Insects have evolved morphological and physiological adaptations in response to selection pressures inherent to their ecology. Consequently, visual performance and acuity often significantly vary between different insect species. Whilst psychophysics has allowed for the accurate determination of visual acuity for some Lepidoptera and Hymenoptera, very little is known about other insect taxa that cannot be trained to positively respond to a given stimulus. In this study, we demonstrate that prior knowledge of insect colour preferences can be used to facilitate acuity testing. We focused on four psyllid species (Hemiptera: Psylloidea: Aphalaridae), namely Ctenarytaina eucalypti, Ctenarytaina bipartita, Anoeconeossa bundoorensis and Glycaspis brimblecombei, that differ in their colour preferences and utilization of different host-plant modules (e.g. apical buds, stems, leaf lamellae) and tested their visual acuity in a modified Y-maze adapted to suit psyllid searching behaviour. Our study revealed that psyllids have visual acuity ranging from 6.3 to 8.7 deg. Morphological measurements for different species showed a close match between inter-ommatidial angles and behaviourally determined visual angles (between 5.5 and 6.6 deg) suggesting detection of colour stimuli at the single ommatidium level. Whilst our data support isometric scaling of psyllids’ eyes for C. eucalypti, C. bipartita and G. brimblecombei, a morphological trade-off between light sensitivity and spatial resolution was found in A. bundoorensis. Overall, species whose microhabitat preferences require more movement between modules appear to possess superior visual acuity. The psyllid searching behaviours that we describe with the help of tracking software depict species-specific strategies that presumably evolved to optimize searching for food and oviposition sites.
The occurrence of insects in a wide array of ecological niches has contributed to the tuning of their visual systems to optimize their capacity to perform key tasks under relevant light conditions (Snyder, 1979; Briscoe and Chittka, 2001). The capacity of visual systems to reliably detect a stimulus is determined, in part, by the spectral sensitivity of their photoreceptors, which governs the wavelengths at which excitatory responses are elicited. Consequently, the salience of a stimulus depends on its reflectance and the background composition with which it creates a chromatic and/or achromatic contrast. In complex natural environments, other variables such as stimulus size, shape and distance from the eye also strongly influence behavioural outcomes. Visual acuity (also termed spatial resolution) is constrained by morphological characteristics of the eye: the number and size of facets (influencing light sensitivity) and the angle between facets (influencing spatial resolution) act to determine spatial limitations of object detection under given light conditions (Kirschfeld, 1976; Snyder, 1979; Land, 1997; Land and Chittka, 2013). Small inter-ommatidial angles offer better acuity and therefore allow insects to detect relatively small stimuli located at greater distance (Land, 1997; Land and Chittka, 2013). Conversely, for an equal eye size, lower inter-ommatidial angles are acquired at the expense of an increase of the number of ommatidia competing for photons, resulting in a loss of light sensitivity. Although this can be compensated by a greater ommatidial diameter, this will translate into a loss of resolution. Hence, adaptations in eye morphology and physiology appear to be driven by evolutionary trade-offs that are responses to light environments and the tasks insects need to achieve to enhance survival (Snyder, 1979; Briscoe and Chittka, 2001; Spaethe and Chittka, 2003).
Visual acuity is of particular interest for understanding insect search behaviour and strategy as it influences the spatial basis of stimuli detection in their natural environment (e.g. prey, flowers or leaves). Insect visual acuity has been determined for species of Lepidoptera and Hymenoptera by taking advantage of conventional learning paradigms that condition insects to respond to specific stimuli associated with a nutritional reward (von Frisch, 1967; Lehrer and Bischof, 1995; Kinoshita et al., 1999; Takeuchi et al., 2006; Dyer et al., 2008; Galizia et al., 2011; de Ibarra et al., 2014). Indeed, the utility of conditioning techniques has meant that these insects have been used as model species to understand insect vision. The use of psychophysics has facilitated demonstration of the ability of honeybees, Apis mellifera, to use both colour and intensity contrasts in detection, orientation and landing tasks (Srinivasan and Lehrer, 1988; Lehrer and Srinivasan, 1993; Lehrer, 1994; Lehrer and Bischof, 1995). The relative importance of chromatic or achromatic dimensions of honeybee vision has been shown to depend on specific visual tasks. Achromatic vision is used for tasks such as navigation (Srinivasan, 2014), shape recognition (Stach et al., 2004), assessment of size (Avarguès-Weber et al., 2014), optic flow processing (Chittka and Tautz, 2003) and distance estimation (long range), whereas colour vision is mainly used for flower selection over short distances (Giurfa et al., 1996). However, some diurnal Lepidoptera rely more heavily on colour rather than on achromatic contrast (Kelber, 2005) even at visual angles close to that of their inter-ommatidial angle (≈1 deg) (Takeuchi et al., 2006). Interestingly, in spite of relatively comparable spectral sensitivities and eye structures, bumblebees were shown to detect colours at significantly smaller visual angles than honeybees (≈3 versus 15 deg for honeybees) (Dyer et al., 2008; Wertlen et al., 2008) but to not perform as well in colour discrimination tasks (Dyer et al., 2008). This suggests a trade-off between acuity and colour discrimination that may arise as a result of different ecological habitats in which a bee species evolved (Dyer et al., 2008; Bukovac et al., 2013). Energetic costs associated with the maintenance of sensory organs and the coding of visual information may regulate these trade-offs and lead insects to reduce to a strict minimum the morphological and resource allocation to their visual system (Niven et al., 2007).
The inter-ommatidial angle is also used as an indicator of spatial resolution. Predatory insects generally exhibit the smallest inter-ommatidial angles, e.g. 0.24 deg for the dragonfly Anax junius, conferring on them a higher resolution as required for predation. For comparison, the inter-ommatidial angle of lepidopterans and hymenopterans ranges between 0.5 and 2.5 deg, e.g. 1.7 deg for A. mellifera (Land, 1997). However, these measurements are somewhat less informative than behavioural data as they do not necessarily correspond to minimum visual angles required for detection and do not necessarily allow for the assessment of minimum angles for colour detection (Giurfa et al., 1996; Dyer et al., 2008). In addition, inter-ommatidial angles on their own may only indicate physical limitations at the peripheral level (i.e. eye structure) but may not give an insight into subsequent neural processing of visual inputs (Dyer et al., 2011).
Psyllids (Hemiptera: Psylloidea) are small sucking insects, the majority of which are highly host specific for many economically and ecologically important plant species, which they directly (damage caused by feeding) or indirectly (pathogen vector) injure (Hodkinson, 2009; Grafton-Cardwell et al., 2013; Nissinen et al., 2014; Walker et al., 2014). Australian species (i.e. approximately 15% of the world's psyllid fauna) are of particular interest with regard to host specificity because of their explosive radiation on myrtaceous hosts; 90% of these species feed on Eucalyptus alone (Yen, 2002; Austin et al., 2004; Hollis, 2004; de Queiroz et al., 2012). Currently, the mechanisms that underpin psyllid host specificity, such as host finding, nutrition and dispersal, are not well understood. Hence, greater insight into how psyllids perceive and exploit different host cues (such as leaf colour) would be of great value for identifying key factors that determine host range. There is a paucity of physiological, behavioural or morphological data on visual acuity in Hemiptera generally. These studies include behavioural observations of the detection thresholds of ‘yellow’ targets relative to their size by aphids (Moericke, 1955), the measurement of the inter-ommatidial angles of a limited range of Hemiptera including two aquatic (predatory) species, namely Gerris paludum and Notonecta glauca (Land, 1997) and a more recent compilation of morphological optics data on 20 aphid species (Döring and Spaethe, 2009). No study that we are aware of has been able to combine these important methodologies to enable robust between-species trade-off comparisons.
Plants leaves vary in shape, size and colour depending on species and leaf ontogeny. Hence, the capacity of the psyllid's visual system to detect host leaves relies on its perception of leaf colours and its ability to resolve objects in natural conditions. Colour vision is the perception and discrimination of light wavelength radiation independently from intensity (Kelber and Osorio, 2010). The basic requirements to demonstrate colour vision in an organism include the presence of at least two classes of photoreceptor of varying spectral sensitivity, behavioural evidence for intensity-independent colour discrimination and evidence of the ability to process different spectra inputs via colour opponency mechanisms (Kelber and Osorio, 2010; Kemp et al., 2015). Considering colour vision has, to date, only been conclusively demonstrated for a limited number of vertebrate and invertebrate model organisms, a framework was recently presented for contributions pertaining to the possible existence of colour perception in new taxa for which not all information currently exists to definitely prove colour vision (Kemp et al., 2015). Here, we thus use the term ‘colour vision’ with respect to psyllids because: (1) the existence of different classes of photoreceptors (i.e. UV, blue, green) has been demonstrated in a number of closely related hemipteran species (Briscoe and Chittka, 2001), including some insects belonging to the same suborder Sternorrhyncha such as aphids (Kirchner et al., 2005; Döring et al., 2011) and whiteflies (Mellor et al., 1997); (2) a colour opponent mechanism has been shown to exist in aphids (Döring and Chittka, 2007); and (3) there is recent evidence for intensity-independent preferences for colours in Eucalyptus-inhabiting psyllids (Farnier et al., 2014).
The recent study by Farnier et al. (2014) also showed that colour preferences vary between different species of Eucalyptus-feeding psyllids. Specifically, Anoeconesossa bundoorensis and Glycaspis brimblecombei, which share the same host (Eucalyptus camaldulensis) that produces young ‘red’ anthocyanic leaves, are primarily attracted to long wavelength-rich ‘red’ stimuli. In contrast, Ctenarytaina eucalypti (bluegum psyllid) and Ctenarytaina bipartita were shown to prefer ‘yellow’ and ‘green’ stimuli (Brennan and Weinbaum, 2001; Farnier et al., 2014). Thus, we have evidence that psyllid colour preferences are adaptations of their visual system to facilitate their search for food and oviposition sites.
Differences in colour preferences between closely related species may not necessarily be explained by spectral sensitivity. Recently, Telles et al. (2014) gave an example from Lepidoptera for how interactions between the peripheral and higher brain centres up- or down-regulate the sensitivity or weight given to different colour channels, possibly in relation to motivational state (Telles et al., 2014). This shows that physiological data are sometimes inadequate indicators of behavioural outcomes.
Visual acuity is of particular interest for appreciating the distance range from which visual cues are operant with respect to psyllid searching for hosts and to understand how vision impacts their searching behaviour at the ‘between-host’ level during dispersal (and/or host alternation) or within host canopy level to locate their preferred leaf type and sites for egg laying. A strict definition of ‘acuity’ refers to the minimum resolvable angle subtended by the spatial frequency of a pattern such as stripes of a grating. Here, we refer to acuity as the smallest single object detectable by the eye or ‘single object threshold’ which can be interpreted as the ‘minimum visible’ in opposition to the ‘minimum separable’ measured with gratings (Land, 1997).
In this study, we tested visual acuity by assessing the single object threshold of three species of Eucalyptus psyllid (A. bundoorensis, G. brimblecombei and C. eucalypti) using a binary choice assay inspired from the conventionally used Y-maze that we adapted to suit psyllid searching behaviour. We conducted morphological measurements on the eyes and quantified the searching behaviour of four species of psyllid (A. bundoorensis, G. brimblecombei, C. eucalypti and C. bipartita). We discuss our findings in the light of the current knowledge on insect vision and psyllid feeding and microhabitat preferences.
Single object detection threshold
Experiments using a Y-maze revealed inter-specific differences in stimulus detection according to target size. Maximum stimulus detection by C. eucalypti was observed for the greatest stimulus size tested (subtended visual angle of 35 deg) and decreased progressively with stimulus size before dropping abruptly for stimuli sizes from 10 to 5 deg (Fisher's exact test, P=0.0026; see Fig. 1A). In comparison, A. bundoorensis and G. brimblecombei showed constant responses for stimuli at 35 and 15 deg visual angle; however, stimulus detection in A. bundoorensis decreased significantly with stimulus size from 15 to 10 deg (Fisher's exact test, P=0.0297) whereas that of G. brimblecombei remained constant, dropping only for subtended visual angles between 10 and 5 deg (Fisher's exact test, P<0.001; see Fig. 1B,C). We used a critical visual angle for stimulus detection of 60% probability for individuals to detect the stimulus to facilitate some level of comparison with previous studies with honeybee and bumblebee models (Giurfa et al., 1996; Dyer et al., 2008). Using these criteria and based on the assumption of a linear relationship between psyllid response and the visual angle subtended by the stimuli (between 5 and 10 deg), we estimated acuity thresholds of 8.7 deg for C. eucalypti, 6.8 deg for A. bundoorensis and 6.3 deg for G. brimblecombei (note, C. biparita were not available at the time these experiments were conducted). As psyllid responses may not be linear between stimuli, more data for responses to stimuli subtending a larger number of visual angles in the region eliciting 60% response could alter our critical acuity threshold estimates. Separate tests of A. bundoorensis and G. brimblecombei that were simultaneously exposed to a dark achromatic stimulus of similar size revealed they only occasionally oriented to this achromatic stimulus, which suggests that stimulus detection by psyllids is not promoted by achromatic contrast even at smaller visual angles.
All four psyllid species had similar horizontal fields of vision of around 130 deg (F3,62=1.75, P=0.167). However, the number of facets on their eyes varied significantly between species (H3,54=40.84, P<0.001; see Fig. 2A). Specifically, G. brimblecombei possesses a higher number of facets (∼320 ommatidia per eye), followed by A. bundoorensis (∼200 ommatidia), C. eucalypti (∼190 ommatidia) and C. bipartita (∼160 ommatidia). Similarly, the size of the ommatidia differed significantly between species (F3,63=32.59, P<0.001; Fig. 2B). Glycaspis brimblecombei has the largest ommatidial diameter (∼14 µm) followed by C. bipartita and C. eucalypti (∼12–12.5 µm), whereas the ommatidia of A. bundoorensis are significantly smaller than those of the other species (∼11 µm). Not accounting for inaccuracies in estimation due to differences in curvature, significant differences in eye surface area were apparent among species (H3,64=36.52, P<0.001). Predictably, G. brimblecombei eyes were of significantly greater area (0.058±0.001 mm2) than those of A. bundoorensis (0.022±0.001 mm2), C. eucalypti (0.021±0.001 mm2) and C. bipartita (0.021±0.001 mm2); surface area for these last three species did not differ significantly. However, measurements of the eye surface may be inaccurate as a result of the different curvature of psyllid eyes. Inter-ommatidial angles also varied significantly between species (F3,60=21.32, P<0.001; Fig. 2C). The two Ctenarytaina species have comparable inter-ommatidial angles (6.6±0.10 deg for C. eucalypti and 6.3±0.1 deg for C. bipartita), which are significantly larger than those of A. bundoorensis (5.7±0.2 deg) and G. brimblecombei (5.5±0.2 deg). The calculation of eye parameter revealed no difference between C. eucalypti, C. bipartita and G. brimblecombei (1.4±0.1 rad µm). In contrast, the markedly smaller eye parameter estimated for A. bundoorensis (1.1±0.1 rad µm) suggests a trade-off between light sensitivity and acuity in favour of the latter in this species (F3,58=16.23, P<0.001; Fig. 2D).
Psyllid searching behaviours varied between species (Fig. 3). Specifically, C. eucalypti and C. bipartita exhibited relatively similar patterns of searching characterized by directed trajectories, short searching durations (<120 s) and distances (∼11 cm), and low numbers of turns (Table 1). In contrast, the searching behaviour of A. bundoorensis and, even more obviously, that of G. brimblecombei was more complex. Anoeconesossa bundoorensis exhibited directed movements comparable to those observed in C. eucalypti and C. bipartita but took significantly more time to choose a stimulus (>264 s). Unlike the three other species, G. brimblecombei displayed more sinuous movements, which translated into numerous turns, greater distances traversed (>42 cm) and longer searching durations (>420 s).
This study is the first to quantify the visual acuity of psyllids using a standard maze bioassay. We demonstrate that knowledge of innate responses to artificial stimuli can be used to overcome the absence of learning capacities as is often the case in more basal insect taxa. In this way, we could modify a conventional method to assess psyllid visual acuity to infer the subtended angle value at which particular colour targets are perceived and attract different species. Although a lack of information about psyllid spectral sensitivities remains a challenge for more fully elucidating the existence of colour vision in psyllids, the current study together with our previous work on innate preferences, and modelled likely photoreceptor distributions (Farnier et al., 2014), provide good reason to investigate the possibility of physiological colour processing mechanisms in these ecologically and biologically important models along the lines of the framework provided by Kemp et al. (2015). Our findings reveal that psyllids possess relatively poor visual acuity – comparable to that of aphids (Land, 1997; Döring and Spaethe, 2009).
Interestingly, behavioural estimations of psyllid visual angle in our Y-maze experiments are consistent with our measurements of inter-ommatidial angle (e.g. C. eucalypti: 8.7 deg in Y-maze versus 6.6 deg measured inter-ommatidial angle; A. bundoorensis: 6.8 deg versus 5.6 deg; and G. brimblecombei: 6.3 deg versus 5.5 deg). From these results we infer a coarser colour detection by psyllids than that by some butterflies, which resolve colour-associated tasks at smaller visual angles of about 1 deg (Takeuchi et al., 2006). However, it is remarkable in comparison to the larger honeybee, for which colour detection only occurs at visual angles greater than 15 deg, whilst stimuli containing colour and green contrast are detected at visual angles greater than 5 deg (Giurfa et al., 1996). Our data suggest little or no pooling of signals arising from the ommatidia and therefore the detection of colour is probably close to the limit imposed by individual species’ optics. The preference of A. bundoorensis and G. brimblecombei for red stimuli over black in Y-maze experiments was not influenced by the size of the stimuli. This concurs with previous work that demonstrated a pronounced preference for the ‘red’ stimulus over achromatic stimuli of higher intensity contrast with the background and suggests that psyllid innate colour responses are consistent independent of stimulus size (Farnier et al., 2014).
Morphological measurements of psyllid eyes also revealed prominent differences. Predictably, G. brimblecombei, the largest species (2.5–3.1 mm body length), has the greatest number of ommatidia and the largest ommatidial diameter of the four species. Likewise, C. eucalypti, the larger of the two Ctenaryaina species (2.5–2.8 mm), has more and larger ommatidia than C. bipartita (1.2–1.8 mm). However, the absence of significant differences in the eye parameter of C. eucalypti, C. bipartita and G. brimblecombei suggests that inter-specific differences in eye morphology are primarily size related. This is consistent with positive correlations between body size, number of ommatidia and inter-ommatidial angle as well as conserved eye parameters between species of bees of varying size, i.e. isometric scaling (Jander and Jander, 2002). Interestingly, our results seem to differ from previous studies conducted on aphids. Döring and Spaethe (2009) found no correlation between the number of ommatidia, inter-ommatidial angle and body length in aphids, i.e. allometric scaling. They concluded that light sensitivity might be a limiting factor for aphids, which therefore prioritize facet diameter to the detriment of spatial resolution (Döring and Spaethe, 2009). Although our study focused on a limited number of species, there is no indication of a similar compromise between light sensitivity and visual acuity in psyllids. On the contrary, measurements of the eyes of A. bundoorensis provide an example of a trade-off where light sensitivity is sacrificed in favour of visual acuity. Despite its small body length (1.7–2.4 mm), A. bundoorensis possesses a relatively large number of ommatidia; it has more ommatidia than either C. eucalypti or C. bipartita. The large number of ommatidia in this species is associated with small facet diameter. The existence of such a trade-off is further supported by a significantly smaller eye parameter than that found in the other species. Similar trade-offs were also observed in a limited number of aphid species by Döring and Spaethe (2009), who attributed such adaptation to the greater mobility of these species in response to predation (e.g. insectivorous birds) and the subsequent necessity to return to the host. Escape response is unlikely to explain this trade-off in psyllids as G. brimblecombei is arguably the species most exposed to predation as a consequence of its larger size and the sugary lerps the nymphs build, which birds consume (Paton, 1980; Pereira et al., 2012; Steinbauer et al., 2015). However, A. bundoorensis is unique because it oviposits and feeds on different plant modules, i.e. feeding occurs on leaves and oviposition occurs in crevices on stems. Such resource preferences are associated with higher mobility than evident in the other species.
Our observations of psyllid searching behaviours revealed significant differences between species, which appear to be influenced by their microhabitat preferences. For instance, the sedentary bud-dwelling species C. eucalypti and C. bipartita (Steinbauer, 2013) exhibit similar patterns characterized by short unidirectional paths, consistent with searching acropetally along stems towards apical buds. Likewise, A. bundoorensis, which does not occupy protected microhabitats (Taylor et al., 2013), exhibits similar unidirectional movements. However, longer distances traversed and prolonged searching durations seem to reflect differences in the utilization of host ‘architecture’ of this species, which, unlike the bud-dwelling species, frequently moves between growing branches and young leaves to find either feeding or oviposition sites. The searching behaviour of G. brimblecombei differed markedly from that of the three other species, with intense and protracted sinuous movements possibly in accordance with the utilization by this species of the surface of flat leaves for feeding and oviposition.
In the light of these findings and of the apparent linkage of psyllid colour preference and leaf spectral characteristics, we suggest psyllids have sufficient visual acuity to locate different modules within the canopies of their hosts. Generally, our results suggest that greater acuity is associated with a greater need for higher mobility. The fact that A. bundoorensis, for which morphological trade-offs favouring acuity over light sensitivity were found, spends a substantial amount of time searching branches and stems for oviposition sites supports such an hypothesis. In contrast, more sedentary bud-dwelling species seem to be attracted to brighter stimuli, suggesting constant intensity-dependent responses to visual stimuli. In the latter instance, simple phototactic responses appear ideal for these species to orientate toward the sun-lit apical ends of branches.
MATERIALS AND METHODS
Glycaspis brimblecombei Moore (from Eucalyptus camaldulensis Dehnh.) and Ctenarytaina bipartita (Burckhardt et al., 2013) (from E. kitsoniana Maiden) were collected from populations on the Bundoora Campus of La Trobe University and at Hoddle Range (State of Victoria, Australia), respectively. Ctenarytaina eucalypti (Maskell) were collected from a Eucalyptus globulus Labill. plantation at Clonbinane. Anoeconeossa bundoorensis Taylor and Burckhardt were taken from a glasshouse colony maintained on potted E. camaldulensis. Experiments were conducted between May 2012 and October 2014 during the austral spring–summer, which corresponds to the peak of psyllid activity in the field. Important aspects of the feeding and oviposition behaviours of our model species are available in Moore (1961), Morgan (1984), Burckhardt et al. (2013) and Taylor et al. (2013).
The horizontal directionality of the field of vision appears to be the most relevant to the psyllid's line of approach from a linear shape such as a stem or leaf pedicel as it attempts to climb onto a given plain coloured surface like a leaf. In practice, psyllids did not climb onto colour stimuli unless they intersect with the horizontal plane of the bioassay arena. Stimulus sizes were chosen based on the visual angles of stimuli used in previous experiments (for the largest stimuli) and arbitrarily decreased to sizes at which limits of psyllid visual performance could be observed. The selection of stimulus colours was based on the innate preferences reported in Farnier et al. (2014). Red stimuli (Microsoft RGB settings: R: 255, G: 0, B: 0, see Fig. 4 for reflectance spectra) were presented to G. brimblecombei and A. bundoorensis whereas yellow stimuli (R: 255, G: 255, B: 0) were presented to C. eucalypti. In addition to the preferred coloured stimulus, a dark (black), achromatic stimulus (i.e. G45; R: 45, G: 45, B: 45), which was previously shown to attract A. bundoorensis and G. brimblecombei in the absence of coloured stimuli, of the same size was presented in the other arm as an alternative choice.
Bioassays were conducted as follows: 50 insects (25 males and 25 females) of each of C. eucalypti, A. bundoorensis and G. brimblecombei were tested individually and only once. Psyllids were anaesthetized at −18°C for 3 min and then placed on the white dot in the maze, which was positioned equidistant (i.e. 7 cm) from the ends of the arms of the maze. Psyllids were allowed to orientate in the maze for 10 min and any individual that did not make a choice in that time was excluded. Only insects orientating and subsequently climbing onto a target were recorded as having perceived a stimulus. Psyllids choosing to climb onto any other part of the maze or on the background (including background areas located next to the stimuli) were considered as not having completed the task. Whilst this is a conservative criterion for determining visual angle, it is most ecologically relevant, as the insect must use its vision to correctly find the target. In order to compensate for the absence of training, data were normalized to account for the noise caused by between-species differences at cooperating with the task. Normalization consisted of the calculation of a response score ranging between 0 and 1 relative to the highest number of ‘finders’ of each species at any visual angle tested. Our normalization allows us to take into consideration differences in the magnitude of responses exhibited by different psyllid species and to estimate detection thresholds in a way comparable to that for other model insects such as bees and butterflies. The orientation of stimuli was changed after every four insects to preclude position effects. Different species and sexes were tested in separate mazes. Nominal logistic regressions revealed no influence of sex on psyllid responses for all species. Therefore, male and female responses were pooled for statistical analyses. Fisher's exact tests were used to determine the influence of stimulus size on psyllid detection performance and to compare the choice of the red or dark stimulus for different stimulus sizes.
Psyllids that had been preserved in 70% ethanol were point-mounted on card using entomological glue (Australian Entomological Supplies, Bangalow, NSW, Australia) before being photographed using a Canon EOS 7D digital camera fitted with a LU Plan Fluor 10×/0.30A lense (Nikon, Japan) on a Visionary Digital BK Imaging System (Visionary Digital, USA). Individual images were collated with Zerene automontage software version 4.02 (Synchroscopy, Cambridge, UK). The eyes of eight individuals were photographed for each of the four species in both dorsal and lateral views (Fig. 5). Contrast and sharpness were adjusted using Adobe Photoshop to optimize the definition of the structure of the eyes. Measurements were taken using ImageJ software. The inter-ommatidial angle and the angle subtended by the eye were measured using the dorsal view following the same procedure as in Döring and Spaethe (2009) and Yilmaz et al. (2014). The number of ommatidia on each eye was counted directly using the lateral view. The same view was used to measure the surface area of both eyes. The ommatidial diameter was calculated by drawing a segment across ‘in-focus’ ommatidia in the lateral view and dividing the length of the segment by the number of ommatidia crossed. The measurement of the ommatidial diameter was performed on both eyes and repeated three times. The eye parameter was calculated by multiplying the inter-ommatidial angle by the mean ommatidial diameter (Snyder, 1979). Data for each measurement were tested for normality and analysed using ANOVA test of variance followed by Tukey's post hoc tests (95%) for pairwise comparisons. The number of ommatidia and surface area of the ommatidia were analysed using non-parametric Kruskal–Wallis tests followed by Mann–Whitney tests for pair-wise comparisons.
Psyllid movement tracking
Psyllid movement was recorded in different experiments conducted under similar environment and illumination conditions as those described for the Y-maze experiments. Psyllids were placed in arenas formed by a 15.5 cm diameter glass Petri dish positioned above a grey coloured cardboard sheet. The inner walls of the dish were encircled with a 5 cm wide strip of the same grey cardboard (160 gsm; K. W. Dogget; colour edited, R: 166, G: 166, B: 166; ∼35% reflectance) on which four 4×4 cm colour stimuli were preprinted (no use of glue) and arranged opposite one another. Psyllids were placed in the centre of the arena and allowed to search until they eventually climbed onto one of the stimuli. Psyllid movements were recorded (at 30 frames s−1) with a video camera (540TVL high resolution, EVO series, Pacific Communications, Australia) positioned approximately 60 cm above the arena and connected to a digital video recorder (PDRH-800e, Pacific Communications). Footage length for each insect was reduced using Windows Movie Maker (Microsoft) to only include sequences where active searching characterized by steady and continuous motion was evident. Psyllid positions were digitized using digitizing software (Hedrick, 2008) in MATLAB R2013a (MathWorks Inc.). The x–y coordinate data were then used to summarize searching behaviour, including search duration, distance traversed and number of turns defined by deviations exceeding a 30 deg angle from the initial trajectory, using custom-written functions in Matlab R2013a (R. Peters, La Trobe University). Psyllid searching behaviour was compared in the presence of three colour stimuli as described in Farnier et al. (2014). As data for distance traversed, duration searching and number of turns failed normality tests, non-parametric Kruskal–Wallis tests followed by multiple Mann–Whitney tests for pairwise comparisons were used to compare species' responses.
We thank Ms Renae Forbes for helping with the acquisition of searching behaviour data and Mr Rob Evans for helping to maintain plants supporting our psyllid colonies.
K.F. designed, conducted the experiments, analysed the data and wrote the manuscript. A.G.D. provided equipment for the behavioural assays and contributed to data interpretation and analyses and the writing of the manuscript. G.S.T. provided the equipment and guidance for the acquisition of close-up high-resolution photographs of psyllid eyes. R.A.P. supplied equipment, wrote the program for the tracking of insect movement and contributed to writing the manuscript. M.J.S. provided funding and guidance, and contributed to the writing of the article.
The research was funded by an Australian Research Council (ARC) Future Fellowship [FT100100199] to M.J.S. with associated Australian Postgraduate Award (APA) scholarship with top-up from La Trobe University to Kevin Farnier and an ARC Queen Elizabeth II (QEII) Research Fellowship [DP0878968] to A.G.D.
The authors declare no competing or financial interests.