not clearly identified but could include weaker defences against herbivores for Pool species and/or the indirect impact of grazing on aquatic vegetation (through higher mechanical disturbance of submersed sediment, impact on water transparency…) [7]. No significant effect of grazing on the species richness per pool was found, nor on the number of Annuals, Perennials, Pool and Opportunistic species. This result contrasts with the presence of 2-times more exclusive species in Ungrazed (47 species) than in Grazed (23 species) pools. The number of species was not correlated with the size of the pools, which can probably be explained by various factors such as the diversity of micro-habitats, the shape of the depression or the intensity of competition. The vegetation density was low even in the Ungrazed pools, and no negative correlation was found between species richness and biomass, suggesting a weak intensity of competition. The density of species varies significantly along topographic (hydromorphic) gradients, but the analysis of the species richness at the pool level could not integrate that factor. The surface area is unlikely to play a major role in the species richness of pools [43].

At the local scale (intra-pool), grazing had a lower but significant influence on species richness (Table 4) than topography (belts) and time (which explains mainly inter-annual hydrological variations). The impact of grazing on species richness (Total, Annuals and Perennials) was only significant in 2009, which was a wet year. Grazing thus interacts with hydrology, which had a smaller effect on the species richness in 2009 (significant only for the Total species richness and number of Annual species) (Table 3). The combined effect of hydrology and topography is usually the main variable that explains the species richness and the distribution of plant species in wetlands [25,44,45], the number of species declining when hydromorphy increases.

The effect of grazing was stronger on species abundance particularly that of Annuals (Table 4). The effect of grazing on the abundance of Perennials per quadrat showed a significant interaction between the grazing regime and years (Table 4) while the effect of year was not significant. In contrast, the impact of years on the abundance of annuals was much higher than that of grazing. These results show a stronger impact of grazing than of years (contrasted in terms of hydrology) or belts on the abundance of Perennials probably resulting from their long exposure to grazing. The long exposure of Perennial species to grazing and the restriction of the production by drought stress during the dry Mediterranean summers strongly affect their competitive ability [46]. However, the role of competition in the structure of plant communities is probably weak under dry Mediterranean climate. This was shown by the positive correlation found for 2008 between the abundance of Perennial species and the number of Annual species, as well as with the Total species richness. Similarly, the positive correlation between biomass and species richness (Fig. 2) suggests a limited role of competition in dry years even in the absence of grazing (pools within reserve). This highlights the role of abiotic factors, notably climate, in the control of the dominance of species [19–21] and thus in creating regeneration niches within plant communities that can be occupied by a large number of Annuals. In 2009 (wet year), the increased productivity changed the competitive relationships; this explains the non-significant correlation between the richness in Annuals and Perennials. No evidence of a potential reduction of the species richness by submersion stress or by higher production was found. This year corresponded probably for the plant community of the temporary pools to an intermediate disturbance (“Humped-back” model [47]). During dry years, the species richness was restricted by the low productivity.