Arid steppes, with scattered vegetation, are commonly qualified as degraded ecosystems. Their natural vegetation cover is affected by both abiotic stresses and human activities and, therefore, suitable managements are needed for their sustainability. This study was carried out in the mountain rangeland chain of Matmata (southern dryland of Tunisia) during the springs of 2 rainy yr, 2019 and 2020. Our aims were to evaluate the impact of three grazing regimes (grazing exclusion [over 42 ys; 1978–2020]; seasonal [protected in spring and summer and grazed in autumn and winter, since the 1960s]; and continuous grazing [control; grazed over time]) on the cover of some plant life forms (Raunkiaer, Noy-Meir, and Grime types), under various conditions (sites, aspects, and elevations). To achieve this goal, the pseudoreplicated-nested design was applied, taking into account the site locations (grazing regimes), aspects (exposure to sunlight), and elevations. The cover of the retained plant life forms was measured using the quadrat point method. The main results showed that plant life forms are strongly affected by the grazing regime since the chamephytes (Ch_R), arido-active (AA_NM), and competitive stress-tolerant species (CS_G) are higher under seasonal grazing than control. All these species are qualified as less resistant to grazing. However, arido-passive (AP_NM), ruderal stress-tolerant (RS_G), and therophytes (Th_R) seem to be more resistant to grazing. Studying the impact of grazing management on the plant life form types provides useful information to find the most adapted species for rangeland sustainability all around the world.
The central goal of comparative ecology is to understand the diversity of functional and life history strategies among co-occurrence species (Ackerly 2004). In terrestrial ecosystems, many studies have been conducted to identify some adaptive strategies of plants to survive under harsh conditions. The arid steppe vegetation, with climatic and edaphic determinism, is characterized by a set of communities owing their physiognomy and composed mainly with grasses, forbs, and shrubs. This composition is essentially marked by the abundance of some perennial grasses (Stipa tenacissima L., Lygeum spartum Loefl. Ex. L…) and shrubs (Artemisia herba-alba L., Haloxylon scoparium Pomel…) growing in sparse tufts with irregular distribution of annual species (Aïdoud et al. 2006; Benaradj 2009). In this ecosystem, vegetation cover is subject to several constraints related to edaphoclimatic conditions and human activities that affect both the biophysical environment and the plant tufts. Tunisian drylands, with a mean annual rainfall < 200 mm, are currently facing degradation and desertification, mainly in the most arid rangelands. To conserve biodiversity and maintain the autoregeneration capacity (i.e., ecosystem resilience potential), defined as maintaining their capacity in the face of disturbances (Adger 2000), it is important to use appropriate management techniques. Fencing is one of the main management practices used to protect the rangeland plant communities. It significantly improves the aboveground vegetation productivity but decreases plant diversity and density after long protection (Wu et al. 2011). Among many authors, Ouled Belgacem et al. (2019) reported that seasonal grazing, including strategic rest periods, is generally better for rangeland vegetation than long-term animal exclusion. The adaptive strategies of the existing plants in already managed/protected areas can be assessed in order to select the more adaptive species and use them in future rehabilitation programs of very degraded or unresilient rangelands.
The Raunkiaer biologic types are considered as a flora adaptation strategy to climate (Raunkiaer 1934) Noy-Meir (1973). classified plants to arido-passive species (AP_NM), without photosynthetic activity during the dry period, and to arido-active species (AA_NM), which maintain such activity, even reduced, during this same period. To assign components of stress tolerance (S_G), survival under disturbance (R_G), and competitive ability (C_G), Grime (1974) developed a simplified determination key based purely on species characteristics. Investigating all these plant life forms is strategic to know the flora diversity and adaptation and to propose management programs for conservation of natural resources (Encina et al. 2016). In this context, the present study aims to understand the vegetation response by assessing plant life forms' distribution under the effects of several factors including human disturbances (i.e., grazing regimes), exposure to sunlight (north and south aspects), and elevation levels in arid montane rangelands. The main questions are what are the better life forms that describe the rangeland situation under various grazing regimes, on one hand, and which life forms can be grouped to indicate the resilience of these ecosystems, on the other hand?
Materials and Methods
Three study sites belonging to the Toujane region (Matmata mountain chain, southeast of Tunisia) were studied (Fig. 1). The mean annual precipitation in the region is about 150 mm with notable interannual variations. The physical environments of the three sites are similar (same soil substratum, topography, and geology), but they differ in altitudinal range and grazing regime (Fig. 2). Zmerten (site 1, 33.435332°N, 10.124727°E, 3 000 ha) is located in the middle of the mountain chain and has been strictly protected from grazing for 42 yr. Its natural vegetation is dominated by Juniperus phoenicea L., S. tenacissima, and Rosmarinus officinalis L. Brighith (site 2, 33.489832°N, 10.200800°E, 100 ha) is located on the northeastern border of the chain. This site is dominated by S. tenacissima and is devoted to seasonal grazing (protected in spring and summer and grazed in autumn and winter) with controlled animal charge (100 animals, sheep and goats, in the whole site). The control (site 3, 33.487485°N 10.199511°E) is closed to site 2 and dominated by S. tenacissima. It is continuously open to grazing (100 animals, sheep and goat, all over the year).
In each site, three 20-m line-transects, 100 m apart, are installed at three elevation levels (low, medium, and high) and two aspects to sunlight (south [high], north [low]) of the mountain hill (18 lines per site in total) in order to measure the floristic composition using the quadrat point method (Daget and Poissonet 1971; Floret 1988). A metal fine pin (1-m length) was entered perpendicularly (above-down) into the vegetation every 20 cm along the line (in total 100 hits per line). Each plant species touching the pin was recorded. Unknown plant species were identified using the Flora books (Alapetite 1979, 1981; Le Floc'h et al. 2010). The methodological approach is a pseudoreplicated-nested design, with the three grazing treatments (exclusion, seasonal, and continuous grazing) confounded with sites, within each of which is a nested aspect (north aspect vs. south), within which again is nested altitude (low, medium, high). Measurements were carried out in the springs of 2019 and 2020 (each site was sampled once per year). These 2 yr were rainy (190 mm and 322 mm, respectively, for 2019 and 2020).
Since the 2 studied yr are rainy, the data collected during the two springs were combined to obtain the mean cover of Noy-Meir, Raunkiaer, and Grime types (n = 6). The log transformation was used when data normality is nonrespected. Multiple separate one-way analyses of variance (ANOVAs) were undertaken to compare these types within and among sites. The multivariate general linear model was used to analyze the combined effect of elevation and aspect in the same site, using two-way ANOVA, on one hand, to study the combined effect of sites, elevations, and aspects and using three-way ANOVA, on the other hand. When an ANOVA was significant, the Duncan test was applied for classification. All these statistical analyses were performed using SPSS 20.0 software. The Origin Pro software (version 2021, Origin-Lab Corporation, Northampton, MA) was used for nonmetric multidimensional scaling analysis. Ordination distances between plant life forms and their relationship with sites were measured by Kruskal's stress.
Raunkiaer biological types
Chamephytes, hemicryptophytes, and therophytes are the dominant life forms in the studied sites (> 80%). The dominant Chamephyte species are Argyrolobium uniflorum (Decne.) Jaub. & Spach, Gymnocarpos decander Forssk. and Teucrium polium L. The hemicryptophyte species are mainly composed of S. tenacissima, Herniaria fontanesii J. Gay., and Helianthemum semiglabrum auct. The dominant therophytes are Fagonia glutinosa Delile, Plantago ovata Forssk, and Stipa capensis Thunb (Table 1).
Variations within sites
Changes in Raunkiaer biological types within the three sites are shown in Figure 3. After long-term protection (Zmerten, site 1), chamaephytes (Ch_R), hemicryptophytes (He_R), and therophytes (Th_R) covered 19.05%, 31.44%, and 10.33%, respectively, in the northern aspect. Their respective covers are 14.94%, 27.94%, and 10.67% in the southern aspect. The Ch_R and Th_R covers did not differ between the north and south aspects nor between elevations on the same aspect (P > 0.05). In the northern aspect, He_R showed high significant differences (P= 0.004) between elevations. The interaction between aspect and elevation was not significant for all types (P > 0.05).
Under seasonal grazing (Brighith, site 2), Ch_R, He_R, and Th_R covered 18.89%, 59.28%, and 0.28%, respectively, in the northern aspect. Their respective covers in the southern aspect are 24.94%, 39.67%, and 1.22%. Ch_R did not differ between elevations in the southern aspect nor between the two aspects for all elevations (P > 0.05). A significant difference was noted between the south and north aspects at low elevation (P = 0.012) for He_R. At the same time, a significant effect of elevation is found in the northern aspect (P = 0.013). Very low cover of Th_R was noted in all elevations for both aspects.
In the control site, Ch_R, He_R, and Th_R showed no significant variations between elevations in the northern aspect. However, there was a significant effect on the He_R group between the north and south aspects at low elevation (P = 0.015).
Variations between sites
Th_R participated with 10.5%, 0.75%, and 4.67% in the total plant cover of sites 1 (livestock exclusion), 2 (seasonal grazing), and 3 (continuous grazing), respectively. These species varied significantly between sites at low, medium, and high elevations in the north aspect (P = 0.037, P = 0.02, and P = 0.031, respectively). The interaction between sites and aspects was significant at low elevation (P = 0.038, Table 2).
Average cover (%) of the dominant plant species in the three studied sites during the springs 2019 and 2020.
General linear model analysis showing the interaction of sites, elevation, and aspects on the plant life form types (n = 36). Only significant effects are mentioned here.
The highest Ch_R and He_R covers were obtained in sites 1 and 2. For Ch_R, no significant difference between sites at both aspects was noted. In the northern aspect, He_R differed significantly between sites at low (P= 0.000) and medium elevations (P= 0.027). In the southern aspect, He_R varied significantly between sites only at low elevations (P= 0.036). High significant interaction between sites and elevations was noted for He_R in the northern aspect (P= 0.001, see Table 2).
Variations within sites
Figure 4 presents variation of Noy-Meir types covers. The highest covers were recorded in the northern aspects of site 1 (51.89% and 23.39%, respectively, for AA_NM and AP_NM) compared with those of the southern side (41.22% and 16.39%). No significant difference was noted for both AA_NM and AP_NM between elevations (P > 0.05), aspects (P > 0.05), and their interactions (P > 0.05).
In site 2, AA_NM covers are higher in the northern aspect (75.67%) compared with those of southern (43.56%). Inversely, AP_NM showed the highest covers in the southern aspects. AA_NM covers varied significantly between aspects in all elevations (P < 0.05). It appears that aspects influenced the AP_NM covers mainly in low elevation (P = 0.012). AA covers varied significantly according to elevations in the northern aspect (P = 0.027), where no significant interaction between aspects and elevations is noted for both AA_NM and AP_NM species.
In site 3, AA_NM covers are higher in the northern aspect with 45.78% compared with 31.83% in the southern one. The corresponding covers of AP_NM species are around 6.11% and 10.44%. The difference between aspects is highly significant (P= 0.007) at low elevation for AA_NM and at medium elevation for both AA_NM (P= 0.045) and AP_NM (P= 0.004). No elevation effects were noted for AA_NM and AP_NM in the northern aspect.
Variations between sites
The highest covers of AA_NM species are recorded under seasonal grazing (59.61%), followed by those of livestock exclusion (46.56%) and control (38.81%). In southern aspects, AP_NM covers varied significantly between sites at low elevation (P= 0.049). In northern aspects, high significant differences of AA_NM covers are noted between sites in low (P= 0.001) and medium elevations (P= 0.019). In this same aspect, the variation of AP_NM covers between sites is also significant at medium (P= 0.046) and high elevation (P= 0.03). A high interaction between sites and aspect is noted for both AA_NM and AP_NM species at low elevation (P < 0.01, see Table 2). Likely, the interaction between sites and elevation was significant for AP_NM in southern aspects (P= 0.041).
Variations within sites
Grime-type covers in the three studied sites are shown in Figure 5. The CRS_G species showed the lowest cover in the three sites. In site 1, CS_G, RS_G, and S_G species cover, respectively, were 28.17%, 11.39%, and 16.5% in the southern aspect and 26.61%, 16.22%, and 29.94% in the northern one. CS_G covers varied among elevations in the northern aspect (P= 0.002), but only RS_G varied between aspects at low elevation (P= 0.04).
In site 2, CS_G and S_G covers were higher in the northern aspect (61.94% and 13.72%, respectively) when compared with the southern one (41.44% and 9%). The effect of elevation was significant only for RS_G in the southern aspect (P= 0.012). A significant effect of aspects was noted for CS_G (P= 0.006) at low elevations and RS_G (P= 0.006) at high elevations. The interaction between aspects and elevations is only significant for RS_G (P= 0.012).
In site 3, the highest covers of CS_G, CRS_G, and S_G are recorded in the northern aspect (29.17%, 0.56%, and 19.94%, respectively). Their respective covers in the southern aspect are 20.0%, 0.22%, and 11.27%. Aspects were significant for CS_G (P = 0.029) and RS_G (P = 0.025) at low and medium elevations, respectively. In the southern aspect, the elevation was significant for CS_G (P = 0.022). No significant interaction between aspects and elevations was obtained.
Variations between sites
The covers of CS_G and CRS_G are lower under open grazing (24.58% and 0.39%, respectively) when compared with livestock exclusion (27.39% and 1.64%) and seasonal grazing (51.69% and 3.39%). The ANOVA showed high significant difference between sites for CS_G (P < 0.001) and S_G (P= 0.017) in low elevation and for CS_G (P= 0.016) in medium elevation in the northern aspect. In the southern one, a significant difference between sites is noted for CRS_G (P= 0.015), CS_G (P= 0.009), and RS_G (P= 0.024) in low, medium, and high elevations, respectively. According to the two-way ANOVA, the interaction between sites and elevations is highly significant for CS_G (P < 0.001, see Table 2) in the northern aspect and for RS_G (P= 0.004) in the southern one. A significant interaction between sites and aspects is noted for CS_G (P= 0.042) and RS_G (P= 0.021) at low elevation.
Plant life forms ordination
The nonmetric multidimensional scaling analysis combined the plant life forms and sites (Fig. 6). It shows difference in plant life forms within and between sites. Three main life forms group can be identified. The first group contains He_R, Ch_R, AA_NM, CRS_G, and CS_G. The second contains Th_R, AP_NM, and RS_G. The last one includes only S_G species. These groups hold life forms with same signification, so it seems that using one plant life form type from each group can explain the studied approach about the ecosystem resilience and plant adaptations.
The plant life forms were generally used as descriptive tools to characterize the vegetation communities' adaptation to different stress and disturbances. The main results showed that Ch_R, AA_NM, and CS_G are the most dominant types in the three sites, mainly in the seasonal grazing one (see Fig. 6). It appears that these plants are particularly sensible to both high disturbance and/or long protection. Pastoral plants in arid free-grazed rangelands presented lower covers compared with rested areas (Ouled Belgacem et al. 2019). Our results confirmed this and showed that under seasonal grazing, the Ch_R, AA_NM, and CS_G covers are higher compared with the control site. Similar results are obtained by Vecchio et al. (2019) showing an increase in plant species as a response to seasonal grazing after improvement of soil characteristics. In fact, seasonal grazing, including short grazing periods alternated by fencing during the growing season, modified the soil by increasing its organic matter content and improving its physical and chemical properties (Vecchio et al. 2018). We showed that He_R species (mainly S. tenacissima) varied strongly among the tested management regimes. Its specific cover was higher in the fenced site compared with the control due to its high ability to compete for water and nutrients in absence of disturbances. In fact, protection favors the development of more competitive and stress-tolerant species regardless of their palatability (Louhaichi et al. 2012; Tarhouni et al. 2017). The S. tenassicima cover is higher under seasonal grazing than control and livestock exclusion sites. Short fencing periods (provided by seasonal grazing) enhanced the fast-growing and competitive species (especially the highly preferred plants by animals) to develop, flower, and disperse their seed (Vecchio et al. 2019). Our results corroborate these findings and showed that seasonal grazing with controlled animal charge seems to be better for natural vegetation than long-term protection or continuous grazing. To conclude, improved species cover, in the Brighith site (seasonal grazing), concerns mainly He_R and CS_G species. These plant life forms are principally presented by S. tenacissima (highest cover) and some other palatable, less palatable, and unpalatable grasses. The better resilience ability noted in this site permitted both ecological and pastoral benefits.
Aspect and altitudinal effects on Ch_R, AA_NM, and CS_G are masked under livestock exclusion, despite the slightly higher covers at the low and/or medium elevations on the southern aspect. These results can be related, in part, to improvements in soil characteristics at these two elevations (probably better organic matter content in soil, which improves its water balance). In fact, at high elevations the soil surface is more exposed to rainfall, runoff, and erosion (by water and wind). Such situations can negatively affect the plant growth and cover. The aptitude of some perennials plants for the vegetative multiplication and their resistance to drought could be the origin of their high covers on the southern aspects characterized by strong sun radiation (Neffati 1984). This may indicate the higher capacity of mountain ecosystems to withstand drought and disturbance (Ben Salem et al. 2009). He_R species, mainly the S. tenacissima key species, showed the highest covers on southern aspects. The same results were stated by Gasque and Garcıa-Fayos (2004) showing that direct effect of sun exposure favors seedlings and regeneration of S. tenacissima due to its high ability to compete for water and nutrients. The high covers of protected annuals plants (Th_R, AP_NM, and RS_G) on the northern aspects compared with those of the south are mainly due to shadows and cool, humid winds from the north. In southern exposure, more opened to erosion and intensive wind and temperature, the evaporation increases and soil moisture decreases (Broza et al. 2004).
Grazing services provided by an ecosystem are not necessarily related to its biodiversity. In cases of overgrazed areas, biodiversity and species richness can be highly favored by unpalatable and/or grazing-resistant plants (Tarhouni et al. 2017; Msadek et al. 2021). In fact, biodiversity can be increased by the grazing-resistant species despite their lower grazing value. Hence, the rangeland will be highly diversified (high environmental value) without palatable species (low economical value). Despite their high palatability, annual species (AP_NM, RS_G, and Th_R types) are higher in seasonal and continuous grazing areas, confirming their grazing resistance and qualification as indicators of disturbance in arid rangelands (Tarhouni et al. 2017). However, these species cannot ensure rangeland sustainability due to their high dependence to climate conditions. Otherwise, we suggest that perennials grouped in AA_NM, CS_G, and He_R types can help to achieve biodiversity and safeguard sustainability of resistant grazing systems thanks to their adaptive capacities, despite their low palatability.
It seems that the combined effects of climatic drought and human disturbance are more obvious in the southern aspect where the covers of AA_NM, CS_G, CRS_G, and S_G are lower in the grazed sites (continuous and/or seasonal) compared with the northern one. Contrasting results are shown for AP_NM, RS_G, and Th_R, which are preferred first over perennials to be grazed by livestock (Ouled Belgacem et al. 2019). The groups obtained from the ordination of various plant life forms can facilitate the assessment of vegetation. In fact, only one classification can be used from each group to understand the plant adaptation under the studied conditions. For example, in this study, only Grime's classification can be considered since the third group contains only S_G species. The similarity in covers of plant species during the two studied seasons, mainly in the grazed sites (continuous and/or seasonal), can be explained by their regenerations after a succession of rainy years Ben Salem et al. (2009). showed a remarkable germination of plants species during the humid seasons. Our results prove that arid mountain ecosystems are endowed with an important capacity for resilience under good rainfall conditions, mainly with low disturbances. Further measurements during dry years (< 150 mm) are necessary in the future to study the impact of drought on the growth and resilience/regenerative capacity of plant species in these regions.
Understanding the ecological context and complexity of the species' adaptive strategies can provide new insights into the mechanisms governing plant responses to grazing. Our results suggest that seasonal grazing, with controlled animal charge, enhances the majority of plant species types compared with the control and livestock exclusion sites. This indicates a higher capacity for the montane ecosystems to withstand drought and reasonable disturbance. Seasonal grazing regimes should therefore be developed and generalized for the sustainable use of these natural rangelands. The adopted life-form types could also be used to predict the response of natural vegetation to climate changes and variations. We suggest the combination of the three types of classification in order to offer more information about multiple functional characteristics of inventoried species as a response to disturbance factors (climate and human activities). This can be valuable to understand vegetation physiognomy and resilience ability of the ecosystems and to assess the impact of the applied managements on arid ecosystems. Hybrid classification between Raunkiaer and Grime classifications can be also useful for such purpose.
This study was funded by the Arid Regions Institute of Medenine (IRA, Tunisia) with support from the International Center for Agricultural Research in the Dry Areas (ICARDA) and the Consultative Group on International Agricultural Research (CGIAR). The authors would like to thank anonymous reviewers for their valuable contributions to this manuscript. The opinions expressed in this work belong to the authors and do not necessarily reflect those of IRA, ICARDA or CGIAR.