Polarization of foliar reflectance: novel host plant cue for insect herbivores

Abstract

Insect herbivores exploit plant cues to discern host and non-host plants. Studies of visual plant cues have focused on colour despite the inherent polarization sensitivity of insect photoreceptors and the information carried by polarization of foliar reflectance, most notably the degree of linear polarization (DoLP; 0–100%). The DoLP of foliar reflection was hypothesized to be a host plant cue for insects but was never experimentally tested. Here, we show that cabbage white butterflies, Pieris rapae (Pieridae), exploit the DoLP of foliar reflections to discriminate among plants. In experiments with paired digital plant images, P. rapae females preferred images of the host plant cabbage with a low DoLP (31%) characteristic of cabbage foliage over images of a non-host potato plant with a higher DoLP (50%). By reversing the DoLP of these images, we were able to shift the butterflies' preference for the cabbage host plant image to the potato non-host plant image, indicating that the DoLP had a greater effect on foraging decisions than the differential colour, intensity, or shape of the two plant images. Although previously not recognized, the DoLP of foliar reflection is an essential plant cue that may commonly be exploited by foraging insect herbivores.

1. Introduction

Locating, feeding, and laying eggs on suitable host plants enable insect herbivores to maximize their fitness and that of their offspring [1]. Foraging for suitable plants, insects exploit plant cues with visual, infrared, olfactory, tactile, or gustatory characteristics [2–4]. Studies of visual plant characteristics have largely focused on plant colour, brightness (intensity of perceived reflected light), or shape [2,5]. Yet, differential polarized reflections from plant foliage have long been hypothesized to guide plant-foraging insects [6,7].

For polarized reflections from plant foliage to serve as a host plant indicator for insect herbivores, three criteria must be met: (i) the DoLP of foliar reflections must differ between host and non-host plants, (ii) the insects’ photoreceptors must be capable of sensing and processing plant-derived polarized light, and (iii) the specific DoLP of host plant foliage reflections must inform plant selection decisions by foraging insects. We investigated these criteria with our model organism, the cabbage white butterfly, Pieris rapae.

Like water or glass surfaces, plant foliage surfaces can polarize sunlight in a direction parallel to the surface, through specular reflection [8] (figure 1a). This direction, or axis of polarization (AoP, 0–180°), is dependent upon the relative positions of the Sun, the reflecting leaf surface, and the foraging insect (figure 1b). The DoLP, the fraction of the light that is polarized in the predominant AoP (figure 1b), is also affected by many leaf characteristics that differ among plant species, such as pigmentation, pubescence, epicuticular waxes, and surface undulations, or even by viral infection [10,11]. Females of P. rapae lay eggs on brassicaceous plants including cabbage and rutabaga which possess an epicuticular wax layer giving their leaves a matt appearance [2] and a lower DoLP compared to many other plants (electronic supplementary material, figure S4).

Figure 1.

Diagrams depicting polarization by reflectance and an ommatidium (photoreceptor unit) of the P. rapae compound eye. (a) Unpolarized light (light vibrating equally in all directions) from the Sun is polarized via reflection from surfaces such as water or plant foliage. Light vibrating in the direction parallel to the surface is preferentially reflected resulting in polarization. (b) Light reflections from cabbage (bottom) or potato (top) foliage (note colour and shape differences); associated compass diagrams show the distribution in vibration direction of waves composing each light ray. The predominate direction of vibration, or AoP (0–180°), is represented by the direction of the compass needles. The DoLP as a measure of the anisotropy of vibration directions is depicted as the amount of spread around this predominate direction and by the size of the compass needle. (c) Diagram of an ommatidium. (d) Cross-sectional diagram of an ommatidium showing the eight photoreceptors (R1–8). Diagrams in c and d modified from [9]. (e) Electron micrograph showing the parallel microvilli of photoreceptors R1–4 composing the rhabdom. (f) The resulting modulations in sensitivity, with changes in the AoP of incident light, of the indicated photoreceptors. AoP, axis of polarization; DoLP, degree of linear polarization. (Online version in colour.)

The photoreceptors of insects and other arthropods including those of P. rapae can sense polarized light. Both the AoP and the DoLP of light affect photoreceptor responses through differential absorbance by photopigments [12] embedded within the cellular membrane of microvilli composing the ommatidial rhabdom (figure 1c–e). This arrangement makes these photoreceptors more sensitive to light with an AoP that is parallel to the microvilli (figure 1f). Increasing the DoLP of a stimulus light increases the differential response of photoreceptors to the AoPs of light. It is through these mechanisms that both the AoP and the DoLP affect the responses of insect photoreceptors. The visual system of P. rapae has been extensively studied [9,13–15]. Electrophysiological recordings and electron microscopy demonstrated that the blue-sensitive and red-sensitive photoreceptors, and a subset of green-sensitive photoreceptors, are sensitive to vertically, obliquely, and horizontally polarized light, respectively [16].

2. Material and methods

Details of the materials and methods are fully described in the electronic supplementary material.

3. Results

(a). Depolarizing-filter experiment

To investigate whether the DoLP of foliar reflections informs plant selection decision by foraging P. rapae, we ran a series of behavioural experiments. When we offered female P. rapae a choice between a live host plant (cabbage, Brassica oleracea) with a low DoLP (31%; electronic supplementary material, figure S3) and a live non-host plant (potato, Solanum tuberosum) with a high DoLP (50%) in the absence of plant odour (electronic supplementary material, figure S1), we observed a strong preference for the cabbage host plant (figure 3a). To determine whether the differential DoLP of the stimulus plants informed the females' plant choice, we added a depolarizing filter to both stimulus windows of the bioassay arena, thereby reducing and equalizing the DoLP of the two stimulus plants. With the information conveyed by the DoLP removed, P. rapae females failed to select their cabbage host plant (figure 3a), demonstrating the importance of the DoLP as an essential host plant cue.

Figure 3.

DoLP affects plant choice by P. rapae. (a) Without access to plant odours, females prefer a live cabbage host plant (right) over a live potato non-host plant (left) when polarized light cues are intact (top bar). This preference disappears when these cues are removed with a depolarizing filter (bottom bar). (b) Females also prefer the image of a cabbage host plant (right) over the image of a potato non-host plant (left) when presented with a DoLP matching that of live plants. The preference could be removed (bottom row) or even reversed (middle row) by changing the DoLP of the images. Numbers of females responding to each stimulus are shown within bars. The asterisk(s) either indicate(s) a percentage deviating from 50% or a significant difference between two percentages (χ² test, *p < 0.05, **p < 0.01, ***p < 0.001). DoLP, degree of linear polarization. (Online version in colour.)

(b). Liquid crystal display monitor proof-of-concept experiment

Digital plant images (relative to live plants) offer greatly enhanced opportunities of independently manipulating visual characteristics to tease apart their potential roles in host plant foraging. We therefore designed a novel combination of λ/4 retarder films and liquid crystal displays (LCDs) to modify and display static plant images. Previous uses of LCDs or projectors were either limited in their ability to modify DoLP and intensity [8] or limited in their ability to display colour [17]. Our set-up allowed for pixel-level control of colour and intensity, and global control of both AoP and DoLP. We used a bioassay arena (figure 2) where LCD monitors displaying plant images replaced live plants. LCDs emit highly polarized light (greater than 95%) due to linear polarizers used in their construction. We manipulated the AoP and DoLP of plant images by rotating the LCDs and counterrotating the images, and by changing the alignment between the λ/4 retarder film and the AoP of the LCD, respectively. We used photo polarimetry to both generate the potato and cabbage plant images tested in bioassays and to characterize the AoP and DoLP of both plant species. The DoLP of potato foliage (50%) and cabbage foliage (31%) differ markedly, but the species are similar in their AoP (electronic supplementary material, figures S3 and S4).

Figure 2.

Bioassay set-up to test behavioural responses of female P. rapae to image stimuli. (a) Diagram of experimental arena. (b) Exploded view of the arrangement of components between the LCD monitor and the stimulus windows. LCD, liquid crystal display. (Online version in colour.)

When we offered P. rapae females a choice between images of cabbage or potato plants, each matching the mean DoLP and the modal AoP of the corresponding plant species, most females selected the cabbage image (figure 3b, top bar), thus demonstrating the feasibility of testing plant images, instead of live plants, for behavioural responses of bioassay insects. Therefore, we proceeded to isolate and test the exclusive effect of the DoLP on the insects’ responses. When we offered P. rapae females a choice between a cabbage image with a DoLP (50%) approximating that of a potato plant and a potato plant image with a DoLP (31%) approximating that of a cabbage plant, most females selected the potato plant image (figure 3b, middle bar). By simply reversing the DoLP of the two images, we were able to make the virtual non-host potato plant as attractive (47 : 19 preference ratio) as the virtual cabbage host plant with its typical DoLP (47 : 18 preference ratio). This astounding result indicated that the DoLP was a more important cue in these bioassays than the differential colour (although differences were small), intensity, and shape of the two plant images. Moreover, when we kept the DoLP of both the cabbage and the potato plant image unnaturally low (less than 15%), the distinct shape and colour of the cabbage host plant were insufficient to attract P. rapae females (figure 3b, bottom bar), emphasizing again the importance of the DoLP as a host plant cue.

(c). Degree and axis of polarization preference experiments

To determine the range of the AoP and the DoLP of foliar reflections that remain attractive to P. rapae females, we offered females a choice between a cabbage image that varied in either AoP or DoLP, and a cabbage control image with a fixed DoLP of 31% and a fixed AoP of 90° (figure 4). In these experiments, cabbage images with an AoP at or near 45° and 135° proved repellent, whereas cabbage images with any other AoP were equally attractive. Furthermore, most cabbage images with a DoLP less than, or greater than, the DoLP (31%) indicative of cabbage were repellent to P. rapae females. Combined, these results indicate that P. rapae females are attracted to a DoLP indicative of a host plant (electronic supplementary material, figure S4) but are relatively indifferent to the AoP of plants, except for repellency to AoPs near 45° and 135°. The indifference of P. rapae females to most AoPs greatly enhances the utility of the DoLP as a foraging cue because the AoP of plant reflections will vary considerably depending on the position of the insect and the Sun relative to the plant. Furthermore, unlike the DoLP, the AoP is largely unaffected by foliage surface characteristics, as shown by our polarimetry [8] (electronic supplementary material, figure S4).

Figure 4.

Both the DoLP and the AoP affect plant choice by P. rapae. (a) A cabbage image with an AoP of 45° or 135° was repellent to females. (b) Most images with a DoLP above or below that typical of cabbage (31%) were discriminated against by females. Responses to a treatment image with a DoLP and an AoP identical to those of the control image were assumed to be 50%. Numbers of females responding to each treatment are shown within bars. The asterisk(s) indicate(s) either a proportion deviating from 50% or a significant difference between two percentages (χ² test, p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001). AoP, axis of polarization; DoLP, degree of linear polarization. (Online version in colour.)

4. Discussion

This is the first study documenting that the polarization of foliar reflectance serves as a host plant cue for insect herbivores. Based on our data, and considering the typically small differences in foliage colour between plant species [18], it seems that relative differences in DoLP among plants could be more informative host plant cues than plant shape, foliage colour, or intensity. Many insects exploit polarized light during navigation, and aquatic insects use horizontally polarized light to locate bodies of water for oviposition [12]. Most non-aquatic insects were once thought to lack polarization-sensitive photoreceptors in the ventral portion of their compound eyes. However, more recent histological and electrophysiological work indicates that this type of polarization sensitivity could be widespread among insect taxa [19–23] including herbivores other than P. rapae.

Most of the insect visual systems studied to date are incapable of independently perceiving DoLP and AoP because these systems rely on information from a single polarization-sensitive photoreceptor or from the comparison between two such photoreceptors [24]. To fully disentangle the effects of DoLP, AoP, intensity, and colour as foraging cues, comparison among at least three photoreceptors with similar spectral sensitivity, but with sensitivity to distinct AoPs, would be required [25]. It is therefore likely that P. rapae does not perceive differences in DoLP in isolation from other characteristics of light. The neurological mechanism(s) in P. rapae underlying the observed discrimination of stimuli with contrasting DoLP remain(s) unknown. Papilio butterflies perceive polarization differences as colour or intensity differences depending on the behavioural context [6,26]. Both mechanisms are plausible for P. rapae, but specific behavioural experiments are needed to determine the photoreceptors that are involved and how they perceive DoLP differences.

The fitness benefits foraging insects accrue by exploiting polarization host cues will depend upon the specificity of these cues. Our measurements (electronic supplementary material, figure S4) and those of a previous study [10] revealed significant variation among plant species. Species within genera (most prominently Brassica spp. and Solanum spp.) have a similar DoLP, whereas genera within a plant family (e.g. Brassica and Sinapis in the Brassicaceae) have distinctly different DoLPs. These findings suggest that polarization host cues have the greatest utility for insect herbivores that specialize on a single plant genus or on several closely related genera. However, any differences in polarization host cues among genera are not absolute, in that the DoLP also pertains to the viewing angle of the foraging insect [8].

The complementary information conveyed by plant-derived polarization cues could help insect herbivores locate and select optimal host plants. As optimal hosts confer significant fitness benefits to plant herbivores [27], it follows that there might be strong selection pressure for foraging insect herbivores to exploit plant polarization cues. The preference of P. rapae for DoLPs approximating those of both matt host plants (Brassica spp. approx. 30%) and shiny host plants (Sinapis approx. 70%) supports the concept that the additional information provided by the DoLPs of foliar reflections confer fitness benefits. Avoiding areas with DoLPs below 30% may be adaptive in that these areas are more likely to be shaded, and without direct solar illumination will lack the host information provided by polarized specular reflections. The benefits of these cues are further evident by improved larval performance on wild cabbage plants with a bluish appearance [28] that presumably had a significant epicuticular wax layer and thus a low DoLP of foliar reflections [18]. The failure of P. rapae females to visually discern among cabbage host plants [28], or between cabbage host plants and lettuce (Lactuca sativa) non-host plants [29] when polarization cues were not considered, also points to the DoLP of foliar reflections as an important host plant cue.

A sound understanding of how polarized light cues inform host plant selection decisions by insect herbivores will present pest managers and plant breeders with new options to lower the ‘apparency’ of host plants. For example, breeding plant lines with reduced leaf wax [30] or spraying plants with kaolin clay suspensions [31] will modify foliar surface characteristics and polarizations of their reflections, thus rendering plants less apparent to specific insect herbivores. However, the many potential trade-offs of these types of interventions (e.g. changes in leaf surface affecting water-use efficiency or resistance to generalist insect herbivores or pathogens) will require in situ, system-specific investigations prior to large-scale implementations of any intervention.

Data accessibility

Data are available from the Dryad Digital Repository: https://dx.doi.org/10.5061/dryad.xgxd254bs [32].

Authors' contributions

M.C.G., G.S.H., and H.G. performed the bioassays. M.C.G., S.C., and A.J.B. preformed polarimetry. A.J.B., G.G., and M.C.G. designed experiments. G.G. supervised the project. A.J.B. and G.G. wrote the paper.

Competing interests

The NSERC-Industrial Research Chair to G.G. was supported by Scotts Canada Ltd. as the industrial partner.

Funding

This study was supported by an Alexander Graham Bell Canadian Graduate Scholarship to A.J.B., NSERC Undergraduate Student Research Awards to M.C.G., G.S.H., H.G., and S.C., and by an NSERC-Industrial Research Chair to G.G.