Environmental Drivers Override Host Phylogeny in a Locoweed–Endophyte Symbiosis

Abstract

Plant endophytes, often termed the “second genome”, critically shape host adaptability. However, the complexity of their interactions, regulated by microbial traits, host species, and environment, has limited both our understanding of symbiosis and the application of beneficial endophytes. The symbiosis between locoweeds (Oxytropis and Astragalus species) and the endophyte Alternaria sect. Undifilum, which produces the neurotoxin swainsonine, serves as an ideal model for investigating these relationships. Through extensive national surveys (2021–2023) across China’s major locoweed habitats, combining field sampling with cultivation, molecular, quantitative, and modeling approaches, a central question emerged: To what extent are the distribution and function of this symbiosis shaped by the contemporary environment versus host evolutionary history? The results showed that: (1) Among 32 surveyed species of Oxytropis, Astragalus, and Sphaerophysa, the endophyte Alternaria sect. Undifilum colonized 11 species. In colonized plants, endophyte loads ranged from 0.02 to 58.87 pg/ng total DNA, and swainsonine concentrations varied from 0.00003% to 1.00%. (2) Environmental factors, rather than host phylogeny, were the key driver governing the geographical distribution and expression of the symbiosis. (3) Low temperature and drought stress regulated the symbiotic relationship and chemical defense through both direct effects on the symbionts and indirect pathways involving grazing pressure. This study demonstrates that the environment is the core force dominating the geographical pattern and functional expression of the locoweed–endophyte symbiosis at ecological scales. These findings provide new perspectives for understanding the general principles of plant–endophyte symbiosis and establish a scientific foundation for predicting and utilizing endophyte resources in changing environments.

1. Introduction

The diverse interactions between plants and microorganisms, ranging from mutualism to antagonism, are key forces driving the functioning and evolution of terrestrial ecosystems [1,2]. As core members of the plant microbiome, endophytes shape host adaptability by providing nutrients [3], enhancing stress resistance [4], and synthesizing defensive compounds [5,6]. These benefits make them a potential resource for sustainable agriculture [7]. However, as the “second genome” of plants [8], endophytes’ interaction patterns with hosts are more complex compared to other microorganisms (such as arbuscular mycorrhizal fungi) [9], which limits their stable application in agricultural production [10]. Therefore, elucidating the ecological and evolutionary drivers of plant–endophyte symbiosis is critical for both theory and practice [11].

Currently, understanding of plant–endophyte symbiotic mechanisms largely stems from a few symbiotic systems, such as grass–Clavicipitaceae endophyte (C-endophyte) symbiosis [12]. This symbiosis exhibits high host specificity, enhances plant stress resistance, produces toxins that deter herbivores [5], and its functional benefits are jointly regulated by host species and environmental conditions [13,14]. Similarly to C-endophytes, the Alternaria sect. Undifilum endophyte, which is widely distributed on locoweeds (primarily Oxytropis and Astragalus species) across Asia, Europe, and the Americas, has a narrow host range [15,16,17]. Alternaria sect. Undifilum endophyte was first isolated from O. sericea and O. lambertii in the United States and confirmed to be toxigenic [18]. Subsequently, seven species within this group have been reported from locoweeds of the genera Astragalus, Oxytropis, and Swainsona in the United States, Australia and New Zealand, including Al. oxytropis, Al. fulvum, Al. cinereum [19], as well as Al. wetherii, Al. pubentissima, Al. pubentissimoides, and Al. swainsonii [20]. However, in contrast to this relatively high global diversity, available evidence indicates that the endophytic fungi carried by major locoweed species in China are comparatively species-poor and dominated by Al. oxytropis [21]. This symbiosis is of paramount ecological and economic importance, as locoweeds infested with these endophytes are among the most devastating toxic plants on the rangelands of both China and the United States, causing massive and persistent livestock losses [11,13,14,15,16,17,18]. This is because Alternaria sect. Undifilum synthesizes swainsonine, a toxin that causes neurological poisoning in livestock [22,23].

Yet, unlike the relatively well-defined C-endophyte system, the ecological drivers of the locoweed-Alternaria symbiosis remain widely debated. The traditional view of mutualistic symbiosis between endophytes and hosts is challenged in the case of Alternaria sect. Undifilum. Multiple studies have shown that Alternaria sect. Undifilum does not significantly promote host plant growth or drought resistance [24,25,26]. Instead, due to the delayed onset of swainsonine poisoning symptoms, it may lead to sustained grazing by livestock, thereby increasing the risk of herbivory for the plants [27,28,29]. However, there is also evidence suggesting potential benefits in enhancing plant disease resistance and promoting lateral root development [21,30]. These conflicting results strongly indicate that the outcomes of locoweed–endophyte interactions are not fixed but are jointly regulated by the host and environment [26]. Consequently, the locoweed-Alternaria symbiosis presents a powerful model system to dissect a central question in microbial ecology: to what extent are the distribution and functional expression of a specialized symbiosis governed by contemporary environmental filters versus historical evolutionary constraints (host phylogeny)? Resolving this question would advance a fundamental understanding in fungal symbiosis ecology, offering broader insights into how environment and evolutionary history jointly shape the dynamics of plant–fungal associations. This question is particularly salient given the significant geographic variation in toxicity, coupled with the high diversity of Chinese locoweeds and the specificity of their dominant endophyte, Al. oxytropis. Therefore, clarifying the core drivers governing the establishment and maintenance of this symbiotic relationship has become key to understanding its ecological function and evolutionary logic.

To address these questions, we conducted large-scale surveys across major locoweed distribution areas in China from 2021 to 2023. Using field ecology, microbial cultivation, molecular quantification, and environmental modeling, this study aimed to: (1) systematically determine the host range and geographical distribution of the swainsonine-producing endophyte in China; (2) quantify variation in endophyte colonization and swainsonine accumulation, and evaluate the relative roles of host phylogeny versus environmental factors; and (3) identify the key ecological drivers, particularly abiotic stressors, that influence this plant–fungus relationship. Our work provides a comprehensive ecological framework for understanding the maintenance and drivers of the locoweed–endophyte symbiosis, offering new insights into the context-dependent nature of plant–microbe interactions.

2. Materials and Methods

2.1. Plant Materials

Field surveys for Oxytropis, Astragalus, and Sphaerophysa plants were conducted during their flowering period (June to August) from 2021 to 2023 across major locoweed distribution regions in China, including Tibet, Xinjiang, Gansu, Ningxia, Inner Mongolia, and Qinghai (Figure 1A). The above-ground parts of 10 to 20 plants were collected from each habitat.

2.2. Plant Species Identification

2.2.1. Morphological Identification

Morphological identification was performed by Dr. Wu Yuhu, a researcher at the Northwest Institute of Plateau Biology, Chinese Academy of Sciences. The voucher specimens are deposited in the Herbarium of Lanzhou University, with the specimen numbers and collection information listed in Supplementary Table S1.

2.2.2. Molecular Identification

DNA extraction: Plant materials described in Section 2.1 (using three plants per habitat as replicates) were rinsed with tap water to remove excess soil and dried with absorbent paper. Approximately 50–100 mg of fresh tissue was placed in a 2 mL centrifuge tube with a sterilized steel bead and ground for 4–7 min using a grinder. Genomic DNA was extracted using the Ezup Column Plant Genomic DNA Extraction Kit (Sangon Biotech, Shanghai, China) according to the manufacturer’s instructions. The extracted DNA was amplified via PCR in a 25 µL reaction mixture (1 µL DNA template, 1 µL each of forward and reverse primer, 9.5 µL ddH₂O, and 12.5 µL Mix enzyme) using a single primer pair: ITS1-f/ITS4-r (f′-TCCGTAGGTGAACCTGCGG, r’-TCCTCCGCTTATTGATATGC) [31]. PCR reaction conditions were initial denaturation at 95 °C for 5 min, cycle denaturation at 95 °C for 45 s, annealing at 58.6 °C for 1 min, extension at 72 °C for 30 s, 29 cycles, and final extension at 72 °C for 5 min.

Electrophoresis: PCR products were checked by agarose gel electrophoresis at 120 V for 25 min using a Bio-Rad PowerPac Basic system (Bio-Rad, Hercules, CA, USA) to confirm successful amplification. Products showing clear bands were sent to Sangon Biotech (Shanghai) for sequencing.

Phylogenetic tree construction: The obtained sequences were assembled using DNAstar v11.1 software. All newly generated sequences in this study have been uploaded to NCBI, with detailed information provided in Supplementary Table S1. The assembled sequences were aligned with those from the NCBI database to construct a phylogenetic tree. The aligned locoweed sequences were imported into MEGA X software, processed using the integrated ClustalX tool for alignment, and trimmed at both ends to remove redundant regions while retaining intron sequences. A Maximum Likelihood (ML) tree was constructed with 1000 bootstrap replicates; the K2P model was selected as the nucleotide substitution model.

2.3. Detection of Endophyte Alternaria Section Undifilum in Oxytropis, Astragalus, and Sphaerophysa Plants

2.3.1. Detection of Endophyte Alternaria Section Undifilum Using Isolation Method

From each habitat collected in Section 2.1, 10–20 plants of Oxytropis/Astragalus/Sphaerophysa were selected, and the presence of endophyte was detected using the tissue isolation method. Ten health stem segments were taken from each plant, rinsed under running water, surface-sterilized with 75% ethanol for 30 s, followed by sterilization with 1% sodium hypochlorite for 3–5 min, and then rinsed three times with sterile water. The last rinse of sterile water was spread on a PDA plate to observe whether any colonies grow. If no colonies appear, it indicates that the subsequent tissue isolation would yield endophytes. The surface moisture was absorbed with sterilized filter paper, and the segments were cut into small pieces and vertically inserted into water agar (WA) medium for cultivation. After 24–48 h, the cut was observed under a stereomicroscope. The presence of wavy, short, white hyphae (characteristic of the Alternaria section Undifilum hyphae) [32] at the cut indicated the presence of endophyte, recorded as E+. The absence of such hyphae indicated no endophytic fungi, recorded as E-. The endophyte isolation rate was calculated as follows:

$$\mathrm{E}\mathrm{n}\mathrm{d}\mathrm{o}\mathrm{p}\mathrm{h}\mathrm{y}\mathrm{t}\mathrm{e} \mathrm{c}\mathrm{a}\mathrm{r}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{g}\mathrm{e} \mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e} \mathrm{b}\mathrm{y} \mathrm{i}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\left(\%\right)={\displaystyle \frac{\mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{p}\mathrm{l}\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{s} \mathrm{w}\mathrm{i}\mathrm{t}\mathrm{h} \mathrm{i}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{d} \mathrm{e}\mathrm{n}\mathrm{d}\mathrm{o}\mathrm{p}\mathrm{h}\mathrm{y}\mathrm{t}\mathrm{e}}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{p}\mathrm{l}\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{s} \mathrm{i}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{d}}}\times 100$$

2.3.2. Detection of Endophyte Alternaria Section Undifilum Using Specific Primers Method

Plant DNA extracted in Section 2.2.2 was amplified by PCR using the specific primer Omtssu [26,33]. This primer set exhibits high specificity for Alternaria sect. Undifilum, as it reliably amplifies target DNA from the endophyte while showing no cross-reactivity with host plant DNA or other common fungi [34]. The PCR reaction was performed in a 25 μL system (1 μL DNA template, 1 μL each of forward and reverse primers, 9.5 μL ddH₂O, and 12.5 μL Mix enzyme). The Omtssu primer sequences were OmtssuF (CATAGAAAAAAAAATAAACAAACTG) and OmtssuR (TGTCTGCCCAGGTTACGG). The PCR conditions for OmtssuF_OmtssuR were: initial denaturation at 95 °C for 10 min, followed by 30 cycles of 94 °C for 45 s, 52 °C for 45 s, and 72 °C for 45 s, with a final extension at 72 °C for 10 min, ending at 4 °C. After the PCR reaction, the products were analyzed by agarose gel electrophoresis to observe the target bands, and the endophyte carriage rate was recorded through photography:

$$\mathrm{E}\mathrm{n}\mathrm{d}\mathrm{o}\mathrm{p}\mathrm{h}\mathrm{y}\mathrm{t}\mathrm{e} \mathrm{c}\mathrm{a}\mathrm{r}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{g}\mathrm{e} \mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e} \mathrm{b}\mathrm{y} \mathrm{s}\mathrm{p}\mathrm{e}\mathrm{c}\mathrm{i}\mathrm{f}\mathrm{i}\mathrm{c} \mathrm{p}\mathrm{r}\mathrm{i}\mathrm{m}\mathrm{e}\mathrm{r}\mathrm{s}\left(\%\right)={\displaystyle \frac{\mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{p}\mathrm{l}\mathrm{a}\mathrm{n}\mathrm{t} \mathrm{P}\mathrm{C}\mathrm{R} \mathrm{p}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{s} \mathrm{w}\mathrm{i}\mathrm{t}\mathrm{h} \mathrm{b}\mathrm{a}\mathrm{n}\mathrm{d}\mathrm{s}}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{p}\mathrm{l}\mathrm{a}\mathrm{n}\mathrm{t} \mathrm{P}\mathrm{C}\mathrm{R} \mathrm{p}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{s}}}\times 100$$

2.4. Quantitative Analysis of Endophyte Alternaria Section Undifilum by Quantitative Real-Time PCR (qPCR) in Tested Plants

2.4.1. Plant DNA Extraction

From each habitat collected in Section 2.1, five plants were selected. Dry plant samples (25 mg each) were weighed equally, and plant DNA was extracted (method same as Section 2.2.2). The DNA concentration was detected using a NanoDrop 2000 (micro-UV spectrophotometer, Thermo Fisher Scientific, Waltham, MA, USA). Qualified samples were stored at −20 °C for later use.

2.4.2. Preparation of Standard Curve for Endophyte Alternaria Section Undifilum Expression

DNA extracted from the hyphae of the endophyte Alternaria Section Undifilum isolated from O. ochrocephala plants was used as a template. The DNA was serially diluted 10-fold with ddH₂O to concentrations ranging from 2.3 to 2.3 × 10⁻⁸ ng/μL. Each template concentration was set up with four parallel replicates. Additionally, the PCR reaction included one NTC reaction, where double-distilled water replaced DNA, to confirm the absence of reagent contamination in the reaction system. The logarithm of the initial template DNA concentration was used as the X-axis, and the cycle number (Cq value) when the fluorescence detection signal in the PCR reaction tube reached the machine detection threshold was used as the Y-axis. A standard curve was constructed to obtain the linear regression equation (Supplementary Figure S1).

2.4.3. Quantitative Detection of Endophyte Alternaria Section Undifilum in Plant Samples

The stored plant sample DNA (uniformly diluted to 50 ng/μL) was used as a template, and qPCR amplification was performed using primers SwnK-q-S/SwnK-q-AS (SwnK-q-S: ACATATGCCTTCGAGCGAGA, SwnK-q-AS: TACCAAATGGAGGTCGCACT) (three technical replicates were set for each sample) by Guan et al. [21]’s method. The positive control was the DNA of the endophytic fungal strain Alternaria Section Undifilum, and sterile water replaced the DNA template as a negative control to confirm the absence of contamination in the system. The Cq values of the samples obtained from qPCR detection were substituted into the standard curve and converted into the endophytic fungal content in the plant sample DNA (pg/ng) through calculation.

2.5. Quantitative Analysis of Swainsonine Content in Plants by Ultra Performance Liquid Chromatography–Tandem Mass Spectrometry (UPLC-MS/MS)

2.5.1. Preparation of Swainsonine Standard

0.1 mg of swainsonine standard (purity ≥ 98%, Sigma-Aldrich, St. Louis, MO, USA) was dissolved in 1 mL of water to prepare a SW standard stock solution. 10 μL of the stock solution was mixed with 1 mL of water to prepare a standard solution with a concentration of 1 × 10³ μg/mL. The solution was then serially diluted with water to prepare a series of solutions with mass concentrations of 200, 100, 50, 25, 12.5, 6.25, 3.125, 1.5625, 0.78125, 0.390625, and 0.1953125 μg/L. The standard solutions were stored at −30 °C for later use.

2.5.2. Extraction of Swainsonine from Tested Plant Samples

Swainsonine was extracted from the specimens collected in Section 2.1 using the method described by Gardner and Cook [35]. Five plants from each habitat were taken as replicates. Using an analytical balance, 20 mg of dried ground plant material was weighed into a 2 mL centrifuge tube, and 1 mL of 2% acetic acid was added for oscillating extraction for 16 h (overnight). During this process, the samples were thoroughly mixed with the acetic acid solution. After extraction, the samples were centrifuged for 5 min, and the supernatant was transferred to a new tube, capped, and stored at −20 °C. 500 μL of the supernatant was aspirated, diluted to 5 mL with H₂O, and filtered through a microporous filter into a 3 mL autosampler vial.

The UPLC-MS/MS chromatographic conditions were (a) column: CAPCELLPAK C18 (100 mm × 2.1 mm, 2 μm); (b) mobile phase: A: 5% methanol in water, B: 20 mM ammonium acetate in water; (c) gradient elution: 0–0.5 min, 90% A; 0.5–3 min, 90% A → 5% A; (d) flow rate: 0.3 mL/min; (e) column temperature: 30 °C; (f) injection volume: 2 μL.

The UPLC-MS/MS mass conditions were: (a) ion source: ESI+; (b) scan mode: positive ion mode, multiple reaction monitoring (MRM); (c) ion source voltage: 5.5 kV; (d) ion source temperature: 550 °C; (e) curtain gas pressure: 241.3 kPa; (f) collision gas pressure: medium; (g) nebulizer gas pressure (GS1) and auxiliary heating gas pressure (GS2): 344.7 kPa. Quantitative ion pairs for swainsonine: m/z 174.3/156.0, collision energy 19 V; qualitative ion pairs: m/z 174.3/138.2, collision energy 26 V; optimized declustering potential (DP): 60 V.

2.6. Statistical Analyses

2.6.1. Detection of Host Phylogenetic Signal in Endophytic Fungal Colonization and Swainsonine Concentration

Host Phylogenetic Signal in Species Level: A plant species was scored as 1 (host) if the locoweed endophyte was detected in at least one habitat, otherwise 0 (non-host). The average swainsonine content across habitats was calculated for each species and categorized as: 1 (≥0.1%), 2 (>0% but <0.1%), or 3 (undetectable). Using the ITS sequences from Section 2.2.2, Blomberg’s K and Pagel’s λ for both traits were computed with the R package phytools [36].

Host Phylogenetic Signal in Population Level: A population was classified as 1 (host) if it contained endophyte-colonized plants, otherwise 2 (non-host). The average swainsonine content per population was categorized as above (1, 2, or 3). The following calculation is consistent with that at the species level.

2.6.2. Correlation Between Locoweed Endophytic Colonization and Environmental Heterogeneity

Literature Data Collection: To complement our systematic survey of endophytes and swainsonine in Chinese Oxytropis, Astragalus, and Sphaerophysa species, we compiled additional records from the literature. Host species identified in our survey were included, but O. giraldii, A. pseudoscaberrimus, and S. salsula were excluded due to limited new data. A CNKI search on 5 June 2025, using key locoweed species names yielded 892 results. We extracted species, location, coordinates, endophyte status, and swainsonine content from 100 qualifying papers (see Supplementary Table S2).

Environmental Data Processing: Distribution points were cleaned in R (v4.3) using CoordinateCleaner and spThin. We obtained 19 bioclimatic variables (WorldClim v2.0, https://www.worldclim.org/), elevation, and mean grazing intensity (1980–2020) for analysis [37]. To address multicollinearity among the 21 initial variables, we performed Pearson correlation analysis (psych package) and iteratively removed less important variables (assessed by andomForest) where |r| ≥ 0.9, resulting in 12 retained variables (

Table 1. Information on the 21 environmental variables included in the study.

Abbreviation	Full Name
bio1 *	Annual mean temperature (°C) *
bio2	Mean diurnal temperature range (°C)
bio3 *	Isothermality *
bio4	Temperature seasonality
bio5 *	Maximum temperature of warmest month (°C) *
bio6 *	Minimum temperature of coldest month (°C) *
bio7 *	Temperature annual range (°C) *
bio8	Mean temperature of wettest quarter (°C)
bio9	Mean temperature of driest quarter (°C)
bio10	Mean temperature of warmest quarter (°C)
bio11 *	Mean temperature of coldest quarter (°C) *
bio12 *	Annual precipitation (mm) *
bio13 *	Precipitation of wettest month (mm) *
bio14	Precipitation of driest month (mm)
bio15	Precipitation seasonality (mm)
bio16	Precipitation of wettest quarter (mm)
bio17 *	Precipitation of driest quarter (mm) *
bio18	Precipitation of warmest quarter (mm)
bio19 *	Precipitation of coldest quarter (mm) *
elevation *	elevation (m) *
graze *	Grazing intensity (SU·ha−1) *

* indicates the variables that were ultimately included in the model after de-linearization.

).

Environmental Factor PCA: To assess environmental differences among locoweed species, we performed Principal Component Analysis (PCA) with PERMANOVA using the R package vegan (v4.3.3) [38]. Environmental variables were standardized prior to PCA to visualize distribution patterns. PERMANOVA tested the significance of overall environmental differences among species groups.

Environmental Heterogeneity Quantification: To avoid confounding effects from interspecies habitat preferences, we analyzed within-species relationships between environmental heterogeneity and endophyte traits. For each species (excluding O. glacialis due to its small number of habitats, n = 3), we calculated an environmental heterogeneity index from the PCA results [39]. This index, derived as the multivariate standard deviation (square root of summed variances across PC1–PC5), represents species dispersion in multidimensional environmental space. Calculations used dplyr in R.

Variation Analysis of endophyte colonization and swainsonine levels: We calculated coefficients of variation for endophyte colonization rate and swainsonine levels per species, using established classification criteria (Section 2.6) [40].

Heterogeneity Correlation: Spearman correlation analysis assessed relationships between species’ environmental heterogeneity indices and coefficients of variation for both endophyte colonization and swainsonine levels [41]. All analyses were conducted in R v4.3.2.

2.6.3. Correlation Between Ecological Factors and Endophyte Colonization/Swainsonine Concentration

To systematically assess ecological drivers of the locoweed–endophyte symbiosis, we analyzed correlations at two scales: (1) overall correlations ignoring species taxonomy to identify broad patterns, and (2) species-specific correlations to detect unique responses.

Overall Correlation: We performed Spearman correlation analysis between all ecological factors and both endophyte colonization rate and swainsonine concentration across all samples [42]. The relative importance of factors was compared using standardized regression coefficients. Analyses used the stats package in R.

Species-Specific Correlation: Due to limited variation in endophyte colonization status, Spearman correlation analysis was performed specifically on swainsonine levels using the stats package in R. This analysis was conducted separately for key species (O. ochrocephala, O. glabra, and O. falcata) that met the criteria of sufficient sample size (n > 10) and substantial trait variation.

2.6.4. Partial Least Squares Path Modeling (PLS-PM)

We constructed a PLS-PM to quantify causal relationships among host species, key ecological factors, endophyte colonization, and swainsonine concentration [43]. The model, built in R (v4.3.2) using the plspm, psych, and lavaan packages, utilized the pre-screened dataset from Section 2.6.2. We defined six latent variables with corresponding manifest variables: species; temperature: bio5, bio6, bio11; precipitation: bio12, bio17; endophyte (colonization status); swainsonine (concentration level); grazing Intensity.

The model was fitted with plspm(). Model quality was assessed by the Goodness-of-Fit index and path coefficient significance (tested via 1000 bootstrap iterations). Standardized path coefficients were visualized using innerplot().

3. Results

3.1. Distribution and Identification of Chinese Oxytropis, Astragalus, and Sphaerophysa Species

3.1.1. Plant Distribution and Morphological Identification

To determine the national distribution pattern and host range of the swainsonine-producing endophyte, an extensive survey was conducted across the major distribution areas of Oxytropis, Astragalus, and Sphaerophysa species in China (Figure 1A). The survey revealed 32 legume species across 115 habitats within six provinces/autonomous regions: the Tibet, Xinjiang, Qinghai, Gansu, Inner Mongolia, and Ningxia. Morphological identification confirmed these included 17 species of Oxytropis, 14 species of Astragalus, and 1 species of Sphaerophysa (Figure 1B–AF).

3.1.2. Molecular Identification of Plants

A maximum likelihood (ML) phylogenetic tree constructed based on the ITS sequences of the collected Oxytropis, Astragalus, and Sphaerophysa plants is shown in Figure 2. The results revealed that all tested samples were clearly divided into three highly supported (bootstrap values > 90%), monophyletic clades corresponding to the genera Oxytropis, Astragalus, and Sphaerophysa, supporting their distinct generic-level taxonomic status. The phylogenetic relationships indicated that Oxytropis and Astragalus are closely related, forming sister groups, while Sphaerophysa was placed outside these two genera and showed closer affinity to the genus Swainsona.

Regarding the infrageneric phylogenetic structure, the genus Astragalus exhibited high interspecific resolution. All tested species (e.g., A. variabilis, A. scaberrimus, etc.) formed independent, highly supported monophyletic branches, indicating that the ITS sequences effectively distinguished different species within Astragalus. Furthermore, the sequences could even differentiate subspecies, as seen between A. polycladus and its variety A. polycladus var. parvicarpus, suggesting significant genetic differentiation. In contrast, the genus Oxytropis showed lower interspecific resolution. However, its subgeneric phylogenetic structure was very clear. The ITS sequences effectively distinguished the subgenera Oxytropis (Subgen. Oxytropis), Orobia (Subgen. Orobia), and Tragacanthoxytropis (Subgen. Tragacanthoxytropis), with each subgenus forming an independent, highly supported clade.

In summary, the ITS-based phylogenetic results were consistent with the morphological identifications, confirming the accuracy of the plant species identification.

3.2. Qualitative Detection of Endophytic Fungus Alternaria Sect. Undifilum in Plants

Using tissue isolation and a specific primer detection method, the presence of Alternaria Sect. Undifilum endophytic fungi in plants from Northwest China was determined. The isolation results showed that among the plants collected from 115 habitats, the endophytic fungus Alternaria Sect. Undifilum was isolated from 28 habitats (Supplementary Figure S2,

Table 2. Infection Rate of Alternaria Sect. Undifilum Endophytic Fungi, Endophyte Content, and Swainsonine Content in Tested Plants.

Species	Number	Plant Carrier Rate (%)		Endophyte (pg/ng Total DNA)		Plant Swainsonine (%)
		Isolation Detection	Specific Primer Detection	Mean	RSE (%)	Mean	RSE (%)
O. ochrocephala	MHLZU6818	100.00	100.00	5.32	28.02	0.17	25.97
	MHLZU6549	100.00	100.00	0.07	96.70	0.03	76.23
	MHLZU6800	100.00	100.00	5.18	25.98	0.42	25.47
	MHLZU6801	75.00	80.00	6.03	53.05	0.61	27.72
	MHLZU6799	92.86	100.00	9.72	34.19	0.52	31.40
	MHLZU6823	84.00	100.00	10.61	31.68	0.61	18.73
	MHLZU6798	75.00	100.00	2.16	26.07	0.07	74.01
	MHLZU6824	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6825	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6826	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6827	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6828	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6829	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6830	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6852	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6532	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
O. falcata	MHLZU6805	100.00	100.00	2.53	6.87	0.38	9.53
	MHLZU6817	100.00	100.00	19.67	15.25	1.00	19.39
	MHLZU6804	100.00	100.00	2.81	16.77	0.28	14.39
	MHLZU6547	90.00	100.00	3.95	38.43	0.40	18.14
	MHLZU6802	75.00	80.00	6.21	27.65	0.04	26.25
O. glabra	MHLZU6814	80.00	100.00	38.77	29.99	0.51	17.91
	MHLZU6815	100.00	100.00	19.20	-	0.26	-
	MHLZU6831	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
O. glacialis	MHLZU6820	100	100	33.14	26.42	0.75	26.98
O. deflexa	MHLZU6806	100	100	29.76	6.04	0.61	8.93
	MHLZU6819	90	100	34.86	37.60	0.54	32.62
O. sericopetala	MHLZU6822	90	100	58.87	15.26	0.61	14.28
O. giraldii	MHLZU6781	50	66.6	0.05	71.13	0.00037	83.93
	MHLZU6551	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6832	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6834	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6835	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6857	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6536	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
A. pseudoscaberrimus	MHLZU6812	40	100	0.02	2.52	0.00003	10.69
A. variabilis	MHLZU6807	100	100	55.60	19.61	0.56	19.11
	MHLZU6808	100	100	37.45	27.62	0.32	25.46
	MHLZU6809	100	100	44.43	24.64	0.49	21.60
	MHLZU6810	100	100	39.59	19.89	0.39	19.92
	MHLZU6811	80	80	10.71	47.93	0.11	50.54
	MHLZU6813	70	100	2.60	47.97	0.03	47.67
A. strictus	MHLZU6821	100	100	8.12	16.39	0.24	15.97
	MHLZU6855	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
	MHLZU6856	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
S. salsula	MHLZU6816	100	100	23.46	24.54	0.3731	28.63

Note: This table only displays plant species in which the endophytic fungus Alternaria Sect. Undifilum was detected. “n.d.” indicates that the endophyte and/or swainsonine was not detected in the corresponding sample.

). No endophytic fungi were isolated from the remaining plants. This is the first time Alternaria Sect. Undifilum endophytes have been isolated from A. pseudoscaberrimus and S. salsula.

Furthermore, the infection rate detected by specific primers was 0–7.14% higher than that determined by the isolation method (Supplementary Figure S3, Table 2). Some plants from which no endophytes were isolated could be detected using specific primers, indicating that specific primers are more sensitive than the isolation method. In subsequent sections, plants from which Alternaria Sect. Undifilum endophytes were isolated would be referred to as “locoweeds”.

3.3. Quantitative Analysis of Endophytic Fungus Alternaria Sect. Undifilum in Locoweeds

The qPCR results indicated differences in fungal content among different species (Table 2). Species with relatively high endophyte content included O. sericopetala, A. variabilis, O. glabra, O. glacialis, O. deflexa, and S. salsula, while species with lower content included A. pseudoscaberrimus and O. giraldii. There was significant variation in fungal content among different individuals of the same species. Additionally, the measurement precision varied across samples. The residual standard error (RSE) for each sample ranged from 2.52% to 96.70%, with most samples having RSE values between 15% and 50%.

3.4. Swainsonine Concentration Levels in Locoweeds

This study employed the multiple reaction monitoring (MRM) mode of an ultra-high performance liquid chromatography–tandem mass spectrometry (UPLC-MS/MS) system to analyze swainsonine in plants (Figure 3B). The MRM chromatograms showed that plants not carrying the endophytic fungus exhibited no distinct chromatographic peaks at the target compound’s elution position, with a stable baseline and effective exclusion of plant matrix interference (Figure 3B). In contrast, plant samples carrying the endophytic fungus displayed distinct characteristic swainsonine chromatographic peaks at retention times identical to those of the standard (Figure 3C–M), confirming the presence of swainsonine in these plants.

Quantitative results (Table 2) revealed substantial variations in swainsonine content both among different species and among different populations of the same species, ranging from undetectable levels (0%) to as high as 1.0038%. At the species level, plants of the genus Oxytropis generally exhibited higher toxin contents, with significant intraspecific variation. For example, swainsonine content varied greatly among different populations of O. ochrocephala, from completely toxin-free (e.g., eight populations including MHLZU6532) to high levels (e.g., MHLZU6801, 0.6123%). O. falcata showed the highest toxin content among all species (MHLZU6817, 1.0038%), and all its tested populations tested positive for the toxin, with relatively low standard errors and residual standard errors, indicating stable colonization by toxin-producing endophytes and consistent toxin production. Notably, in O. giraldii, only the MHLZU6781 population contained trace amounts of swainsonine (0.00037%), while the other five populations tested negative.

Furthermore, swainsonine levels in plants of the genus Astragalus also varied considerably. A. variabilis was the primary toxin-producing species within this genus, with concentrations ranging from 0.0283% to 0.5619% across different populations. In A. strictus, only the MHLZU6821 population tested positive for swainsonine (0.2449%), while other populations tested negative.

3.5. Relationship Between Endophyte Alternaria Sect. Undifilum and Swainsonine Content with Host Geographical Distribution, Species, and Phylogeny

Significant differences (p < 0.05) were observed in the contents of both Alternaria Sect. Undifilum and swainsonine among different host species and among plants from different geographical locations (Figure 4A–D). This suggests that the endophyte content and swainsonine levels in locoweed may be influenced by both the host plant species and the geographic location.

This study examined the phylogenetic signals of endophytic colonization and swainsonine concentration in locoweeds at the species and population levels using Blomberg’s K (Figure 4E) and Pagel’s λ (Figure 4F). The results revealed that at the species level, the phylogenetic signal for endophyte colonization was Pagel’s λ = 0.0001 with p ≥ 0.05, indicating an almost absent phylogenetic signal. This was further supported by a Blomberg’s K = 0.0589 (p ≥ 0.05), suggesting no significant phylogenetic signal for endophyte colonization at the species level. Similarly, the phylogenetic signal for swainsonine was negligible at the species level, also indicating no significant phylogenetic conservatism in swainsonine concentration among species. In contrast, at the population level, both endophyte colonization and swainsonine concentration exhibited significant phylogenetic signals (p < 0.01), indicating a high degree of phylogenetic conservatism. This significant phylogenetic conservatism among populations suggests that different populations of the same species maintain similar colonization status and swainsonine concentrations.

3.6. Significant Correlation Between Environmental Heterogeneity and Variation Coefficients of Endophyte and Swainsonine in Locoweeds

To investigate whether environmental factors influence endophyte colonization and swainsonine production, the coefficients of variation (CV) for environmental heterogeneity, endophyte colonization, and swainsonine content were calculated for each host plant (Figure 5A–F). Given the significant habitat disparities among species (Figure 5, p = 0.001), subsequent analyses were conducted separately for each host plant to avoid within-group heterogeneity arising from differences in habitat preferences. Correlation analysis revealed a significant positive relationship between environmental heterogeneity and both endophytic fungal colonization (p = 0.03) and swainsonine concentration (p = 0.01, Figure 5E,F). These results suggest that variation in the host plant’s habitat may significantly influence the colonization of endophyte and the accumulation of swainsonine in locoweeds.

3.7. Environmental Factors Influence Swainsonine Production Through Direct Effects and Indirect Promotion of Endophyte Colonization

To investigate the environmental factors influencing the colonization of Alternaria Sect. Undifilum and swainsonine production in locoweeds, the correlations between various environmental factors and both endophytic colonization and swainsonine content was further analyzed (Figure 6). Overall, endophytic colonization showed significant positive correlations with mean annual temperature (Bio1), annual precipitation (Bio12), and precipitation of the driest quarter (Bio17), while it was significantly negatively correlated with maximum temperature of the warmest month (Bio5), minimum temperature of the coldest month (Bio6), mean temperature of the coldest quarter (Bio11), precipitation of the wettest month (Bio13), elevation, and grazing intensity (Figure 6A). Swainsonine content was significantly negatively correlated with Bio5, Bio6, Bio11, and grazing intensity.

The relationship between swainsonine content and environmental factors in three major locoweed species (sample size n > 10) was further analyzed (Figure 6B). This analysis was restricted to swainsonine due to limited intraspecific variation in endophyte colonization. The results indicated that swainsonine content in these three locoweeds was significantly negatively correlated (p < 0.05) with the minimum temperature of the coldest month (Bio6) and the mean temperature of the coldest quarter (Bio11) (Figure 6B). Additionally, the relationship between swainsonine and ecological factors exhibited species-specific characteristics: swainsonine content in O. ochrocephala and O. glabra was also significantly correlated with the maximum temperature of the warmest month (Bio5) and grazing intensity, respectively.

Structural equation modeling revealed that temperature factors—including maximum temperature of the warmest month (Bio5), minimum temperature of the coldest month (Bio6), mean temperature of the coldest quarter (Bio11)—had significant negative effects on both endophytic colonization and swainsonine synthesis (Figure 6C). Precipitation factors, such as annual precipitation (Bio12) and precipitation of the driest quarter (Bio17), also showed significant negative effects on swainsonine content. Endophytic colonization significantly promoted the accumulation of swainsonine. Although locoweed species had no direct significant effect on endophytic colonization or swainsonine concentration, they significantly influenced grazing intensity (Figure 6D). Meanwhile, both temperature and swainsonine concentration also exhibited significant effects on grazing intensity.

4. Discussion

4.1. Comprehensive Survey Reveals Alternaria Sect. Undifilum Endophyte Colonization and Toxicity Levels in Chinese Locoweeds

This nationwide survey, documenting 32 plant species across 115 habitats in northwestern China, provides the most systematic distribution map to date for Oxytropis, Astragalus, and Sphaerophysa species (Figure 1). Their wide ecological adaptability, spanning alpine meadows to desert steppes [44], explains their potential to become dominant locoweeds in degraded grasslands, contributing to widespread livestock poisoning [45]. This study systematically clarifies the colonization levels of swainsonine-producing endophytes and the corresponding swainsonine in these plants [46], addressing a significant knowledge gap left by prior research focused on a few main species [47,48,49].

Our results also validate and refine the established chemotype framework for locoweeds. Following the classification of American locoweeds into Chemotype 1 (swainsonine > 0.1%) and Chemotype 2 (swainsonine < 0.01%) [50], we confirmed that O. ochrocephala, O. glabra, O. falcata, O. deflexa, O. glacialis, O. sericopetala, A. variabilis, and A. strictus consistently fall into Chemotype 1 (Table 2). This aligns with previous domestic studies. For instance, Guo [51] classified O. glacialis and O. deflexa as Chemotype 1 highly toxic locoweeds; Wang et al. [49] detected swainsonine in 92.5% of O. glabra samples from Inner Mongolia.

Notably, our extensive survey also included species for which the association with the swainsonine-producing endophyte was previously unconfirmed or ambiguous. We successfully isolated and molecularly identified the toxigenic endophytic fungus Alternaria sect. Undifilum from A. pseudoscaberrimus and S. salsula, significantly expanding the known host range of this symbiotic system. While the detailed discovery and risk assessment for S. salsula are presented in a dedicated paper [52], its confirmation here significantly expands the known host range of this symbiotic system within our ecological framework. In contrast, A. pseudoscaberrimus, while harboring the endophyte, exhibited extremely low toxin levels (<0.001%), providing a reasonable explanation for their historical absence from major locoweed classifications.

Conversely, O. kansuensis, traditionally reported as toxic [32], tested negative for both the endophyte and swainsonine across all six surveyed habitats in Qinghai and Gansu [53,54], including three geographically distinct populations from the Tianzhu region of Gansu—a key area where high infection rates were previously documented [53,54] (Supplementary Figures S3D and S4). The stark contrast between our results and these historical reports can be explained by the significant challenges in species identification. At our Tianzhu sites, O. ochrocephala and O. kansuensis co-occur in mixed stands and are morphologically highly similar. More importantly, their ITS sequences are phylogenetically proximate, complicating reliable discrimination even with molecular tools—a factor that likely contributed to Guo et al.’s [54] findings. To ensure accurate delineation in this complex habitat, we performed simultaneous and targeted sampling of both species at three distant Tianzhu locations. The results were clear and consistent: while O. ochrocephala (MHLZU6799, MHLZU6800, MHLZU6801) exhibited high endophyte colonization (80–100%), all sampled O. kansuensis individuals (MHLZU6533, MHLZU6535, MHLZU6842) tested negative. This strongly suggests that in previous studies conducted in Tianzhu, endophyte-positive O. ochrocephala plants may have been misidentified as O. kansuensis, thereby overestimating the endophyte carriage rate for the latter in this region. The endophyte-free O. kansuensis samples collected from Qinghai further indicates that the species may generally exhibit a low endophyte colonization rate in natural populations, though more systematic surveys are still needed.

Furthermore, the fact that some plants tested positive only by PCR without successful endophyte isolation can be attributed to several factors. Firstly, PCR is capable of amplifying DNA fragments from both viable and non-viable fungal cells [55], meaning that genetic material may be detected even when living fungi cannot be cultured. Secondly, the endophyte load in these samples may have been relatively low. Notably, in all plants where isolation failed but PCR was positive, the endophyte content measured by qPCR was below 0.05 pg/ng. This indicates that specific primer-based PCR offers higher detection sensitivity, while qPCR provides reliable quantitative evidence for assessing endophyte load.

In summary, these results provide, to our knowledge, the most comprehensive dataset to date on the host range and toxicity of Chinese locoweeds, resolving long-standing ambiguities and establishing a crucial foundation for future ecological and management studies.

4.2. Environmental Factors Rather than Host Phylogeny Govern Locoweed–Endophyte Symbiosis Formation

Plant–microbe symbioses typically involve both host genetics and environmental factors [56,57]. Swainsonine-producing endophytes, which form critical chemical defense partnerships with plants [58,59], serve as a powerful model for dissecting this interplay. However, the relative roles of environmental selection and host phylogeny in their distribution and evolution remain unclear [60]. This question remains a central debate, as highlighted in recent reviews which identify the ecological drivers and evolutionary context of this symbiosis as key knowledge gaps [34].

First, the molecular phylogenetic analysis based on the ITS region in this study provided insights into the genetic underpinnings of potential locoweeds (Oxytropis, Astragalus, and Sphaerophysa). As an effective molecular marker for plant population genetics, nuclear ITS has been widely used in plant identification and genetic structure analyses [61,62]. While it effectively confirmed generic boundaries and showed high resolution among Astragalus species, ITS exhibited limited power to differentiate some closely related Oxytropis species (e.g., O. ochrocephala and O. kansuensis). This pattern suggests differing evolutionary histories: Astragalus may have undergone more gradual diversification, whereas Oxytropis likely experienced rapid recent radiation or maintains higher levels of interspecific gene flow, hindering the accumulation of fixed diagnostic mutations in this nuclear marker [63,64]. This evolutionary context is crucial for interpreting current ecological patterns. Notably, Oxytropis constitutes the majority of toxic locoweeds in China [21], contrasting with the North American context where Astragalus is more prominent [46]. The inferred rapid expansion potential of Oxytropis may be a key factor enabling its dominance in degraded grasslands, linking its evolutionary trajectory to contemporary ecological impacts [45]. Future studies employing chloroplast genes in conjunction with ITS could improve species delimitation within Oxytropis [65]. Furthermore, within the ITS phylogeny, Astragalus and Oxytropis are sister genera, while Sphaerophysa is placed outside this clade and shows closer affinity to Swainsona—an endemic Australian locoweed genus [20]—thereby establishing the genetic basis for the toxic nature of Sphaerophysa.

Our integrated analysis, combining extensive field surveys with literature data, demonstrates that environmental filtering, not host phylogeny is the primary force structuring the locoweed–endophyte symbiosis. This conclusion is robustly supported by the near absence of phylogenetic signal in endophyte colonization at the species level (Pagel’s λ ≈ 0, p ≥ 0.05; Figure 4F), indicating that host phylogenetic background is poor predictors of symbiotic association. The phylogenetic tree further revealed clear differences in symbiotic status among closely related species, likely stemming from independent loss of endophytes in specific lineages [66]. Although horizontal transmission remains theoretically possible, existing evidence supports vertical transmission as the primary mode in this system [26,67]. Therefore, the absence of phylogenetic signal at the species level is more likely due to repeated independent losses of endophytes driven by environmental pressures during evolution [68]. This inference is plausible given that long-term common garden experiments have shown the endophyte confers negligible to limited fitness benefits to its hosts, supporting the idea that the symbiosis may drift towards loss in the absence of strong environmental selection for its maintenance [24].

Although the vertical transmission mechanism ensures the stability of intraspecific symbiotic relationships (population-level Pagel’s λ ≈ 1, p < 0.01) [26], significant colonization differences were still observed among different geographical populations of the same species, consistent with the findings of Davis et al. [69]. Correlation analysis indicated that this intraspecific variation showed a significant positive correlation with environmental heterogeneity (Figure 5E,F). This finding confirms that within the stable framework provided by vertical transmission, environmental conditions become the key factor regulating symbiotic expression, shaping the micro-geographic patterns of endophyte colonization. Furthermore, we observed considerable residual standard error (RSE) in endophyte load and swainsonine concentration among individuals within the same population (Table 2). This significant intra-population heterogeneity indicates that even under the same population designation, the symbiotic status of individual plants is highly differentiated. This phenomenon is consistent with the findings of Cook et al. [50] and may not only stem from individual differences in the efficiency of vertical transmission [26,70] but also suggests that local microenvironmental heterogeneity (microhabitat effects) within a population could be a potential key factor driving the differentiation in symbiotic expression among individuals, a hypothesis that warrants further investigation.

In conclusion, this study establishes the predominant role of environmental filtering in shaping the locoweed–endophyte symbiosis: while phylogenetic relationships define the potential range of symbiosis, environmental conditions ultimately determine its actual expression. This clear demonstration of environmental factors overriding phylogenetic control at a macro scale offers a novel ecological perspective for understanding the assembly of specialized symbiotic systems, challenging the assumption of tight co-evolution.

4.3. Environmental Factors Regulate Symbiotic Expression and Chemical Defense Through Multiple Pathways

Having established the primary influence of the environment, this study further dissected the specific pathways through which ecological factors regulate the symbiotic relationship (Figure 6). Both the overall correlation analysis (Figure 6A) and the structural equation model (Figure 6C) jointly revealed the mechanisms of key environmental factors. Low-temperature stress, particularly the minimum temperature of the coldest month (Bio6) and the mean temperature of the coldest quarter (Bio11), significantly promoted both endophyte colonization and toxin synthesis (path coefficients: −0.43 and −0.39, respectively). It indicated that moderate low temperatures may facilitate the establishment and maintenance of the symbiotic relationship. Our finding that low-temperature stress promotes the symbiosis aligns with and generalizes earlier observations of temporal variation. Achata et al. [71] and Cook et al. [72] both documented seasonal fluctuations in swainsonine content in O. sericea, indicating that environmental conditions influence toxin production over time. Notably, Achata et al. [71] further reported a strong correlation between average daily temperature and swainsonine content within specific populations of O. lambertii and A. mollissimus. Our study extends this critical insight from a temporal correlation at fixed sites to a spatial driver across continental-scale biogeographic gradients. We demonstrate that cold conditions (Bio6, Bio11) are a key environmental filter not only for toxin concentration but also for the broader establishment and maintenance of the endophyte symbiosis. In contrast, precipitation factors showed a significant negative correlation with swainsonine content, consistent with Guan et al. [21], who reported a negative correlation between annual humidity and swainsonine levels. Furthermore, indoor experiments by Klypina et al. [25] demonstrated that drought stress amplified toxicity by 25% to 30% in Alternaria-colonized A. mollissimus. Collectively, this evidence suggests that drought stress may be a key environmental factor promoting toxin synthesis. Furthermore, the endophyte load significantly influenced swainsonine levels, which is consistent with the findings of Grum et al. [73].

The structural equation model further uncovered a complex interaction network involving the environment, plants, and herbivores. Swainsonine concentration had a significant negative effect on grazing intensity, while also being regulated by temperature and locoweed species (Figure 6C). This result confirms the chemical defense function conferred to host plants by the Alternaria endophyte. However, this defensive effect exhibited clear species specificity: swainsonine significantly reduced grazing intensity only for O. ochrocephala and O. glabra, with no significant effect observed for O. falcata (Figure 4B). This aligns with poisoning incidence: frequent cases are reported for O. ochrocephala and O. glabra, while none are documented for O. falcata [26]. The underlying reason may lie in the habitat preference of O. falcata, which primarily grows in non-grazing areas such as scree slopes and alpine meadows [74], resulting in limited livestock exposure. This suggests that plant environmental preferences also influence grazing pressure. On the other hand, climate warming-induced rapid growth of warm-season grasses reduces livestock consumption of locoweeds [75], thereby diminishing the ecological defensive value of swainsonine and leading to a relative increase in grazing intensity. This mechanism explains how temperature indirectly regulates grazing intensity by altering the availability of food resources.

In summary, environmental factors finely regulate the locoweed–endophyte symbiotic relationship through temperature and water stress, with the ecological effects of chemical defense being shaped jointly by swainsonine level, environment and host niche preferences. By disentangling these direct and indirect pathways, our study moves beyond identifying correlates to propose a mechanistic framework for how abiotic stress translates into ecological outcomes in a plant–microbe–herbivore system. It provides large-scale ecological support for the hypothesis that the locoweed–endophyte association may have evolved primarily as a stress-response system [34], a concept with general relevance for understanding how climate shapes symbiotic communities [76].

5. Conclusions

In summary, this multi-scale ecological study demonstrates that environmental factors, rather than host phylogeny, are the dominant drivers of the locoweed–endophyte symbiosis. We show that: (1) this first systematic national survey clarifies the host species range, endophyte colonization levels, and swainsonine concentrations in Chinese locoweeds; (2) environmental filtering served as the primary determinant of the symbiosis’ geographical distribution; (3) specific abiotic stresses—notably low-temperature and drought conditions—are strongly associated with enhanced fungal colonization and swainsonine biosynthesis, while also indirectly modulating the symbiotic outcome through pathways involving grazing pressure. While the available evidence suggests that vertical transmission may maintain symbiotic stability within populations, environmental conditions ultimately govern its expression across spatial scales. These findings advance our understanding of plant–microbe symbioses by revealing how environmental forces shape symbiotic outcomes, underscoring the importance of incorporating both contemporary ecological patterns and historical evolutionary processes in future symbiosis research.