Study site and sample collection

The research was undertaken in the municipality of Tzucacab, Yucatan, Mexico (Figure 1). This is a region of seasonal tropical deciduous forest with average temperatures from 24-28 °C and 1000-1200 mm total annual precipitation, most of it falling during the summer wet season; the main soil groups are Luvisols (42.97%), Vertisols (22.01%), Phaeozems (19.66%) and Leptosols (14.43%) (^{INEGI, 2009}). This is a karstic environment where no superficial flowing water exists (^{Estrada-Medina, Jiménez, Álvarez, and Barrientos (2019)}, and soils are calcium-rich, promoting P insolubility (^{Avnimelech, 1980}).

Figure 1: Study sites. AS15 = agroforestry system 15-year-old; AS30 = agroforestry system 30-year-old; SF15 = seasonal forest 15-year-old; SF30 = seasonal forest 30-year-old. 

Much of the area includes mixed-species homegardens for fruit and fiber, and shifting agriculture of abandoned and active maize fields or Milpas (a traditional Mayan agricultural system, integrated by a polyculture based on maize, bean, and squash).

For this study, six agroforestry systems (AS) were chosen, three that have been established for 15 years and another three that have reached a 30 year maturation. In addition, six secondary vegetation sites were selected, three with 15 years and three with 30 years of regeneration; these sites were previously utilized for maize cropping in a milpa system for up to three years. These patches of vegetation belong to regeneration sites of the region’s original seasonal forests, so we will refer to them as seasonal forests (SF) for the purposes of this study.

Agroforestry systems consist of home gardens dominated by citrus, fruit-bearing, and timber trees. Additionally, species from the Burseraceae, Euphorbiaceae, Fabaceae, Polygonaceae, and Rubiaceae families predominantly inhabit the secondary vegetation areas (^{Zamora, García, Flores, and Ortiz, 2008}).

Bulk soil samples were collected at 0-30cm depth after removing surface litter from the soil and then placed into plastic bags for transport and subsequent laboratory processing. In each of the 12 study sites, plots of 100 × 100m were selected, nine samples were collected in each plot systematically, and the sampling was carried out during both the dry (April) and wet (August) seasons (Figure 1).

A total of 216 samples were collected, encompassing 2 systems (AF and SF) × 2 ages (15 and 30 years) × 3 replicates of the systems × 2 seasons (dry and wet) × 9 samples per plot.

Samples processing and laboratory analysis

Soils were air-dried and sieved (2mm) to separate soil particles and aggregates from fine roots, with coarse organic matter materials predominating. Each soil sample was divided into two subsamples, and the fine root samples were separated for further analysis. Following sieving, the first soil sub-samples were analyzed for organic C by the Walkley-Black method (^{Nelson and Sommers, 1996}).

The second soil subsample was used for analyzing Glomeromicotyna AMF spores, which were rehydrated and isolated by sucrose flotation from 10 g of soil (^{Allen, Moore, Christense, and Stanton 1979}), counted under a dissecting microscope at 10X magnification, and morphologically identified to rhizophilic and edaphophilic taxa by taxonomic keys (^{Weber et al., 2019}).

The rhizophilic AMF guild is comprised of the families Glomeraceae, Claroideoglomeraceae, and Paraglomeraceae, and the edaphophilic guild includes the Gigasporaceae and Diversisporaceae. A third guild, the ancestral AMF, is also recognized by ^{Weber et al., (2019)}, and consists of Archaeosporaceae, Ambisporaceae, Pacisporaceae, and Acaulosporaceae. However, this guild is based on phylogeny rather than function, and the total spore numbers we observed from this guild were < 1 spore g^-1 soil. Therefore, we did not include the ancestral guild in the statistical analyses.

Finally, fine roots (< 2mm) obtained from each soil sample were cleared with KOH (10%) and washed twice with distilled water until white; then, roots were rehydrated, stained with trypan blue, and 100 intersections per slide were microscopically observed for mycorrhizal colonization at 100x magnification; of which, depending on the number of colonized intersections, the percentage of colonization was obtained (^{Allen, Swenson, Querejeta, Egerton, and Treseder, 2003b}).

Statistical analyses

Glomeromicotyna spore numbers were calculated as spores g^-1 of dry soil for rhizophilic and edaphophilic species. Differences in number of spores and soil organic carbon (SOC) among land uses (AS and SF) and system ages (15, 30 years) per season (wet, dry) were tested by a two-way ANOVA upon verifying the assumption of homogeneity of variances using the Levene’s test. Tukey’s post hoc test was employed to examine significant effects.

A principal component analysis on the correlation matrix of the categorical (codified as dummy variables) and numerical variables was used to explore patterns (similarities and differences) within the dataset. Relationships between SOC and number of spores were analyzed by a linear regression for each season and guild. All analyses were run in PAST Ver. 4.01 (^{Hammer, Harper, and Ryan 2001})

Soil organic carbon

Soil Organic Carbon (SOC) varied according to land use and system age, showing lower values in AS than SF and lower values in younger systems (15 years) than older ones (30 years) (Table 1). The interaction (land use age / seasons) of the two-way ANOVA was not significant; thus, only the land use differences are shown.

As expected, SOC contents show different patterns in the studied land uses in the Yucatan Peninsula, which could be related to the management practices in each system. SOC is usually greater in a forest than any other land use, but it varies by region (^{Ali, Begum, Hayat, and Bohannan, 2017}; ^{Muktamar, Nurliana, Aningtias, and Anugrah, 2021}). SOC values in the SF were lower than those found in drier deciduous forests of central and northern Yucatan, with 14.6% in a 12-year-old SF in Hocaba, Yucatan (^{Shang and Tiessen, 2003}) and 27.9% in a 15-year-old SF in Dzibilchaltun, Yucatan (^{Ceccon, Olmested, and Campo, 2002}), respectively.

This trend was also observed in a study conducted by ^{Álvarez-Rivera et al. (2021)}, in which they investigated homegardens and successional forest sites (> 30 years of fallow) in Tzucacab, reporting an average SOC of 6.24 and 14.61%, respectively.

On the contrary, management practices applied in AF systems can have diverse impacts on the physicochemical properties of the soil, particularly on SOC. It has been reported that land use changes, can have a negative effect on SOC reservoirs (^{Matos, Freese, Mendonça, Slazak, and Hüttl, 2011}), while specific management practices i.e., (manure amendments, crop rotations) can foster an increase in SOC content (^{Wolka, Biazin, Martinsen, and Mulder, 2021}).

Broadly, the studied AF sites had lower SOC compared to SF, as they are former citrus home gardens turned into mixed species homegardens (^{Padilla-Vega, Jiménez, and Estrada, 2015}) with a history of practices that reduce soil cover and SOC accumulation (weeding control, litter burning, agrochemical application, biomass pruning for forage, etc.). But AF sites also showed increasing SOC with site age, ranging from 6% at 15 years to 7-8% at 30 years. This pattern has also been reported by ^{Lawrence (2005)} in tropical dry forests of southern Mexico.

Despite the AF sites having lower SOC than the SF sites, they are showing the same increase with age of maturation or regeneration. The increase in SOC reservoirs can be influenced by various factors (such as climate, microbial activity, vegetation composition, and pH, among others). Nevertheless, it has been reported that as vegetation matures, SOC also tends to increase, until the vegetation reaches climax conditions, and SOC stabilizes or might experience a slight decline (^{Guan et al., 2019}).

Mycorrhizal colonization and glomeromicotyna spore abundance

For all samples, hyphae accounted for 80% or more of the mycorrhizal colonization, complemented by vesicles and infrequent occurrence of arbuscules. Mycorrhizal colonization was lower in AS during both the dry season (21%) and wet season (32%) than in SF (34 and 38% in both seasons respectively). There were no significant differences in percent colonization due to site age (values are averages of site ages). Therefore, no additional statistical analyses were done, as there was no relationship between the age of the site and mycorrhizal colonization within a land use type.

The most notable differences in colonization were observed among the younger systems (15-years-old), which experience higher disturbance levels. Among these, the AF systems exhibited the lowest mycorrhizal colonization. On the other hand, both age groups of SF systems and the 30-year-old AF systems exhibited higher AMF colonization resilience, as mycorrhizal colonization remains consistent between seasons. Nonetheless, a trend can be observed wherein higher colonization occurs during the rainy season, with this pattern being more pronounced in AF systems. A high mycorrhizal colonization has also been documented in Cacao agroforestry systems in Brazil (^{de Oliveira and de Oliveira, 2005}) and seasonal forest of the Pacific coast of Mexico (^{Allen, Rincón, Allen, Pérez, and Huante, 1998}), associated with an increment in microbial diversity due to the less disturbed conditions of these systems (^{Matos et al., 2022}; ^{Polo-Marcial, Solís, Murillo, Ávila, and Andrade, 2023}).

Seasonal variations in mycorrhizal colonization have been explored in previous studies. Several investigations have documented heightened mycorrhizal colonization during the dry season e.g., (^{Sivakumar, 2013}; ^{Chareesri, De Deyn, Sergeeva, Polthanne, and Kuyper, 2020}). However, in this study, we observed an opposing trend. This pattern might be attributed to soil moisture and available phosphorus, both of which are pivotal factors influencing AMF responses (^{Chu et al., 2020}; ^{Chudaeng and Teamkao, 2020}; ^{Liu, Shen, Li, Xiao, and Song, 2021}); due to the region’s environmental and karstic conditions, moisture and P are limited during the dry season.

Statistical differences were detected in the number of rhizophilic spores between AS and SF land uses, with differences between AS15 and AS30 only during the wet season; for edaphophilic spores, the only difference identified was among AS15 and the other three systems in both seasons (Table 2). The number of spores was consistently higher in the 30-year-old than 15-year-old systems. Spore abundances of rhizophilic AMF were higher during the wet than the dry season, but not statistically different. The mean rhizophilic Glomeromicotyna AMF spore abundance was 7.17 spores g^-1 in the wet season versus 6.67 in the dry season (P = 0.037, n = 108). The mean of edaphophilic Glomeromicotyna AMF spore abundance was 1.37 spores g^-1 in the wet season versus 1.43 in the dry season (P = 0.037, n =108). The limited number of spores found in these soils is probably not only related to the high contents of organic carbon but also the pH and P content in these soils, as a low number of fungi spores is often associated with a combination of high organic carbon contents, acid pHs and low available P contents (^{Alori, Fawole, and Akanji, 2020}).

The PCA biplot of the first two principal components indicates a contrary gradient between seasons (dry, wet) and a better relationship of SF to SOC than AF (Figure 2). Within systems, a stronger relationship of 30-year-old systems with SOC compared to 15-year-old systems is shown. AS have a lower importance than SOC content. However, both edaphophilic and rhizophilic AMF spores are more related to SF than to AS. Edaphophilic AMF are more related to SF30 whereas rhizophilic AMF are more related to SF15.

Figure 2: Principal components analysis biplot. Data points represent the average value of the 9 replicates per plot. AS15 = agroforestry system 15-year-old; AS30 = agroforestry system 30-year-old; SF15 = seasonal forest 15-year-old; SF30 = seasonal forest 30-year-old; SOC = soil organic carbon; DRY = dry season; WET = wet season. 

Rhizophilic species spore abundance was sensitive to land use changes and system age but not to seasonality. On the other hand, edaphophilic AMF spore abundance did not show a clear relationship to SOC suggesting that they were more tolerant to the perturbations imposed than rhizophilic AMF. This response was opposite to our expectations. Based on taxonomic classifications, the rhizophilic species fall under the previous Glomeraceae classification, while the edaphophilic species are in the Gigasporaceae (^{Weber et al., 2019}), and can be compared with previous studies reporting taxon names.

Rhizophilic AMF tend to be early successional taxa that have well-developed root internal hyphal development, while edaphophilic are late successional species with extensive external hyphal networks (^{Allen, Allen, Egerton-Warburton, Corkidi, and Gómez, 2003c}, ^{Weber et al., 2019}). Rhizophilic species tend to produce a higher density of spores, especially many with small diameters (50-100 µm), while edaphophilic species tend to produce fewer spores with larger diameters (150-200 µm), as we observed here and as well as at another Yucatecan seasonal forest (^{Allen et al., 2003c}).

Based on their late successional status, we expected edaphophilic spores to increase as well as rhizophilic spores with increasing SOC. Instead, we found that spores of rhizophilic AMF species were more sensitive to changes in SOC than those of edaphophilic species, showing more spore abundance as SOC increases. ^{Van der Heyde et al. (2017)} found that Gigasporaceae species (edaphophilic) are less tolerant to disturbances than species of Glomaceae (rhizophilic). An increase in Gigasporaceae spores was indeed observed during the recovery of a recently burned site in the Yucatan, where both spore density and organic matter recovered to levels observed in mature forests after only three years (^{Allen et al. 2003c}).

A recently established homegarden was not available for this study, and edaphophilic AMF may have already established their hyphal network (and can sporulate) in 15-years-old vegetation. Nevertheless, only the rhizophilic spores continued to increase sporulation with increasing SOC beyond 15 years. The shallower soils and lower precipitation in these sites promote SOC accumulation, conditioning a different relationship between AMF and SOC than ^{van der Heyde et al., (2017)} observed. By contrast, the percentage of mycorrhizal colonization was not related to site age or SOC. This is likely because colonization percentage varies with the observation date, while spores persist longer in the soil (^{Allen et al., 2003c}).

SOC content and glomeromicotyna spore number relationships

Regression analyses show linear relationships between SOC content and the number of rhizophilic AMF spores for both wet and dry seasons (Figure 3). No relationships were found between SOC and the number of spores of edaphophilic AMF in any season (data not shown).

Figure 3: Soil organic carbon (SOC) vs number of rhizophilic AMF spores, linear regression. AS15 = agroforestry system 15-year-old; AS30 = agroforestry system 30-year-old; SF15 = seasonal forest 15-year-old; SF30 = seasonal forest 30-year-old. Remove “Lineal” from figure. 

We postulate that the Glomeromicotyna spore abundance of rhizophilic AMF species could be used as an indicator of soil health at any time of the year. SOC and AMF have intimate feedback, but they must still be better understood. AMF contribute to the stable organic pool. The implications of change rates of rhizophilic and edaphophilic AMF are still unknown. Rhizophilic AMF are presumed to provide pathogen protection but also predominate in arid environments, such as during the dry season in these forests. Edaphophilic AMF are postulated to be especially important in nutrient uptake (^{Weber et al., 2019}). These fungi are needed consistently at all stages of vegetative regrowth. In contrast, pathogen protection and drought stress become increasingly crucial as root mortality and leaf area for transpiration increase with increasing plant density and leaf area of older vegetation.

Importantly, this generalization must be validated for other sites before use because soil conditions and vegetation type can change the relationships found here. For example, ^{Ceccon et al., 2002} found SOC values of 27.9 and 22.0% in 15 and 60-year-old deciduous forest plots in northern Yucatan, suggesting other factors than age affect SOC. In this case, do the rhizophilic AMF spore numbers show the same regression with time, or SOC, or is this a site-specific relationship? Thus, a preliminary examination is mandatory in other sites to check for the relations found here and whether this assessing tool is site-specific. A comprehensive community analysis based on sequencing approaches would also provide important insights into the structure of AMF communities, as this analysis of spores only revealed the composition of AMF in the soil, and not in the roots.