Spectral sensitivity of retinal photoreceptors of tortricid moths is not tuned to diel activity period

Vision is a vital sense for the majority of insect species. The light environment exerts a strong pressure on the evolution of their visual system, leading to substantial differences in eye structure and retinal properties between diurnal and nocturnal species. The irradiance and light spectra that reach the surface of the Earth fluctuate vastly between day and night. The mean light irradiance in the 300–700 nm range on a sunny day is of the order of 10¹⁴ photons cm⁻² s⁻¹ nm⁻¹, decreasing by two orders of magnitude when the sun is at the horizon, and dropping to 10⁶ photons cm⁻² s⁻¹ nm⁻¹ on moonless nights (Cronin et al., 2014). In addition, the spectral properties of the light vary depending on the time of the day: at midday, the irradiance spectrum is relatively flat between 400 and 700 nm, during the day–night transition the red–orange part dominates the spectrum, and after the beginning of the ‘true’ night, the short wavelengths prevail (Warrant and Somanathan, 2022).

Colour vision is based upon photon absorption by the different photopigments. These are composed of a G-protein coupled receptor called opsin and a linked chromophore, which is the light-sensitive part. Absorbance spectra of the photopigments depend on the sequence of amino acids that form the opsin (Cronin et al., 2014). In its ancestral form, insect vision in both diurnal and nocturnal species is based on the expression of three types of photopigments with peak absorbance in the ultraviolet (UV), blue and green wavelengths, between 300 and 700 nm (Lebhardt and Desplan, 2017). Species may adapt to different environmental light conditions by changing spectral sensitivity, the number of opsins and the expression of additional filtering pigments in the receptors, co-expressing different opsins in the same photoreceptor, and varying anatomical structures such as lenses or screening pigments (Lebhardt and Desplan, 2017). Consequently, differences in spectral sensitivity or temporal resolution are expected in related species that live in environments with different light conditions. Recent studies demonstrate multiple effects of the light environment on the evolution of colour vision in Lepidoptera. For instance, high light availability leads to an increase in opsin diversity and evolution rate in Lepidoptera (Sondhi et al., 2021), while a study in diurnal and nocturnal sphingid moths has revealed the accelerated evolution of opsin genes and subtle shifts in spectral sensitivity, presumably fine-tuned according to the nocturnal or diurnal lifestyle. A long-wavelength shift in absorption of the green-absorbing opsin was found in nocturnal moths, which presumably results in a higher overall photon capture, but at the expense of colour vision as a result of a reduced overlap between green-sensitive and blue-sensitive photoreceptors (Akiyama et al., 2022).

The eyes of nocturnal insects have evolved to maximize photon capture in order to compensate for photon shot noise (i.e. voltage fluctuations due to integration of random photon responses), transducer noise (introduced by random interactions in the signalling cascade) and dark noise (caused by high amplification gain in the visual pathway) that are associated with vision in dim light (O'Carroll and Warrant, 2017; Honkanen et al., 2017). These adaptations in nocturnal insects include superposition optics, larger apertures, shorter focal lengths and larger rhabdoms than in diurnal species (Warrant, 2017). In addition, photoreceptors of nocturnal species usually have slower responses, higher gain and wider spatial reception fields than those of diurnal ones. Additional adaptations to dim light include reduced spatial and temporal resolution to improve contrast detection (Meece et al., 2021), and neural summation to increase the signal to noise ratio (Stöckl et al., 2016). Taken together, these adaptations endow nocturnal insects with highly sensitive eyes, which can support colour vision even at very dim light levels, where human colour vision ceases (Warrant and Somanathan, 2022).

The order Lepidoptera includes about 20,000 species that are mostly diurnal (butterflies) and about 160,000 species that are mostly nocturnal (moths). Diurnal Lepidoptera commonly have apposition eyes and a variable number of different spectral photoreceptors peaking at different wavelengths (Stavenga and Arikawa, 2006), resulting in trichromatic or tetrachromatic colour vision (Koshitaka et al., 2008; Zaccardi et al., 2006). Nocturnal Lepidoptera generally have three different photoreceptor classes peaking in the UV (340–360 nm), blue (440–480 nm) and green (520–540 nm) wavelengths (Van Der Kooi et al., 2021), and superposition optics, with the rhabdoms separated from the dioptrical apparatus by a large clear zone (Exner, 1891). Apposition and superposition eyes are, however, associated with different phylogenetic lineages, which have predominantly diurnal and nocturnal lifestyles, but are not exclusive to ‘butterflies’ and ‘moths’, respectively (Kawahara et al., 2018). Some nocturnal insects, at least the large hawkmoths, can discriminate colours (Kelber et al., 2002) and navigate using landmarks or the sky as references (Warrant et al., 2004; Reid et al., 2011; Dacke et al., 2003).

Although moths are overly active at night, crepuscular or diurnal pheromone communication has been described in at least eight species from five families (Groot, 2014), and it is relatively frequent in the family Tortricidae (i.e. leafrollers), which comprises 10,000 species. For example, adults of Grapholita molesta, Lobesia botrana and Cydia pomonella (the main pests of peach, grape and apple crops worldwide, respectively; Benelli et al., 2023a,b; Knight et al., 2019) are active before sunset, at sunset and after sunset, respectively (Batiste et al., 1973; Lucchi et al., 2018; Kim et al., 2011; Fig. 1). The diurnal and nocturnal species (G. molesta and C. pomonella, respectively) are closely related and belong to the same tribe, Grapholitini, whereas the crepuscular species (L. botrana) belongs to the tribe Olethreutini (Fagua et al., 2017). As in other phytophagous lepidopterans, adults of these species do not feed because they rely on larval nutritional reserves (Wäckers et al., 2007; Amat et al., 2022). Consequently, reports of flower visiting are occasional and only in L. botrana (Benelli et al., 2023a), and so detecting flower cues does not appear to be essential to them. However, they use vision to fly and to navigate in odour plumes (Navarro-Roldán et al., 2019), and probably to recognize and land on the host plants. Vision could also be involved in mating and oviposition behaviour (Yang et al., 2020, 2022). Colour has some effect on the capture of males in pheromone traps (Knight and Miliczky, 2003; Knight and Fisher, 2006; Zhao et al., 2013; Padilha et al., 2018; Rayegan et al., 2016), and UV light appears to influence male but not female captures in semiochemical traps (Adasme, 2022). Insects, including Lepidoptera also rely on the detection of polarized light to evaluate the host plants or mates (Yadav and Shein-Idelson, 2021), meaning that the detection of polarized reflections from objects might play a role in moth behaviour. A comparative study of the visual system of these three moth species could provide valuable information regarding insect evolutionary adaptations to diurnal and nocturnal conditions. Furthermore, there are only four other studies on the spectral sensitivity in tortricids moths, on C. pomonella (Pristavko et al., 1981), L. botrana (Crook et al., 2022), Cydia strobilella (Jakobsson et al., 2017) and Adoxophies orana (Satoh et al., 2017). In addition, the exploration of the visual physiology of these economically important insect species could provide valuable insights into how to exploit their visual system for pest control (Shimoda and Honda, 2013).

In the present study, we explored several aspects of the visual physiology of adults of C. pomonella, G. molesta and L. botrana, which all have compound eyes with superposition optics (the typical superposition deep pseudopupil can be seen in the specimens in Fig. 1). Spectral sensitivity in both sexes and two eye regions (dorsal and ventral) was determined using electroretinography (ERG) and selective chromatic adaptation. ERG results were further supported by intracellular recordings from single photoreceptors. Adaptations for vision in dim light include slower photoreceptor kinetics and neural spatio-temporal summation (Warrant and Dacke, 2011; Honkanen et al., 2017); therefore, we expected to find differences in the speed of vision among the three species occupying different time niches. Thus, visual temporal acuity was determined by measuring the diminishment of ERG responses upon stimulation with flickering light at high frequencies, i.e. flicker fusion frequency (FFF) measurements. All three species have trichromatic retina with highly conserved spectral sensitivity and similar temporal acuity.

Grapholita molesta (Busck), Lobesia botrana (Denis & Schiffermüller) and Cydia pomonella (L.) larvae were reared on a semi-artificial diet (Ivaldi-Sender, 1974) at 25°C under a 16 h:8 h light:dark photoperiod. Pupae were sexed and shipped from Spain to Slovenia. Adults were kept at 18°C with unrestricted access to 10% sucrose in water. Before electrophysiological recordings, adults were anaesthetized with ice. All insects tested were between 1 and 3 days old.

Moths were immobilized in plastic pipette tips with a mix of melted beeswax and resin, and their eyes were pre-oriented in a mini goniometer to yield frontal illumination and a lateral electrode trajectory in the recording set-up. A 50 µm Ag/AgCl wire was inserted into the thorax as a reference electrode. The mini goniometer was positioned with the eye in the centre of rotation into a large goniometric stage, which also carried the micromanipulator (Sensapex, Oulu, Finland) with the recording electrode, which allowed us to illuminate only the desired part of the recorded eye (i.e. dorsal or ventral). For ERGs, the eye was gently impaled just below the cornea with a blunt glass electrode filled with insect saline. For intracellular measurements, a small hole was cut in the cornea with a razorblade chip and sealed with silicon vacuum grease. A high-resistance electrode was manufactured from 1.00/0.50 mm outer/inner diameter borosilicate tubing with a P-2000 laser puller (both from Sutter Instrument Company, Novato, CA, USA). It was loaded with 3 mol l⁻¹ KCl, yielding a resistance of 100–150 MΩ. The electrode tip was advanced by ∼250 µm through the clear zone, before it impaled the photoreceptors. At best, one or two cells could be recorded before the electrode was broken by the tracheolar sheath surrounding the rhabdoms.

Monochromatic stimulation was delivered from a custom LED array (Belušič et al., 2016) or from a 75 W xenon arc lamp (Cairn Research, Faversham, Kent, UK), filtered with a monochromator (B&M Optik, Limburg an der Lahn, Hesse, Germany), a motorized graded neutral density filter on a fused silica substrate (Thorlabs, Bergkirchen, Bavaria, Germany) and a motorized UV-capable polarizing filter (Knight Optical, Harrietsham, Kent, UK). Second-order UV beams from the monochromator, occurring at >600 nm, were blocked with a UV-blocking filter. Both light sources were calibrated with a Flame spectroradiometer (Ocean Optics, Orlando, FL, USA) to yield equal photon flux (∼10¹⁴ photons s⁻¹ cm⁻²) at each wavelength between 300 and 700 nm. For selective adaptation, monochromatic light from the LED array was shone coaxially with the test stimuli from the monochromator. The LED intensity was a maximum of ∼2 magnitudes brighter than the isoquantal series, but was arbitrarily adjusted to yield minimal responses at the adapting wavelength, while avoiding excess response suppression at the other wavelengths. For the FFF measurement, stimuli were delivered from a 525 nm, 350 mA green LED (Roithner, Wien, Austria), driven by a Cyclops current driver (Open Ephys, Atlanta, GA, USA). A function generator (Hewlett Packard, Palo Alto, CA, USA) was programmed to create sinusoidal stimuli with ±0.5 contrast, with a frequency sweeping of 1 s stimulus from 10 Hz to 50 Hz in steps of 5 Hz, and from 60 Hz to 270 Hz in steps of 10 Hz. The LED intensity was adjusted to evoke large but subsaturating ERG responses.

The signals were amplified and filtered at the Nyquist frequency with a SEC-10LX amplifier (NPI, Tamm, Ludwigsburg, Germany) and a CyberAmp 320 signal conditioner (Molecular Devices, San José, CA, USA), digitized with a Micro1401 MkII (CED, Milton, Cambridge, UK) and acquired with WinWCP software v. 5.5.4. For the FFF experiments, the signals were amplified and filtered with a DAM50 amplifier (WPI, Worcester, MA, USA), digitized with a Powerlab 25T and acquired with Labchart 8.0 (AD instruments, Bella Vista, NSW, Australia). This program was also used to analyse the signals with fast Fourier transform and measure the amplitude of its real component.

Macro-photos of the eyes of immobilized insects, mounted to a coarse micromanipulator (Narishige, Tokyo, Japan), were obtained with a USB microscope (Edge AM4515ZT, AnMo Electronics, New Taipei City, Taiwan). The eyes were illuminated with the ring LED illuminator of the microscope and the surface reflections were attenuated with built-in crossed polarizers. A stack of photos at different focal planes was merged into an image with an extended depth of focus in Adobe Photoshop CS5 (Adobe Systems Inc., San José, CA, USA).

To analyse the effects of species, sex, eye region (dorsal or ventral) and spectral band (UV, blue and green) on ERG amplitude (N=3), generalized linear models (GLM) with gaussian family link were run in R 4.1.2 software (http://www.R-project.org/). The chosen three spectral bands included values ±10 nm around the main peaks obtained by the template fitting (355, 455 and 530 nm). Within each spectral band (UV, blue, green), we took the sensitivity values for the peak and the four nearest points (Δλ=5 nm) so that N=15 for each species/sex/eye location (dorsal or ventral) (Fig. S1). Model selection was performed starting with a non-interaction model with main effects, and adding different higher order interactions until no significant difference between models was detected with ANOVA, while inspecting the Akaike information criterion (AIC) of each successive model. Pairwise comparisons of model terms used the package ‘emmeans’ in R (https://CRAN.R-project.org/package=emmeans). To compare FFF among species (sexes were combined), a non-parametric Kruskal–Wallis test was used in Matlab R2022a.

Spectral sensitivity was first scanned in the ventral and in the dorsal areas of dark-adapted eyes. Spectral sensitivities could be well fitted with a weighed sum of three rhodopsin absorbance templates (Stavenga et al., 1993) with peak absorbance wavelengths (λ_max1,2,3) and relative amplitudes (A_1,2,3) as free parameters (Table S1). The main absorbance peak was in the green part of the spectrum (525–530 nm) and a secondary peak was in the UV part (350–355 nm). Additionally, an intermediate peak in the blue (450–460 nm) was hidden in this curve, being more easily visible in the recordings from the ventral retina in dark-adapted individuals (Fig. 2A,E,I,M,Q,U). The green peak was probably due to the most abundant receptor class, having broad-band sensitivity and peaking at ∼530 nm, while the UV and blue peaks were due to the UV and blue receptors, respectively. The GLM model that best fitted ERG responses included the second-order interactions with 10 parameters (Table S2A). The main factor contributing to ERG amplitude was the spectral band (UV, blue or green) accounting for 95.11% of the variance (Table S2B). It was followed by eye region (dorsal or ventral; 1.95% of the variance), the interaction between spectral band and eye region (1.47% of the variance), and the interaction between spectral band and species (0.86% of the variance). The other variables contributed a maximum of 0.18% of the variance each. Pairwise comparisons of eye region and spectral band showed that estimated mean sensitivities significantly differed between regions in the blue and UV part of the spectrum (Tukey's test P<0.0001; Table S2C, Fig. S2).

To further analyse eye sensitivity, we recorded the spectral response of retinae selectively adapted with constant UV, blue and green light (Fig. 2). The adapting light saturated the response in the corresponding spectral receptors and isolated the response of the remaining receptors, which were less sensitive to the adapting light. In eyes illuminated with constant monochromatic light, the response was reduced in the spectral band of the adapting light and the remaining responses showed two or three sensitivity peaks. Using green adaptation, the green peak could not always be abolished in L. botrana and C. pomonella (e.g. Fig. 2B,F,R,V), probably because of the residual response by the abundant green receptors. Furthermore, the blue peak was far more visible in the ventral retinae. Using blue adaptation, the blue peak disappeared and the dorsal and ventral eye sensitivities became indistinguishable from each other, except in G. molesta (Fig. 2K,O). Using UV adaptation, the UV peak disappeared in all cases and the blue peak was slightly reduced, except in L. botrana, where it disappeared completely (Fig. 2T,X).

The ERG results strongly suggested that the retinae have three sets of receptors, maximally sensitive in the UV, blue and green ranges. To verify these results, we obtained recordings from single retinula cells in the species with the largest eyes, C. pomonella (Fig. 3). As expected, broadband-sensitive cells peaking in the green were encountered most frequently (N=4 in males and N=6 in females). A single blue and two UV-sensitive cells were impaled in three females, and a single UV cell was impaled in a male. Peak sensitivities (340, 440 and 535 nm) closely matched the values obtained with ERG recordings. The UV cells had a tail of sensitivity in the green part of the spectrum. The blue cell showed negative responses to stimuli in the green part of the spectrum. These responses could be abolished or increased with current injection (Fig. 3D).

In addition to the spectral sensitivity, the intensity–response relationship (the ‘V–log I’ curve) and polarization sensitivity (PS) were measured in all impaled cells (Table 1). The intensity–response relationships were fitted with a Hill function with Hill slope (n), half-maximal intensity and maximal response as free parameters. All cells had a relatively steep Hill slope (n≈0.81–1.22). The green and blue cells had low PS, while the UV cells had moderate PS, and the sensitivity maximum aligned with the dorso-ventral axis (not shown).

The speed of vision was determined by measuring the ERG response to a flickering green LED, sinusoidally modulated at frequencies between 10 and 270 Hz. The responses were analysed via fast Fourier transform and the amplitudes of the real part were normalized and plotted against stimulus frequency (Fig. 4). Each response showed two maxima: one at the lowest frequency (∼10 Hz) due to the low-pass property of the photoreceptors, and a second one at around 40–50 Hz due to the high-pass property of the lamina monopolar cells (Rusanen and Weckström, 2016). The FFF was defined as the point of the curve where the Fourier amplitude was 5% of the second maximum.

There was no significant difference among species in FFF, with a mean value in the range 100–120 Hz (P=0.5364, Kruskal–Wallis test, χ²=1.25, sexes were combined, N=9 for G. molesta and C. pomonella, N=10 for L. botrana).

Using extracellular and intracellular recordings, we have shown that the three tortricid moth species have the retinal substrate for colour vision, based on UV, blue and green photoreceptors, and that it does not differ among species or between sexes. Similar results were obtained in other studies on Tortricidae (Satoh et al., 2017; Jakobsson et al., 2017; Crook et al., 2022). The studied tortricid moths thus have a generalist trichromatic colour vision substrate and do not show any fine tuning in the spectral sensitivity, in contrast to sphingid moths (Akiyama et al., 2022). However, the three species seem to have evolved differently sized compound eyes (Fig. 1) with superposition optics, with the nocturnally active C. pomonella having the largest and the diurnally active G. molesta the smallest eyes. A larger eye size can support a larger superposition aperture and consequently higher sensitivity (Kirschfeld, 1974; Horridge, 1978; Land, 1981; Warrant and McIntyre, 1990), but the aperture size and the anatomical parameters will have to be measured in order to confirm this point quantitatively. Additional adaptations of the peripheral visual system could exist at the level of the pupillary mechanism, mediated by the pigment granules in the distal retina, which give rise to the brown eye colour (Fig. 1). It is tempting to speculate that the differences in eye colour between L. botrana and the other two species are due to the different sensitivities and dynamics of the pupil, but a single macro-photo, obtained at different subjective circadian periods in different species, is insufficient to draw conclusions. Furthermore, cultured tortricid moths can have altered screening pigments due to genetic drift (Satoh et al., 2017).

Because of the clear visibility of the blue peak that appears in ventral recordings supported by the pairwise comparisons obtained from the factor band:position of the GLM, we can conclude that the compound eyes in all three species are regionalized, with the dorsal and ventral retina enriched with UV and blue receptors, respectively, similar to sphingid moths (Bennet et al., 1997; White et al., 2003). Eye regionalization with a similar pattern has been found in insects from different orders (Briscoe et al., 2003; Henze et al., 2012), and it probably fine-tunes the compound eyes to optimize object contrast against the long wavelength-enriched and short wavelength-enriched ventral and dorsal hemispheres of the visual environment, respectively (Nilsson and Smolka, 2021). Ommatidial architecture in Tortricidae appears to be very similar to that in other Lepidoptera, with two larger distal receptors R1 and R2 and proximal R5–8, but the basal R9 may be missing (Satoh et al., 2017). The most abundant class of receptors are the broad-sensitive green receptors, which probably terminate in the first optical ganglion, the lamina, and are the main input into the motion vision pathway (Matsushita et al., 2022). These cells are most likely allocated to retinula positions R3–8. The UV and blue receptors probably terminate in the second optical ganglion, the medulla, and are likely to be the main input into the colour and polarization vision pathway. These cells are probably allocated to retinula positions R1–2, as inferred from their maximal sensitivity to vertically polarized light (Arikawa and Uchiyama, 1996). Additionally, the sensitivity to polarized light in these cells indicates that polarized reflections might influence tortricid host plant selection, similar to diurnal butterflies (Kinoshita et al., 2011; Blake et al., 2019). The blue-sensitive cells form opponent pairs with green-sensitive cells, and being able to increase or decrease this effect with current injection indicates that the hyperpolarizing responses were caused by anionic (chloride) current through the histaminergic channels on receptor axons (Chen et al., 2019), postsynaptic to the green receptor axons. Direct inter-photoreceptor opponency is found in the retina of diurnal butterflies and flies (Schnaitmann et al., 2018; Chen et al., 2020; Matsushita et al., 2022), which are insects with true colour vision (Van der Kooi et al., 2021). This indicates that Tortricidae are capable of processing colour similar to other insects.

The regionalization of the tortricid moth retina, with the UV and blue receptors found in the dorsal and ventral parts, respectively, is very similar to that in the sphingid moth Manduca sexta (Bennett et al., 1997). In previous studies of tortricid moths, researchers have either failed to show the blue receptors (C. pomonella; Pristavsko et al., 1981) or take into account different parts of the eye (L. botrana; Crook et al., 2022), or have reported various levels of UV sensitivity across species from different photic environments (Eguchi et al., 1982), with increased UV sensitivity sometimes being attributed to the demands of nocturnal navigation (Ogawa et al., 2015; Barta and Horváth, 2004). We suggest that the results from these studies should be treated with caution, and that more attention should be paid to potential effects associated with the recording position in the retina. Our intracellular recordings, however, provide a definitive proof of the existence of the three spectral classes of photoreceptors, with polarization sensitivity in the UV part of the spectrum and opponent processing of colours in the blue–green range.

A very similar substrate for colour vision exists in diurnal and nocturnal hawkmoths (Kelber et al., 2003). Both diurnal and nocturnal species are trichromats (UV, blue, green) and have superposition eyes. Nocturnal hawkmoths can discriminate flowers in dimmer light but flowers that are usually visited by these nocturnal insects tend to offer a strong achromatic contrast, suggesting that they mainly use achromatic cues (Kelber et al., 2003). Similarly, G. molesta (Yang et al., 2020) and C. strobilella (Jakobsson et al., 2017) principally use achromatic cues to locate oviposition places on their host plants, probably using vision as a secondary cue after olfaction, which is a prominent sense in moths. Yet, some species of noctuid and crambid moths have, in addition to UV, blue and green, also violet-sensitive (Belušič et al., 2017) or red-sensitive photoreceptors (Langer et al., 1979), but the behavioural function of the putative tetrachromatic retina is so far unknown.

All three tortricid moth species in this study had similar flicker fusion resolution. Higher FFF is expected in diurnal species because higher photon abundance requires less integration time for good temporal resolution. However, a more precise method for evaluating the speed of vision might still reveal subtle differences. Similar to our study, Chatterjee et al., (2020) compared the FFF in species from a single family with different diel activity periods and failed to identify any differences between diurnal and nocturnal species under identical experimental conditions as in our study. Yet, some moths are certainly capable of tuning the temporal and spatial properties of their vision at the level of the retina and the optical ganglia, according to the available ambient light (Stöckl et al., 2017).

Tortricids show preference when given a choice of traps of different colours (Knight and Miliczky, 2003; Myers et al., 2009; Zhao et al., 2013; Rayegan et al., 2016). In addition, mated females of G. molesta show colour and light intensity oviposition preferences under laboratory conditions (Yang et al., 2020, 2022). Tortricid moths are an economically important group for which the alternative to insecticide control is mostly based on odour-baited traps and mating disruption with sex pheromones (Knight et al., 2019). Taken together, our results indicate that both colour and polarization cues can be detected by the tortricid moths and may be exploited in the design of more effective traps (Shimoda and Honda, 2013).

We are grateful to Marko Ilič for his help with the data analysis.