Our large-scale investigation of the AMF communities of the plants and soils in the region of AlUla revealed several interesting findings discussed below: i) reports of AMF in the roots of multiple plant species known as non-mycorrhized, ii) low mycorrhization rates of major crops (date palms and Citrus spp.), iii) very low numbers of AMF spores in the soils, iv) absence of several AMF families (e.g. Gigasporaceae) usually commonly found in farms and natural ecosystems, v) the slight impact of land uses on AMF diversity and soil parameters, and vi) the importance of using several approaches (metabarcoding, histology, and spore counting) on a large scale to improve our knowledge of AMF found in hyper-arid environments. With an average pH of 8.86 ± 0.53, up to 10.40, the hyper-arid region of AlUla is also characterized by hyper-alkaline soils (Table 1). We highlight the small effect on AMF community composition of land uses compared to passive dispersal and soil physical and chemical parameters in this desert region. Finally, we discuss the importance of a vast sampling analysis dispositive with a variety of methods to study the AMF at a large scale. Arid and hyper-arid lands are sensible and rather unknown ecosystems, understanding further these ecosystems is crucial for both agriculture and land restoration in the context of climate change and increase in desertification.
3.1. Mycorrhizal status of hyper-arid ecosystems plants
In the present study we collected plant roots from 46 plant species (belonging to 25 families) and reported their mycorrhizal status, with mutualistic symbiosis with AMF found in at least one individual plant in 29 species (63%). We queried the scientific literature to determine if these results were in line with those of other previous studies (Table S1). Three main findings, as discussed below, emerged from this comparison: i) we filled a knowledge gap by clarifying the previously unknown mycorrhizal status of some plant species, ii) we challenged this status when the plant was previously described as being non-mycorrhizal, and iii) we did not observe AMF in some plant species previously described as being compatible with arbuscular mycorrhization.
To the best of our knowledge, this is one of the first studies to report the presence of arbuscular mycorrhizae in the roots of the 10 following plant species: Astragalus sieberi (Fabaceae); Astragalus spinosus (Fabaceae); Convolvulus spicatus (Convolvulaceae); Ficus cordata (Moraceae); Ficus palmata (Moraceae); Forsskaoleae tenacissima (Urticaceae); Ochradenus baccatus (Resedaceae); Pergularia tomentosa (Asclepiadaceae); Picris cyanocarpa (Asteraceae) and Searsia tripartita (Anacardiaceae). Some species of the Astragalus, Convolvulus, Ficus, Pergularia, Picris and Searsia genera are already known to be compatible with AMF (Table S1). We identified arbuscular in eight plant species from these genera, potentially filling a knowledge gap regarding their unknown mycorrhizal status. Moreover, this is one of the first reports of mycorrhization detection in Forsskaolea and Ochradenus plant species. We have also reported the presence of AMF in Chenopodiastrum (ex: Chenopodium) and Rumex genera, which are known for their variable mycorrhizal status (Fontenla et al. 1998; Karanika et al. 2008; Daisog et al. 2012; Safari Sinegani and Elyasi Yeganeh 2017; Kellogg et al. 2021). Some plant families are commonly known to be mycorrhizal, e.g. Asteraceae, Fabaceae, Poaceae and Solanaceae (Wang and Qiu 2006). All of the Asteraceae and Fabaceae plant species investigated here were mycorrhized, thereby confirming previous findings (Table S1). However, our data regarding Solanaceae and Poaceae species revealed poor mycorrhization rates (1/3 species mycorrhized, 1/8 individuals and 2/4 species, 2/5 individuals, respectively), which was much lower compared to the findings of other studies (Table S1). Amaranthaceae plant species are known to be compatible with AMF (Chaudhry et al. 2005; Wang and Qiu 2006; Santhoshkumar, et al. 2019), yet their mycorrhization rates are considered to be low to null (Tester et al. 1987). Here we reported much higher rates, with evidence of AMF root colonization in 3/7 Amaranthaceae species and in 12/39 individuals.
These differences in plant mycorrhizal status when comparing our data with previous reports could be explained by multiple factors, such as: i) driven by environmental conditions, ii) outdated, incomplete data, or iii) simply a lack of knowledge and investigations. Non-mycorrhizal plants are more common and diverse in arid lands (Soudzilovskaia et al. 2017; Brundrett and Tedersoo 2018). However, we could expect that plants growing in extreme desert conditions (e.g. extreme drought, low soil nutrients, discontinuous vegetation cover, high heat and soil alkalinity in the AlUla region) could greatly benefit from this mycorrhizal interaction feature which can help plants mobilize water and nutrients and survive heat stress (Duc et al. 2018; Harkousse et al. 2021; Madouh and Quoreshi 2023). Another example of such plant adaptations in extreme environments was reported in the highly polymetallic soils in New Caledonia, where Cyperaceae plants (known as non-mycorrhizal) were found to unexpectedly develop mycorrhization (Lagrange et al. 2011), which is in line with our observations in AlUla, where AMF was detected in Cyperus conglomeratus. Primary descriptions of the plant mycorrhizal status often come from temperate climatic conditions and unconstrained environments, while our results, along with those of other studies (O’Connor et al. 2001; Agwa and Al-Sodany 2003; Chaudhry et al. 2005; Lagrange 2009), were obtained in arid environments under harsh conditions (aridity, nutrient shortage, extreme pH, etc.). Environmental conditions, e.g. biotic stress, climate and soil nutrient content, are important factors underlying the establishment of AMF mutualistic symbiosis (Muthukumar et al. 2004; Lagrange 2009), plant taxonomy alone may not be discriminant enough to know the true mycorrhization potential of a plant species. For example, salinity, heat and drought can affect plant mycorrhization (Diatta et al. 2014; Duc et al. 2018). Land use, pH and geological parent material may also explain the mycorrhizal distribution in plants (Menzel et al. 2016).
Another interesting point concerns the energy cost of plant-AMF symbiosis in arid environments, where more energy is required to develop a functional mutualistic relationship (Koltai and Kapulnik 2010). Plants must feed their AMF symbionts with sugars produced by photosynthesis (which involves atmospheric carbon dioxide and water consumption). Overuse of this trade-off may be deleterious to plants in arid environments where water is already a scarce resource (Chaves et al. 2002). To conclude this section, our study challenged current views on the mycorrhizal status of desert plants, as we noted that the hyperarid and hyperalkaline environmental conditions in AlUla may be conducive to the development of plant-AMF mutualistic symbiosis.
3.2. Unexpected low AMF levels in crops
In this study, we observed mycorrhizal colonization of two major crops, i.e. date palm (P. dactylifera; Arecaceae), and Citrus spp. (Rutaceae). Date palm is: i) the main crop in the AlUla region, ii) adapted to desert conditions (e.g. drought, pH, heat and salinity), and iii) characterized by high mycorrhizal rates in other locations (Ramoliya and Pandey 2003; Abohatem et al. 2011; Bouamri et al. 2014; Arab et al. 2016; Hazzouri et al. 2020). Citrus spp. plants are often jointly cropped with date palms in AlUla orchards (Gros-Balthazard et al. 2023), and are highly dependent on mycorrhizal colonization (Wu and Xia 2006; Song et al. 2015). In our study, no AMF structures were found in any of the collected roots of 13 individual Citrus plants. Regarding date palm, only 23% of the 135 individuals were colonized, and none of the 8 individuals collected in natural ecosystems displayed any root AMF. In comparison, Bouamri et al. (2014) found mycorrhized roots in all of the 10 date palms sampled in farms at Tafilalet (Morocco). Environmental DNA revealed high variability in AMF proportions across all date palm samples (ranging from 0 to 95% of the total reads issued from 18S sequencing). AMF reads were detected inside and on the surface of date palms roots, except in two individuals growing in natural conditions. These AMF mainly concerned the Glomeraceae family, i.e. mostly Glomus, Rhizophagus and Sclerocystis genera.
Drought stress alleviation is one of the main potential roles of AMF to benefit plant crops. Many studies have obtained evidence on the effectiveness of AMF in enhancing date palm growth and improving the tolerance of these trees to abiotic and biotic stress in date palm farms (Jaiti et al. 2007; Evelin et al. 2009; Abohatem et al. 2011; Al-Karaki 2013; Ait-El-Mokhtar et al. 2019). Water management is crucial in these desert regions, yet water supplies have been jeopardized. Over time, the use of non-renewable water resources from aquifers gradually increased the soil salinity, so farmers had to find alternatives, such as sea water desalinization (Shahin and Salem 2014; Zemni et al. 2022; Seraphin et al. 2022; Odnoletkova and Patzek 2023).
Arbuscular mycorrhizae in cultivated date palms was found to be reduced when water access is scarce (Baslam et al. 2014) which could explain the absence of mycorrhizae in natural conditions. However, date palm mycorrhization is generally low in the region, even on intensely irrigated farms. This observation is specific to date palms, because the mycorrhization rate was checked on plant species growing in the vicinity of date palms (i.e. Chenopodiastrum murale, Sonchus oleraceus and Plantago lanceolata). These adventitious plants were found to have an average of 71% arbuscular mycorrhizal structures, with a mean frequency of 81%, thereby indicating that the water scarcity and environmental conditions did not hamper the mycorrhization of these species. The absence of mycorrhizae noted in Citrus spp. trees and the low AMF colonization rate of date palms, i.e. two crops that are usually colonized by AMF, could have been the result of the extreme soil conditions. Indeed, even though water stress may be alleviated by irrigation, the soils remain nutrient-poor and alkaline, which could specifically affect interactions between AMF and these plants, while having less of an impact on other species (Oehl et al. 2017). However, poor soils (mainly lacking phosphorus) are commonly reported to favor plant-AMF interactions (George et al. 1995; Ma et al. 2023). Regarding water scarcity, it is unclear if this factor enhances or reduces AMF colonization, and this may vary according to the plant species, fungal species, environmental conditions, and water stress intensity (López-Ráez 2016).
3.3. Environmental selection of spores and alternative AMF dispersal strategies in arid environments
AlUla soils displayed an extremely low quantity of spores, regardless of the sampling location and conditions, with only 58 spores/kg of soil detected. Similar studies on arid or semi-arid desert soils (rocky to sandy soils) in natural ecosystems or farms, reported much higher spore numbers in soils, ranging from 200 spores/kg of soil in a semi-arid ecosystem (Requena et al. 1996) to up to 18,400 spores/kg of soils under date palms in farms (Bouamri et al. 2006). Despite this considerable variability in spore quantities in desert soils, to our knowledge this is the first time that such a low number of AMF spores has been reported in these soils.
The low quantity of spores in AlUla soils could be attributed to two non-mutually exclusive processes: i) high spore degradation in these soils due to environmental factors (e.g. heat, UV radiation, sand abrasion), or ii) the natural selection in the region in favour of AMF species producing low spore numbers. AMF fitness may be linked to the spore morphology and characteristics. The first hypothesis that could be put forward is that hyperarid environments are naturally conducive to the selection of small thick-walled spores (e.g. Archaeospora, Rhizophagus), while other spores that are unfit for survival in desert ecosystems are quickly disappearing and/or just absent. Requena et al. (1996), for instance, reported that many spores they observed in a Spanish semi-arid region were degenerated. Regarding AMF natural selection, AMF spores and cells harbor many nuclei (Hijri and Sanders 2005; Kokkoris et al. 2020) that may be genetically distinct and could be selected and transferred (or not) to the offspring enabling them to rapidly select traits like production of spore (Verbruggen and Kiers 2010).
Despite the low spore abundance in AlUla soils we monitored, our results showed that plants were still able to establish plant-AMF interactions (especially in herbaceous species). In Namibia, the spore content in soils could not be correlated with the arbuscular colonization of plants (Uhlmann et al. 2004), and it is unlikely that spores are important for inducing AMF colonization of plant roots (López-Sánchez and Honrubia 1992). These considerations suggest that spores may not be the main AMF propagation pathway in hyperarid regions. There are other means of AMF survival and propagation, such as via hyphae or root fragments colonized by AMF, through the so-called propagules (Salomon et al. 2022). In our case, these propagules might be the most probable way of AMF dispersion, with colonized root fragments offering shelter and nutrients for the fungi (Müller et al. 2017), whereas the low extent of plant cover on the soils may indicate that there was a low mycelium hyphae density in the soils.
3.4. Predominance of Glomeraceae and absence of Gigasporaceae in the hyperarid AlUla region
Overall, AMF detected in AlUla soils and plants showed high dominance of Glomeraceae species, mostly with 18S, in addition to Diversisporaceae and Claroideoglomeraceae, which were observed throughout the ITS2 datasets (Fig. 3; 4). Datasets sequenced with 18S markers were almost exclusively composed of Glomeraceae taxa, with the main genera being Glomus and Rhizophagus, in addition to Sclerocystis in date palm roots. Glomeraceae, mostly Glomus, is often predominant in soils exposed to abiotic stress as they tend to be opportunistic while bearing features that are advantageous in adverse environments (e.g. ability to rapidly colonize plant roots). This predominance of Glomeraceae has been reported in numerous studies in different arid regions worldwide using 18S sequencing and spore identification (Symanczik et al. 2014; Qiang et al. 2019; Harrower and Gilbert 2021; Adenan et al. 2021; Zhao et al. 2022). Diversisporaceae are also regularly identified in deserts (Qiang et al. 2019; Vasar et al. 2021; Harrower and Gilbert 2021; Zhao et al. 2022). Other AMF families detected in these arid environments include Paraglomeraceae (Zhao et al. 2022), Claroideoglomeraceae (Symanczik et al. 2014; Zhao et al. 2022), Acaulosporaceae (Zhao et al. 2022), and Scutellosporaceae (Gigasporales) (Harrower and Gilbert 2021).
AMF diversity in the date palm rhizosphere has until now mostly been described on the basis of spore morphotypes. Our study is one of the first in which AMF communities in date palm roots were investigated using a metabarcoding method based on the 18S. Several studies on the date palm rhizosphere reported a dominance of Glomeraceae taxa, mostly Glomus spores along with recurrent Gigasporales (Gigasporaceae, Scutellosporaceae and Racocetraceae) (Al-Yahya’ei et al. 2011; Khirani et al. 2020; Dalal and Solanki 2021). The dominance of Glomeraceae, especially Glomus detected via both 18S sequencing and spore observation, is consistent with our results. However, authors of other studies have often reported detecting spores from the Gigasporales order in date palm orchards, which is inconsistent with our metabarcoding results (only two OTUs on soils of farms in the Grid dataset, representing 0.015% of the AMF reads; and in date palm roots in the Palm dataset, representing 0.8% of the AMF reads), and is also out of line with our soil spore extractions which mostly revealed small sized spores. Gigasporales spore sizes reportedly range from 120 to 640 µm (Souza 2015) yet we extracted spores between sieves with meshes of 40 µm and 1 mm. Our spore extraction method is thus suitable for trapping Gigasporales spores. In cases where no spores of this order were found, it could be hypothesized that this order was either absent, or highly preyed upon by other organisms. This scant presence of Gigasporales taxa in dryland habitats has already been described (Adenan et al. 2021), with only 1 OTU found to belong to Scutellospora. Gigasporales were reported to be more preyed on and degraded in dunes than in temperate environments (Lee and Koske 1994). Taxa of other orders such as Archaeosporales, Entrophosporales and Paraglomerales were also seldom found in AlUla, in line with the findings of other previous studies conducted in desert environments (Adenan et al. 2021).
AMF identification via metabarcoding also depends on the database used for the taxonomic assignment (SILVA for 18S and UNITE for ITS2). This difference of database may play a role in the differences in community composition observed between markers. Datasets sequenced with ITS2 showed more Unidentified Glomeromycotina than those sequenced with 18S, but in both datasets those sequenced mostly belonged to the Glomerales and Diversisporales orders. The high proportion of unidentified AMF obtained reflected an overall knowledge gap that exists with regard to desert ecosystems, particularly in reference databases. Studies like ours, involving high sequencing depth and the combined use of ITS2 and 18S, could enhance the overall understanding through the detection of a broader range of AMF taxa.
3.5. Soil parameters and location rather than land use shape AMF communities in AlUla
Despite the fact that there were no significant differences in soil characteristics, shifts in AMF communities were identified between land uses. Although the alpha-diversity (Hill numbers) was mostly similar under all conditions, the beta diversity (PERMANOVA on Bray-Curtis distances) differed between agrosystems, natural ecosystems and urban areas. These results are consistent with those reported by Yu et al. (2022), where the microbial composition was more sensible than the diversity index. Our R² PERMANOVA values explained < 0.01% of the community differences, but even the slightest difference can be significant when the sample number is sufficient. Overall, our results suggest that the fungal community composition obtained for abandoned agrosystems was an intermediate pattern between the active agrosystem and the natural ecosystem land uses, thereby suggesting that agriculture had a legacy effect on the microbial communities, as described in (Maurice et al. 2024a). Moreover, our definition of abandoned agrosystems included farms that had been abandoned recently or longer ago, while having different extents of past anthropic pressure. Compared to natural ecosystems, these past anthropic disturbances potentially modified the soils: i) in terms of hygrometry, ii) lowering the pH, and iii) fertilizer and pesticide use, patterns also described by Tchabi et al. (2008). In the light of all of these elements, we expected that farming and past anthropic disturbances would lead to marked disparities in AMF communities between the four land uses. The small differences in AMF communities between land use is inconsistent with the findings of other studies where AMF were highly affected by land uses and agricultural practices, mainly tillage (Thiele-Bruhn et al. 2012). However, tillage and other mechanical treatments are not required in date palm cultivation, so there should be very little alteration in the soil structure. There might be a link between the lack of differences in soil parameters between land uses (Fig. S5) and the small differences between land uses for AMF community composition. This lack of a marked effect of land uses on AMF communities has also been reported by Jansa et al. (2014).
The main drivers of AMF community structure in AlUla were generally the longitudinal sample distribution, along with the proportion of silicon detected as compared to other measured elements and pH, i.e. mostly pHKCl. The high effect of the geographical distribution on the map i.e longitude, latitude (Fig. 5) unveils that soils collected at close distances tend to show more similar AMF community. Microbial dispersal is considered to be mostly passive, which implies that geographically close points would tend to be affected by the surrounding microbial communities due to the limited dispersal. In our results, the extent of the distance effect on the AMF community structure could thus be an indicator of the importance of dispersal limitation (Zhou and Ning 2017).
The soil pH is a very common AMF community structuring driver (Tedersoo et al. 2014, 2022; Davison et al. 2015, 2021). The high Si concentration, as opposed to that of other elements, is an indication of the sandy texture of soils. Compared to other soil types, sandy soils are characterized by high porosity, warmer temperatures, high dryness and low nutrient contents (De Deyn et al. 2008; Laity 2009). In the study of Al-Ghamdi and Jais (2013), the AMF colonization rate increased along with the soil sand content. The effect of the soil texture, notably the sand content, is considered to be an important factor governing the AMF community composition (de Oliveira Freitas et al. 2014). Overall, the desert microbial community composition differs when comparing biomes in temperate or tropical regions, and the communities display a lower phylogenetic diversity (Murgia et al. 2019). Davison et al. (2015) highlighted low AMF richness in semi-desert areas in comparison with other environments, such as in forests, grasslands and shrublands. Other key parameters are known to shape AMF communities, including temperature, precipitation and organic carbon levels (Davison et al. 2015). There were no noteworthy temperature and precipitation differences in our study because all samples were collected in the same region (AlUla). However, the temperature was often lower and the humidity higher in active farms because of irrigation and the date palm vegetation cover (de Grenade 2013). Organic carbon was very low in all of the analyzed soils sampled in farms and natural ecosystems (data not shown). In these environments, however, changes at the microenvironment scale can have a marked impact on plant and microbial communities by creating fertility islands, which could partly explain the variability in composition and diversity of AMF communities that we noted at the regional scale (Maurice et al. 2023).
3.6 Combined use of AMF metabarcoding, root histology and spore extraction in a desert region with diverse landscapes and land uses
Based on molecular data and direct observations, we conducted a large-scale study to characterize AMF in the AlUla region (Saudi Arabia) and took into account how the different land uses (natural ecosystems, active/abandoned agrosystems, and urban areas) and soil parameters shaped the fungal microbiome and environment. The combined use of different methods of analysis allowed us to describe directly and indirectly the arbuscular mycorrhization in both soils and roots in a variety of conditions and plant species. Note that in our study—spanning an arid 5,000 km² area—we used a total of 7,620,047 AMF reads, characterized mycorrhization in 46 different plant species with 277 root samples, while extracting spores from 12 soils. By comparison, other large-scale studies that have been carried out to describe microbial communities often involved distant sampling areas, while focusing on specific environments and extrapolating findings to infer intermediary conditions (O’Connor et al. 2001; Davison et al. 2015; Tedersoo et al. 2022). Multiple methods of analysis mixing high throughput analysis and plant and soil characterization Other large-scale kilometric grid sampling has also been carried out, e.g. in the studies of Karimi et al. (2019) and Djemiel et al. (2023), who mapped bacteria and fungi throughout France, with samples collected 16 km apart. Menzel et al. (2016) sampled AMF across Germany based on 130 km² sampling cells. Our kilometric grid had a more accurate 1 km² mesh, covering an area of 1,039 km². Although challenging, this region-wide sampling effort generated useful information regarding AMF variations at regional and local scales. We highly recommend using this method in future studies in arid deserts to gain greater insight into these specific ecosystems and to gather data that could be of use in further research carried out the current climate change setting.