Native Perennial Plants Colonizing Abandoned Arable Fields in a Desert Area: Population Structure and Community Assembly

: In recent years, the phenomenon of abandonment of arable ﬁelds has increased in Saudi Arabia due to low soil fertility, drought, low rainfall, high levels of evapotranspiration, soil salinization, and low level of groundwater. We evaluated the e ﬀ ect of agricultural land abandonment on soil properties, perennial vegetation composition, and population structure in the Al-Kharj region, Saudi Arabia. A total of 11 perennial plant species belonging to 9 families and 11 genera were detected in the di ﬀ erent abandoned ﬁelds of the study area. Four plant communities were identiﬁed after the application of the detrended correspondence analysis (DCA) ordination. The indicator species were (1) Seidlitzia rosmarinus — Zygophyllum hamiense , (2) Traganum nudatum — Seidlitzia rosmarinus , (3) Traganum nudatum — Prosopis farcta , and (4) Calligonum comosum—Pulicaria undulata. Results of the soil analysis showed signiﬁcant di ﬀ erences in soil texture, pH, salinity, and nutrient content among the four recognized plant communities. Demographic analysis indicated that populations of Zygophyllum hamiense and Calligonum comosum tended to be either inverse J-shaped or positively skewed which may have indicated rapidly-growing populations with high reproductive capacity. Conversely, the size–frequency distribution of Traganum nudatum , S. Rosmarinus, and Prosopis farcta was approximately symmetrical (i.e., bell-shaped). The present study sheds light on the necessity of managing abandoned agricultural ﬁelds for restoring and improving rangelands with native species that are adapted to the local conditions such as low water demand.


Introduction
The abandonment of agricultural fields is a common occurrence worldwide due to changes in economic circumstances, water resource shortage, and climate change [1][2][3]. Many non-crop plant species spontaneously colonize arable fields and compete with crops for nutrients, light, and space [4]. These species are considered as weeds from an agronomic perspective (segetals), but they can play a vital ecological role by contributing significantly to supporting biodiversity [4][5][6]. Farmland abandonment leads to a mixture of social, economic, and ecological effects, especially in arid areas [7][8][9]. Despite the negative impacts of land abandonment on the environment, there are also some positive consequences, including opportunities for revegetation [10,11].
Abandoned farmlands undergo natural succession or passive restoration [12] toward their original vegetation. Passive means of restoration is an important way to rehabilitate abandoned fields and improve biodiversity and ecosystem services. Several examples have shown that active restoration does not always surpass passive restoration [13]. However, passive restoration is a slow process. For example, it was found that some abandoned farmlands in the USA had only 75% of plant diversity relative to surrounding natural vegetation nearly a century after abandonment [14]. Managing for plant diversity by means of active restoration has not been widely practiced in restorations, as evidenced in the dominant use of monocultures in global revegetation efforts [15,16].
Changes in land use over a long time affect vegetation structure and habitat in many regions [17][18][19][20]. The abandoned farmlands are open to colonization by the previously-shrubby vegetation, which is often of high biological and environmental value [17,21,22]. The abandoned fields are considered as semi-natural habitats, where some plant species, such as grasses and woody species, are dependent on the land-use disturbance regimes [23,24]. Plant diversity can be more strongly affected by many factors such as the soil seed bank or the biogeographic species pool, land-use history, and type of management [23,25].
In the late-1970s, many fields in the Al-Kharj region, southeast of Riyadh, Saudi Arabia were cultivated with diverse agricultural crops, and thus the area was considered a first-class agricultural region. However, many of the arable fields were abandoned due to a shortage of water sources and government instructions to reduce cultivation in the area. Therefore, we monitored the changes in the plant communities 40 years after post-cultivation abandonment and addressed the following questions: (i) What are the main changes in the vegetation composition after abandonment? (ii) What is the status of the population structure of the dominant plant species?

Study Area
The study area is located in the Al-Kharj region, southeast of Riyadh, Saudi Arabia, between 24 • 00 N-47 • 00 E, 24 • 00 N-48 • 00 E and 24 • 50 N-47 • 00 E-24 • 50 N-48 • 00 E, and lying at about 400-500 m.a.s.l. (Figure 1). The study area is surrounded by rangelands that were usually grazed by domestic livestock. Many crops were cultivated in this area, such as wheat, alfalfa, and Rhodes grass, but these croplands were abandoned 40 years ago. The soil of the central region, in general, can be considered as "aridisols", with no accumulation of clays or organic matter [26]. On the contrary, the deposits in the wadies or valley basins are deep and fine-textured with a firm, flat vegetation-covered mud surface, supported by occasional rocky, sandy, or phytogenic mounds [27].
Summers are usually very hot and winters are relatively cold. The average annual temperature is 20.1 • C. January is the coldest month with an average low of 10 • C, while August is the hottest month with an average high of 45 • C. Rainfall is scanty and unpredictable. The average annual precipitation is 81 mm/year, mostly falling during the winter months [28].

Vegetation Sampling
A total of 20 sampled stands were selected in the abandoned fields from February to April, at least one km from each other ( Figure 1). Four quadrats (5 × 5 m) were placed in each stand, with the consideration given to representing a wide range of physiographic and environmental variation within each stand [29]. The vegetation sampling involved a listing of all plant species at the sample stands. The plant species density was estimated as the number of individuals of each species per unit area, and the plant cover of each species was estimated according to ordinal cover classification systems by Braun-Blanquet [29]. The importance value of each species was calculated as a summation of its relative density and relative cover. The plant species were identified according to Collenette [30] and Chaudhary [31].

Soil Analysis
Soil samples were collected from a depth of 0-50 cm at three random sites at each stand and mixed as a composite sample. The samples were air-dried and sieved through a 2 mm sieve. The organic matter was determined based on loss-on-ignition at 450 °C [32]. Soil texture was determined by the hydrometer method [32]. Soil water extract was prepared (1:5) using distilled water, and the pH and electrical conductivity (EC) were immediately measured. Soil CaCO3 was estimated according to Loeppert and Suarez [33]. Soil HCO3 -, Cl -, and SO4 2− were estimated by titration methods. The contents of K and Na were determined by flame photometry (Jenway PFP7 Flame Photometer, Chelmsford, Essex, England), while Ca +2 and Mg +2 were estimated using an atomic absorption spectrometer (A Perkin-Elmer, Model 2380, Waltham, MA, USA) according to Allen, et al. [34]. The contents of P and N were determined spectrophotometrically using a spectrophotometer (Thermo Scientific, model ICP MSEOS 6000 Series, Cambridge CB5 8BZ, United Kingdom). All procedures are outlined by Allen, et al. [34].

Vegetation Sampling
A total of 20 sampled stands were selected in the abandoned fields from February to April, at least one km from each other ( Figure 1). Four quadrats (5 × 5 m) were placed in each stand, with the consideration given to representing a wide range of physiographic and environmental variation within each stand [29]. The vegetation sampling involved a listing of all plant species at the sample stands. The plant species density was estimated as the number of individuals of each species per unit area, and the plant cover of each species was estimated according to ordinal cover classification systems by Braun-Blanquet [29]. The importance value of each species was calculated as a summation of its relative density and relative cover. The plant species were identified according to Collenette [30] and Chaudhary [31].

Soil Analysis
Soil samples were collected from a depth of 0-50 cm at three random sites at each stand and mixed as a composite sample. The samples were air-dried and sieved through a 2 mm sieve. The organic matter was determined based on loss-on-ignition at 450 • C [32]. Soil texture was determined by the hydrometer method [32]. Soil water extract was prepared (1:5) using distilled water, and the pH and electrical conductivity (EC) were immediately measured. Soil CaCO 3 was estimated according to Loeppert and Suarez [33]. Soil HCO 3 − , Cl − , and SO 4 2− were estimated by titration methods. The contents of K and Na were determined by flame photometry (Jenway PFP7 Flame Photometer, Chelmsford, Essex, England), while Ca +2 and Mg +2 were estimated using an atomic absorption spectrometer (A Perkin-Elmer, Model 2380, Waltham, MA, USA) according to Allen, et al. [34]. The contents of P and N were determined spectrophotometrically using a spectrophotometer (Thermo Scientific, model ICP MSEOS 6000 Series, Cambridge CB5 8BZ, United Kingdom). All procedures are outlined by Allen, et al. [34].

Vegetation Data Analysis
The data were arranged in an 80 quadrat × 11 species matrix, using cover values. The matrix was subjected to cluster analysis (two-way indicator species analysis, TWINSPAN) for determination of the vegetation groups. Detrended correspondence analysis (DCA) was applied to the matrix data set in order to obtain an efficient graphical representation of the ecological structure of the identified vegetation groups [35]. DCA was performed using the PC-ORD software program, version 5.5 [36]. The variation in soil variables in relation to the plant community was assessed via a one-way analysis of variance (ANOVA). To determine the relationships between the four determined plant communities and soil variables, Spearman rank correlation coefficient was used, based on density, cover, and richness. Both ANOVA and correlation were performed using the SPSS software program, version 24.0 (IBM Corp. Released 2016. IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY, USA: IBM Corp).

Species Size Structure Analysis
In the 20 sampled stands, the height (H) and mean crown diameter (CD) of each individual of the dominant plants (with high frequency and abundance) were measured within the studied quadrats. The diameter was measured for three crown diameters depending on the uniformity of the individual. The plant size index was calculated according to the modified form of Crisp and Lange [37] as (H + CD)/2. The size index value of plant per each species was used to determine the frequency distribution of the different age (size) cohorts. Size index values were used to classify the plant populations into ten size classes. The first and second classes were considered as seedlings and the third class was assigned to the juvenile stage. In order to evaluate the characteristics of each size class, the total density of the plant species was also calculated. The mean density, height, crown diameter, volume, size index, and the correlation between height and diameter of each population were also calculated.

Vegetation-Soil Relationship
Analysis of variance revealed that all the measured soil variables were significantly different among the four recognized communities ( Table 2)

Vegetation-Soil Relationship
Analysis of variance revealed that all the measured soil variables were significantly different among the four recognized communities ( Table 2)  Plant density and species richness were positively related to sand and organic matter content, while they were negatively correlated with pH, EC, and cation and anions concentrations (Table 3). On the other hand, the plant cover had a significant negative correlation to plant density, sand percentage, and organic matter content, while it showed a significant positive correlation with pH, EC and Cl − and SO 4 2− , cation and anion concentrations.

Demographic Analysis of the Dominant Plant Species
Demographic analysis indicated that the populations of Seidlitzia rosmarinus had a largest mean diameter (181.73 cm) and size index (128.93 cm), and the lowest height/diameter ratio (0.46 cm) ( Table 4). The population of Calligonum comosum had the highest average height (106.88 cm), height/diameter ratio (0.82) and volume (4.06 m 3 ). The population of Prosopis farcta had lowest mean values of most size-structure characteristics. The relationship between height and diameter for most species was significantly positive except for the population of Prosopis farcta, for which it was not significant.  The populations of Traganum nudatum, Seidlitzia rosmarinus, and Prosopis farcta had approximately symmetrical (i.e., bell shaped) size distribution across all sizes (Figures 3 and 4). Contrary to this, the size-frequency distribution of Zygophyllum hamiense and Calligonum comosum tended to be either inverse J-shaped or positively skewed which may suggest the presence of rapidly-growing populations with high reproductive capacity ( Figure 5).

Discussion
The dominance of the succulent shrubs of the Chenopodiaceae family in the study area reflects the nature of the arid salinized soil of the abandoned field habitats in the Arabian desert, as these species are morphologically and physiologically adapted to dry and saline habitats [38,39]. The absence of annual species in the surveyed vegetation could be due to low rainfall during the sampling season. These results are comparable with some previous studies [17,40]. Castellanos, et al. [40] observed the dominance of succulent shrubs in the abandoned fields of the Sonoran Desert of Mexico. Moreover, El-Sheikh [17] pointed out that in late succession stages, the abandoned fields in Egypt were dominated by shrubs and trees. However, Bonet [41] showed that annuals attained higher values of abundance in the abandoned fields in south-eastern Spain.

Discussion
The dominance of the succulent shrubs of the Chenopodiaceae family in the study area reflects the nature of the arid salinized soil of the abandoned field habitats in the Arabian desert, as these species are morphologically and physiologically adapted to dry and saline habitats [38,39]. The absence of annual species in the surveyed vegetation could be due to low rainfall during the sampling season. These results are comparable with some previous studies [17,40]. Castellanos, et al. [40] observed the dominance of succulent shrubs in the abandoned fields of the Sonoran Desert of Mexico. Moreover, El-Sheikh [17] pointed out that in late succession stages, the abandoned fields in Egypt were dominated by shrubs and trees. However, Bonet [41] showed that annuals attained higher values of abundance in the abandoned fields in south-eastern Spain.
The classification of vegetation stands resulted in the identification of four plant communities. The floristic composition of these communities is hamiense, which is comparable to that of plant communities found in the neighboring natural areas of Al-Kharj such as wadis (ephemeral watercourses), raudhas (meadows), and sand dunes [42,43]. However, xerophytic and succulent halophytes are much more represented in the study area [44]. This is due to the ingression of species from surrounding areas in former arable fields [17,45]. The separation of the four communities was good evidence that the first axis of the DCA diagram represented the succulent halophytes shrubs (VG-I, II).
On right side of the DCA, the community of Calligonum comosum-Pulicaria undulata was separated. This represented the old succession at abandoned sandy dry fields, where Calligonum comosum shrubs growing on dry sandy soils and sand dunes. This shrub is very resistant to adverse soil and moisture conditions. It produces root suckers and is easily propagated by cutting and layering. The soil of the succulent halophyte communities (VG-I: Seidlitzia rosmarinus-Zygophyllum hamiense) had high silt and clay content, high salinity and pH, and high SO 4 2− , chloride, and mineral concentrations.
These results are consistent with findings by Knops and Tilman [46] in the Minnesota sandplain, USA, Zeller, et al. [47] in the Passeier Valley in northern Italy, Dunjó, et al. [48] in the Serra de Rodes catchment, Spain, and Zhang, et al. [49] in the Qinling Mountains, China. Conversely, McLauchlan [50], Knops and Tilman [46], and Du, et al. [51] pointed out that after a long time of the field abandonment, total nitrogen and organic carbon increased and accumulated in the soil. This could explain the assemblage of VG III: Traganum nudatum-Prosopis farcta, where the roots of Prosopis farcta shrub have the ability to increase the fertility of the soil by fixation of nitrogen [52,53]. The relationship between plant density and total nitrogen and organic carbon content was negative as communities with low density have high values of plant cover due to the interspecific and intraspecific competition between species [54]. Likely, plant cover was positively correlated with the presence of soil minerals because most halophytic species such as Seidlitzia rosmarinus, Zygophyllum hamiense, and Traganum nudatum are well-adapted to salinized habitats [55].
The plant communities in the abandoned fields of the study area were characterized by a higher diameter than height. This is a strategy of many desert shrubs to reduce the high temperature [17,56,57]. The size structure of the Zygophyllum hamiense and Calligonum comosum populations was inverse J-shaped or positively skewed, which may represent rapidly-growing populations with high reproductive capacity. This may not be surprising, as small plants usually have higher reproduction rates and colonization ability. On the other hand, the populations of Seidlitzia rosmarinus, Prosopis farcta, and Traganum nudatum had approximately symmetrical (i.e., bell-shaped) size structure distribution, meaning that the percentage of small and old individuals was comparable with a high percentage of the medium-size individuals.

Conclusions
Agricultural land abandonment allows the regeneration of wild native perennial plants in the Al-Kharj region. Four plant communities were recognized in the study area, where soil salinity and nutrients were the main effective soil parameters determining species composition. Also, the population structure of the indicator species of these communities showed substantial differences. Prosopis farcta was the most dominant species after the abandonment of the arable lands. Therefore, we expect that this plant could colonize and dominate the abandoned fields in the future. The current study highlights the requirement for monitoring the abandoned agricultural fields and for restoring and improving rangelands with native species that are adapted to the local conditions such as low water demand. Therefore, governmental policy should be directed to take benefits and services from this habitat/ecosystem, using their resources without altering their natural characteristics.