Viral Infection Risk Driven by Ecological Trait Susceptibility to Habitat Heterogeneity

Adrian D Zamfir; Bisola M Babalola; Miguel A Mora; Aurora Fraile; Michael J McLeish; Fernando García‐Arenal

doi:10.1111/mec.70489

. 2026 Jul 28;35(14):e70489. doi: 10.1111/mec.70489

Viral Infection Risk Driven by Ecological Trait Susceptibility to Habitat Heterogeneity

Adrian D Zamfir ¹, Bisola M Babalola ¹, Miguel A Mora ¹, Aurora Fraile ¹, Michael J McLeish ^1,^✉, Fernando García‐Arenal ^1,^✉

PMCID: PMC13408557 PMID: 42516027

ABSTRACT

Traits for host exploitation that drive the transmission of disease and pathogen emergence are expressed in heterogenous environments. To advance the development of accurate forecasting of infection risk and emergence, determining constraints on ecological strategies for resource use by pathogens over a range of host communities is essential. Infection risk, that is, the likelihood of infection, is approximated by incidence. Here, we examine virus host use by analysing incidence responses to habitat heterogeneity for 18 plant viruses with different modes of horizontal transmission. We estimate virus incidence responses to habitat heterogeneity and compare evolutionary histories and transmission mode of the viruses as drivers of infection risk among four habitats of an ecosystem subject to contrasting levels of anthropic disturbance. Results indicate that variation in plant community composition among the habitats shaped virus incidence responses. In addition to ecological drivers, transmission mode and shared evolutionary history contributed to virus incidence responses to habitat heterogeneity. Evidence of both ecological and evolutionary determinants of host exploitation is an important step towards forecasting infection risk.

Keywords: disease ecology, disturbance, ecological fitting, host range, plant viruses, resource utilisation, transmission

1. Introduction

Infection risk depends on parasite and host ecology and the evolutionary relationships among them (Hosseini et al. 2017; Johnson et al. 2019). Understanding what determines infection risk by parasites is necessary to forecast their emergence, which is relevant to plant viruses that have severe impacts on plant production, community assembly and composition (Anderson et al. 2004; Aranda and Freitas‐Astúa 2017). Central to infection risk and emergence are changes in the number of host species used by a virus (Elena et al. 2014; Woolhouse and Gowtage‐Sequeria 2005), that is, changes in host range (McLeish et al. 2018; Morris and Moury 2019). Host range evolution depends on factors intrinsic to the virus and host, such as genetic traits that determine host specificity, and extrinsic factors such as ecological traits that inform on ecological processes, (Violle et al. 2007) that determine virus use of host plant resources (McLeish et al. 2018; Lefeuvre et al. 2019). Intrinsic factors have been much analysed through experiments aimed at understanding host‐associated variation in virus fitness, which have highlighted trade‐offs across hosts in constraining host range evolution (Bedhomme et al. 2015; Wang et al. 2024). There is also evidence that host intrinsic factors such as phylogenetic relatedness (Holmes 2022; Gougherty and Davies 2021; Gilbert and Parker 2016) or a short life span (Cronin et al. 2010; Halliday et al. 2021; Johnson et al. 2012) contribute to infection risk. By contrast, experimental analyses of extrinsic ecological factors in host range evolution (e.g., Moreno‐Pérez et al. 2016; González et al. 2019; González et al. 2021) are scarce. Investigations of host range and resource use that result in trait responses to heterogeneity in wild plant communities are even more limited (Cooper and Jones 2006; Roossinck and García‐Arenal 2015; Lefeuvre et al. 2019).

Transmission dynamics involve interactions among different groups of species (plants, viruses, vectors, etc.), heterogeneity in resources and spatially structured communities (Halliday et al. 2021; Jones 2018; McLeish et al. 2017). Host range is a plastic trait (McLeish et al. 2018) and changes to host range that lead to infection risk and possibly emergence are not independent of community composition and the abiotic and biotic environment (Fitzpatrick et al. 2020). Thus, host range plasticity is expected to contribute to virus trait responses to variation in resource availability. Changes to the spatial structuring and species diversity (e.g., host richness and abundance) among habitats (habitat heterogeneity) due to anthropogenic disturbance alter resource availability, and modify opportunities for the movement and encounter rates of plant viruses between hosts (Jones 2018; Laine 2023; Parratt et al. 2016). Thus, so‐called ecosystem simplification is considered a key driver of plant virus emergence (Roossinck and García‐Arenal 2015; Stukenbrock and McDonald 2008). The impact of host diversity on infection risk has been predicted to be either positive (amplification effect) or negative (dilution effect) (Keesing et al. 2006; Ostfeld and Keesing 2012). However, biodiversity‐infection risk relationships depend on the life history of the parasite and on a wide range of ecological, environmental, spatial and temporal factors, rather than diversity per se (Rohr et al. 2020). More work is necessary to understand the relationship between habitat heterogeneity and infection risk.

To understand and ultimately forecast infection risk and emergence, it is necessary to analyse multi‐host—multi‐pathogen interactions within ecosystems that are characterised by host resource heterogeneity (Johnson, Ostfeld, and Keesing 2015; Woolhouse and Gowtage‐Sequeria 2005). Here, we analyse extrinsic ecological and intrinsic virus traits that underlie host resource use and infection risk in a highly mosaicked ecosystem. The analysis of spatial structuring among four dominant habitat types of distinct plant communities (McLeish et al. 2021) and high throughput sequencing (HTS) identification of 158 virus operational taxonomic units (OTUs) that infected 119 plant species in this ecosystem (McLeish et al. 2024) has been the basis for examining trait responses to habitat heterogeneity. McLeish et al. (2024) addressed the relationship between virus community structuring and habitat heterogeneity. The study found that habitat heterogeneity had a role in virus community assembly, and that virus trait responses to virus species' distributions across habitats varied with the spatial scale of observation. A majority of the 158 virus OTUs analysed by McLeish et al. (2024) exhibited habitat specificity, with communities connected by key generalist viruses and host reservoirs, and therefore it is expected that habitat heterogeneity also shapes infection risk. Here, we analyse factors that determine infection risk rather than virus community assembly. For this, we focus on a subset of 18 of the 158 virus OTUs, which had broad host ranges and occurred among the habitats. These OTUs have different modes of horizontal transmission that together represent different evolutionary histories among nine genera. We use RT‐PCR that is performed with individual plants to validate virus presence/absence in each sample and to allow the estimation of their incidence, that is, the proportion of infected hosts in a population (Campbell and Madden 1990), across the four habitats. Infection risk, defined as the likelihood of infection, is approximated by incidence as far as hosts do not recover from infection and infection does not induce host mortality (Rodríguez‐Nevado et al. 2020), two conditions met here. With these data we address three hypotheses relative to determinants of infection risk: (1) intrinsic factors of evolutionary history and mode of transmission are independent of virus realised host range among habitats; (2) extrinsic factors associated with anthropic disturbance and habitat heterogeneity modify virus incidence; and (3) intrinsic virus factors associated with evolutionary history and mode of transmission modify virus incidence. Regression models were designed to determine whether extrinsic or intrinsic factors had a stronger effect in predicting incidence responses.

2. Material and Methods

2.1. Study Area and Plant Sampling

Sampling of field plants was performed between July 2015 and June 2017 in the Vega del Tajo‐Tajuña agricultural region of the Tagus River Basin in the South‐Central plateau of the Iberian Peninsula (Spain) (Figure S1). Repeat sampling was conducted at multiple sites of each of four habitats (Table S1) subject to different degrees of disturbance (Appendix S1). Oak forests (Oak) are native communities dominated by evergreen Quercus spp. with understory vegetation of sclerophyllous shrubs and grasses; Wasteland includes undisturbed patchy areas of successional shrubland, that had an agricultural use in the past but have been abandoned for the last 10 years; Crop are habitats where annual monocultures share the space with assemblages of wild plant species (weeds) that reassemble every year, and Edge are the narrow borders that separate crop fields, with relatively permanent communities that benefit from nutrient and water supplemented to adjacent crops. Oak and Wasteland are considered as non‐anthropic habitats, with presently undisturbed communities, while Crop and Edge together formed the anthropic habitats. A total of 78 collections were done at four sites each of Oak, Wasteland and Edge and 11 sites of Crop. Collections were conducted at sites of Oak and Wasteland in spring and autumn, at sites of Crop when the summer or winter crops were present, and at sites of Edge in spring and autumn and when the adjacent crops were sampled. The higher number of Crop sites aims to better characterise the variation expected from the Crop communities as cultivated fields are subject to crop rotation and fallow periods, and involve different crop species. Sites represented the main crops of the area, maize (2 sites) and melon (4 sites) as summer crops, and barley (3 sites) and brassicas (2 sites) as winter crops. The mean distance between sites was 28.8 km (range 0.1–90 km), and distances between adjacent Edge and Crop sites varied from 0.1 to 0.5 km (McLeish et al. 2021). Plants were sampled systematically along fixed itineraries, regardless that they showed symptoms of diseases or herbivory. The sampling design (described in detail in Appendix S1) maintained the relative species richness and abundance of plant species distributed among the four habitats. Rarefaction analyses showed near asymptotic relationships between the number of samples and the expected number of species at each collection (McLeish et al. 2021). Previous work has demonstrated differences in plant species diversity among the communities that characterise each habitat, with positive spatial autocorrelation at broad scales (i.e., ecosystem) driven by the environment (McLeish et al. 2021; Table 1). To estimate plant diversity in each habitat, Tsallis entropy (S _q) was calculated:

S_{q} = \frac{1 - \sum_{i = 1}^{W} p_{i}^{q}}{q - 1}

where W is the number of states (species), p _i is the probability of the state i, and q is a real parameter. Note that S _q is a family of entropies and q indicates the degree of deviation from the usual Shannon entropy that corresponds to q → 1 (Mendes et al. 2008). Significant negative spatial autocorrelation at fine‐scales showed community structuring among adjacent study sites of the different habitat categories. As study sites were generally not contiguous, with intervening patches surrounding the sites, explicit consideration of spatial scale, geographic distance and connectivity measures between sites was given precedence over habitat area and is critical to the interpretation of biological process.

TABLE 1.

Number of host species per virus and habitat as detected by RT‐PCR (first value in fraction) and HTS2C (second value in fraction) detection. The plant species richness and diversity (Tsallis entropy) of each habitat are displayed in the last rows.

Virus genus	Virus species	Mode of transmission	Ecosystem	Crop	Edge	Oak	Waste
Anulavirus	PZSV	Seed and pollen	35/67	2/9	26/52	2/12	8/12
Betacarmovirus	TCV	Contact, coleoptera	5/33	0/0	2/10	0/17	3/10
Cucumovirus	CMV	Aphid, non‐persistent	51/85	8/13	20/47	19/31	13/20
Cucumovirus	TAV	Aphid, non‐persistent	31/38	4/7	25/28	6/6	1/3
Ilarvirus	PMoV	Pollen	8/46	0/2	8/43	0/2	0/0
Luteovirus	BYDV	Aphid, persistent non‐propagative	6/10	2/3	3/8	1/1	0/0
Polerovirus	BChV	Aphid, persistent non propagative	11/32	2/3	10/28	0/3	0/0
Polerovirus	BMYV	Aphid, persistent non propagative	12/31	2/2	10/25	2/3	0/2
Polerovirus	CABYV	Aphid, persistent non propagative	16/23	0/1	14/21	1/1	2/3
Polerovirus	TuYV	Aphid, persistent non propagative	21/33	2/2	15/26	3/7	2/3
Potyvirus	PPV	Aphid, non‐persistent	11/11	2/2	6/6	2/2	1/1
Potyvirus	TuMV	Aphid, non‐persistent	17/20	0/2	8/9	2/3	7/7
Potyvirus	WMV	Aphid, non‐persistent	22/37	4/8	10/17	8/14	2/3
Sobemovirus	RuCMV	Contact	30/62	7/12	16/39	3/9	8/15
Tobamovirus	PMMoV	Contact, soil	23/41	3/4	12/28	7/17	3/8
Tobamovirus	TMGMV	Contact, soil	43/75	7/9	12/29	13/34	12/33
Tobamovirus	TMV	Contact, soil	21/57	2/7	8/29	4/26	7/19
Tobamovirus	YoMV	Contact, soil	26/38	1/2	4/8	11/14	15/22
Plant species richness ^a				48	77	78	89
Tsallis entropy (S _q) ^a				1.481	3.085	4.236	4.545

Open in a new tab

^{^a}

Data from McLeish et al. (2021).

2.2. Detection of Virus Host Interactions

Total RNA from individual plant species was extracted (Appendix S1) and RNA extracts of the same plant species and collection were pooled to obtain a single HTS library (Appendix S1). Both read orientations of each pool were run in the same sequencing lane, and each pool was run in a separate lane. All libraries included a step for rRNA depletion using the RiboZero kit. Library preparation and high throughput sequencing with Illumina HiSeq platforms (Appendix S1) were as in McLeish et al. (2022).

Detection of operational taxonomic units (OTUs) of viruses was as in McLeish et al. (2024) (Appendix S1). Local BLAST queries were conducted with Blast+ version 2.2.29 (Camacho et al. 2009) against a database of plant virus genomic references retrieved from the NCBI Viral Genome Browser (https://www.ncbi.nlm.nih.gov/genomes/, accessed December 2018). To standardise the BLAST query matches across libraries as well as minimise both the frequency of false positives and the exclusion of true positive detections, reads were retained only if: (1) the query coverage was 100%; (2) the alignment length was greater or equal to 125 nt. These two criteria (HTS2C) were implemented to optimise sequence similarity between read and reference, while still allowing for mismatches and gaps expected from the presence of viruses that are divergent from the available reference genomes.

The occurrence in libraries and individual plants of 18 virus OTUs that were selected under the HTS2C criteria was confirmed by RT‐PCR using sets of specific primers that amplified regions of 70–174 nucleotides in the coat protein gene of the reference virus species (Table S2 and Appendix S1). The specificity of the primers was checked by RT‐PCR assays in libraries in which the virus OTU was not detected, but where a virus from the same genus or family was, and by Sanger sequencing of the amplicons from at least five different putative host plant species for each virus OTU (Appendix S1). Virus species names of the genomic reference were given to OTUs confirmed by RT‐PCR assays. The 18 virus species were: Pelargonium zonate spot virus (PZSV), cucumber mosaic virus (CMV), tomato aspermy virus (TAV), parietaria mottle virus (PMoV), barley yellow dwarf virus‐PAV (BYDV‐PAV), beet chlorosis virus (BChV), beet mild yellowing virus (BMYV), cucurbit aphid‐borne yellows virus (CABYV), turnip yellows virus (TuYV), plum pox virus (PPV), turnip mosaic virus (TuMV), watermelon mosaic virus (WMV), rubus chlorotic mottle virus (RuCMV), pepper mild mottle virus (PMMoV), tobacco mild green mosaic virus (TMGMV), tobacco mosaic virus (TMV) and Youcai mosaic virus (YoMV) (Table S3). Information on the virus species taxonomy at the genus level and on their mode of transmission is given in Table 1.

The number of plant species in which a virus was detected by either HTS2C or RT‐PCR defined its host range.

2.3. Network and Modularity Analyses

Plant–virus interactions were analysed by means of bipartite networks in which nodes represent either a virus or a host species. For each node, the number of links (node degree, k) describes the number of interactions occurring between a host species and virus species (Blüthgen et al. 2008). Networks can be divided into modules, which represent clusters that have more interactions among nodes within them than between nodes of other clusters (Valverde et al. 2020). Modules were quantified by the function modularity (Q). The nestedness metric is based on overlap and decreasing fill (NODF) and is used to describe patterns of species interactions in bipartite networks. High NODF indicates that the hosts in which a virus species is detected are subsets of hosts of virus species with wider host ranges. Connectance (C) measures the density of links, or the percentage of non‐zero values in the distribution. We calculated the mean k, Q, NODF and C of the network topology, and used z‐scores to standardise the indices for comparison. To test whether the observed index value differed from chance, a null set of 1000 networks of the same size as the observed was simulated, and a t‐test was conducted to compare the observed index value with the mean of the null set. Network indices and modules were estimated from undirected graphs, as traits of both viruses and hosts determine interactions. All network analyses were conducted with the R packages igraph and bipartite (Dormann and Strauss 2014).

2.4. Analyses of Virus Incidence

For each virus, incidence, defined as the proportion of infected hosts in a population (Campbell and Madden 1990), was estimated at two scales. In the first scale ‘population’ was defined as the population of plants of species in which the virus was detected by RT‐PCR, yielding the ‘incidence within the virus host range’. In the second scale, ‘population’ was defined as that of all plant species growing in a sampling site, regardless that the virus was detected or not in each of them, yielding the ‘incidence within the whole plant community’. Thus, two variables were defined for the analysis of incidence of each virus. Incidence within the virus host range (i) was defined as:

i = \frac{1}{j} \sum_{j} i_{j} = \sum_{j} \frac{n_{j}}{t_{j}}

where i _j is the incidence in host j, n _j is the number of individual plants that tested positive by RT‐PCR, and t _j is the total number of individual plants tested by RT‐PCR for host j. Incidence in the whole plant community (I) was computed as follows:

I = \frac{\sum_{j} (i_{j} \times N_{j})}{T}

where i _j is the incidence in the jth host; N _j is the number of individuals in libraries of that host HTS2C positive for that virus OTU that were and were not analysed by RT‐PCR; and T is the sum of all individual plant samples in the HTS libraries (positive analysed and not analysed by RT‐PCR, and HTS‐negative) for a specific virus OTU.

2.5. Statistical Analyses

Dissimilarities in the host ranges of each virus were inferred by hierarchical clustering of the host species richness per plant family. The input matrix was scaled using the R function scale. The Euclidean distance method was selected with the Ward's (‘ward.D2’) option, implemented in the R package hclust, which uses agglomerative hierarchical clustering appropriate for uneven variances. Differences among host ranges were analysed by least significance differences (LSD) tests.

For the incidence variables i and I, normality and homoscedasticity were tested using Shapiro–Wilk's test and Levene's test, respectively. While i had a normal distribution, I had an exponential distribution as determined from QQ plots, histograms, and probability curves adjusted to the data to find the best fit as determined by Akaike's information criterion using the R (v.4.3.1) (R Core Team 2023) package rriskDistributions (Belgorodski et al. 2017). The contribution of different factors to the variance of incidence was analysed by full factorial generalised linear models (GLMs) in which factors were considered as fixed. The unit of analysis was site for all models, except for those including season in which the unit of analysis was collection, and those including plant family in which the unit was the plant family. Statistical significance was analysed using Wald's chi‐squared (Wald χ ²) test as implemented in the R package stats. To test whether traits were significantly different among classes within each factor, LSD or Dunn tests were used, depending on the distribution of the data. The relative importance of a given estimator included in the best‐ranked model was calculated by decomposing the R ² value of the model into components that corresponded to each estimator.

3. Results

3.1. Virus Detection

Based on the analysis of 323 HTS libraries of 2037 pooled individual samples that represented 119 plant species (McLeish et al. 2024), 18 virus OTUs detected in plant species from at least three habitats were selected. These OTUs matched references of virus species of nine genera with positive‐sense ssRNA genomes (Table 1 and Table S3) and were transmitted horizontally by aphids in a non‐persistent or a persistent non‐propagative manner, by pollen, by contact or by Coleoptera. The OTUs differed in genome structure and evolutionary history, as represented by genera, and OTUs from different genera shared strategies of horizontal transmission. Table 1 shows the number of plant species per habitat in which these OTUs were detected by HTS2C. To verify that the OTUs corresponded with the reference, RT‐PCR with virus‐specific primers was performed in a subset of randomly selected libraries in which each OTU was detected by HTS2C, with the condition that at least 50% of the libraries from each habitat were represented. A total of 4170 RT‐PCR reactions were performed in 1931 individual plant samples that represented 94 host species.

The correlation between detection by HTS2C and RT‐PCR (Table S4) was good, except for PMoV and TCV (R ² = 0.90, p > 0.0001 across viruses and habitats; PMoV and TCV not included in the correlation analysis). The host range was narrower for RT‐PCR than for HTS2C detections (Table 1). This difference derives partly from detection by RT‐PCR occurring in fewer libraries than detection by HTS2C, which was partly due to heterogeneity in the preservation of RNA extracts across plant species and partly in that not all HTS2C host species were represented in the RT‐PCR analyses. Further analyses of host range were based on the RT‐PCR data set.

3.2. Host Ranges, Taxonomy and Modes of Horizontal Transmission

Host ranges varied from 51 plant species in 18 families for cucumber mosaic virus (CMV) to five plant species in four families for TCV (Table 1 and Figure S2). A hierarchical cluster analysis of host range based on species richness per plant family (Figure 1) showed that the four poleroviruses (BChV, BMYV, CABYV and TuYV) had the most similar host ranges. Host ranges of the other 14 viruses did not cluster according to taxonomy or mode of transmission (H1). For example, one cluster included the host ranges of TuMV (potyvirus), CMV (cucumovirus), both transmitted by aphids non‐persistently, as well as the contact‐transmitted tobamoviruses, TMV and YoMV; another cluster included the host ranges of the contact‐transmitted tobamoviruses PMMoV and TMGMV and the aphid non‐persistently transmitted potyvirus WMV. All 18 viruses were detected in species from taxonomically distant plant families of Monocots, Magnoliids or Eudicots. The distribution and abundance of species of the different plant families varied largely among habitats (Table S5), but species of families Brassicaceae, Asteraceae and Poaceae were each relatively abundant in all habitats (Table S5), and these were the families with highest host species richness across viruses (Figure 1). For all viruses except BChV, the host species richness in each plant family (Table S6) was positively correlated (R ² > 0.34, p < 0.002) with the abundance of host species from that family in the study area (Figures S3–S7). The slopes of the correlations ranged between 0.018 for PMoV and 0.002 for TuMV, indicating a minor influence on the viruses' host range on the abundance of the species of a plant family.

Normalised hierarchical clustering of host ranges based on the RT‐PCR data used to infer dissimilarities in host use. The heatmap was generated from a matrix of the number of species per plant family in the host ranges of each virus. The name of each virus species analysed is given. Dark to light blue cells indicate weighted broad to narrow host ranges of each virus OTU relative to plant family.

The strength of species interactions among the 18 viruses and 94 host species was evaluated with bipartite network indices (Figure 2). The nestedness (NODF), modularity (Q) and connectance (C) were significantly higher than expected by chance (t ₉₉₉ = −523.58, p < 0.0001; t ₉₉₉ = −1610.10, p < 0.0001; t ₉₉₉ = −278.05, p < 0.0001 for NODF, Q and C, respectively), indicating a highly connected network of nested subsets of the viruses as host range increases. The network also provided evidence of modules that related to habitat. Four modules were found at the ecosystem scale (Figure 3). The first module groups interactions of the four poleroviruses with hosts mostly from Edge. The second module grouped interactions of four viruses with different lifestyles (the pollen transmitted PZSV and PMoV, plus BYDV and TAV transmitted by aphids persistently and non‐persistently, respectively) with hosts from Edge. The third and fourth modules comprise hosts from all habitats with viruses that were mostly transmitted by contact (PMMoV, TMGMV, TMV, RuCMV, PPV; third module) or that differed in transmission mode (CMV, TCV, TuMV, WMV and YoMV; fourth module). Thus, modules were organised according to habitat, virus taxonomy and transmission mode.

Plant–virus interaction network at the ecosystem scale. Grey nodes represent virus species (n = 18), as detected by RT‐PCR assays. Green nodes represent host species (n = 94). Node size is proportional to its degree. Links (593 interactions) are coloured according to the habitat, Crop, Edge, Oak or Wasteland, in which the interaction occurs. Background colours differentiate interactions occurring mostly in the anthropic (crop, Edge) or non‐anthropic (Oak, Wasteland) habitats. Virus nodes are indicated by the virus species acronyms.

Estimated network modules as an approximation of ecological compartments in the whole ecosystem. Modules were obtained from an undirected graph and are indicated with red lines. Host species occurring in Edge are indicated in blue.

Consistent with these results, the realised host range of each virus varied broadly among habitats (Figure S2). For each virus, each habitat had different host species richness, with few host species occurring in more than one habitat, as shown in Tables S7–S24. For example, one of 46 host species of TMGMV was a host in all four habitats, and five species were hosts in two habitats (Table S7). For most viruses (12/18), the realised host range was largest in Edge, which was not the habitat with highest plant species richness (Table 1; McLeish et al. 2021). Two separate GLMs with virus genus and mode of transmission as fixed factors showed that host range depended on both virus genus (Wald χ ² _8,63 = 165.57, p < 0.0001), which explained 30.40% of the variance, and mode of transmission (Wald χ ² _3,68 = 43.42, p < 00001), which explained 12.75%. According to virus genus, host ranges constituted four size classes: (1) cucumoviruses and anulaviruses (mean host range across sites 15.6 and 14.2 species, respectively); (2) sobemoviruses and tobamoviruses (mean host range 11.2 and 9.0, respectively); (3) potyviruses, poleroviruses and ilarviruses (mean host range 4.8, 4.5 and 2.5, respectively), and (4) luteoviruses and carmoviruses (mean host range 1.8 and 1.2 respectively) (p < 0.040 in a LSD test across virus species and sites). Pollen, aphid non‐persistent and contact‐transmitted viruses had broader host ranges than those persistently‐transmitted by aphid (p < 0.0003).

3.3. Host Use Among Habitats

As RT‐PCR was performed with RNA extracts of individual plants, the incidence of each virus in each host was estimated to inform on the distribution of virus infection across hosts. The incidence across hosts (i) is shown for all viruses in Tables S9–S24. For all viruses, incidence varied largely among hosts. Incidence could differ among habitats for any host that occurred in more than one habitat. For example, the incidence of TMGMV in Artemisia herba‐alba was 50% in Oak but 100% in Wasteland, in Lithospermum arvense was 78% in Crop and 50% in Edge, and in Rubia peregrina 33% in Crop and 0% in Oak (Table S7). Similarly, the incidence of CMV in Convolvulus arvensis was very high in the anthropic habitats (100% in Crop, 90% in Edge) and 37% in Wasteland (Table S8). These data indicated incidence across hosts that was habitat dependent.

Regression was used to analyse incidence i across habitats (Figure 4), with virus species and habitat as fixed factors (Model 1, Table 2). The analyses excluded BYDV, PMoV and TCV as the sample sizes were insufficient. Incidence i varied according to virus species, to habitat and to the interaction between both factors, which explained 55.20%, 14.80% and 30.0% of the variance of i, respectively. Pairwise comparisons showed that CMV (72.3%), WMV (62.7%), TMGMV (61.8%) and RuCMV (56.8%) had significantly higher i than the other virus species (t ₁₅₀ > 1.98, p ≤ 0.027). i was significantly higher in Crop (55.6%) and Edge (54.7%) than in Oak (34.1%) and Wasteland (32.1%) (t ₁₅ > 1.98, p < 0.0036). As the interaction between virus species and habitat was significant, one‐way ANOVAs were performed for each virus i across habitats; i was highest in Crop for BChV, TuYV and TuMV (p < 0.033) and in both Crop and Edge for TuMV (p < 0.0020). For the other viruses, i was not significantly different among habitats.

Incidence over hosts (i, upper panel) and incidence over plant communities (I, lower panel). Data are for 18 virus species in four different habitats (Crop, Edge, Oak, Wasteland) shown in the tight. Standard errors derive from estimates at the different sites of each habitat.

TABLE 2.

Summary of generalised linear models for the incidence response of each virus across hosts (i) and for the incidence within the plant community (I), showing the R ² and Wald's χ ² for each factor and interaction.

Incidence (i) across hosts
Model	R ²	Habitat	Virus species	Virus genus	Mode transm.	Plant family	Season	Interaction
(1) i ~ virus species + habitat + interaction + ε	67.97	χ ² ₃₁₅₀ = 42.00 p < 0.0001	χ ² _17,150 = 170.62 p < 0.0001	—	—	—	—	χ ² _51,150 = 95.50 p = 0.0001
(2) i ~ virus genus + habitat + interaction + ε	35.26	χ ² ₃₁₅₆ = 31.33 p < 0.0001	—	χ ² ₅₁₅₆ = 34.50 p < 0.0001	—	—	—	χ ² _15,156 = 19.49 p = 0.19
(3) i ~ mode transm. + habitat + interaction + ε	31.04	χ ² ₃₁₈₅ = 33.40 p < 0.0001	—	—	χ ² ₃₁₈₅ = 40.03 p < 0.0001	—	—	χ ² ₉₁₈₅ = 11.96 p = 0.22
(4) i ~ plant family + habitat + interaction + ε	26.68	χ ² ₂₁₉₅ = 16.27 p = 0.0003	—	—	—	χ ² ₄₁₉₅ = 16.18 p = 0.0028	—	χ ² ₈₁₉₅ = 38.50 p < 0.0001
(5) i ~ virus genus + season + interaction + ε	29.25	—	—	χ ² ₇₂₈₁ = 104.45 p < 0.0001	—	—	χ ² ₁₂₈₁ = 2.91 p > 0.09	χ ² ₇₂₈₁ = 4.82 p > 0.19
(6) i ~ mode transm. + season + interaction + ε	15.85	—	—	—	χ ² ₃₂₈₉ = 46.50 p = < 0.0001	—	χ ² ₁₂₈₉ = 2.44 p = 0.12	χ ² ₃₁₅₀ = 95.50 p = 0.0001

Incidence (I) within the plant community
Model	R ²	Habitat	Virus species	Virus genera	Mode transm.	Plant family	Season	Interaction
(7) I ~ virus species + habitat + interaction + ε	62.29	χ ² ₃₁₅₀ = 45.75 p < 0.0001	χ ² _17,150 = 111.67 p < 0.0001	—	—	—	—	χ ² _51,150 = 74.76 p = 0.017
(8) I ~ virus genus + habitat + interaction + ε	40.87	χ ² ₃₁₅₆ = 29.17 p < 0.0001	—	χ ² ₅₁₅₆ = 18.44 p = 0.002	—	—	—	χ ² _15,156 = 36.15 p = 0.001
(9) I ~ mode transm. + habitat + interaction + ε	8.96	χ ² ₃₁₈₅ = 35.19 p < 0.0001	—	—	χ ² ₃₁₈₅ = 11.30 p = 0.010	—	—	χ ² ₉₁₈₅ = 17.23 p = 0.045
(10) I ~ plant family + habitat + interaction + ε	14.50	χ ² ₂₁₉₅ = 5.72 p = 0.057	—	—	—	χ ² ₄₁₉₅ = 12.73 p = 0.013	—	χ ² ₈₁₉₅ = 14.57 p = 0.068
(11) I ~ virus genus + season + interaction + ε	18.74	—	—	χ ² ₇₂₈₁ = 49.70 p < 0.0001	—	—	χ ² ₁₂₈₁ = 3.16 p > 0.08	χ ² ₇₂₈₁ = 12.03 p = 0.10
(12) I ~ mode transm. + season + interaction + ε	6.29	—	—	—	χ ² ₃₂₈₉ = 10.71 p = 0.013	—	χ ² ₁₂₈₉ = 1.58 p > 0.20	χ ² ₃₂₈₉ = 3.46 p > 0.33

Open in a new tab

To identify factors that underlie host use, full factorial GLMs were performed in which intrinsic virus traits associated with a common evolutionary history, represented by virus taxonomy (genus) and mode of transmission, plus host taxonomy at the plant family level, were compared among the habitats. Models that integrated all the factors were specious, and thus binary models with either virus genus, model of transmission or host plant family, plus habitat, were performed. The models included six genera, with the betacarmoviruses, luteoviruses and ilarviruses excluded from the analyses, as their incidence was zero or very low in at least two habitats. The unit of analysis, number of units and the model fit according to AIC are shown in Table S25, and the results of the analyses in Table 2. Incidence i depended on the virus genus (Model 2, Table 2), which explained 40.4% of i variance, and on the habitat, which explained 36.7%, but not on the interaction between genus and habitat. Pairwise comparisons showed that i of cucumoviruses, potyviruses and sobemoviruses was greater than anulaviruses, tobamoviruses, and poleroviruses (t ₁₅₆ > 1.98, p < 0.0412). Incidence i in Crop and Edge was greater than in Oak and Wasteland (t ₁₅₆ > 1.98, p < 0.0092).

The role four transmission modes (contact, aphid non‐persistent, aphid persistent and pollen) had on virus incidence was tested (Model 3, Table 2). The mode of transmission explained 46.80%, and habitat 38.90% of the variance in i. The interaction between transmission and habitat was not significant. i was higher for aphid non‐persistent than for contact, pollen and aphid persistent transmission (t ₁₈₅ > 1.97, p < 0.0001).

As the species richness of each plant family used by each virus differed, incidence i was modelled with plant family and habitat as fixed factors (Model 4, Table 2). i varied according to plant family, which explained 22.8% of its variance, habitat, which explained 22.9%, and to the interaction between plant family and habitat, which explained 54.3% of the variance. Across families, i was higher in Rubiaceae, Poaceae and Asteraceae than in Convolvulaceae and Lamiaceae (t ₁₉₅ > 1.97, p < 0.018). Across habitats i was higher in Edge than Oak and Wasteland (t ₁₉₅ > 1.97, p < 0.0016). As the interaction between habitat and plant family was significant, incidence i for each plant family across the four habitats was tested with one‐way ANOVAs, which showed that for the Convolvulaceae and Poaceae, i was highest in Edge than Wasteland, and higher in Wasteland than Oak (t ₃₉ > 2.02, p < 0.0067).

Regressions with virus genus or mode of transmission, and season (autumn vs. spring collections) (Models 5 and 6, Table 2), showed that i did not vary according to season or the interaction between season and virus genus, or between season and mode of transmission.

Last, a GLM with the host plant life cycle, annual or perennial, as a factor showed that i varied according to the host plant being annual or perennial (Wald χ ² _1,485 = 5.42, p = 0.019), where life cycle explained 3.7% of the variance of i.

All the analyses in Table 2 were also performed after eliminating hosts with sample sizes of 1 that produced extreme per‐host incidence estimates and introduced noise to aggregated incidence measures. This resulted in the elimination of 29 out of 94 host plant species involved in 106 out of a total of 789 virus‐host interactions across hosts and habitats. The elimination of these hosts reduced the high variance in the estimates of incidence but did not affect the main conclusions gained from patterns of incidence (Table S26).

3.4. Determinants of Infection Risk

The I variable, which considers all plants, hosts and other plants in which infection was not detected, in the community, informs on the probability of infection of any host within the plant community (Figure 4). Virus species and habitat were modelled as fixed factors with I as the response (Model 7, Table 2). Incidence I depended on virus species, which explained 48.2% of the variance, habitat, which explained 21.6%, and the interaction between virus species and habitat, which explained 30.2%. I was higher for CMV and TMGMV than for WMV, RuCMV, PZSV, TAV, TMV, YoMV and PMMoV, and lowest for TuYV, BChV, PPV, BMYV, TuMV, CABYV, PMoV, BYDV and TCV (p < 0.0260). I was higher in Crop (23.3%) and Edge (14.4%) than in Oak (7.4%) and Wasteland (6.6%) (p < 0.0036). The I of each virus species across the four habitats was further tested by one‐way ANOVAs, which showed that I was highest in Crop for PMMoV, YoMV, BChV, TuYV and WMV (p < 0.027), and for PZSV in Edge than in the remaining habitats (p < 0.0005).

As for variable i, full factorial GLMs were performed in which intrinsic virus traits, plus host plant family, were compared among the habitats. Incidence I (Model 8, Table 2) depended on virus genus, which explained 28.0% of the variance, habitat, which explained 43.3%, and on the interaction between genus and habitat, which explained 28.8%. I was higher for cucumoviruses, sobemoviruses and tobamoviruses than for potyviruses, anulaviruses and poleroviruses (p < 0.022), and was higher in Crop and Edge than in Oak and Wasteland (p < 0.008). The I of each virus genus across the four habitats was further tested by Kruskal–Wallis rank sum tests, which showed that I was highest in Crop for the poleroviruses, potyviruses and tobamoviruses (p < 0.03), and in Crop and Edge for the anulaviruses (p < 0.03).

Incidence I also depended (Model 9, Table 2) on mode of transmission, which explained 20.90% of the variance, habitat, which explained 58.60%, and the interaction between mode of transmission and habitat, which explained 25%. I was higher for aphid non‐persistent and contact transmission than for aphid persistent and pollen (p < 0.022), and lowest in Oak than in the other habitats (p < 0.035). Kruskal–Wallis rank sum tests showed that I was highest in Crop for aphid non‐persistent and aphid persistent (p < 0.035), and in Crop and Edge for contact‐transmitted viruses (p < 0.058).

Model 10 (Table 2) showed that I depended on plant family, which explained 38.6% of the variance, and habitat, which explained 17.3%, but not on the interaction between plant family and habitat. I was higher in Rubiaceae (12.9%) and Poaceae (12.7%) than in Lamiaceae (3.7%) (t ₁₉₅ > 1.97, p < 0.004) and in Edge and Wasteland than in Oak (t ₁₉₅ > 1.97, p < 0.02).

Models 11 and 12 (Table 2) showed that I did not vary according to season (autumn and spring) or the interactions between season and virus genus or mode of transmission.

4. Discussion

To forecast infection risk and emergence it is necessary to identify the drivers of host use in ecological communities (Johnson, de Roode, and Fenton 2015). A community ecology approach is a powerful tool used to elucidate on infection risk in multi‐host multi‐virus systems (Johnson et al. 2013; Woolhouse and Gowtage‐Sequeria 2005). Here we consider incidence as a host‐use trait of 18 RNA viruses and examine its response to habitat heterogeneity among plant communities and 94 host plant species. A previous metagenomic study established that 158 virus OTUs displayed high habitat specificity and community structuring of host range and transmission traits (McLeish et al. 2024). That study analysed the relationship between virus community structuring, habitat heterogeneity, and virus resource utilisation. However, whether habitat heterogeneity translated to variation in infection risk via virus common evolutionary history or mode of transmission was not addressed. In the present study we selected 18 viruses detected by HTS in at least three habitats to test the general hypothesis that virus intrinsic and extrinsic ecological factors are drivers of incidence, and therefore, infection risk. RT‐PCR of individual plant samples and verification with virus‐specific primers were used to estimate incidence across communities of the four habitats.

Host range was weakly dependent on intrinsic virus factors categorised by species identity, evolutionary history (i.e., taxonomy at the genus level), and mode of transmission (H1). Genetic factors contributed host use strategies as viruses with three‐partite genomes (cucumoviruses and anulaviruses) had the largest host ranges, as predicted (Moury et al. 2017). The effect of mode of transmission on host range agrees with conclusions from a meta‐analysis of mostly experimental host ranges (Moury et al. 2017). The large host ranges of the contact‐transmitted sobemo‐ and tobamoviruses suggest mechanisms other than plant‐to‐plant contact may contribute to their horizontal transmission (Zamfir et al. 2023). The narrowest host ranges of viruses transmitted persistently by aphids may be explained by the high specificity of virus‐vector interactions (Fereres and Raccah 2015), and by plant–aphid interactions (Power and Flecker 2003). Although not unexpected, each virus included in its host range taxonomically distant plant families, a surprising outcome according to classical analyses of host range that were based on virus transmission (Gibbs et al. 2020). Host range depended less on specificity towards plant family than on extrinsic ecological factors. For instance, the positive correlation between the host species richness of each virus and the abundance of plants from the corresponding plant family, showed that the likelihood of encountering a host modulated the realised host range of a virus. This conclusion is contrary to genetic studies of host range evolution that showed across‐host fitness trade‐offs constrain the taxonomic diversity of virus' hosts (Elena et al. 2014; García‐Arenal and Fraile 2013). Given that host range is augmented by habitat heterogeneity, and weakly influenced by evolutionary history and transmission mode, variation in virus incidence should be subject to extrinsic ecological factors. Interactions among species that occur without adaptive evolution, so‐called ecological fitting (Agosta and Klemens 2008), might be a pervasive influence on resource use traits of plant viruses (Peláez et al. 2021; Zamfir et al. 2023) and infection risk.

The network of host–virus interactions of the 18 generalist viruses and 94 host plant species was both nested, as expected for generalist interactions, and modular, as expected due to the realised host range variation of the viruses among the four habitats. Concurrent network modularity and nestedness in generalist plant‐virus networks are theoretically a consequence of ecological traits, and where nestedness depends on community assembly (Valverde et al. 2020). Modules were organised by host resource use differences among the habitats as well as species identity and mode of transmission (Figure 3). For instance, the four poleroviruses (BChV, BMYV, CABYV and TuYV) with hosts mostly from Edge conformed to a module. The network of host–virus interactions reported for 158 virus OTUs and 119 host plant species in the same ecosystem was highly modular with no evidence of nestedness, due to most interactions being habitat specific (McLeish et al. 2024). The differences in nestedness reported between the two studies are consistent with distinctions in the size of the community examined in each. By extension, as additional species are introduced during community assembly (or, as sample size changes), so too should nestedness among species vary.

Incidence in hosts (i) informs on how infection is distributed among the detected hosts, and may be a proxy for host susceptibility and competence. Values of i showed that each virus had a distinct host use strategy, with plant species that were preferentially used according to habitat‐ and virus‐dependencies, as shown by the effects of the factor habitat and the interaction between virus species and habitat on the variance of i (Table 2). Thus, incidence i depended on habitat (H2) and on virus intrinsic factors (H3). This type of host use supports the hypothesis of facultative generalism (Shipley et al. 2009), where habitat heterogeneity promotes relatively narrow resource use according to community composition and is a factor in plant susceptibility and competence (Malpica et al. 2006; McLeish et al. 2017). Facultative generalism is a resource exploitation strategy that implies conditional responses to environmental heterogeneity and the existence of niche differences among hosts and viruses (McLeish et al. 2017; Seabloom et al. 2015). Fitness trade‐offs establish niche differences that promote coexistence within the same habitat and where trait‐mediated interactions between species that alter phenotypes are dependent on habitat (Valverde et al. 2020).

Intrinsic factors also had an influence on incidence i. For instance, viruses in genera with aphid non‐persistent or contact transmission (cucumo‐, poty‐ and sobemoviruses) had higher incidence (i) over their hosts than the other genera (H3). This result is consistent with adaptive traits of particular viruses having an advantage over extrinsic and intrinsic traits of other viruses. It is also interesting that relatedness among hosts had a relatively minor influence on host use, contrary to expectations (Gilbert and Parker 2016; Moury et al. 2017). Similarly, the effect of plant life cycle on incidence i, predicted by theory and shown by experiments (Cronin et al. 2010; Johnson et al. 2012; Halliday et al. 2021; Hily et al. 2014; Malmstrom et al. 2005) was small: although in annuals i (51.7%) was significantly higher than in perennials (43.5%; p < 0.05), this trait explained a low fraction of the variance of i. Both lines of evidence suggest plasticity in host use by generalist viruses is partly determined by the availability of resources and non‐adaptive mechanisms.

Incidence in the plant community (I) informs on the probability that any one individual plant is infected, that is, on infection risk. This estimate of infection risk is consistent with analyses showing that both community diversity and composition have an effect on infection distributions, as the presence of non‐host species affects the probability of transmission to susceptible hosts (Roossinck and García‐Arenal 2015; Rohr et al. 2020). The estimate of I and, hence, infection risk, is sensitive to uncertainties in host range determination inherent to any method of virus detection. However, our study underscores community dynamics and is not focused on fine scale details such as whether a plant is a host or not, but rather on whether the entire community of plants that distinguishes habitats shapes infection risk at coarser scales (H2). Incidence in the plant community (I) was a better predictor of the likelihood of infection than incidence in hosts (i). Incidence (I) derived from the entire plant community was consistent with incidence on the host species (i). The major difference was that I, but not i, depended on the interactions between virus genus and habitat and between mode of transmission and habitat. Incidence I response of viruses that shared both evolutionary history and mode of transmission was dependent on habitat heterogeneity (H2). This result supports the hypotheses that ecosystem simplification is linked to pathogen infection risk (Jones 2009; Roossinck and García‐Arenal 2015; Stukenbrock and McDonald 2008). For instance, viruses in four of six analysed genera had higher incidence (I) in the anthropic habitats Crop and Edge than in Wasteland and/or Oak. An earlier study (Bernardo et al. 2017) found higher virus incidence and diversity at the virus family taxonomic level in cultivated habitats, and an association between virus family and habitat. However, the four of six genera in our study comprised only 6 of 15 virus species analysed, and indicated that the effect of ecosystem simplification on incidence was not uniform across virus taxa. This is consistent with particular taxa exhibiting stronger adaptive responses to habitat heterogeneity than others, and that non‐adaptive responses might be determined by community composition as well.

In summary, our results show the role of extrinsic ecological factors, such as plant community composition, on the incidence of generalist plant viruses. Intrinsic factors of shared evolutionary history and transmission mode had a minor role in determining host range, where the availability of plants was important (H1), and is consistent with plasticity in this trait. Network modularity and nestedness implied that although there is evidence of host range plasticity across habitats, intrinsic factors of some viruses might contribute to host use strategies (H3). This notion was supported by evidence of significant interactions between both shared evolutionary history and transmission mode with habitat on incidence responses by some viruses (H2 and H3). A major contribution of this study is to show that viruses exhibit incidence responses that are determined by intrinsic and extrinsic factors, where the influence of either is taxon‐dependent. Thus, virus incidence can be viewed as a taxon‐dependent outcome of adaptive and non‐adaptive mechanisms, each of whose individual contribution to infection risk is subject to the plant community in which they occur. Evidence of virus‐specific, ecological trait susceptibility to habitat heterogeneity is an important step towards accurately forecasting infection risk and emergence.

Author Contributions

A.D.Z., B.M.B. and M.A.M. obtained the data; A.D.Z. analysed the data; A.F., M.J.M. and F.G.‐A. designed the study; M.J.M. and F.G.‐A. wrote the initial version of the manuscript, and all authors contributed to the final one.

Funding

This work was supported by Ministerio de Ciencia e Innovación, Spain (PID2021‐124671OB‐I00) and HORIZON EUROPE Marie Sklodowska‐Curie Actions (813542T INEXTVIR).

Disclosure

Benefit‐Sharing Statement: Benefits Generated—Benefits from this research accrue from the sharing of our data and results on public databases as described above.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Appendix S1: Supporting methods.

Figure S1: Geographic distribution of the sampling sites. For each site, habitats are indicated: Crop (Cr, orange dots), Edge (Ed, green dots), Oak (Oa, blue dots), Wasteland (Wl, purple dots). The colour scale indicates the elevation. Obtained from McLeish et al. (2021).

Figure S2: Host range of each virus species based on the RT‐PCR detection assays. The different habitats are indicated by the legend colour codes.

Figure S3: Correlation between number of host species per plant family and host abundance for each of the four tobamovirus species, Pepper mild mottle virus (PMMoV), Tobacco mild green mosaic virus (TMGMV), Tobacco mosaic virus (TMV) and Youcai mosaic virus (YoMV) at the ecosystem scale. Pearson's coefficient of correlation, the regression equation and the p‐value are provided.

Figure S4: Correlation between number of host species per plant family and host abundance for Barley yellow dwarf virus (BYDV), Parietaria mottle virus (PMoV), Rubus chlorotic mottle virus (RuCMV) and Turnip crinkle virus (TCV) at the ecosystem scale. Pearson's coefficient of correlation, the regression equation and the p‐value are provided.

Figure S5: Correlation between number of host species per plant family and host abundance for Beet mild yellowing virus (BMYV), Cucurbit aphid‐borne yellows virus (CABYV) and Turnip yellows virus (TuYV) at the ecosystem scale. Pearson's coefficient of correlation, the regression equation and the p‐value are provided.

Figure S6: Correlation between number of host species per plant family and host abundance for Cucumber mosaic virus (CMV), Tomato aspermy virus (TAV) and Pelargonium zonate spot virus (PZSV) at the ecosystem scale. Pearson's coefficient of correlation, the regression equation and the p‐value are provided.

Figure S7: Correlation between number of host species per plant family and host abundance for the virus indicated in each panel.

Table S1: Locations of the sampling sites collected during the years 2015, 2016 and 2017 in Madrid province, Spain.

Table S2: Primer pairs, their sequence and the size of the amplicon produced for each virus species.

Table S3: Names of the virus genus and reference species, type of genome, transmission mode and NCBI accession number of the viral species analysed.

Table S4: Comparison of HTS and RT‐PCR detection of virus OTUs. Percentage of detection is given in parenthesis as the ratio of RT‐PCR detection/HTS2C detection. NA means not analysed because the virus OTU was not present in a specific habitat.

Table S5: Abundance of the different plant families in each of the four habitats.

Table S6: Number of host species from the corresponding plant family infected by each of the plant virus species analysed. Host abund. indicates the abundance of each host species at the ecosystem scale.

Table S7: TMGMV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S8: CMV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S9: PMMoV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S10: RuCMV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S11: TCV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S12: BMYV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S13: TuYV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S14: TMV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S15: YoMV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S16: BYDV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S17: CABYV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S18: PMoV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S19: TAV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S20: PZSV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S21: PPV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S22: TuMV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S23: WMV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S24: BChV incidence (i) in its different hosts across all habitats and at the ecosystem scale.

Table S25: Summary of generalised linear models for the incidence response of each virus across hosts (i) and for the incidence within the plant community (I), showing the number of units in the analysis (N), the nature of the unit (Scale of the analysis) and the model fit according to the Akaike information criteria (AIC). Models (Mod.) are as in Table 2.

Table S26: Summary of generalised linear models for the incidence response of each virus across hosts (i) showing the R ², the Akaike information criteria (AIC) and Wald's χ ² for each factor and interaction. Libraries with a sample size of 1 were not included in the analyses.

MEC-35-e70489-s001.docx^{(11.8MB, docx)}

Acknowledgements

This work was funded in part by Plan Estatal de Investigación Científica, Técnica y de Innovación, Ministerio de Ciencia e Innovación, Spain (grant PID2021‐124671OB‐I00). A.D.Z. and B.M.B. were supported by funding from the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska‐Curie grant agreement no. 813542T INEXTVIR.

Contributor Information

Michael J. McLeish, Email: mcleish.michale@gmail.com.

Fernando García‐Arenal, Email: fernando.garciaarenal@upm.es.

Data Availability Statement

Data are available from the Dryad repository at https://doi.org/10.5061/dryad.612jm6487, and as Supporting Information.

References

Agosta, S. J. , and Klemens J. A.. 2008. “Ecological Fitting by Phenotypically Flexible Genotypes: Implications for Species Associations, Community Assembly and Evolution.” Ecology Letters 11: 1123–1134. [DOI] [PubMed] [Google Scholar]
Anderson, P. K. , Cunningham A. A., Patel N. G., Morales F. J., Epstein P. R., and Daszak P.. 2004. “Emerging Infectious Diseases of Plants: Pathogen Pollution, Climate Change and Agrotechnology Drivers.” Trends in Ecology & Evolution 19: 535–544. [DOI] [PubMed] [Google Scholar]
Aranda, M. A. , and Freitas‐Astúa J.. 2017. “Ecology and Diversity of Plant Viruses, and Epidemiology of Plant Virus‐Induced Diseases.” Annals of Applied Biology 171: 1–4. [Google Scholar]
Bedhomme, S. , Hillung J., and Elena S. F.. 2015. “Emerging Viruses: Why They Are Not Jacks of All Trades?” Current Opinion in Virology 10: 1–6. [DOI] [PubMed] [Google Scholar]
Belgorodski, N. , Greiner M., Tolksdorf K., and Schueller K.. 2017. “rriskDistributions: Fitting Distributions to Given Data or Known Quantiles. R Package Version 2.1.2.” http://www.bfr.bund.de/cd/52158.
Bernardo, P. , Charles‐Dominique T., Barakat M., et al. 2017. “Geometagenomics Illuminates the Impact of Agriculture on the Distribution and Prevalence of Plant Viruses at the Ecosystem Scale.” ISME Journal 12: 173–184. [DOI] [PMC free article] [PubMed] [Google Scholar]
Blüthgen, N. , Fründ J., Vázquez D. P., and Menzel F.. 2008. “What do Interaction Network Metrics Tell Us About Specialization and Biological Traits.” Ecology 89: 3387–3399. [DOI] [PubMed] [Google Scholar]
Camacho, C. , Coulouris G., Avagyan V., et al. 2009. “BLAST+: Architecture and Applications.” BMC Bioinformatics 10: 1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
Campbell, C. L. , and Madden L. V.. 1990. Introduction to Plant Disease Epidemiology. Johm Wiley and Sons. [Google Scholar]
Cooper, I. , and Jones R. A.. 2006. “Wild Plants and Viruses: Under‐Investigated Ecosystems.” Advances in Virus Research 67: 1–47. [DOI] [PubMed] [Google Scholar]
Cronin, J. P. , Welsh M. E., Dekkers M. G., Abercrombie S. T., and Mitchell C. E.. 2010. “Host Physiological Phenotype Explains Pathogen Reservoir Potential.” Ecology Letters 13: 1221–1232. [DOI] [PubMed] [Google Scholar]
Dormann, C. F. , and Strauss R.. 2014. “A Method for Detecting Modules in Quantitative Bipartite Networks.” Methods in Ecology and Evolution 5: 90–98. [Google Scholar]
Elena, S. F. , Fraile A., and Garcia‐Arenal F.. 2014. “Evolution and Emergence of Plant Viruses.” Advances in Virus Research 88: 161–191. [DOI] [PubMed] [Google Scholar]
Fereres, A. , and Raccah B.. 2015. “Plant Virus Transmission by Insects.” In Encyclopedia of Life Sciences, 1–12. [Google Scholar]
Fitzpatrick, C. R. , Salas‐González I., Conway J. M., et al. 2020. “The Plant Microbiome: From Ecology to Reductionism and Beyond.” Annual Review of Microbiology 74: 81–100. [DOI] [PubMed] [Google Scholar]
García‐Arenal, F. , and Fraile A.. 2013. “Trade‐Offs in Host Range Evolution of Plant Viruses.” Plant Pathology 62: 2–9. [Google Scholar]
Gibbs, A. J. , Hajizadeh M., Ohshima K., and Jones R. A. C.. 2020. “The Potyviruses: An Evolutionary Synthesis Is Emerging.” Viruses 12: 132. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gilbert, G. S. , and Parker I. M.. 2016. “The Evolutionary Ecology of Plant Disease: A Phylogenetic Perspective.” Annual Review of Phytopathology 54: 549–578. [DOI] [PubMed] [Google Scholar]
González, R. , Butković A., and Elena S. F.. 2019. “Role of Host Genetic Diversity for Susceptibility‐to‐Infection in the Evolution of Virulence of a Plant Virus.” Virus Evolution 5: vez024. [DOI] [PMC free article] [PubMed] [Google Scholar]
González, R. , Butković A., Escaray F. J., et al. 2021. “Plant Virus Evolution Under Strong Drought Conditions Results in a Transition From Parasitism to Mutualism.” Proceedings of the National Academy of Sciences of the United States of America 118: e2020990118. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gougherty, A. V. , and Davies T. J.. 2021. “Towards a Phylogenetic Ecology of Plant Pests and Pathogens.” Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences 376: 20200359. [DOI] [PMC free article] [PubMed] [Google Scholar]
Halliday, F. W. , Jalo M., and Laine A.‐L.. 2021. “The Effect of Host Community Functional Traits on Plant Disease Risk Varies Along an Elevational Gradient.” eLife 10: e67340. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hily, J. M. , García A., Moreno A., et al. 2014. “The Relationship Between Host Lifespan and Pathogen Reservoir Potential: An Analysis in the System Arabidopsis thaliana ‐Cucumber Mosaic Virus.” PLoS Pathogens 10: e1004492. [DOI] [PMC free article] [PubMed] [Google Scholar]
Holmes, E. C. 2022. “The Ecology of Viral Emergence.” Annual Review of Virology 9: 173–192. [DOI] [PubMed] [Google Scholar]
Hosseini, P. R. , Mills J. N., Prieur‐Richard A.‐H., et al. 2017. “Does the Impact of Biodiversity Differ Between Emerging and Endemic Pathogens? The Need to Separate the Concepts of Hazard and Risk.” Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences 372: 20160129. [DOI] [PMC free article] [PubMed] [Google Scholar]
Johnson, E. E. , Escobar L. E., and Zambrana‐Torrelio C.. 2019. “An Ecological Framework for Modeling the Geography of Disease Transmission.” Trends in Ecology & Evolution 34: 655–668. [DOI] [PMC free article] [PubMed] [Google Scholar]
Johnson, P. T. J. , de Roode J. C., and Fenton A.. 2015. “Why Infectious Disease Research Needs Community Ecology.” Science 349: 1259504. [DOI] [PMC free article] [PubMed] [Google Scholar]
Johnson, P. T. J. , Ostfeld R. S., and Keesing F.. 2015. “Frontiers in Research on Biodiversity and Disease.” Ecology Letters 18: 1119–1133. [DOI] [PMC free article] [PubMed] [Google Scholar]
Johnson, P. T. J. , Preston D. L., Hoverman J. T., and LaFonte B. E.. 2013. “Host and Parasite Diversity Jointly Control Disease Risk in Complex Communities.” Proceedings of the National Academy of Sciences of the United States of America 110: 16916–16921. [DOI] [PMC free article] [PubMed] [Google Scholar]
Johnson, P. T. J. , Rohr J. R., Hoverman J. T., Kellermanns E., Bowerman J., and Lunde K. B.. 2012. “Living Fast and Dying of Infection: Host Life History Drives Interspecific Variation in Infection and Disease Risk.” Ecology Letters 15: 235–242. [DOI] [PubMed] [Google Scholar]
Jones, R. A. 2009. “Plant Virus Emergence and Evolution: Origins, New Encounter Scenarios, Factors Driving Emergence, Effects of Changing World Conditions, and Prospects for Control.” Virus Research 141: 113–130. [DOI] [PubMed] [Google Scholar]
Jones, R. A. C. 2018. “Plant and Insect Viruses in Managed and Natural Environments: Novel and Neglected Transmission Pathways.” Advances in Virus Research 101: 149–187. [DOI] [PubMed] [Google Scholar]
Keesing, F. , Holt R. D., and Ostfeld R. S.. 2006. “Effects of Species Diversity on Disease Risk.” Ecology Letters 9: 485–498. [DOI] [PubMed] [Google Scholar]
Laine, A.‐L. 2023. “Plant Disease Risk Is Modified by Multiple Global Change Drivers.” Current Biology 33: R574–R583. [DOI] [PubMed] [Google Scholar]
Lefeuvre, P. , Martin D. P., Elena S. F., Shepherd D. N., Roumagnac P., and Varsani A.. 2019. “Evolution and Ecology of Plant Viruses.” Nature Reviews Microbiology 17: 632–644. [DOI] [PubMed] [Google Scholar]
Malmstrom, C. M. , McCullough A. J., Johnson H. A., Newton L. A., and Borer E. T.. 2005. “Invasive Annual Grasses Indirectly Increase Virus Incidence in California Native Perennial Bunchgrasses.” Oecologia 145: 153–164. [DOI] [PubMed] [Google Scholar]
Malpica, J. M. , Sacristán S., Fraile A., and García‐Arenal F.. 2006. “Association and Host Selectivity in Multi‐Host Pathogens.” PLoS One 1: e41. [DOI] [PMC free article] [PubMed] [Google Scholar]
McLeish, M. , Peláez A., Pagán I., Gavilán R., Fraile A., and García‐Arenal F.. 2021. “Structuring of Plant Communities Across Agricultural Landscape Mosaics: The Importance of Connectivity and the Scale of Effect.” BMC Ecology and Evolution 21: 173. [DOI] [PMC free article] [PubMed] [Google Scholar]
McLeish, M. , Peláez A., Pagán I., Gavilán R. G., Fraile A., and García‐Arenal F.. 2024. “Plant Virus Community Structuring Is Shaped by Habitat Heterogeneity and Traits for Host Plant Resource Utilisation.” New Phytologist 244: 1585–1596. [DOI] [PubMed] [Google Scholar]
McLeish, M. , Sacristán S., Fraile A., and García‐Arenal F.. 2017. “Scale Dependencies and Generalism in Host Use Shape Virus Prevalence.” Proceedings of the Royal Society B: Biological Sciences 284: 20172066. [DOI] [PMC free article] [PubMed] [Google Scholar]
McLeish, M. J. , Fraile A., and García‐Arenal F.. 2018. “Chapter Nine—Ecological Complexity in Plant Virus Host Range Evolution.” Advances in Virus Research 101: 293–339. [DOI] [PubMed] [Google Scholar]
McLeish, M. J. , Zamfir A. D., Babalola B. M., Peláez A., Fraile A., and García‐Arenal F.. 2022. “Metagenomics Show High Spatiotemporal Virus Diversity and Ecological Compartmentalisation: Virus Infections of Melon, Cucumis melo , Crops, and Adjacent Wild Communities.” Virus Evolution 8: veac095. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mendes, R. S. , Evangelista L. R., Thomaz S. M., Agostinho A. A., and Gomes L. C.. 2008. “A Unified Index to Measure Ecological Diversity and Species Rarity.” Ecography 31: 450–456. [Google Scholar]
Moreno‐Pérez, M. G. , García‐Luque I., Fraile A., and García‐Arenal F.. 2016. “Mutations That Determine Resistance Breaking in a Plant RNA Virus Have Pleiotropic Effects on Its Fitness That Depend on the Host Environment and on the Type, Single or Mixed, of Infection.” Journal of Virology 90: 9128–9137. [DOI] [PMC free article] [PubMed] [Google Scholar]
Morris, C. E. , and Moury B.. 2019. “Revisiting the Concept of Host Range of Plant Pathogens.” Annual Review of Phytopathology 57: 63–90. [DOI] [PubMed] [Google Scholar]
Moury, B. , Fabre F., Hébrard E., and Froissart R.. 2017. “Determinants of Host Species Range in Plant Viruses.” Journal of General Virology 98: 862–873. [DOI] [PubMed] [Google Scholar]
Ostfeld, R. S. , and Keesing F.. 2012. “Effects of Host Diversity on Infectious Disease.” Annual Review of Ecology, Evolution, and Systematics 43: 157–182. [Google Scholar]
Parratt, S. R. , Numminen E., and Laine A.‐L.. 2016. “Infectious Disease Dynamics in Heterogeneous Landscapes.” Annual Review of Ecology, Evolution, and Systematics 47: 283–306. [Google Scholar]
Peláez, A. , McLeish M. J., Paswan R. R., Dubay B., Fraile A., and García‐Arenal F.. 2021. “Ecological Fitting Is the Forerunner to Diversification in a Plant Virus With Broad Host Range.” Journal of Evolutionary Biology 34: 1917–1931. [DOI] [PubMed] [Google Scholar]
Power, A. G. , and Flecker A. S.. 2003. “17. Virus Specificity in Disease Systems: Are Species Redundant?” In The Importance of Species, edited by Peter K. and Simon A. L., 330–346. Princeton University Press. [Google Scholar]
R Core Team . 2023. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. https://www.R‐project.org/. [Google Scholar]
Rodríguez‐Nevado, C. , Gavilán R. G., and Pagán I.. 2020. “Host Abundance and Identity Determine the Epidemiology and Evolution of a Generalist Plant Virus in a Wild Ecosystem.” Phytopathology 110: 94–105. [DOI] [PubMed] [Google Scholar]
Rohr, J. R. , Civitello D. J., Halliday F. W., et al. 2020. “Towards Common Ground in the Biodiversity–Disease Debate.” Nature Ecology & Evolution 4: 24–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
Roossinck, M. J. , and García‐Arenal F.. 2015. “Ecosystem Simplification, Biodiversity Loss and Plant Virus Emergence.” Current Opinion in Virology 10: 56–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
Seabloom, E. W. , Borer E. T., Gross K., et al. 2015. “The Community Ecology of Pathogens: Coinfection, Coexistence and Community Composition.” Ecology Letters 18: 401–415. [DOI] [PubMed] [Google Scholar]
Shipley, L. A. , Forbey J. S., and Moore B. D.. 2009. “Revisiting the Dietary Niche: When Is a Mammalian Herbivore a Specialist?” Integrative and Comparative Biology 49: 274–290. [DOI] [PubMed] [Google Scholar]
Stukenbrock, E. H. , and McDonald B. A.. 2008. “The Origins of Plant Pathogens in Agro‐Ecosystems.” Annual Review of Phytopathology 46: 75–100. [DOI] [PubMed] [Google Scholar]
Valverde, S. , Vidiella B., Montañez R., Fraile A., Sacristán S., and García‐Arenal F.. 2020. “Coexistence of Nestedness and Modularity in Host–Pathogen Infection Networks.” Nature Ecology & Evolution 4: 568–577. [DOI] [PubMed] [Google Scholar]
Violle, C. , Navas M.‐L., Vile D., et al. 2007. “Let the Concept of Trait Be Functional!” Oikos 116: 882–892. [Google Scholar]
Wang, X. M. , Muller J., McDowell M., and Rasmussen D. A.. 2024. “Quantifying the Strength of Viral Fitness Trade‐Offs Between Hosts: A Meta‐Analysis of Pleiotropic Fitness Effects.” Evolution Letters 8: 851–865. [DOI] [PMC free article] [PubMed] [Google Scholar]
Woolhouse, M. E. , and Gowtage‐Sequeria S.. 2005. “Host Range and Emerging and Reemerging Pathogens.” Emerging Infectious Diseases 11: 1842–1847. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zamfir, A. D. , Babalola B. M., Fraile A., McLeish M. J., and García‐Arenal F.. 2023. “Tobamoviruses Show Broad Host Ranges and Little Genetic Diversity Among Four Habitat Types of a Heterogeneous Ecosystem.” Phytopathology 113: 1697–1707. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials