Building an Integrative and Testable Hypothesis on How Edaphic Factors Ultimately Inuence the Occurrence and Quality of Ophiocordyceps Sinensis, the Vegetable Caterpillar

Background The yield of commercially harvested “vegetable caterpillar” Ophiocordyceps sinensis has dramatically plummeted in the last few decades, while market demand has increased. Besides controlling the obvious overexploitation of this species, understanding how edaphic factors inuence this system may improve the chances of successful cultivation and thus support the conservation of O. sinensis in the wild. Our study investigates how the presence/absence and the quality of O. sinensis may be linked to a series of edaphic factors pertinent to its microhabitat, including enzyme activity, nutrients, moisture, pH and nematode population. In order to provide a preliminary hypothesis on the relationships among edaphic factors and their inuence on O. sinensis, we performed a principal component analysis and structural equation modelling despite limited replication. Results Soil samples containing O. sinensis were more moist and contained a higher concentration of nutrients and enzyme activity than control samples collected nearby, where the species was absent. Preliminary analyses indicated that enzyme activity may be crucial and appeared to be affected by a number of other soil factors. We found that O. sinensis would occupy microhabitats with a relatively higher soil fertility and a more persistent enzyme activity, where the values of total nitrogen and catalase are especially important. Otherwise, with the exception of organic matter and enzyme activity, mean values did not suggest any other factors potentially corresponding to a better quality of O. sinensis. Based on these preliminary ndings and a further literature review, we formulated the rst integrative hypothesis (network of interactions) on how soil factors may inuence each other and O. sinensis. Finally, we indicate how this hypothesis may be tested in the future, in order to increase the chances for successful cultivation and thus promote the conservation and sustainable harvesting of O. sinensis. strong results,


Background
Ophiocordyceps sinensis (Berk.) G. H. Sung, J. M. Sung, Hywel-Jones and Spatafora (Syn. Cordyceps sinensis) is an endemic entomopathogenic fungus species distributed in the region of the Qinghai-Tibet Plateau. It is composed of the fruiting body of an entomophagous fungus and the larva of a host, of which several species may be colonized depending on the fungus variety [1]. There are three stages characterizing the parasitic complex between O. sinensis fungi and the Hepialus larva (Lepidoptera, Hepialidae) [2] these include; the primary infection, followed by a parasitic phase and nally a saprophytic phase [3]. Under favourable environmental conditions, the latter phase ultimately results in the fruiting body of the fungus (which germinates out of the soil from the oral cavity of the dead host), at which point commercial harvesting can occur.
For centuries, medicinal value has been conferred to O. sinensis, resulting in an increasingly high economic importance [4]. Among other medicinal properties, O. sinensis is widely used against liver, kidney and lung diseases, as well as for its anticancer effects, and even its dermatological properties [5,6]. Nevertheless, some studies have warned about its potential toxicity [7][8][9]. Available harvest estimates range from 85 to 185 tons annually throughout the distribution range area of O. sinensis (the Himalayas, Hengduan Mountains and eastern Qinghai-Tibet Plateau) [10]. Although the exploitation of the "vegetable caterpillar" represents a signi cant income for local communities, harvesting practices have been deemed unsustainable leading to an obvious decline in the abundance of wild O. sinensis [11], even causing severe disturbances to its habitat [12,13]. Aside from natural predators and pathogenic microorganisms, habitat alteration deriving from anthropogenic activities (e.g., herding) affects the distribution and abundance of wild O. sinensis [14]. Additionally, climate change is likely to have caused a vertical displacement of this species´ distribution [15,16]: a study showed that the upper limit of the prime habitat of the caterpillar fungus shifted upwards by 200-500m (from 3900-4400 to 4400-4600) in two decades [10], resulting in a contraction of the distribution range of O. sinensis.
In addition to distribution and phylogenetic work [17], most studies on O. sinensis have focused on diverse aspects related to medicinal use or to commercial applications, including chemical composition analysis, arti cial larval and mycelium cultures, among others [18][19][20][21]. Furthermore, researchers and high-tech commercial companies have been attempting to cultivate O. sinensis in controlled environments (in-vitro), as this could theoretically represent a rapid and effective way not only to meet the increasing demands of the market, but also to promote the conservation of this species [22][23][24]. After decades of effort, many achievements have been made, such as the successful arti cial cultivation of fungi and host larvae separately, and the taxonomic delineation and variability of the pathogenic fungus and host larvae [25]. Other challenges that have not yet been fully overcome, include the di culties of simulating the natural habitat in laboratories, and the technical complexity of arti cially infecting the host larvae with the fungus O. sinensis [26]. Some of these shortcomings are clearly the result of a lack of understanding of the factors which are crucial to the natural environment of O. sinensis (i.e., alpine meadows), justifying more in-depth research in the ecology of the species. More speci cally, soil factors are likely to be most relevant for this system, since most of the life cycle of the Hepialus larva, the O. sinensis fungus as well as the inoculation and hyphal development of the O. sinensis complex occur underground. Indeed, the Hepialus larva feeds on plant roots [27], and the O. sinensis fungus depends on microbial reserves in the soil [28,29]. After being infected (in the soil) by the fungus, the caterpillar remains alive for a long period (5-12 months in laboratory, may be longer in the wild) and keeps developing until O. sinensis mummi ed [30,31]. Upon its death, the caterpillar moves near the surface of the soil (2-5 cm depth) and places itself vertically, with the head facing upward, allowing the stromata to grow out of the soil and release the ascospores [32]. Thus, investigating the soil microhabitat around O. sinensis appears most relevant.
Focusing on the microhabitat of O. sinensis, we investigated the relationships among edaphic factors, by measuring enzyme activity (speci cally catalase, urease, sucrase, nitrate reductase (NR), acid phosphatase (ACP) and cellulase), nutrient content (including total nitrogen (TN), total phosphorus (TP), total potassium (TK), organic matter (OM), available phosphorus (AP), rapidly available potassium (Kppm)) and nematode population in the soil. Not only soil samples need to be collected directly around O. sinensis and thus are of a small volume, but also such samples are di cult to come by because of the secrecy surrounding collection sites. As a result, our study ultimately suffers from limited replication. Nevertheless, in addition to a literature review, we simulate the network of soil factors using structural equation modelling (SEM). By doing so, we aim at establishing a working hypothesis in the form of a network of potential interactions among soil factors and O. sinensis. We then indicate how this hypothesis may be tested in future studies.

Results
In general, mean values rarely differed much among quality categories of O. sinensis. However, larger differences were observed between soil samples that contained O. sinensis and soil samples that did not. Preliminary analyses indicates that enzyme activity may be crucial, as it appears to be affected by a number of other soil factors. Our results need to be interpreted with caution; nevertheless, they allowed us to build up the rst testable hypothesis integrating a large array of soil factors realistically important to O. sinensis (represented visually in Fig. 1

Pearson correlation between environmental variables and eigenvalue of O. sinensis
In order to build up our hypothesis, we needed to identify potential correlations. Hence, we tentatively performed statistical analyses despite pseudoreplication. These results need to be cautiously evaluated and the p-values provided are only indicative. Pearson correlation was applied to analyse interrelation of each environmental variable and eigenvalue of O. sinensis (Table 2). We found that the polypide, stroma and the fresh weight (FW) of O. sinensis were likely to be strongly correlated with each other (indicative p < 0.01), and that some environmental variables may be correlated to all three of them. These environmental variables may include catalase, TN, TP, OM, AP, Kppm, and moisture. Additional, yet probably weaker correlations (indicative p < 0.05), may exist between the eigenvalue of O. sinensis and other edaphic factors, such as cellulase, NR, TK, and nematode populations. As expected, based upon the mean values themselves (see above), the eigenvalue of O. sinensis had no distinct relationship with pH, but a positive correlation may be visible with some soil enzyme activity, moisture, and all measured soil nutrients. Among these factors, some may be correlated. For example, the respective correlations between catalase and urease with other enzymes appeared signi cant (indicative p < 0.01). Also, each type of soil nutrient appeared to have a signi cant correlation with each other, and correlations between enzymes and nutrients are also mostly signi cant, soil moisture did not show any signi cant correlations with most of the enzyme activity, but did with most of the nutrients (p < 0.01). In addition, the pH appeared to be positively correlated only to sucrase and TK, two factors that did not vary much among groups. Nematodes population showed opposite correlations as those observed for moisture (indicative p < 0.05), and nematode populations were negatively correlated (p < 0.01) with moisture. In conclusion, the quality of O. sinensis showed relationships with most of the soil enzyme activity, all soil nutrients, as well as moisture and nematode populations. Based on previous studies [33][34][35] and our analysis, we developed an initial model by assuming that all the variables were directly correlated with the quality of O. sinensis, and that there were connections between nutrients and enzyme activity, as well as causal relationships between soil moisture, pH and soil nutrients, enzyme activity and nematodes. After tting the model, we made progressive modi cations to the original design with the remainders of the parameters until an optimized model t was achieved (χ 2 = 4.249, p = 0.643). The nal model tting is presented in Fig. 1 and Supplementary Table S4 [see  Additional le 1], and constitute the testable hypothesis we provide for future research.
In this path analysis, soil moisture, nutrients and enzyme activity were associated with the higher quality of O. sinensis. First, we assumed that soil nutrients would have a strong direct effect, yet it appeared that soil nutrients possibly affected the quality of O. sinensis by altering enzyme activity. The path coe cient from nutrients to O. sinensis was only 0.10 (indicative p = 0.719), and the path coe cient from nutrients to enzyme activity was 0.92 (indicative p < 0.001). Soil moisture was also positively associated with soil nutrients (path coe cient = 0.93, p < 0.001), in addition to having a direct effect on the quality of O. sinensis. Moreover, lower pH values were associated with higher nutrient content, which in turn was associated with larger nematode populations. The values of pH did not appear to have any noticeable relationships with nematode populations.
Finally, in order to investigate the relationship between soil nutrients and O. sinensis in more details, we disassembled the one-dimensional nutrient data and estimated the contribution of the different nutrient components (which included TN, TP, OM, AP and Kppm) to O. sinensis. A similar procedure was conducted with data on soil enzyme activity (which included catalase, urease, NR and cellulase). The results indicated that TN made a large contribution to the quality of O. sinensis (92.59%) and that the remaining nutrient variables only accounted for small fractions (Kppm = 6.20%, AP = 0.90%, OM = 0.31%, TP < 0.01%). Enzyme activity, catalase and NR were found to potentially have a major contribution to the quality of O. sinensis (i.e., catalase = 74.57%, NR = 22.78%), with the remaining enzymes contributed less than 3.0% altogether (i.e., urease = 1.73%, cellulase = 0.92%). These seemingly strong results, as all of our results, are however only indicative.

Discussion
Ophiocordyceps sinensis is a notorious component of traditional medicine, and is widely used across the region of the Qinghai-Tibet Plateau. Its commercial harvest has plummeted in the last few decades partly due to over-exploitation. In order to preserve the natural populations of O. sinensis, as well as to bypass problems related to soil contamination by heavy metals (rendering O. sinensis improper to consumption), researchers have mainly focused on improving arti cial cultivation of the species [25,36,37]. Until now, such in vitro studies have made a series of major breakthroughs for the production of mycelia and the cultivation of the fungus, as well as for rearing host larvae and even the cultivation of the complex post-infection. However, challenges still persist for largescale cultivation, the most prominent of them being the initial inoculation of the host larvae by O. sinensis under controlled environment [25]. This crucial step may strongly depend on edaphic conditions in alpine meadows, the natural habitat of O. sinensis [38]. Admittedly, collecting O. sinensis in phases prior to its fructi cation is challenging because this organism is so elusive. Investigating soil conditions around the fructi cation is, for now, the best possible proxy.
Because integrative studies on soil properties of the native habitat of O. sinensis are lacking, a working and testable hypothesis on their relationship is needed.
To develop such a hypothesis, we performed an experiment targeting the effect of soil factors on the commercial quality of that species. Since the statistical power of our experiment is limited, our aim is only to provide a hypothesis for future studies to test. Generally speaking, we nd that soil moisture, soil nutrients and soil enzyme activity are likely to be crucial factors for the quality of O. sinensis. In the following, we will discuss the potential network of relationships among soil factors and their effect on the quality of the "fungus caterpillar". We nally cross-check and combine our ndings with the available literature.

Edaphic factors and their effects on O. sinensis
Soil enzymes are crucial to soil ecology, and to the development of a variety of organisms [39], which appears to be also the case for O. sinensis. We nd that catalase, cellulase, urease and NR (simpli ed as one factor in a PCA) are likely to affect the quality of this species positively, with catalase having possibly the strongest effect, followed by NR. Hence, we believe that O. sinensis requires a dynamic soil micro-habitat, with a high ability for redox and denitri cation, as expected by the functions of the catalase and the NR, respectively [37,40]. The interaction network (our hypothesis, see Fig. 1) also suggests that most environmental parameters (among others, nutrient and soil pH; see Fig. 1) may ultimately in uence O. sinensis by indirectly affecting soil enzymes. These results potentially con rm that soil enzyme activity could act as an indicator for soil functions in this ecosystem. Our results corroborate other studies showing that soil enzyme activity is an important biochemical indicator of soil quality, and an indicator of nutrient dynamics in general [41,42]. For example, soil enzymes participate in the formation and evolution of components leading to soil fertility [43]. Furthermore, soil enzyme activity appears to be in uenced by a series of other soil parameters acting in concert. In the following, we will discuss the different indirect mechanisms by which these factors may affect soil enzyme activity, and ultimately the quality of O. sinensis.
Although soil nutrients do not appear to be associated in a direct manner with the quality of O. sinensis in the SEM analysis (path coe cients = 0.10, p = 0.719), they should not be neglected in this network ( Fig. 1): nutrients act indirectly on the vegetable caterpillar by modifying enzyme activity. More speci cally, TN may be instrumental, and the relationship between O. sinensis and NR is probably strong. Together, these results suggest that O. sinensis develops best in relatively nitrogen-rich habitat. Our study thus aligns with the results of Wu et al. [44], who observed a direct effect of soil nutrients (incl. hydrolysable nitrogen) on the spatial distribution of O. sinensis (although that study did not investigate enzyme activity). On the contrary, our results may, at least at rst sight, contrast with a series of other studies which revealed an inhibitory effect of nutrients on enzyme activity [45][46][47]. However, these papers only investigated the direct effect of a given nutrient on a particular enzyme in vitro, while our approach was designed to include simultaneously a range of nutrients as well as a range of enzymes. This constitutes the strength of our hypothesis, as the dynamics between nutrients and enzymes should be, if possible, investigated together [42] within a near natural setting. The positive effect of nutrients on enzyme activity may in fact only be apparent when a substantial portion of this system is investigated at once.
Moreover, our results tend to show that nutrients availability was directly associated with pH values, as expected. We found that pH values (for the soil samples containing O. sinensis) corresponded to previous investigations as 5.5 to 7.5 [48,49], and were more often slightly acidic. The possible association between pH and nutrients availability obviously depends on the nutrient considered, yet in our samples, slightly lower pH values always corresponded to overall greater nutrient availability (considering all four nutrients investigated). Although soil pH is not the primary in uence on O. sinensis, a suitable range of pH is important to the quality of O. sinensis and steady pH over time is bene cial [19,48,49]. Furthermore, our study may indicate that nutrient availability might have an effect on the nematode population, with different trophic groups of nematodes (e.g., bacterivores, fungivores) responding differently back to front. For example, bacterivores are known to be positively affected by a nitrogen increase, whereas fungivores populations would decline [50]. However, in general, unbalanced amounts of soil nutrients are expected to negatively impact the nematode community, which has repercussions at the ecosystem level [51]. Moreover, the destabilization of soil nutrients content could lead to changes in the micro ora, which then re ect back into the nematode community [52,53]. Overall, our study suggests that increased nutrient content in the soil negatively affects nematodes populations, which matches the results of Li et al. [51]. Even though the nematode population appears to occupy a marginal location in our network of interactions, it may still indirectly in uence O. sinensis, because of its strong correlation with soil nutrients and the micro ora in general. Clearly, trophic groups of nematodes should be further investigated both individually and in concert to grasp the mechanisms dominating their interaction with soil nutrients. Yet, overall, our study does nd trends which do align with other studies even despite a limited dataset.
Clearly, some additional data are required to fully understand the dynamic interactions between O. sinensis and its micro-habitat, and not only in terms of replication for the factors that we investigated. First, there are feedbacks between vegetation (plant coverage, abundance, species richness, community, aboveand below-ground biomass etc.), nutrients, enzyme activity and even the nematode community [42,44,50]. Thus, in order to fully unravel interactions among soil nutrients, enzyme activity, and vegetation, as well as their effect on the quality of O. sinensis, biomass data would need to be included in future research. Some studies have revealed that below-ground plant biomass can directly affect both the host larvae (as food source) and the fungus [37,54], indicating that above-or below-ground plant biomass may play an important role in this network. Another component of this dynamic system would be the population of microorganisms, which is not only important for the entire ecosystem (as decomposers, producers of soil enzymes and components of soil nutrients), but also for O. sinensis speci cally. Indeed, in the study of , communities of microorganisms have been associated with the development and metabolic process of O. sinensis [55]. Microorganisms also play an important role in soil food webs, for example as a food source for nematodes [34], which we investigated here. By considering the potential role of biomass and microorganisms, our hypothesis is the most integrative to date, and represents a necessary rst step towards incorporating more parameters and gathering a better understanding of the interactions between O. sinensis and its immediate environment.
Limitations of the study design in time and space: perspectives Our study was performed in only one locality involved in the harvesting of O. sinensis, thus limiting its power for generalization. Although we believe our study site is highly representative and bears some potential for broader conclusions, we cannot exclude that the results may vary if study sites were replicated throughout the distribution range of the vegetable caterpillar. Hence, investigating a variety of geographically distant sites would be necessary in future studies. In addition, conducting a series of control experiments either in-situ or in an arti cial environment would allow identifying thresholds (maximum and minimum), beyond which several factors of the soil would become detrimental to O. sinensis. Among others, nitrogen enrichment may restructure plant and nematode communities and exacerbate the toxic effect of ammonium and aluminium [50]. Therefore, nitrogen intake beyond such threshold would modify the equilibrium of available nutrients, probably affecting the quality of O. sinensis in an indirect manner (via enzyme activity), as shown in our study. Identifying such thresholds would be instrumental in establishing locally adapted strategies for sustainable harvesting in concert with strategies aiming at preventing overexploitation [13]. The network of interaction we produced as a testable hypothesis can thus serve as a base to investigate how the role of edaphic factors may vary spatially.
Furthermore, our study only represents a snapshot in time of the relationship between O. sinensis and edaphic factors, which corresponds to the sporulation phase and the optimal harvest time. Although our study depicts conditions under which high commercial values may be reached, gaining an understanding of the interaction between the vegetable caterpillar and soil components at earlier phases of the development of this complex (including the primary infection and the parasitic phase) would be necessary. So far, our study can only assume that the effect of soil factors remains stable throughout the growing season. In controlled conditions, the failure to infect the host larvae with O. sinensis (whereas infected young larvae captured in the wild would develop normally in controlled conditions) has only been attributed to missing cryptic environmental factors [26]. However, no-one can exclude that soil conditions included in our hypothesis may vary through the life cycle of O. sinensis and may favor or prevent inoculation depending on the season. Our research, providing the rst integrative framework of edaphic factors on the sporulation phase of this parasitic complex, thus represents a base to which other phases of the lifecycle of O. sinensis can be compared. Hence, future studies should investigate a more comprehensive set of soil parameters (e.g., also including plant biomass and microorganisms) at different developmental phases of O. sinensis, and from replicated populations in time and space.
Edaphic conditions, climate change and pasture management: towards assembling the big picture We found that a higher quantity of nutrients may be favorable to O. sinensis, but we could not identify an upper threshold beyond which the quantity of nutrients in the soil would become detrimental to O. sinensis. Identifying this threshold would be crucial, because it may be attained when pastoral practices are causing an excess of grazing or fertilizing, both of which being critical for nutrient cycling and soil fauna biodiversity [56]. Our hypothesis could represent a good base to investigate differences in pastoral practices and their impact on O. sinensis and more generally, its habitat. We argue that further studies should investigate in detail the complex interactions between grazing intensity (such as from nomadic to static herding), edaphic factors (especially soil enzyme activity), soil biomass, and the abundance of O. sinensis, using a geographically and temporally replicated design. In addition, climate change not only has a direct impact on edaphic conditions by affecting the temperature and precipitation [57,58], but also leads to an upward displacement of O. sinensis indirectly [10,[59][60][61]. Therefore, such studies should obviously project their ndings into the future, using available climate scenarios.

Conclusions
In this study, we investigated the relationships among enzyme activity, nutrient content, moisture, pH, nematode population and the quality of O. sinensis in the soil community in southeastern edge of the Qinghai-Tibet Plateau. Despite a limited statistical power due to the di culty to gather su cient samples, we developed an integrative and testable hypothesis (see. Figure 1 Soil samples containing O. sinensis were collected in April 2014, which corresponds to the early fruiting season of the fungus and to the optimal collection time associated with the highest commercial value [10]. Samples of O. sinensis were harvested randomly throughout the study site, in the natural habitat of this species on the private land of local expert collectors. In total, 40 samples were collected, each of them containing one individual of O. sinensis. Samples were immediately handed over to us, in compliance with local customs, including the secrecy of collection spots, which ultimately limited the number of samples we could obtain. Each soil sample containing O. sinensis was approximately 160 cubic centimetres and 100 grams, and the distance from the head of the polypide to the surface of the soil ranged from 0 cm to 4.6 cm. As control group, additional soil samples (N = 10) which did not contain O. sinensis were also collected randomly from the same alpine meadow, and were characterized by the same surrounding oristic composition.

Quality estimation of O. sinensis samples
After measuring the physical characteristics of each sample (i.e., length, width, height, and weight), we separated O. sinensis specimens from the soil. Meanwhile, we measured the growth depth, the length of the polypide and stroma, the FW of the entire parasitic complex. We then used a cluster analysis in IBM SPSS statistics to separate O. sinensis specimens into three groups corresponding to three different levels of quality (i.e., high-quality (A), general-quality (B) and poor-quality (C)). Estimation of the quality of O. sinensis specimens was based on the length of polypide, stroma and FW of the entire parasitic complex, as well as on additional visual examination (robustness and shape) (Supplementary Tables S1, S2). We categorized the 40 O. sinensis specimens as follows: high-quality (i.e., larger and heavier specimens, group A, 13 specimens), general-quality (group B, 14 specimens), and poor-quality (group C, 13 specimens) (Supplementary Table S1, S2 in Additional le 1). We followed a well-established categorizing strategy for the quality of O. sinensis that is commonly used for commercial purposes [63,64]. We then carefully separated O. sinensis specimens from the soil, and pooled (mixed) the different soil samples according to the quality of the O. sinensis they contained (e.g., all soil samples that contained a high-quality O. sinensis specimen were pooled together), resulting in three experimental composite samples and one control composite sample. Thus, the four different composite samples contained each between 10 and 14 soil samples, which we believe covers the natural variation of the soil at the study site. Since collecting soil samples makes sense only in the immediate surrounding of O. sinensis specimens, but the soil samples per O. sinensis specimen would have been too small to measure all variables of interest, hence pooling (mixing soil samples) was necessary. Finally, we divided each of these composite sample into three equal parts to act as replicates to account for measurement error for the different factors we investigated. The use of composite samples is often applied in rhizosphere research, in order to determine the effect of diverse factors on communities or target organisms, as for example the relative effects of soil nutrients, plant species, microbial activity and rhizosphere bacterial community on strawberries (Fragaria ananassa Duch.), oilseed rapes (Brassica napus L.), and cucumbers (Cucumis sativus L.) seedlings [65,66]. For each of the 12 sub-samples, we rst used approximately 100 grams of soil and conducted the nematode population experiment immediately. Next, we divided the remaining soil samples into two parts, one part of which was rst used for soil moisture measurement and the rest was stored at 4°C to measure enzyme activity, the second part was dried under natural air in a cool and ventilated place for the pH and nutrient tests.
Soil enzyme activity, physicochemical, and nematode population analysis Soil moisture was calculated as percentage of weight loss by using the fresh soil samples immediately, after the samples were dried in an oven for 18h at 105°C (until constant mass was achieved). Catalase, urease, sucrase, NR, ACP and cellulase were determined by different methods as shortly outlined hereafter. We used titanous sulphate-colorimetry for catalase´s activity, expressed as the mass ratio (mg/g) of hydrogen peroxide hydrolyse in desiccated soil after one hour [67]. Urease activity was measured via phenol sodium-sodium hypochlorite-colorimetry, expressed as the mass ratio (mg/g) of ammonium nitrogen (NH 4 + -N) in desiccated soil (after 24 hours drying). Invertase activity was estimated with the 3,5-dinitrosalicylic acid-colorimetry and expressed as the mass ratio (mg/g) of glucose in desiccated soil (after 24 hours). Then, 2,4-dinitrophenol-colorimetry was used for nitratase, with the mass ratio (mg/g) of nitrate nitrogen (NO 2 − -N) in desiccated soil after 24 hours [68]. The disodium phenylphosphate-colorimetry was used for ACP, expressed as the mass ratio (mg/g) of phenol released from soil after one hour. Finally, the 3,5-dinitrosalicylic acid-colorimetry was performed for cellulase, expressed as the mass ratio (mg/g) of glucose in desiccated soil (after 72 hours).
Potentiometry (GB 7859 − 1987) was used to measure soil pH with 1:5 (w/v) soil-to-water ratios [69]. The semi-micro Kjeldahl method (GB 7173 − 1987) was used to determine soil TN content, which included both nitrate-nitrogen and nitrite-nitrogen [70]. We used the Mo-Sb colorimetric method to determine TP after high-temperature melting in sodium hydroxide (GB 7852 − 1987) [71]. The ame photometry method was performed to determine TK after high-temperature melting in sodium hydroxide (GB 7854 − 1987) [72]. The determination of Kppm was also measured with the ame photometry method after the extraction of samples using ammonium acetate (GB 7856 − 1987) [73]. Colorimetry was used to determine AP after acid leaching by 0.05 mol/L HCl-0.025 mol/L1/2 H 2 SO 4 (GB 7853 − 1987), which was suited to acidic soil [74]. The soil OM was determined by potassium dichromate titrimetric (GB 9834 − 1988) [75]. Finally, we used the Baermann funnel method for 24 hours to separate nematodes, which were stored in a 5% formaldehyde solution until counting under the anatomical lens (Leica DM4000 B) was completed [76].

Statistical Analysis
Limitations in sampling resulted in a lack of replication in our experiment, thus strongly diminishing the power of statistical analyses. However, our goal is to provide a rst network of interactions as a working hypothesis for future studies to test. Thus, although questionable, we believe our approach represents a rst step in describing how edaphic factors are intertwined and affect O. sinensis.
We rst identi ed the soil variables that showed a signi cant difference among the four groups based on the result of one-way analysis of variance (ANOVA) and Pearson correlation analysis. The ANOVA was used for testing the variation among each group and the Pearson correlation analysis was applied for investigating the correlation relationship among each soil variable and the three O. sinensis characters (Table 1 and Table 2), both of them were calculated using SPSS. Then we used a principal component analysis (PCA) [77] to reduce the dimension on multiple targets with characteristic indexes for these selected soil variables and O. sinensis containing the three characters, which means, multiple nutrient variables were aggregated into one, the same for soil enzyme activity and O. sinensis. Meanwhile, we applied Bartlett's test of sphericity on these variables. The results indicated that the variables were suitable for factor analysis. Therefore, the enzyme indicators that contained catalase, urease, NR and ACP were simpli ed as one factor, as were the nutrients that contained TN, TP, OM, AP, Kppm and the characteristic targets of O. sinensis that contained the length of polypide, stroma and FW.
The variable indicators including soil enzymes, nutrients, moisture, pH, nematode populations and the characteristic targets of O. sinensis were used in the following SEM analysis, all the four groups (including control group) are included. Before the SEM, we conducted an ANOVA analysis to check the relationship between these indicators, especially the effects of environmental indicators on O. sinensis. According to the ANOVA calculated for each variable beforehand (Supplementary Table S3, see Additional le 1), only the nematode populations showed a non-signi cant effect, but in consideration of the theoretical relationship among nematode, enzyme activity and O. sinensis, we still retained all the data points in the analysis. Structural equation modelling [78,79] was conducted to quantify direct and indirect effects between environmental variables and the quality of O. sinensis. The initial SEM was built on the basis of prior theoretical knowledge, which means the relationship between variables needed to be set up in advance. Structural equation modelling was propitious to show the theoretical causal relationships between inter-correlated variables; thus, it was used to evaluate potential multivariate relationships [80]. Generally, the χ 2 test indicated whether the model was suitable for tting [80]. We used IBM SPSS Amos 22.0 (Amos Development Corporation, Crawfordville, FL, USA) to operate the SEM model. According to the modi cation indices, we added new paths to improve the model adaptability, which could also be legitimately explained. The modi ed model adequately t these data (χ 2 = 4.249, p = 0.643), and the other model t information are shown in Supplementary