As obligatory parasites, viruses require the host cell machinery to accomplish their infection cycle. Consequently, viruses are not able to replicate and to produce progenies outside permissive cells of compatible hosts. This strict dependency on living organisms represents a challenge for crop viruses for which cultivated hosts are absent during weeks/months following plant removal at harvest. For the maintenance of these parasites in the environment several strategies exist, including abilities i) to be vertically transmitted through the accumulation in seeds, tubers and vegetative organs of host plants (Zinga et al., 2013; Cieniewicz et al., 2017), and ii) to infect several host species, in both cultivated or non-cultivated areas, that have overlapping growing periods over the year (Alexander et al., 2014). Horizontal transmission of viruses that infect rooted hosts could rely on direct contacts (i.e. mechanical inoculations) between neighboring plants and/or on the involvement of vectors (e.g. nematodes, fungi and insects; (Brault et al., 2010)). The biology of these organisms and the nature of their interactions with viruses, allow transmission from short (few meters) to long (up to kilometers) (e.g. transmission of Plum pox virus by aphids; (Pleydell et al., 2018)) distances for few minutes/hours (e.g. aphid-borne viruses transmitted in a non-persistent manner; (Brault et al., 2010)) to years (e.g. long-term maintenance of viruliferous status of nematodes; (Demangeat et al., 2005)).

Host infection leads to the production of viral proteins which participate in all steps of the infectious process, including replication, short- and long-distance movement in planta and hijacking of both plant defences and metabolism (Alcaide et al., 2022). The efficiency of each of these steps, which depends on the ability of the virus to use the resources of the infected plant, determines the intensity of systemic infection which constitutes the fitness in planta of the pathogen (Elena et al., 2014; Alcaide et al., 2020). For plant viruses, fitness varies depending on biotic (host (species, genotype, age…) or pathogen (viral species and isolate, multiple infections…)) and abiotic (linked to the environment (e.g. water, temperature…)) parameters. The wide and intensive network of virus dynamics in the environment exposes host plants to many viral species, strains and isolates. Moreover, host ranges of plant viruses frequently include several species (Moury et al., 2017). This leads to the frequent occurrence of viral co-infections (e.g. Zinga et al., 2013). In co-infections, resources of infected plant are used by each pathogen partner (Elena et al., 2014). Consequently, the fitness in planta of a viral entity that infects a host can be i) unchanged (i.e. neutral interaction), ii) increased (i.e. synergistic interaction) or iii) decreased (i.e. antagonistic interaction) by the presence of a co-infecting virus. In return, the fitness in planta of the latter is impacted (i.e. symmetric interaction) or not (i.e. asymmetric interaction) by the competitor virus (Alcaide et al., 2020). In addition to raw changes in viral fitness in planta, plant physiology can be differentially impacted by single/co-infections (Alcaide et al., 2022) leading to the expression of more or less severe symptoms (Moreno and López-Moya, 2020). For insect-borne viruses, these changes in plant phenotype and physiology may impact performances and/or behaviour of the vectors (Domingo-Calap et al., 2020; Jia et al., 2022). Changes of virus fitness (qualitative and/or quantitative variations of infection rate and/or virus accumulation) and/or vector traits may result in modifications of the frequency and/or the efficiency of transmission events (Péréfarres et al., 2014). Thus, by impacting a wide range of plant-virus-vector interactions, co-infections can alter the infectious process, the severity and/or the spread of viral diseases.

Wheat (Triticum aestivum) is one of the four major staple foods cultivated worldwide. This plant species accounts in average for approximately 20% of the calorie intakes in the worldwide human diet (Shiferaw et al., 2013). Therefore, for numerous countries wheat production represents alimentary, economic and geopolitical interests, suggesting that wheat yield should at least be maintained at the current level to match food security. However, wheat is exposed to abiotic (e.g. drought, nutrient deficiency, UV…) and biotic (e.g. bacteria, fungi, viruses, insects…) stresses that can impact grain production. Among viral diseases affecting wheat, yellow dwarf disease (YDD) and wheat dwarf disease (WDD) can induce yield losses of up to 80 and 90%, respectively (Perry et al., 2000; Lindblad and Waern, 2002). YDD can be caused by ten virus, among which the barley yellow dwarf virus-PAV (BYDV-PAV; species: Luteovirus pavhordei; genus Luteovirus; family Tombusviridae (Scheets et al., 2020)) is described to be the most prevalent in France (Lister and Ranieri, 1995). Wheat dwarf virus (WDV; family Geminiviridae, genus Mastrevirus (Abt and Jacquot, 2015)) is the etiological agent of WDD. These viruses are transmitted in a persistent, non-propagative manner (Brault et al., 2010) by different insect vectors. BYDV-PAV is exclusively transmitted by aphids (mainly Rhopalosiphum padi (Linnaeus, 1758) and Sitobion avenae (Fabricius, 1775); (Parry, 2013)), whereas, WDDV is transmitted by leafhoppers (Psammotettix alienus; (Abt and Jacquot, 2015)). For the last three decades, these viral diseases were not considered an important threat for cereal growers. Indeed, to limit the incidence of YDD and WDD in cereals fields, insects (i.e. aphids and, at a lesser extent, leafhoppers) were efficiently managed in cereal fields with insecticides (Mc Namara et al., 2020). However, the recent ban of neonicotinoids in the EU (Jactel et al., 2019) associated with the description of Sitobion avenae clones resistant to pyrethroids in European countries (Fontaine et al., 2023) had a considerable impact on the availability of chemical resources to control insect-borne viral diseases in cereal fields. Thus, the current management methods against YDD and WDD are based on cultural practices (e.g. late sowing, management of volunteers (Östman et al., 2001; Lindblad and Waern, 2002)), the use of limited virus-resistance/tolerance resources (Pfrieme et al., 2022; Pichon et al., 2022) and/or the use of natural-based substances (Pichon et al., 2022). Despite these alternatives, the absence of neonicotinoids-coated seeds (that represented in 2016 in France 35% and 75% of sown wheat and barley seeds, respectively) could increase the exposure of cereals fields to insect-borne viral diseases (Mc Namara et al., 2020).

Wheat plants co-infected by BYDV-PAV and WDV have already been reported from field surveys (Jarošová et al., 2018; Liu et al., 2020). While main biological parameters involved in YDD and in WDD epidemiology have been studied, particularly for YDD (Van den Eynde et al., 2020), little is known about the outcome of BYDV-PAV – WDV interactions in small grain cereals. Based on our knowledge, Jarošová et al. (2018) provided a first attempt to study BYDV-PAV – WDV interactions by monitoring virus load in co-infected plants. Monitoring viral accumulation in infected plants (i.e. a proxy of virus fitness in planta) is an important step to study interactions between viruses. However, this approach does not allow per se a comprehensive understanding of the impact of co-infections on epidemiology of the disease induced by each viral partner. The aim of our study was to estimate the impact of co-infections on biological parameters involved in the epidemiology of BYDV-PAV and WDV. For each virus, monitored parameters (i.e. infection rate, virus accumulation and transmission efficiency) were evaluated at different steps of infection using single and competitive (i.e. simultaneous and sequential) inoculations.

Analyses of the epidemiology of plant viral diseases have long been based on data collected from biological, evolutionary and ecological properties of each partner involved in studied pathosystems. Co-infection of hosts, frequently observed in cultivated and uncultivated areas (Zinga et al., 2013; Jarošová et al., 2018; Liu et al., 2020), can change nature and intensity of the interactions between viruses, vectors and hosts, and thus modify our understanding of ecology and epidemiology of plant viruses (Alcaide et al., 2020). Plants co-infected with BYDV-PAV and WDV have been reported in the literature. However, little is known about if and how the sympatry between these two viruses can benefit or constrain their epidemiology. In this study, we evaluated the impact of co-infections on parameters of the epidemiology of YDD and WDD diseases. Experiments were carried out to evaluate and compare, through different competition scenarios (i.e. single- and co- (simultaneous and sequential) inoculations), efficiency of BYDV-PAV and WDV to infect, to accumulate in and to be spread by their respective insect vector from wheat plants. This deciphering of the epidemiological processes associated with these two viral diseases allowed to accurately describe that BYDV-PAV and WDV achieve their infection cycle and their plant-to-plant transmission with different efficiencies, and interact asymmetrically during co-infection of wheat. Moreover, the impact of competition scenarios on biological parameters of these two viruses was evaluated at different stages of the infection cycle and using plants at different ages. Results showed that BYDV-PAV – WDV interactions lead to different phenotypes ranging from synergism ( Figure 4B , Figure 4D , Figure 6 and supplementary Figure 1A ) to antagonism ( Figure 4D , Figure 5 , Table 4 and supplementary Figure 1A ).

For insect-borne viral diseases, the epidemiological process begins with the migration of viruliferous vector(s) from infected reservoirs towards crop fields. The intensity of primary inoculations relies on several parameters including abundance, sanitary status and behaviour (e.g. feeding and preferences) of insect vectors, and on the infection rate associated to virus inoculation (Jones et al., 2010). In western Europe, with winter wheat sown from early September to late October, YDD and WDD introductions in fields mainly occur in autumn. The aphid R. padi is the most abundant vector of BYDV-PAV in cereal fields in autumn (Gillet et al., 1990). The abundance and the population dynamics of this aphid species have been shown to predict YDD risk (Fabre et al., 2003a). The leafhopper P. alienus, the vector of WDV, is observed in winter cereal fields from sowing to the end of the year when temperatures are cold (Mehner et al., 2002). It has been shown that the abundance R. padi and P. alienus and the proportion of viruliferous individuals vary with space and time and are linked to the composition of the agroecosystem (Östman et al., 2001; Lindblad and Arenö, 2002; Mehner et al., 2002; Fabre et al., 2005). Insect behaviour (e.g. feeding, settling and mobility) is an important parameter determining which plants and how many plants are visited by each individual vector. Numerous factors, such as i) the plant species/cultivars (Tholt et al., 2015) and/or of competitor insect species (Alla et al., 2001), can impact preferences and/or feeding of insects. Moreover, it has been shown that behaviour traits of insects can be also influenced by their sanitary status and the sanitary status of visited host plants (Mauck et al., 2018). Finally, insects are able to re-take off several times after first landing leading to inoculation of several plants by a single insect (Parry, 2013; Abt and Jacquot, 2015). The abundance, the sanitary status and/or the behaviour of vectors have been extensively monitored and incorporated in predictive modelling studies targeting persistently transmitted plant viruses (Jones et al., 2010). For these viruses, the infection rate introduced in models is frequently considered as a constant close to 100% ((Jones et al., 2010) but see Thackray et al., 2009 for exception). However, under our experimental conditions infection rates of BYDV-PAV and WDV in a susceptible cultivar were i) below 93% and ii) can decreased with plant age at inoculation (e.g. BYDV-PAV; from 7 to 19 days old inoculated plants). These results suggest that during the first weeks after sowing, wheat plants show age-dependent level of susceptibility to these two viruses. This may lead to a range of efficiencies for virus introduction in wheat fields. Positive (Zhang et al., 2021) and negative (Lindblad and Sigvald, 2004) impacts of plant age on host susceptibility to pathogens have been reported for several pathosystems, suggesting that host plant development at inoculation represent a key parameter to understand the ecology and epidemiology of viral diseases (Ashby and Bruns, 2018). Together with these previously published works, our results highlight that, when modelling the intensity of the introduction of a given viral disease in fields, infection rate should not be considered as a constant either for a group of viruses (e.g. persistently transmitted viruses) or for a virus species (e.g. BYDV-PAV and WDV).

From sowing to harvest, several species of insects and/or viruses are introduced in and spread within fields. This continuous biological process leads to frequent virus co-inoculations of plants (Zinga et al., 2013; Tollenaere et al., 2016). The ecology of P. alienus (vector of WDV) and R. padi (vector of BYDV-PAV), and the report of cereal plants co-infected with BYDV-PAV and WDV (Jarošová et al., 2018; Liu et al., 2020) imply that these vectors occur in sympatry in cereal fields. However, under laboratory conditions, Alla et al. (2001) showed that simultaneous deposition on cereal plantlet of R. padi and P. alienus i) increased mortality, ii) slowed down the development and iii) modified the behaviour of the leafhopper P. alienus. These data are in favor to antagonistic interactions between these two insects. However, simultaneous inoculations of BYDV-PAV and WDV carried out on wheat plants of different ages (this work) is associated with an increase of the total infection rate of BYDV-PAV for 15-and-19 days old inoculated plants indicating that BYDV-PAV and WDV interact synergistically in favor to BYDV-PAV. Moreover, simultaneous inoculations led to either similar (i.e. 7-and 12-days old plants), lower (i.e. 9 days old plants) or higher (i.e. 15-and 19-days old plants) occurrence of co-infections compared with expectations from independency between these partners. This indicates that, depending on the age of inoculated plants, BYDV-PAV – WDV interactions resulting from simultaneous inoculations vary from antagonism to synergism, but the effect of these interactions on the total infection rate of viruses only benefit to BYDV-PAV.

For two insect-borne viruses transmitted by two different vectors, simultaneous inoculations require that insects land and inoculate viruses on the same plant in a short time window, which limit the probability of occurrence in fields. However, co-infections can also result from sequential inoculations of the two viruses. In that scenario, the pre-infecting virus has the opportunity to systemically infect the host which may impact the physiology of the latter (Alcaide et al., 2022). Then, compared with a healthy host, a pre-infected plant may constitute a different environment for a competitor virus modifying its capacity to replicate and/or to accumulate (Chávez-Calvillo et al., 2016). Depending on the delay between the two inoculation steps, pre-infection has neutral to beneficial effects on the infection rates of BYDV-PAV and WDV. Improvement of the infection rate of WDV occurred earlier than for BYDV-PAV, suggesting that the synergy between these viruses is asymmetric in favor to WDV. This indicates, as reported for other pathosystems (Alcaide et al., 2020; Moreno and López-Moya, 2020), that the order of inoculation of these two viruses is an important factor for virus-virus interactions in plant and their consequences on the epidemiology of associated diseases. Overall, our results show that the sympatry between BYDV-PAV and WDV impacts their infection rates in complex ways depending of plant age, pre-infection duration and the order of virus inoculation. The increase of BYDV-PAV infection rate resulting from simultaneous and sequential inoculations, suggests that the intensity of primary inoculations may be impacted by WDV. This could lead to more infected plants from which disease spread can occur within fields.

For WDV and BYDV-PAV, secondary inoculation is a key parameter for disease prevalence in cereals fields (Leclercq-Le Quillec et al., 2000; Lindblad and Sigvald, 2004). Thus, together with the dynamics of vector populations (Fabre et al., 2003a), agricultural practices (Östman et al., 2001; Lindblad and Waern, 2002) and the quality of reservoirs (Van den Eynde et al., 2020), the differences in accumulation of BYDV-PAV and WDV in infected plants could lead to contrasted efficiencies in plant-to-plant transmissions at field scale. While predictive modelling studies on viral disease epidemics are still lacking for WDV, several models exist for BYDV-PAV (Jones et al., 2010 and the reference therein). In these models, the latency period of the host (i.e. delay between infection and infectious status) are frequently considered as a constant varying from days (e.g. 4 days; Kendall et al., 1992) to weeks (e.g. 2.5 weeks; Thackray et al., 2009). Our data indicate that, in the susceptible wheat cv. Rubisko, plants infected with BYDV-PAV reached maximal infectious status before the 7^th day of infection, which is consistent with previously published data (Pichon et al., 2022). This result should be considered for any future improvement of BYDV-PAV epidemiological model.

The fitness of a virus in planta relies mainly on host-virus interactions. However, during co-infections, each virus uses host resources to complete its infectious cycle. This competitive phenomenon can modify the fitness landscape of each competitor (Elena et al., 2014). In this study, co-infections with BYDV-PAV and WDV led to opposite (i.e. positive and negative effects for BYDV-PAV and WDV, respectively) and antagonistic (i.e. against WDV) interactions. Mechanisms leading to changes in viral fitness during co-infections can result from indirect (i.e. involving host-virus interactions) and/or direct interactions between viruses (Tollenaere et al., 2016). Viral silencing is the main host defense mechanism involved in antagonistic and synergistic effects of co-infections (Moreno and López-Moya, 2020). Interactions between viral proteins may lead to negative or positive impact(s) in one or several step(s) (e.g. host range (Abt et al., 2020), replication (Fontenele et al., 2021) and or cellular tropism (Mayo et al., 2000)) of the virus infection cycle. Direct mechanisms require intimate relationship between viral component(s), which is more likely to result from long term and frequent sympatry between viruses. While BYDV-PAV and WDV have been reported in France for several decades, the frequency of co-infections of plants in wheat fields has not been accurately surveyed yet. However, the prevalence of WDV in French fields has been reported to be highly variable in both space and time (Abt and Jacquot, 2015), which seems to reduce the likelihood of frequent co-infection. In this scenario, indirect mechanism(s) may be more likely involved in the changes in the fitness in planta of these viruses during co-infections. This should be confirmed in future studies aiming at deciphering the molecular mechanisms underlying the interactions between BYDV-PAV and WDV in co-infected plants.

Transmission of BYDV-PAV was not impacted by co-infection with WDV. A similar result was found for WDV at early stage of infection, whereas the capacity of co-infected plants to be source of WDV for insect-mediated transmission was reduced at late stage of infection. This asymmetric antagonism is associated with lower WDV load in co-infected plants. These results are consistent with the accumulation-transmission trade off (Froissart et al., 2010) and with previous reports evaluating the effect of co-infections on the accumulation and transmission of other persistently transmitted viruses (e.g. Péréfarres et al., 2014). However, changes in the feeding behaviour of vectors on co-infected plants have also been reported in the literature (Domingo-Calap et al., 2020; Jia et al., 2022). Thus, modifications of P. alienus behaviour on BYDV-PAV – WDV co-infected plants should also be considered as a possible factor for the observed lower WDV transmission rate from co-infected source plants. As a change in the source quality of a host is expected to alter pathogen spread in a healthy host population (Froissart et al., 2010), our results suggest that co-infections may negatively impact the spread of WDV in wheat fields. However, virus spread also relies on the performances and behaviour of insect vectors (Shaw et al., 2017). These parameters can be modified by sanitary status of host plant (Domingo-Calap et al., 2020) and/or the presence of a competitor insects (Alla et al., 2001). To our knowledge, the effect of WDV – BYDV-PAV co-infections on vector traits have not yet been reported yet. In the future, studying how co-infections can impact the different steps involved in plant-to-plant transmission (i.e. acquisition, latency in vector and inoculation parameters) and vector traits may contribute to a better description of the consequences of the sympatry of BYDV-PAV and WDV on the sanitary status of cereal fields.

Numerous studies on viral co-infections have only evaluated viral fitness in planta and/or virulence under a low number of experimental conditions. These approaches do not allow per se to fully understand the effect of co-infections on the epidemiology on viral diseases. To better understand the impact of BYDV-PAV – WDV interactions on the epidemiology of these two viruses, we included several parameters involved in introduction, maintenance and spread of viruses in our experimental design. Our results suggest that the presence of WDV could have neutral to positive effects on biological parameters involved in primary (i.e. infection rate following simultaneous and sequential inoculations) and secondary (i.e. viral accumulation and transmission efficiency) inoculations of BYDV-PAV in wheat fields. Consequently, this suggests that the epidemiology of BYDV-PAV could benefit from the co-occurrence of WDV. For the latter, sequential inoculations with BYDV-PAV may improve the success of primary infections. However, co-infected plants are poor sources of WDV for leafhopper-mediated transmission, which could negatively impact the spread of this virus in a healthy plant population. These data highlights that the sympatry between BYDV-PAV and WDV seems to exert opposite pressures on WDV epidemiology, making it difficult to evaluate the net effect of co-infections on the epidemiology of WDD disease. In the context of the increased exposure of crops to insect vectors (recent ban of neonicotinoids in Europe and climate change), these findings represent a first step in the understanding of the impact of co-infections of BYDV-PAV and WDV on the biological and ecological parameters of diseases caused by these viruses. Next steps would be the characterization of molecular mechanism(s) underlying BYDV-PAV – WDV interactions and the evaluation of the net effect of the sympatry between these viruses and their respective vectors on the epidemiology of YDD and WDD in cereal fields.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

TA, MS and EJ conceived and designed the experiment. TA, MS, LK, KG and EJ performed the experiment. TA analysed the data. TA and EJ wrote the article. All the authors read and approved the submitted version of the manuscript.