Northern and Southern Forests of Socorro Island Harbor Different Communities of Arbuscular Mycorrhizal Fungi

Abstract

Arbuscular-mycorrhizal (AM) fungi affect individual plants and the diversity of plant communities. AM interactions are sensitive to the physical properties and nutrient availability of soil. In contrast to the comprehensive knowledge of AM fungi's diversity in many mainland ecosystems, there is a considerable gap in the understanding of the communities of AM fungi on oceanic islands. We surveyed the community of AM fungi in the wet forests of Socorro Island (above an elevation of 600 m). Because of the influence of northerly winds, the island's windward side is presumably more humid than the leeward side, although, above 600 m elevation, fog is common on both sides of the island. Forty-six percent of AM fungi species occurred on only one side of the island. Also, four species common to both sides showed marked differences in abundance, and the abundances of the AM fungi correlated with the abundance of plant life-forms. Beta diversity was similar in the southern and northern forests. However, the contribution of nestedness to beta diversity was three-fold greater in the southern forests than in the northern forests. Root colonization by AM fungi was higher in the northern forests, where phosphorous content was also high. The prevalence of nestedness in beta diversity and the lower mycorrhizal colonization of roots in the southern forests could reflect known perturbations in the southern forests.

Oceanic islands harbor simple ecosystems compared to mainland ecosystems, but they are the cradle and museums of unique evolutionary processes and the home to many endemic species (Keast and Miller 1996, Gillespie and Roderick 2002). Elevational and environmental gradients are typical in oceanic islands, and many species in oceanic islands occupy broader environmental niches than in the mainlands (Carlquist 1965, Shimizu and Tabata 1991). Because of that, some biotic interactions in volcanic islands are unparalleled in the mainlands (Losos and Ricklefs 2009). Northeastern slopes of oceanic islands in the Pacific Ocean are exposed to the predominant trade winds with particularly strong wind-stress during the winter and spring in the northern hemisphere (Wyrtki and Meyers 1976). Such wind exposure brings more humidity and lowers the temperature of the islands' windward sides, which may affect not only the species' physiological capabilities but also their biotic interactions. In this study, we evaluated the diversity of mycorrhizal fungi in the forests of Socorro Island. We discussed the results in relation to prevailing environmental conditions, nutrient availability, and the potential impacts of introduced sheep in the southern part of the island for over 150 years.

Mycorrhizal interactions are essential in the colonization of new spaces, such as those devastated by volcanic eruptions (Obase et al. 2008) and emerged volcanic cones in the oceans. Glomeromycota species are mutualists of up to 80% of the land plants in the world (Brundrett 2002). The characteristic haustoria in root cortex cells define these interactions as arbuscular mycorrhizal interactions. These fungi are metabolically active when in symbiosis within the roots, but most species endure in the soil as spores (Gazey et al. 1993, Ruiz-Lozano and Azcón 1996). In symbiosis with the roots, mycorrhizal fungi display two mycelial phases. In the plant's roots, hyphae occupy the intercellular spaces in the cortex and penetrate the cells with specialized haustoria (arbuscules, coils, and other forms). The second mycelial phase is extra-radical. The mycelium extends through the soil and transports absorbed soil minerals to the roots. The fungi gain carbon compounds while depositing mineral nutrients though a metabolic nutrient exchange between root cells and haustoria (Bago 2003, Govindarajulu et al. 2005, Jin et al. 2005). Interactions between plants and Glomeromycota-like fungi date back to the time when plants colonized land environments in the Devonian Period (Feijen et al. 2018), and current functionality suggests that these interactions played a central role in early plant diversification (Brundrett and Tedersoo 2018). Mycorrhizal fungi improve plants' nutrition status, survival, and growth (Caravaca et al. 2003, Piñeiro et al. 2013, Manaut et al. 2015). Further, these interactions' relative importance increased in sites with low availability of soil mineral nutrients (especially phosphorus and nitrogen) for plant growth (Treseder 2004). In addition, through dynamic feedbacks, mycorrhizal fungi can drive plant communitiy composition and diversity (Bever 2002).

The number of species in a community (species richness or alpha and gamma diversity depending on the relative spatial scale) is a fundamental estimate of diversity (Gotelli and Colwell 2001). Moreover, changes in species composition and their abundances from one site to another site (beta diversity) is an informative metric of the drivers of community structure, mainly when partitioned into nestedness and species turnover contributions to beta diversity (Mori et al. 2018). Several factors are known to affect the beta diversity of communities at different spatial and temporal scales. Long acknowledged, dispersal limitations mediated by distance and environmental and productivity gradients are significant determinants of beta diversity (Ricklefs 2004, Qian and Ricklefs 2007, Chase 2010, Kraft et al. 2011, Chan et al. 2016, Polato et al. 2018).

Theoretical models predict that when dispersal limitations and extinction rates are higher than colonization rates, species turnover will be a significant component of beta diversity, that is, different species' pools from place to place. In contrast, when dispersion rates predominate, nestedness is the main component of beta diversity, as pools of low alpha diversity sites are subsets of larger species pools (Lu et al. 2019). Also, priority effects and dominance of the space by different, relatively equal competitors may result in the increased importance of species turnover to beta diversity (Kennedy et al. 2009, Viana et al. 2015, Zemunik et al. 2016) because of stochastic shifts in competition capabilities among the species across space. Further, natural (e.g., recurring wildfires) and human-driven habitat disturbances (e.g., extended habitat reduction, fragmentation, and land-use change) are recognized drivers of biotic homogenization (e.g., Olden 2006, Socolar et al. 2016, Soininen et al. 2017). Because of this, beta diversity, especially the turnover of species, is expected to decrease with a consequent increase in the contribution of nestedness to beta diversity (Coelho et al. 2018).

In this study, we surveyed the community of arbuscular mycorrhizal fungi in the forests of Socorro Island. Northeastern forests in Socorro Island are under the direct influence of the trade winds (SEMAR 2020), while the southwestern forests are on the island's leeward side. Introduced sheep have grazed and trampled on the southern side of the island for over 150 years, affecting large areas of natural vegetation (Ortíz-Alcaraz et al. 2016). In this scenario, we aimed to address the following questions. Do the southwestern and northeastern forests in Socorro Island harbor similar communities of arbuscular mycorrhizal fungi? Is there a correlation between mycorrhizal fungi's abundance and the structure of the vegetation and soil chemistry? Do the southern and northern forests maintain a similar level of beta diversity? Are nestedness and species turnover contributions to beta diversity of mycorrhizal fungi equal in the northern and southern forests? Is root colonization by mycorrhizal fungi equal in the southern and northern forests?

Materials and methods

Socorro Island

Socorro Island (18.790625° N, −110.975620°W) is the largest of the four islands in the Revillagigedo Archipelago. The island is located 460 km south of Los Cabos in the Baja California Peninsula and 700 km from the port of Manzanillo in western Mexico (Figure 1). The volcanic cone extends up to 1,030 m in elevation and covers an area of 132 km². The predominant winds are from the northeast, and that side of the island is more humid than the leeward side (Miranda 1960, León de la Luz et al. 1996). The island's southern side is dryer, has deeper soils, and has flatter terrain with extensive woodlands. On the northern side, forests exist along narrow ravines (Walter and Levin 2008) protected from the hurricane winds that predominantly come from the southeast and south (Romero-Vadillo et al. 2007).

Around 1869, early visitors introduced sheep into the southern end of the island, and the animals became feral. Sheep wandered through the island's central-southeastern side and were common up to elevations of 600–800 m, where forests are the predominant vegetation cover. The dense scrub in the lowlands, in addition to the unstable rocky lava deposits and the lack of freshwater springs, may explain why sheep did not prosper in other parts of the island (Walter and Levin 2008, Ortíz-Alcaraz et al. 2016).

Study Sites

In 2008, in the southern forests of Socorro Island, we selected four sites with relatively easy access, and on the north side of the island, we selected three sites. The sites were no less than 250 m away from each other. All sites were in the range of 700 to 850 m elevation (Figure 1). The most common tree species in all sites were Guettarda insularis (Rubiaceae), Ilex socorroensi (Aquifoliaceae), Sideroxylon socorrense (Sapotaceae), and Oreopanax xalapensis (Araliaceae). We permanently marked ten transects (50 × 2 m) spaced by 15 m in each of seven sites on the island.

Spores of AM Fungi in Soil

At the starting and final positions of each transect, we collected a soil sample of approximately 100 g. All samples within a forest site (ten transects) were combined to obtain a final composite soil sample. Soil samples were stored at 4 °C and transported to the laboratory. Once the soil was dry, we took a 100 g sample from each site and processed them by wet sieving (500, 200, 125, 75, and 38 mm sieves). All material recovered in the two finest sieves was centrifugated in a sucrose density gradient following Gerdemann and Nicolson (1963). We then collected the uppermost layer and rinsed it over a sieve of 38 mm, and the sample containing the spores was transferred to 5 ml vials. All the spores contained in the sample were separated and counted under a dissecting microscope (Nikon SMZ800). Spores were mounted on glass slides in polyvinyl lacto-glycerol (PVLG) with and without Melzer's reagent and observed under an optical microscope Nikon Eclipse-E600. All spores were classified by morphological characteristics such as dimension, color, shape, presence of hyphae of support, and the number of wall layers (Oehl et al. 2011, West Virginia University's INVAM 2012).

Figure 1.

Location of Socorro Island in the Revillagigedo Archipelago in the Pacific Ocean west of Mexico (open circle). Details shown are the top of the Evermann volcano (closed triangle), 100 m elevation belts, and the location of the study sites in southern (circles) and northern (squares) forests.

Plant Life-Forms and Soil Chemistry

We recorded all plants along each transect and classified them by life form: trees, shrubs, lianas, herbs, and ferns.

We used a TruSpec Micro automatic analyzer to quantify the carbon and nitrogen content in the soil samples. For phosphorous content, we followed the Bray and Kurtz (1945) procedure. Briefly, we extracted one gram of soil in 10 ml of a solution composed of 0.025 N HCl and 0.03 N NH₄F for ten minutes and then filtrated. We added molybdate ascorbic acid reagent (6.6 mM molybdate ascorbic acid, 0.6 mM antimony potassium tartrate, and 1.8 N sulfuric acid) and read it in a spectrophotometer (wavelength of 882 nm).

For each soil sample for the carbon mineralization rate, we used 100 g of dried soil and a trap of 10 ml 1 N KOH. Soil samples were irrigated and taken to field capacity. The soil sample and the trap were then placed into a 1 l hermetic jar and incubated for 10 days at 26 °C. The CO₂ trap was replaced every day with a freshly prepared KOH solution. Every day when the CO₂ trap was removed from the jar, we immediately stopped the reaction in the trap by adding 5 ml of 1.5 M BaCl₂. Then, the samples were titrated with 1 M HCl.

For the nitrogen mineralization rate, we worked out the differences in NH₄ and NO₃ between the freshly collected samples and the samples after a 25-day incubation at 25 °C. We used 100 g samples, and for incubation, we added enough water to take them to field capacity. We used 2 N KCl₂ to extract NH₄ and NO₃ (Etchevers et al. 1971) and used the method of salicylic acid nitration (Robarge et al. 1983) and that of Nessler (Jackson 1964) to quantify NH₄ and NO₃, respectively.

Mycorrhizal Colonization

Guettarda insularis is common and abundant in all the studied sites and was selected to investigate potential differences in mycorrhizal colonization between Socorro Island's southern and northern forests. We randomly selected a Guettarda tree among those rooted within each transect and extracted fine roots. We collected first- and second-order roots (fine-roots) at a depth ranging between 5 and 10 cm. We fixed all fine roots in 5% acetic acid. For clearing and staining the roots, we followed Phillips and Hayman (1970), though, we made some modifications because of the very dark pigmentation of the roots of G. insularis. We used a 10% (w/v) KOH solution to clear the roots overnight, and we then heated (90 °C) the roots in the KOH solution for 15 to 20 minutes using a water bath. We rinsed roots in tap water, and after removing the excess water, we immersed them in a 10% HCL (w/v) solution for 10 minutes. Finally, we stained the roots with trypan blue (0.05% weight/volume in glycerol) at 100 °C for 5 minutes in a fume hood. Following the hairpin intersection method (McGonigle et al. 1990), we estimated the percentage of root length colonized by hyphae, vesicles, spores, and arbuscules. We also estimated the average length of hyphae of the mycorrhizal fungi in the root cortex (mm/mm), following Vega-Frutis and Guevara (2009).

Data Analysis

We used Whitaker curves to explore changes in the abundance of the species in the communities (Whittaker 1965) and a Monte Carlo randomization to test for species with significant changes in abundance (Guevara et al. in press) between southern and northern forests. For the Monte Carlo test, we used the average abundance for each species, as there were more studied sites on the south side than on the north side of the island. We used GLM with gamma distribution and reciprocal link function to assess differences in nutrient concentrations and their ratios between northern and southern forests. We used coinertia analysis to explore correlations between the abundance of mycorrhizal fungi spores and the environment (soil chemistry and the abundance of different plant lifeforms). In addition, we used principal component analysis on the logarithmic transformation (ln[x + 1]) as a foundation for the co-inertial analysis (Doledec and Chessel 1994, Dray et al. 2003). We used a Monte Carlo permutation test (Heo and Gabriel 1997) to establish the correlations' significance. For beta diversity, we used the index proposed by Bray and Curtis (1957), and for the partitioning of beta diversity into species turnover and nestedness, we followed Baselga (2013) and Legendre (2014). To test the effect of distance on beta diversity, we calculated the geographic distance and two environmental distance estimates based on chemical soil properties and the abundance of plant lifeforms. Environmental distance was the Euclidian distance between the scores of a non-metric multidimensional scaling on each set of data. We used generalized additive models with gamma distribution and the reciprocal link function to account for possible nonlinear patterns.

To analyze differences between the two sides of the island in the average length of hyphae and percentage of root colonization, we used generalized linear models (GLM) with a gamma distribution and the inverse link function. For all of the analyses, we used R 2.1.5.2 (R Core Team 2017); the ade4 package was used for co-inertia (Dray and Dufour 2007), while the betapart package was used for beta diversity index, and its partitioning (Baselga et al. 2017).

Results

Arbuscular Mycorrhizal Fungi Identified in the Soil

We identified 13 species of arbuscular mycorrhizal fungi, including three families: Acaulosporaceae, Glomeraceae, and Claroideoglomeraceae (Figure 2). The Monte Carlo randomization test revealed significant differences in abundance between the southern and northern forests (Figure 3). Acaulospora morrowiae, G. hoii, and Claroideoglomus claroideum were the only species with no difference in abundance between the southern and northern forests. In contrast, Glomus glomerulatum and Glomus sp_2 occurred only in the southern forests, while Acaulospora sp_1, Acaulospora sp_2, Glomus sp_13, and G. tortuosum were observed only in the northern forests. Claroideoglomus etunicatum and Rhizoglomus intraradices occurred on both sides of the island, but their abundance was higher in the southern forests than in the northern ones. In contrast, Acaulospora scrobiculata and Glomus macrocarpum, also common to both sides of the island, were more abundant in the northern forests.

Figure 2.

Arbuscular mycorrhizal fungi of Socorro Island, Mexico. (A) Glomus tortuosum, (B) Glomus macrocarpum, (C) Claroideoglomus etunicatum, (D) Glomus hoi, (E) Claroideoglomus claroideum, (F) Rhizoglomus intraradices, (G) Glomus sp. 2, (H) Glomus glomerulatum, (I) Acaulospora morrowiae, (J) Acaulospora sp. 1, and (K) Acaulospora scrobiculata. Not illustrated, Glomus sp. 13 and Acaulospora sp. 2.

Figure 3.

Rank–abundance curves for spores of arbuscular mycorrhizal fungi Socorro Island, Mexico. (a) Claroideoglomus etunicatum, (b) Rhizoglomus intraradices, (c) Acaulospora scrobiculata, (d) Glomus macrocarpum, (E) Acaulospora sp_1, (F) Glomus glomerulatum, (G) Acualospora sp_2, (H) Glomus sp_2, (I) Glomus tortuosum, (J) Glomus sp_13, (1) Acaulospora morrowiae, (2) Claroideoglomus claroideum, and (3) Glomus hoii. Capital letters indicate species exclusive to one side of the island. Lowercase letters mark those species that change significantly in abundance, and numbers indicate those species that did not change significantly in abundance between sides.

Plant Life-Forms and Soil Chemistry

We recorded 2,214 plant individuals, of which 39% (864) occurred in the southern forests and 1,350 (61%) in the northern forests. The most significant changes in the composition and abundance of life-forms between southern and northern forests were lianas, representing 8% of the total record. However, lianas were recorded only in the northern forests. Also, the abundance of ferns (25% of all plants) was almost 7-fold greater in the northern forests than in the southern. Shrubs represented 3% of all the records and were 13 times more abundant in the southern forests than in the northern ones. Trees (54% of all plants) were slightly more abundant (5%) in the northern forest, while the abundance of herbs (9% of all plants) was 14% higher in the southern forests.

Phosphorous concentration in the northern forests (0.249 mg/g) was five-fold greater (χ² = 15.6, d.f. = 1, P < 0.001) than in the southern forests (0.047 mg/g). In contrast, the nitrogen and carbon contents were 27% and 39% lower in the northern forests (11.86 mg/g and 12.0%, respectively) compared to the southern (16.4 mg/g and 19.7%, respectively) forests of the island (χ² = 7.7, d.f. = 1, P = 0.006 and χ² = 9, d.f. = 1, P =0.003). In consequence, the carbon to nitrogen (χ² = 3.5, d.f. = 1, P = 0.047), carbon to phosphorus (χ² = 15, d.f. = 1, P < 0.001), and nitrogen to phosphorus ratios (χ² = 15, d.f. = 1, P < 0.001) were higher in the southern forests. For a more detailed analysis of the biogeochemistry of Socorro Island, see Ruiz-Guerra et al. (2019).

Correlation of Mycorrhizal Fungi, Soil Chemistry, and Plant Life-Forms

After co-inertia analysis, the Monte Carlo permutation test showed a significant correlation coefficient (RV = 0.85, P = 0.02). There was a strong correlation between the abundances of mycorrhizal fungi and the soil's mineral nutrient content and the abundance of plant life-forms (Figure 4). Soil carbon and nitrogen contents were high in the southern forestscomparedtothenorthernforests.Glomus glomerulatum, Rhizoglomus intraradices, Claroideoglomus etunicatum, and an un-identified Glomus species were characteristic of the southern forests. In contrast, ferns and lianas were more abundant, and soil phosphorus concentrations were higher in northern forests than in

Figure 4.

Co-inertia analysis. (A) Correlation between the composition of the community of arbuscular mycorrhizal fungi (origin of the arrows) and the biotic (plant life-forms) and abiotic (soil chemistry) environments (tips of the arrowheads) in the southern (S1, S2, S3, and S4) and northern (N1, N2, and N3) forests of Socorro Island. (B) Correlation of environmental variables with the multivariate axis. Plant life-forms (gray letters): S = Shrubs, T = Trees, L = Lianas, H = Herbs, and F = Ferns. Soil chemistry (black letters): C = Carbon, N = Nitrogen, C/N = Carbon to nitrogen ratio, Cmr = Carbon mineralization rate, W = water content, P = Phosphorous, pH = Potential hydrogen. (C) Correlation of the abundance of arbuscular mycorrhizal fungi with the multivariate axis: Glgl = Glomus glomerulatum, Clet = Claroideoglomus etunicatum, Clcl = Claroideoglomus claroideum, Acmo = Acaulospora morrowiae, Rhin = Rhizoglomus intraradices and Glsp = Glomus sp, Acsc = Acaulospora scrobiculata, Acsp = Acualospora sp, Acsp_9 = Acualospora sp_9, Glto = Glomus tortuosum, Glho = Glomus hoii, Glma = Glomus macrocarpum, and Glsp = Glomus sp.

Figure 5.

Beta diversity of arbuscular mycorrhizal fungi in Socorro Island. (A) Total beta diversity. (B) Percentage contribution of species turnover to beta diversity. (C) The relation between beta diversity and environmental distance. We used a nonmetric multidimensional scaling analysis on the abundance of trees, shrubs, herbs, ferns, and lianas at each site for the environmental distance. We then calculated the Euclidean distance between all pairs of sites using the first two axes' scores.

southern ones, and Acaulospora species were abundant in the northern forests.

Table 1.

Colonization of Roots of Guettarda insularis by Arbuscular Mycorrhizal Fungi in Socorro Island

Beta Diversity

Beta diversity in the southern forests averaged 0.375, while among the northern forests, it was 0.428, and reached 0.613 across the two sides of the island (Figure 5A). In southern forests, 64.5% of the beta diversity was due to species turnover. In contrast, in the northern forests and between southern to northern forests, the turnover contribution to beta diversity was 90.8% and 90.9%, respectively (Figure 5B). Beta diversity increased with environmental distance measured as the abundance of plant life-forms (F = 18.3, edf = 1, P < 0.001), but no significant effects were observed as a function of soil chemistry differentiation (F = 0.970, edf = 1, P < 0.338) or geographic distance (F = 0.005, edf = 1, P < 0.944).

Mycorrhizal Colonization

The average length of hyphae in the root cortex of G. insularis was 46% higher on the northern side of the island (χ² = 4.01, d.f. = 1, P = 0.045) compared to the southern forests (Table 1). The percentage of colonization by hyphae (χ² = 1.98, d.f. = 1, P = 0.158) and vesicles (χ² = 1.21, d.f. = 1, P = 0.271) did not vary significantly between sides, while the difference in the percentage of incidence of spores was also not significant (χ² = 3.1, d.f. = 1, P = 0.078), and no arbuscules were observed in the root cortex of G. insularis.

Discussion

Mycorrhizal interactions differ between the southern and northern forests of Socorro Island. Species composition and abundance of the mycorrhizal community, the structure of beta diversity, the extent of root colonization, and the abundance of plant life-forms differed between the southern to northern forests.

Six species of mycorrhizal fungi (46%) out of 13 occurred on only one side of the island. Notably, among the most abundant species, Acaulospora sp. 1 and G. glomerulatum occurred only in the northern and southern forests, respectively. Also, C. etunicatum and R. intraradices, although present on both sides of the island, showed marked differences in their abundances. Both species were among the most abundant species in the southern forests, but their abundances were low in the northern forests. In contrast, the abundances of 43% (3 out of 7) of the shared species (A. morrowiae, C. claroideum, G. hoii) were similar in the southern and northern forests. This evidence suggests that species of mycorrhizal fungi with similar abundances in the southern and northern forests may be tolerant of a wide range of environmental conditions or have a wide range of host plants. In contrast, other species appeared to be sensitive to changes in environmental conditions (biotic and abiotic) with marked differences in their abundances between the island's southern and northern forests.

Previous studies suggest that the northern side of the island, facing the trade winds, contains more humidity than the southern side (Miranda 1960, León de la Luz et al. 1996). Nonetheless, nitrogen and carbon mineralization rates in the soils were similar on both sides of the island. These suggest that the southern and northern forests'soils harbor similar microbial communities, at least in functional terms (Biswas et al. 2018). In the laboratory, mineralization rate measures may underestimate actual field rates—the absence of plant roots and their mutualistic symbionts and alterations to the soil's physical and chemical characteristics are some of the factors known to bias mineralization rate estimates under laboratory conditions. Nonetheless, experimental evidence showed that mineralization kinetics has the same curve-shape under field and a range of laboratory conditions (Oburger and Jones 2009). Therefore, the observed mineralization rates in the soils of Socorro Island should be valid for comparison of the northern and southern forests.

The phosphorus concentration was five times higher in the northern forests than in the southern (see Ruiz-Guerra et al. 2019). Copious evidence, including field and experimental studies (Treseder 2004, Guevara and López 2007), showed that root colonization correlates negatively with phosphorus availability in the soil. In contrast, we observed that root colonization by mycorrhizal fungi was higher in the northern forest, where phosphorous availability was also high.

The comparison between southern and northern forests yielded the highest beta diversity values. The contribution of species turnover to beta diversity averaged 55%. This findings agree with patterns of beta diversity for a wide range of organisms and spatial scales (e.g., Soininen et al. 2017, Coelho et al. 2018). As well, a positive correlation between distance and beta diversity is predicted and observed in many studies (Qian and Ricklefs 2007). The geographic distance itself can play a significant role in increasing beta diversity, as differential migration limitation of the species will increase the communities' nestedness. In addition, distant areas tend to differ in at least some environmental dimensions, potentially increasing beta diversity (Leprieur et al. 2011, Keil et al. 2012). Here, we observed that beta diversity increased with environmental distance (abundance of different plant life-forms) but not with geographic distance or soil chemical differentiation. These findings highlight the importance of the plant community for the diversity of mycorrhizal fungi. However, contrary to the known effects of soil nutrients contents on mycorrhizal diversity (Bradley et al. 2006) and functioning (Treseder 2004), soil chemistry had no effects on the beta diversity. However, the observed patterns may concord with some known perturbations in Socorro island's southern forests (see below).

The overall beta diversity was similar on both sides of the island. However, the contribution of species turnover to beta diversity in the southern forests was 28% lower than that in northern forests. These findings suggest that stochastic processes govern much of the beta diversity in the southern forests. Conversely, deterministic processes seem to drive mycorrhizal fungi's beta diversity on the north side of the island.

Habitat perturbations, such as reduction and fragmentation of the habitat and the establishment of invasive species, have been reported as drivers of habitat homogenization (Socolar et al. 2016). Because only a subset of the original communities' species will cope with the new environment, stochastic processes are likely to drive species composition from site to site (Davey et al. 2013, Filgueiras et al. 2016). Some known disturbance factors affecting the island's southern side include direct exposure to hurricanes from the south and southeast (Romero-Vadillo et al. 2007). However, the effects of hurricanes on the island's forests are likely mild, as the forests are well above 500 m in height, above the elevation where most hurricanes present their strongest winds (Franklin et al. 2000). The second source of perturbation, mostly restricted to the island's southern side, is introduced sheep. Herds of this exotic herbivore trampled and overgrazed the soil and vegetation for over 150 years on the island's southern side (Walter and Levin 2008, Ortíz-Alcaraz et al. 2016) and were still abundant when we collected field data for the present study. Sheep perturbation of the forest habitat would be compatible with the increased contribution of nestedness to beta diversity and the low mycorrhizal colonization of roots observed in G. insularis in the southern forests despite the low levels of phosphorous in the soil. Trampling by sheep may disrupt mycelial networks and fungi's capacity to engage in mycorrhizal interactions (Shelton et al. 2014, Soka and Ritchie 2014, Cavagnaro et al. 2019).

Overall, we showed that southern and northern forests harbored different arbuscular mycorrhizal communities. These differences correlated with changes in the vegetation cover, specifically the cover by ferns and lianas. Nestedness contributed over 30% to total beta diversity in the southern forest. In contrast, in the northern forest, nestedness contributed less than 10% to beta diversity. Similarly, root length colonization was lower in the southern forests despite the available phosphorous being five-fold lower than that in the northern forests. Further studies must consider the inclusion of some estimates of plant performance (e.g., chlorophyll content and specific leaf area) to connect mycorrhizal colonization and plant physiology. Whether sheep were a significant driver in the differentiation of southern and northern forests is still uncertain, and long-term studies after sheep are removed from the island may provide additional evidence of the potential effects of introduced ungulates in Socorro Island.