Spectral Sensitivities and Color Signals in a Polymorphic Damselfly

Animal communication relies on conspicuous signals and compatible signal perception abilities. Good signal perception abilities are particularly important for polymorphic animals where mate choice can be a challenge. Behavioral studies suggest that polymorphic damselflies use their varying body colorations and/or color patterns as communication signal for mate choice and to control mating frequencies. However, solid evidence for this hypothesis combining physiological with spectral and behavioral data is scarce. We investigated this question in the Australian common blue tail damselfly, Ischnura heterosticta, which has pronounced female-limited polymorphism: andromorphs have a male-like blue coloration and gynomorphs display green/grey colors. We measured body color reflectance and investigated the visual capacities of each morph, showing that I. heterosticta have at least three types of photoreceptors sensitive to UV, blue, and green wavelength, and that this visual perception ability enables them to detect the spectral properties of the color signals emitted from the various color morphs in both males and females. We further demonstrate that different color morphs can be discriminated against each other and the vegetation based on color contrast. Finally, these findings were supported by field observations of natural mating pairs showing that mating partners are indeed chosen based on their body coloration. Our study provides the first comprehensive evidence for the function of body coloration on mate choice in polymorphic damselflies.

adjusting ND filters to produce equal quantum flux at each wavelength step of the measurement. A shutter (LS6, Uniblitz, USA) was used to control light duration of the flashes. Visual stimuli (50 ms) were presented with an interval of two seconds between measuring steps. To eliminate the potential effect of chromatic adaptation during the experiment, light stimuli from short to long wavelengths and the reverse direction were both recorded.

Electroretinogram recordings
Electroretinogram (ERG) is a quick method to examine the response from the overall visual system. Before ERG recordings, freshly collected individuals (12 males, 10 andromorphs, 11 gynomorphs) were kept in a fridge at 8°C for at least one hour, and dark-adapted for at least 30 minutes before ERG was conducted. Appendages of damselflies were removed and the openings were sealed with Vaseline to prevent the loss of body fluid. Damselflies were placed on a platform and immobilized with a mixture of beeswax and resin (3:1). The platform was then transferred into a Faraday cage with the eyes of the individual located at the center of the light stimulus apparatus and facing horizontally. Two chlorinated silver wires serving as recording electrodes were attached with conductive gel (Sigma gel, Parker laboratory Inc. USA) to the same region of cornea surface of the two compound eyes to compare evoked potential differences between two corneal surfaces after light stimulation of one eye.
A third chlorinated silver wire was inserted into the thorax as the indifferent electrode.
The recorded signal was pre-amplified with an AC differential amplifier (DP-301, Warner Instrument Corp., USA) with band-pass filters from 1 Hz to 1 kHz. The amplified signal was acquired and transferred to a desktop PC for data analysis using a multifunction data acquisition device (NI USB-6009, National Instruments, USA).
Each damselfly was tested within two hours after preparation and three ERG measurements were taken from each subject. The results were averaged and pooled for each morph.

Intracellular recordings
In order to determine the spectral sensitivity of individual photoreceptors in the eyes of I. heterosticta, intracellular recordings were conducted. The animal was mounted the same way as described for the ERG measurements. Using a razor blade, a small The glass electrode was slowly lowered into the tissue using a water-filled hydraulic micromanipulator (MMW-20, Narishige Co. Ltd. Japan). Once the membrane potential dropped below -25 mV, a white light from an external strobe (Q15, Quantaray, Japan) was delivered to test if a depolarized response could be elicited as indication that the electrode had penetrated a photoreceptor. Further white light flashes were delivered from various angles to identify the optimal receptive field of the penetrated photoreceptor. The damselfly was then dark adapted for 30 minutes before measurement.
Response-log stimulus intensity (V/logI) curve was measured (17 white light intensities over 4 log units with 50 ms flash in 5 s interval). A flash method [1] was applied, and each photoreceptor cell was examined three times between 300 to 700 nm in both directions. The recorded signal was amplified by a multipurpose microelectrode amplifier (AXOPROBE-1A, AXON Instruments, USA) and analysed as for the ERG. Spectral sensitivity curves of the photoreceptors were corrected based on the results of V/logI. A hyperbolic function: where I is stimulus intensity in quantal flux; V is the amplitude of the receptor response in mV; Vmax is the saturated response amplitude; R is the intensity yielding a 50% response of Vmax; n is a constant determining the slope of the function were compared to a previously described template for visual pigments [3]. In total, 61 individuals (31 males, 12 andromorphs, 19 gynomorphs) were tested using intracellular recordings, which included 7 UV cells, 17 Blue cells, 52 Green cells.

Spectral reflectance and irradiance measurements
The reflectance spectra of the bodies of the various damselfly morphs were acquired with a miniature spectrometer (USB-4000-UV-VIS, Ocean Optics, Inc., Dunedin, FL).
The samples were illuminated with a 150 W Xenon lamp (Thermo Oriel, USA).
Damselflies from the electrophysiological experiments and field collections were placed on a horizontal platform and measurements were taken from thoraces through an optic-fiber cable (P100-2-UV-VIS, Ocean Optics, Dunedin, Florida, USA). In order to calculate the relative reflectance spectra of the sample, we also took measurements from a WS-1 diffuse reflectance standard (Ocean Optics) under the same conditions. We measured the spectral reflectance from 31 males, and 103

Calculation of chromatic/achromatic contrasts and discrimination values
By using the spectral sensitivities of the photoreceptors and the reflectance spectrum of each morph as well as the spectrum of the green background vegetation under different light irradiances, we calculated the receptor-specific chromatic and achromatic contrasts [4][5][6]. The receptor quantum catches (Q i ) were calculated as: where i denotes the spectral types of receptor (UV, B, G), S i (λ) is the spectral sensitivity function of the receptor i, I(λ) is the illumination spectrum, and R(λ) is the reflectance spectrum of each morph of individuals or green vegetation. The receptorspecific contrast (q i ), which is the quantum catch of each receptor class adapted to its light background was established as, where Q i B is the adaptation coefficient of a receptor to its light environment by using the von Kries transformation [7].
Discrimination values (ΔS) of the trichromatic visual system were calculated according to equation (3). Intensity (brightness) cues are ignored in this model.
where ω i is the noise value of each receptor class (ω UV = 0.207,ω B = 0.244, ω G = 0.217, values were adapted from equation (5) and (6) in Vorobyev et al. [5]. f i = Ln(q i ) is the log transformed receptor-specific contrast and Δf i is the difference in a receptor between two stimuli (pair-wise comparisons between different morphs or with the green vegetation). The units of ΔS are jnd (just noticeable differences).
Achromatic contrast (brightness contrast) was also analysed by using the green sensitive receptor (i) [8.9], Electrophysiological and behavioral evidence to determine the definitive discriminated threshold value in chromatic and achromatic contrasts for damselflies was not available. Therefore we have set an arbitrary jnd value of 1 as discrimination threshold, based on a previous study using honeybees (Apis mellifera) [5]. We have applied ANOVA with Bonferroni cluster to compare chromatic as well as achromatic contrasts between inter-sexual color spectra, intra-sexual color spectra, and color spectra of morphs and vegetation under two light irradiance conditions (twilight and morning light) to determine the statistical difference with respect to the damselfly's visual system when viewing individual color morphs. One-tailed t-tests were applied to investigate whether both chromatic and achromatic contrasts were significantly different compared to the jnd threshold value 1, under both light irradiance conditions.

Observation of morph frequencies from mating pairs
The observations of mating morph frequencies were conducted from 5:00 to 9:00 AM between December 2011 and January 2012 (n= 31 days). Comparisons of female morph numbers between mating pairs and the general female population were conducted to determine female morph preferences of males for mating. Results were analysed by G-test to determine male mate preference.