(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Why aphid virus retention needs more attention: Modelling aphid behaviour and virus manipulation in non-persistent plant virus transmission [1]

['Elin K. Falla', 'Department Of Plant Sciences', 'University Of Cambridge', 'Cambridge', 'United Kingdom', 'Nik J. Cunniffe']

Date: 2024-10

Plant viruses threaten food security and are often transmitted by insect vectors. Non-persistently transmitted (NPT) plant viruses are transmitted almost exclusively by aphids. Because virions attach to the aphid’s stylet (mouthparts) and are acquired and inoculated via brief epidermal probes, the aphid–virus interaction is highly transient, with a very short aphid virus retention time. Many NPT viruses manipulate their host plant’s phenotype to change aphid behaviour to optimise virus transmission. Epidemiological models of this have overlooked a key feature of aphid NPT virus retention: probing or feeding on a plant causes aphids to lose the virus. Furthermore, experimental studies suggest aphids could possibly inoculate multiple healthy plants within one infective period if they do not feed. Consequences of this for virus manipulation of host plant phenotype have not been explored. Our new compartmental epidemiological model includes both behaviour-based aphid dispersal and infectivity loss rates, and the ability of infective aphids to probe multiple plants before virus loss. We use our model to explore how NPT virus-induced host phenotypes affect epidemic outcomes, comparing these results to representative previous models. We find that previous models behave fundamentally differently and underestimate the benefit of an ‘attract-and-deter’ phenotype, where the virus induces increased aphid attraction to infected plants but deters them from prolonged feeding. Our results also highlight the importance of characterising NPT virus retention upon the aphid during probing. Allowing for multiple infective probes increases disease incidence and the effectiveness of virus manipulation, with implications for epidemic prediction and control.

Plant viruses can cause devastating disease epidemics. Non-persistently transmitted viruses are almost always vectored (transmitted between plants) by aphids. Experiments show virus infection can affect whether aphids are attracted to plants (by altering how infected plants ‘smell’), as well as whether aphids settle for an extended feed after a brief initial probe (by altering how infected plants ‘taste’). Since virus transmission requires an individual aphid to briefly probe an infected plant followed by one or more healthy plant(s), this strongly affects disease transmission. However, most studies exploring virus epidemics do not account for how aphid feeding behaviour affects how long an aphid holds the virus for, or that an aphid could infect multiple healthy plants before losing the virus. We use mathematical modelling to dissect how these aspects of aphid feeding behaviour affect virus transmission, particularly when viruses manipulate the ‘smell’ and ‘taste’ of plants. We show how previous studies, by omitting crucial aspects of aphid feeding behaviour, underestimate how viruses can promote their own transmission. We also highlight that there are very few experimental studies exploring the number of plants an aphid with the virus can consecutively infect, which is a key parameter affecting the severity of epidemics.

Introduction

Plant viruses make up an estimated 47% of pathogens that cause plant disease epidemics [1]. Plant viruses can be transmitted horizontally (i.e. from plant to plant) by vectors, which are typically arthropods [2]. Viruses can be classified by vector transmission type, “arguably the key epidemiological characteristic of plant viruses” [2]. This includes where the virus resides in/on the vector, whether the amount of virus increases within the vector, and for how long the vector retains the virus. Non-persistent transmission (NPT) has the shortest virus retention time on the vector, on a scale of minutes to hours [3, 4]. NPT viruses make up an estimated 42% of insect-vectored plant viruses [5], and are nearly exclusively transmitted by aphids [6], the most common plant virus vector [7]. Aphids are well-suited to be vectors as their piercing–sucking mouthparts can transmit viruses without badly damaging the plant [7]. The virus is acquired through brief initial epidermal probes of an infected plant by an aphid [8, 9], and retained on the aphid’s stylet (mouthparts) [3]. However, if the aphid’s stylet then penetrates beyond the epidermis to initiate feeding from the infected plant’s phloem, the aphid is unlikely to retain the virus [3, 10], as the virus is lost rapidly when the aphid ejects saliva from its stylet during this process [8, 9]. Therefore, transmission of NPT viruses depends on the aphid first dispersing from an infected plant after probing, then subsequently probing an uninfected plant to inoculate it with the virus [10].

Over the past decade or so, it has been increasingly recognised that NPT plant viruses frequently alter their host plant’s phenotype to influence aphid vector behaviour and virus transmission [11]. This can involve changing visual/olfactory cues to alter aphid host selection, changing palatability/quality cues to mediate feeding behaviour, and affecting aphid dispersal [11–13]. Virus manipulation of plant phenotype, also called vector preference [14, 15], was first reported for an NPT virus, cucumber mosaic virus, in 2010 [16], and has since been found in numerous experimental studies across a diverse range of NPT pathosystems, for multiple aphid vectors and hosts (see [10] for review).

For NPT viruses, broadly speaking, two different phenotypes have been found. The first, dubbed ‘deceptive’ [16] or ‘attract-and-deter’ [17], involves the virus inducing its host plant to emit volatiles to attract the vector, but also increasing production of distasteful metabolites to reduce likelihood of prolonged feeding after probing (i.e. deterrence). This virus phenotype likely increases transmission at local scales [11, 17–19], because NPT virus transmission is most efficient when aphids briefly probe an infected plant’s epidermis then reject it as unpalatable and disperse to another plant [20, 21]. The other virus phenotype, ‘attract-and-retain’ or just ‘retain’ [17, 22], invokes the opposite response to the ‘deter’ phenotype in the host, increasing palatability of virus-infected plants by inhibiting distasteful metabolite synthesis or increasing nutrient content, sometimes also enhancing aphid performance [23, 24]. This was previously thought to inhibit NPT virus spread [6], but may in fact encourage longer-term wider-scale spread of the virus by increasing the aphid population and promoting development of winged (alate) aphids [17, 22], since alate morphs are often produced in response to high local population densities [25].

Mathematical models have played a key role in understanding virus manipulation of plant phenotype, and knowledge of how vector dynamics affect disease spread [2], avoiding the obvious difficulties of performing field-studies [26]. The first plant-specific model to include epidemiological dynamics of the virus in the vector population, introduced by Jeger et al. [27] and extended to include vector migration by Madden et al. [4], was a linked compartmental (ordinary differential equation, ODE) model with a frequency-dependent transmission function, coupling disease spread in host plants and vectors. These models introduced a theoretical framework allowing all transmission types (NPT, semi-persistent transmission, persistent transmission) to be captured by changing model parameter values. However, this generality comes at the cost of a lack of specificity to any single transmission type.

This particularly applies to how NPT virus transmission is captured by the Madden et al. [4] model. Aphid feeding and dispersal behaviour is important to NPT virus transmission because of the transient nature of virus retention on the aphid. Despite this, the model assumes both (1) a constant aphid dispersal rate between pairs of plants in the same field, and (2) a constant rate of virus (infectivity) loss from the aphid (i.e. a constant aphid infective period; note throughout this paper we use infective/infectivity here as synonymous with the more traditionally-used term viruliferous or virus-bearing). These assumptions of constant rates are particularly restrictive when considering virus manipulation of the host, which alters aphid dispersal, landing and feeding behaviours in a way that potentially depends on the current state of any epidemic [14]. Note that the Madden et al. [4] model did not incorporate virus manipulation of host plant phenotype since the phenomenon had not been experimentally observed at the time the model was developed. However, compartmental models based on the Madden et al. [4] framework have since been commonly used to investigate this topic, although not always for NPT viruses (e.g., [28, 29]). One exception is Cunniffe et al. [30], who used the same model structure to investigate both NPT and persistently-transmitted viruses but distinguished between landing and feeding, while also relaxing the assumption of a constant aphid dispersal rate. However, those authors still assumed a constant aphid infectivity loss rate.

A different ODE model structure was devised by Donnelly et al. [17] that differentiates between aphid vector probing versus feeding behaviour and relaxes the assumption of constant aphid dispersal and infectivity loss rates. The model does this through use of a Markov chain that tracks aphid ‘feeding dispersals’, i.e. orientation and probing behaviours of a single aphid as it moves between successive plants until it settles for an extended feed. The infection process is modelled as a “by-product of aphid probing of plants during feeding dispersals rather than as frequency-dependent contacts between susceptible hosts and infected vectors” [17]. The number of virus transmission events per aphid feeding dispersal depends on the frequency with which vectors probe an infected host followed by a susceptible one. However, unlike in the compartmental ODE models exemplified by Madden et al. [4], aphids can only probe (and therefore infect) one susceptible plant before losing infectivity in the Donnelly et al. [17] model.

This may be a significant omission. Although it is established that prolonged feeding will very likely cause an aphid to lose the NPT virus from its stylet [3, 10], there is little experimental research investigating the potential for infective probes of consecutive susceptible plants by an aphid, if feeding does not occur. However, the small number of experiments that have been done suggest consecutive infective probes (and hence infections) are possible. For example, Watson [31] found that when given 5 minutes to probe 8 consecutive healthy plants, 100% of infective aphids were still infective by the second plant, and 10 of the 16 (62.5%) were still infective by the fourth. More recent quantitative studies on virus population bottlenecks have found similar results, estimating anywhere from 2 to 5+ plant probes before losing infectivity [32, 33]. This strongly suggests that an assumption of only probing a single susceptible plant per aphid infective period may be too severe. This is particularly likely in the case that the plant is a nonhost of the aphid species, as experimental evidence shows aphids will then reject a plant after very few probes [3].

Therefore, previous models of NPT virus transmission can, broadly-speaking, be partitioned into two classes: compartmental ODE models with constant rates of aphid dispersal and infectivity loss that do not consider aphid behaviour (but that can account for aphids infecting multiple plants within an infective period) based on Madden et al. [4], or the more recent Markov chain model of Donnelly et al. [17] that includes behaviour-based aphid rates, but no ability for multiple infective probes. In this study, we create two models based on each of these model structures, the former we refer to as the Multiple Infective Probes (MIP) model and the latter as the Behaviour-based Aphid Rates (BAR) model. We then compare them to a novel model we introduce in this paper, the MIP-BAR model. This model combines the functionality of the MIP and BAR models into a single easily-extensible system of differential equations, and allows us to answer the following three key questions (for convenience these are referenced throughout the following as Q1, Q2 and Q3).

[END]
---
[1] Url: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1012479

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.