Many migrating insects rely on the plane of sky polarization as a cue to detect spatial directions. Desert locusts (Schistocerca gregaria), like other insects, perceive polarized light through specialized photoreceptors in a dorsal eye region. Desert locusts occur in two phases: a gregarious swarming phase, which migrates during the day, and a solitarious nocturnal phase. Neurons in a small brain area, the anterior optic tubercle (AOTu), are critically involved in processing polarized light in the locust brain. While polarization-sensitive intertubercle cells [lobula-tubercle neuron 1 (LoTu1) and tubercle-tubercle neuron 1 (TuTu1)] interconnect the AOTu of both hemispheres, tubercle-lateral accessory lobe tract (TuLAL1) neurons transmit sky compass signals to a polarization compass in the central brain. To better understand the neural network underlying polarized light processing in the AOTu and to investigate possible adaptations of the polarization vision system to a diurnal versus nocturnal lifestyle, we analyzed receptive field properties, intensity-response relationships, and daytime dependence of responses of AOTu neurons in gregarious and solitarious locusts. Surprisingly, no differences in the physiology of these neurons were found between the two locust phases. Instead, clear differences were observed between the different types of AOTu neurons. Whereas TuTu1 and TuLAL1 neurons encoded E-vector orientation independent of light intensity and would thus be operational in bright daylight, LoTu1 neurons were inhibited by high light intensity and provided strong polarization signaling only under dim light conditions. The presence of high- and low-intensity polarization channels might, therefore, allow solitarious and gregarious locusts to use the same polarization coding system despite their different activity cycles.
- insect brain
- visual system
- polarization vision
- desert locust
many navigating animals rely on external visual signals for spatial orientation. Insects use mainly two mechanisms to calculate moving directions during flight or walking. In familiar areas, they are able to use visual landmarks as directional cues, whereas in unknown terrains and during long-distance migrations, compass signals from the sky are more relevant (Giurfa and Capaldi 1999; Collett and Collett 2000). Besides the direct position of the sun, the plane of sky polarization serves as a crucial reference for spatial directions during seasonal migration or homing (Wehner and Labhart 2006). Celestial polarized light signals are detected by photoreceptors in a specialized region of the compound eye, the dorsal rim area (DRA) (Labhart and Meyer 1999). While diurnal insects, including ants, bees, and monarch butterflies, refer to polarized light generated by the sun (Frost and Mouritson 2006; Wehner 1984), nocturnal dung beetles rely on the dim polarization pattern produced around the moon (Dacke et al. 2003, 2004).
Desert locusts (Schistocerca gregaria) perform long-distance migrations in huge swarms throughout North Africa and the Middle East and have been used as model organisms to analyze neural networks underlying the processing of sky compass signals in the brain. Behavioral experiments on tethered flying locusts have suggested that they are able to use polarized light signals from the blue sky to define their course during migration (Mappes and Homberg 2004). Like other locust species, desert locusts occur in two phases: a gregarious phase and a solitarious phase, which show substantial differences in appearance and behavior (Uvarov 1966; Simpson et al. 1999). While gregarious locusts migrate in swarms during the day, solitarious locusts are nocturnal and preferentially fly as individuals during the night (Waloff 1963; Roffey 1963). Both phases can fly long distances, and their movements can be oriented in consistent directions for periods of time. Movement directions are strongly influenced by wind direction, but there is some evidence for diurnal gregarious locusts that flight directions may be influenced by a sun-compass mechanism (Kennedy 1951; Baker 1978). Recent experiments have shown that solitarious locusts have significantly larger eyes compared with gregarious locusts (Rogers et al. 2010), a common way to increase the sensitivity of the visual system to the nocturnal lifestyle (Warrant 2004; Warrant and Dacke 2011). In the brain, considerable differences in the size and proportion of brain areas underlying the processing of visual signals have been found between both phases (Ott and Rogers 2010), and, at the neural level, differences have been demonstrated in the size of the receptive field of a looming-sensitive interneuron (Rogers et al. 2010). However, how the neural network in the brain allows the navigation of solitarious locusts at night and how this network is adapted to dramatically lower light conditions are completely unknown.
Polarized light information is processed in distinct areas in the locust brain (Homberg 2004; Homberg et al. 2011). The anterior optic tubercle (AOTu) is a major relay station for processing polarized light information and transfers polarized light signals from the optic lobe to the central complex (Homberg et al. 2003). The tubercle receives signals from the dorsal rim area of the medulla and layer 4 of the distal medulla via transmedulla neurons (el Jundi et al. 2011). Two classes of polarization-sensitive neurons, intertubercle cells and neurons of the tubercle-lateral accessory lobe tract (TuLAL), have been identified in the AOTu (Pfeiffer et al. 2005). The intertubercle neurons connect the AOTus of both hemispheres and consist of 3 neurons/brain hemisphere: a single lobula-tubercule neuron 1 (LoTu1) cell and two tubercle-tubercle neuron 1 (TuTu1) cells (Pfeiffer et al. 2005). Neurons of the TuLAL (TuLAL1 cells) transfer polarization signals to input neurons of the central complex (Träger et al. 2008) and consist of ∼40–50 neurons/brain hemisphere (Homberg et al. 2003). Spiking activity in most polarization-sensitive neurons is modulated sinusoidally during zenithal stimulation with a rotating polarizer (Labhart 1988). Except for the LoTu1 cell, all polarization-sensitive neurons of the AOTu show polarization opponency, i.e., they are maximally activated at a distinct E-vector orientation (Φmax) and are maximally inhibited at an orthogonal orientation (Φmin) (Pfeiffer et al. 2005). The LoTu1 neuron lacks an inhibitory part at Φmin, suggesting a particular role in the neural network of the AOTu.
Detailed characterization of polarization-sensitive neurons of the AOTu has focused on the intertubercle neurons, LoTu1 and TuTu1. Pfeiffer et al. (2005) showed that both types of intertubercle neurons receive polarization information via the ipsilateral DRA of the compound eye. Corresponding to anatomic evidence, E-vector tuning in LoTu1 was clustered around a Φmax of 134° (soma in the left hemisphere), whereas two tuning types around Φmax = 135 and 175° were found for TuTu1 neurons (somata in the left hemisphere). Neurons with somata in the right hemisphere showed mirror symmetric tuning (Pfeiffer et al. 2005). In addition to polarized light, both cell types are sensitive to unpolarized light stimulation. Whereas zenithal unpolarized light stimulation, especially at high light intensities, inhibits the neurons (Kinoshita et al. 2007; Pfeiffer et al. 2011), stimulation with chromatic light spots at an elevation of 45° leads to azimuth-dependent excitations and inhibitions, suggesting that the neurons use chromatic cues of the sky to distinguish between the solar and antisolar hemispheres (Kinoshita et al. 2007; Pfeiffer and Homberg 2007). While receptive field structures of neurons in several stages of the polarization vision pathway have been determined (Heinze et al. 2009; el Jundi and Homberg 2010; Träger et al. 2011; el Jundi et al. 2011), we still know little about the receptive field properties of AOTu neurons. In addition, the tuning and response characteristics of TuLAL1 neurons to polarized light are still poorly understood.
The present study had two aims: 1) to close these gaps in physiological data on AOTu polarization-sensitive neurons and 2) to reveal adaptations to different lifestyles by comparing data from gregarious and solitarious locusts. We show that receptive fields of the intertubercle neurons are large (>100°) and centered in the contralateral visual field, whereas those of TuLAL1 neurons are very heterogeneous. Intensity-response (I/R) characteristics show that TuTu1 and TuLAL1 neurons are adapted to signal E-vector orientation during the day independent of light intensity, whereas LoTu1 neurons show increased sensitivity and response strength during the night, suggesting optimal signaling of E-vector contrast under twilight conditions. Surprisingly, we found no differences between solitarious and gregarious locusts in the physiological parameters of these neurons. Therefore, gregarious and solitarious locusts might possess similar adaptations for high and low light intensity detection of the sky polarization pattern.
Gregarious desert locusts (Schistocerca gregaria) were raised under crowded conditions at a constant temperature of 28°C on a 12:12-h light-dark cycle. Rearing conditions for solitarious animals followed the procedures of Roessingh et al. (1993). Animals were kept individually in small boxes at 26.5°C and 60% humidity with a 12:12-h light-dark photoperiod and had neither visual nor olfactory contact. In general, full transition to the solitarious phase required three generations of animals kept in isolated conditions. A number of morphological markers were used as indicators for the successful generation of solitarious animals. Solitarious nymphs had a bright green coloration, in contrast to the yellow-dark brown patterning of gregarious nymphs (Simpson et al. 1999). Freshly hatched adults were light green in the solitarious state but had a pinkish coloration when they were gregarious. Sexually mature males were of yellow color with black patches in the gregarious state and were more uniformly brown-gray colored as solitarious animals. Another marker for solitarious adults was a light midline stripe along the dorsal thorax, which was less prominent in gregarious animals.
Preparation and electrophysiology.
Only sexually mature locusts (1–3 wk after imaginal molt) were used for the experiments. Recordings were performed from AOTu neurons during the subjective night [zeitgeber time (ZT) 12–24] and subjective day (ZT 0–12) of the animals. In both cases, preparation of the animals was performed under identical conditions using a cold light source (KL 1500, Leica Microsystems, Wetzlar, Germany) for illumination.
Animals were cold anesthetized for at least 30 min. Legs and wings were cropped, and stumps were closed with glue or wax. Mouthparts were sealed with wax, and animals were mounted with tape to a metal holder held by a ball joint in a vertical orientation. The holder was carefully adjusted so that the light stimulus (see below) had an exact zenithal position relative to the locust head, as shown by Pfeiffer et al. (2005). A ridge of wax was brought up frontally between the mouthparts and the anterior edge of the compound eyes. The head capsule was opened anteriorly, and the fat and trachea surrounding the brain were removed. To obtain stable recordings, the esophagus was cut, the abdomen was opened posteriorly, and the gut was removed from the opened abdomen. The abdomen was sealed with a tightly knotted thread. A wire platform was inserted between the esophageal connectives and was fixed at the ridge of wax to increase stabilization. Electrode penetration was facilitated by removing the neural sheath at the right anterior optic tubercle. During the whole preparation procedure, which lasted for ∼45 min, and during recording of neurons, the brain was immersed in locust saline (Clements and May 1974).
Neurons of the AOTu were recorded intracellularly using sharp microelectrodes (resistance: 60–190 MΩ). The electrodes were drawn from borosilicate capillaries (inner diameter: 0.75 mm and outer diameter: 1.5 mm, Hilgenberg, Malsfeld, Germany) using a Flaming/Brown horizontal puller (P-97, Sutter, Novata, CA). The tips of the glass micropipettes were filled with 4% Neurobiotin (Vector Laboratories, Burlingame, UK) in 1 M KCl and the shanks were filled with 1 M KCl. A silver wire inserted into the hemolymph solution served as the reference electrode. Neural activity of neurons of the AOTu was amplified (10×) with a custom-made amplifier and monitored with an oscilloscope (HM 205-3, Hameg, Frankfurt/Main, Germany). After being digitized at a sampling rate of 5 kHz (CED 1401 plus, Cambridge Electronic Design, Cambridge, UK), signals were stored on a personal computer using Spike2 software (version 6.02, Cambridge Electronic Design). After the recording, a constant depolarizing current was used to inject Neurobiotin iontophoretically into the neurons (2–3 nA, 1–5 min).
Locusts of both phases were stimulated with polarized monochromatic blue light obtained from a xenon lamp (XBO 150W, LOT-Oriel Group, Darmstadt, Germany, photon flux: 6.9 × 1013 photons·cm−2·s−1) after passing through a monochromatic filter (450 nm), a light guide (Schölly Fiberoptic, Denzingen, Germany), and a motor-driven linear polarizer (HNP'B, Polaroid, Cambridge, MA). The polarization filter was rotated through 360° in clockwise (0–360°) and counterclockwise (360–0°) directions with a constant speed of 30°/s. A set of neutral density filters between the light guide and the xenon lamp allowed us to change the light intensity in logarithmic steps. The polarization filter and the end of the light guide were attached to a perimeter device that enabled us to test the neuronal responses to stimulation from various points along the left-right meridian. In one experiment, ocular dominance was tested by shielding one eye from the light source with a handheld piece of cardboard during the stimulation with zenithal polarized light. Recordings were performed under dim ambient light conditions. During I/R measurements, background light was reduced further by covering the front of the Faraday cage with a light-tight curtain.
Zenithal stimulation of the animal was defined as 90° elevation, and lateral stimulations at an angular distance of 90° from the zenith were defined as 0° ipsilateral or contralateral stimulation. The terms “ipsilateral” and “contralateral” refer to the position of the soma of the recorded neuron. The angular size of the stimulus at the locust eye was ∼4.7°. For stimulation with high-intensity polarized white light, the 450-nm monochromatic filter was moved out of the light beam. The maximum light intensity was 1.68 × 1016 photons·cm−2·s−1, measured in the range of 350–880 nm, using a USB 2000+ fiber optic spectrometer (Ocean Optics, Dunedin, FL).
Brains with Neurobiotin-injected neurons were dissected out of the head and fixed overnight in 4% paraformaldehyde at 4°C. Brains were then washed four times for 15 min with 0.1 M PBS (pH 7.4) and were incubated with streptavidin conjugated to Cy3 (1:1,000, Dianova, Hamburg, Germany) in 0.1 M PBS containing 0.3% Triton X-100 (PBT). After an incubation period of 3 days, brains were again rinsed two times in 0.1 M PBT and then in 0.1 M PBS and were dehydrated in an ascending ethanol series (25–100%, 15 min each). After treatment with a solution of ethanol-methyl salicylate (1:1, 15 min), brains were cleared in methyl salicylate for at least 35 min. The whole mount preparations were finally embedded in Permount (Fisher Scientific, Pittsburgh, PA) between two glass coverslips using 10 reinforcement rings as spacers (Zweckform, Oberlaindern, Germany). Neurons were examined and identified using a Zeiss Axioskop epifluorescent light microscope.
For detailed three-dimensional (3-D) reconstructions of selected neurons, brains were rehydrated, sectioned, and again treated with Cy3-streptavidin as described by el Jundi et al. (2010). Briefly, Permount around the brains was removed by incubation with xylene (2–4 h). Brains were then rehydrated in a descending ethanol series, embedded in albumin-gelatin (4.8% gelatin and 12% ovalbumin in demineralized water), and fixed in 4% formaldehyde solution overnight at 4°C. Subsequently, brains were cut into 130- to 250-μm sections using a vibrating blade microtome (Leica VT1200 S, Leica Microsystems). Brain sections were preincubated with 5% normal goat serum (NGS; Jackson ImmunoResearch) in 0.1 M PBT overnight at 4°C. They were incubated for 6 days with a monoclonal mouse antibody against synapsin I [SYNORF1, dilution: 1:50 (Klagges et al. 1996), kindly provided by Dr. E. Buchner, University of Würzburg] and with Cy3-streptavidin (1:1,000) in 0.1 M PBT containing 1% NGS. Finally, after treatment of the brain sections with a secondary antibody, goat anti-mouse conjugated to Cy5 (Cy5-GAM, 1:300; Jackson ImmunoResearch) and with Cy3-streptavidin (1:1,000) in 1% NGS and 0.1 M PBT (for 4 days at 4°C), brain sections were dehydrated, cleared, and embedded in Permount between two coverslips.
Confocal imaging and 3-D reconstruction.
Brain sections were scanned with a confocal laser scanning microscope (Leica TCS SP5) using a ×20 (HCX PL APO ×20/0.70 Imm UV, working distance: 260 μm, Leica) oil objective. The Cy3 signal was scanned using a DPSS (561 nm) laser, and Cy5 fluorescence was detected with a HeNe (633 nm) laser. All neurons were scanned in several image stacks with a resolution of 1,024 × 1,024 (voxel size: 0.75 × 0.75 × 1.5 μm). The obtained image stacks were processed on a personal computer using Amira 5.3.3 software (Visage Imaging). The procedure of merging of the corresponding image stacks and the 3-D reconstruction of brain areas based on anti-synapsin staining have been previously described by el Jundi et al. (2009). 3-D reconstructions of the neurons were performed using the SkeletonTree tool (Schmitt et al. 2004).
Sampled spike trains were evaluated using Spike2 software with a custom-designed script (kindly provided by Dr. K. Pfeiffer, Halifax, NS, Canada). Action potentials were detected through threshold-based event detection. Events were visualized as mean spiking frequency using a gliding average algorithm (moving average of firing rate in window size: 1 s). With few exceptions, background activities of the recorded cells were measured by counting of spikes divided by the respective time (12 s) in a part of the spike train without stimulation (at dim ambient light conditions). To determine the E-vector tuning of the neurons, events during clockwise and counterclockwise rotations of the polarizer were assigned to the corresponding E-vectors, and lists of these angles were analyzed using Oriana 2.02 software (Kovach Computing Services, Anglesey, UK). The angle of the mean vector r averaged from equal numbers of clockwise and counterclockwise rotations of the polarizer was defined as the E-vector tuning (Φmax) of that neuron. The length of r describes the concentration of action potentials around Φmax and is thus a measure for the directedness of the response during rotation of the polarizer (Batschelet 1981; Pfeiffer et al. 2011).
To quantify the modulation strength of the neurons during polarized light stimulation, we calculated the response strength R (Labhart 1996). The stimulation period of the rotating polarizer was divided into 18 consecutive bins of 20°. In each bin, we calculated the difference between the actual spike frequency and the mean spike frequency during the total stimulation period. The sum of the absolute values of all 18 bins was defined as R. Background variabilities of the cells were calculated in the same way in a section of the spike train without stimulation. Relative R values were obtained by normalizing the response strength at a given position of the visual field to the maximum value (Rnorm). The widths of the receptive fields were determined by analyzing the elevations of half-maximal R in relation to the background variability. For visualization, data points of the receptive fields were connected by lines. Φmax distributions within the receptive field were obtained by subtracting the absolute deviation of the Φmax value at each elevation from the zenithal Φmax.
In I/R diagrams, R values were normalized against R at log(I) = 0 (Rnorm). I/R curves were fitted by applying the following modified Naka-Rushton function to the data (Naka and Rushton 1966): where I is the intensity of the stimulus, K is the intensity of the stimulus at 50% maximal Rnorm [Rnorm(max)], and υ is an exponent.
Box plots were created with Origin 6.0 software (Microcal, Northhampton, CA). The median value was indicated through a horizontal line, and boxes denoted 25% and 75% quartiles of the data. The 5% and 95% ranges of the data were visualized through whiskers.
Circular statistics were performed in Oriana 2.02. Responses of neurons to polarized light were analyzed statistically through the Rayleigh test for axial data (Batschelet 1981). Neurons were defined as polarization sensitive if the distribution of angles was significantly different from randomness (significance level: 0.05). The distribution of the preferred orientations of different recordings from the same neuron type was analyzed through Rao's spacing test (significance level: 0.05). To test whether the Φmax distribution of corresponding neurons differed between solitarious and gregarious animals, the Watson-Williams F-test (significance level: 0.05) was used.
Further quantitative comparisons of the data were made using SPSS software (version 11.5). The Shapiro-Wilk test (significance level: 0.05) was used to test for normality of data, and the Levene test (significance level: 0.05) was used to test for homogeneity of variance. For data that were not distributed normally or if the variance was inhomogeneous, the Mann-Whitney U-test (significance level: 0.05) was applied. In the case of a normal distribution of the data and homogeneity of variance, the two samples were analyzed through a Student's t-test (significance level: 0.05). If data were compared from the same recorded neuron, quantitative analysis was performed through a paired Student's t-test (significance level: 0.05). For statistical evaluation of multiple groups, one-way ANOVA combined with a Tukey honestly significant difference (HSD) post hoc test was applied (significance level: 0.05). If the Shapiro-Wilk test or Levene test was significant, ANOVA with Games-Howell post hoc test was used (significance level: 0.05). Linear regressions were calculated using Origin 6.0. The correlation coefficient (Rcorr) was measured, and the significance of regression was tested through a t-test against a slope of zero (significance level: 0.05).
This study presents electrophysiological data from 113 intracellular recordings from polarization-sensitive neurons of the AOTu in the locust brain. Four types of neuron were analyzed. The TuTu1 neuron innervates the lower units of the AOTus and transfers polarization information from the ipsilateral to contralateral AOTu (Fig. 1A). The second type of intertubercle neuron, the LoTu1 neuron, has additional ramifications in the ipsi- and contralateral anterior lobulae (Fig. 2A) (Vitzthum et al. 2002). The other two types of neuron, termed TuLAL neurons, connect the lower unit of the AOTu via the tubercle-accessory lobe tract to the lateral accessory lobe. TuLAL1a neurons connect the AOTu with a subunit of the lateral accessory lobe, the lateral triangle (Fig. 3A). TuLAL1b neurons ramify in the anterior lobula, lower unit of the AOTu, and median olive of the lateral accessory lobe (Fig. 4A) (Pfeiffer et al. 2005).
Receptive field structure and general tuning of AOTu neurons in gregarious and solitarious locusts.
TuTu1 intertubercle neurons were analyzed in 20 recordings from gregarious animals and 13 recordings from solitarious locusts (Fig. 1). TuTu1 neurons responded with polarization opponency to a dorsally rotating polarizer with excitation at Φmax and inhibition at Φmin relative to the background activity (Fig. 1, B and C). TuTu1 neurons from gregarious locusts had a background activity of 25.5 ± 11.4 impulses/s (means ± SD) and a background variability of 38.9 ± 15.0. They showed an average absolute R value of 176.2 ± 98.9 (means ± SD) to polarized light stimulation. Although visual inspection suggested clustering of Φmax values of the gregarious TuTu1 neurons between 10 and 60° and between 100 and 180° (Fig. 1E), the distribution of preferred orientations was statistically not significantly different from a uniform distribution (0.9 > P > 0.5 by Rao's spacing U-test). Receptive field properties of TuTu1 neurons of gregarious animals were analyzed in 19 recordings. In 16 recordings, the bilateral expansion of the receptive fields was analyzed during the subjective day (ZT 0–12), whereas in 3 gregarious locusts, receptive field properties were analyzed at ZT 12–24 (subjective night). No obvious differences between the receptive fields of TuTu1 cells recorded at night or during the day were noted. The averaged receptive field of all 19 TuTu1 neurons had a width of ∼110° and was centered eccentrically at an elevation of 60° in the contralateral hemisphere (Fig. 1D).
TuTu1 neurons of solitarious animals had a mean background activity of 29.1 ± 10.9 impulses/s (means ± SD) and a mean background variability of 44.9 ± 23.5. Both values did not differ significantly between solitarious and gregarious animals (P = 0.38 and P = 0.33, respectively, by Student's t-test). TuTu1 neurons from solitarious locusts had an averaged R value of 144.20 ± 61 (means ± SD), which was not significantly different from that of TuTu1 cells from gregarious animals (P = 0.55 by Mann-Whitney U test). As in gregarious TuTu1 cells, Φmax orientations of TuTu1 neurons were distributed randomly in solitarious locusts (Fig. 1F). Receptive field properties were analyzed in 11 neurons at night and 2 neurons during the day, but as in gregarious animals, no obvious differences were observed between the two groups. The averaged receptive field of all TuTu1 cells of solitarious animals had a width of ∼120°. Similar to the receptive field in gregarious locusts, it was centered eccentrically between 60 and 30° in the contralateral hemisphere (Fig. 1D). No significant differences were observed between solitarious and gregarious locusts at any tested position in the visual field.
The LoTu1 neuron was analyzed in 69 experiments (Fig. 2). In contrast to TuTu1 neurons, the LoTu1 neuron was activated at Φmax (Fig. 2, B and C) but lacked an inhibition at Φmin relative to background activity (Fig. 2C). In 43 recordings, LoTu1 properties were tested in gregarious animals. The neurons had a background activity of 13.3 ± 10.1 impulses/s, mean background variability of 21.3 ± 7.3, and R value of 72.4 ± 27.8 (means ± SD) in the center of the receptive field. Φmax orientations of the recorded neurons in gregarious animals showed a nonrandom distribution (P < 0.01 by Rao's spacing U-test) and ranged (with three exceptions) from ∼76 to 176°, with a mean Φmax orientation at 128.4 ± 31.6° (means ± SD; Fig. 2E). The receptive field structure of the LoTu1 neuron was analyzed in 33 gregarious animals. Twenty-six recordings were obtained during the subjective day, and seven recordings were performed during the night. No significant differences were found between receptive fields of LoTu1 cells in gregarious animals recorded during the day and at night at any of the tested elevations (by ANOVA with Games-Howell post hoc test). Similar to TuTu1 cells, the gregarious LoTu1 neuron had an eccentric receptive field with the strongest response at an elevation of 60° contralaterally (Fig. 2D). The width of the receptive field was ∼130° along the left-right meridian.
Physiological properties of the LoTu1 cell in solitarious locusts were analyzed in 26 animals. The averaged background activity of 14.9 ± 8.1 impulses/s and the mean background variability of 19.4 ± 6.1 (means ± SD) were not different from the corresponding firing properties in gregarious animals (P = 0.34 and P = 0.25, respectively, by Mann-Whitney U-test). The neurons showed an absolute response strength of 85.9 ± 33.16 (means ± SD) in the center of the receptive field, which did not significantly differ from the R value in gregarious locusts (P = 0.1 by Student's t-test). Φmax orientations of LoTu1 neurons in solitarious animals were distributed randomly (Fig. 2F), but, statistically, the mean preferred directions between solitarious and gregarious locusts did not differ (P = 0.07 by Watson-Williams F-test). The receptive field properties of LoTu1 neurons from solitarious animals were studied in 25 recordings (9 neurons during the subjective day and 16 neurons during the subjective night). Again, no significant differences were found in the receptive field properties between gregarious and solitarious locusts that were recorded during the subjective night or subjective day (ANOVA with Games-Howell post hoc test). In all groups, LoTu1 neurons had a receptive field of highly similar width (∼135°) and shape (Fig. 2D). As in gregarious locusts, the strongest response of the LoTu1 neuron in solitarious locusts was centered at an elevation between the zenith and 60° contralaterally.
Owing to the small diameter of TuLAL1 neurites, recordings from these neurons were relatively difficult and, thus, in previous work, these types of neuron were analyzed only rarely. We studied TuLAL1 neurons in 11 recordings (Figs. 3 and 4). The size of the receptive field along the left-right meridian of TuLAL1a cells was analyzed in seven recordings (2 gregarious locusts and 5 solitarious locusts). In all recordings, TuLAL1a neurons showed polarization opponency (Fig. 3, B and C). The background activity of the two gregarious TuLAL1a neurons ranged from 36.2 to 45.5 impulses/s, and the background variability ranged from 16.5 to 34. Both receptive fields were zenith centered and quite narrow (∼60°; Fig. 3E). R values of both cells ranged from 105.49 to 115.74. The Φmax orientation of both neurons was ∼30°, whereas Φmax orientations of the five TuLAL1a neurons from solitarious animals were distributed randomly (Fig. 3G). Without stimulation, neurons in solitarious animals had a mean background activity of 38.7 ± 19.7 impulses/s and a mean background variability of 38.5 ± 9.2 (means ± SD). Solitarious TuLAL1a neurons had an averaged R value of 101.9 ± 57.79 (means ± SD). No obvious differences were observed in R values, background activity, and background variability of TuLAL1a neurons between solitarious and gregarious locusts. The receptive fields of solitarious TuLAL1a neurons varied considerably in bilateral size and position and had centers in the contralateral or ipsilateral hemisphere (Fig. 3F). In one TuLAL1a cell from a gregarious locust, ocular dominance was tested by monocular stimulation of the ipsi- and contralateral eye (Fig. 3D). In contrast to intertubercle neurons (Pfeiffer et al. 2005), the neuron responded with similar R values to ipsilateral, contralateral, and bilateral polarized light stimulation (Fig. 3D).
Recordings from TuLAL1b neurons were obtained from four gregarious animals (Fig. 4). Three of the four neurons showed polarization opponency (Fig. 4, B and C), whereas one TuLAL1b neuron showed E-vector-dependent differences in activity with maximum activation at Φmax but no inhibition at Φmin. TuLAL1b neurons arborized in the median olive of the lateral accessory lobe (Fig. 4A) or showed additional ramifications in the lateral triangle of the lateral accessory lobe. The four neurons had a background activity of 18.5 ± 5.6 impulses/s and a background variability of 34.24 ± 16.9 (means ± SD) in darkness. TuLAL1b neurons had a mean R value of ∼155 ± 46.8 (means ± SD) in the center of the receptive field. As in TuLAL1a neurons, receptive field structures of individual TuLAL1b cells varied substantially in bilateral extension and position of the receptive field along the left-right meridian (Fig. 4D). The cells had receptive field centers in the zenith, ipsilateral, or contralateral hemisphere. Preferred E-vector orientations in the receptive field center were between 130 and 180° in three neurons and ∼5° in one neuron (Fig. 4E).
Taken together, no differences in the general physiological properties and in receptive field structures of polarization-sensitive neurons of the AOTu between gregarious and solitarious locusts were observed. Both intertubercle neurons had large receptive fields centered to the contralateral hemisphere. In contrast, receptive fields of TuLAL1 neurons varied substantially in bilateral size, shape, and position, with the strongest responses at elevations ranging from 30° contralaterally to 30° ipsilaterally.
I/R functions of AOTu neurons in solitarious and gregarious locusts.
Whereas gregarious locusts migrate during the day, solitarious animals preferentially migrate during the night (Walloff 1963; Roffey 1963). We were therefore interested to see whether these different lifestyles are reflected in the polarization vision network in the locust AOTu. I/R functions were obtained by changing the intensity of the polarized blue light stimulus in the center of the receptive field over a range of 4 log units (Fig. 5). Neurons recorded during the subjective day (ZT 0–12) were treated separately from neurons recorded during the subjective night (ZT 12–24).
TuTu1 neurons were analyzed during the subjective day in seven gregarious locusts and six solitarious animals (Fig. 5A). R values of gregarious TuTu1 neurons were saturated between log(I) = 0 and log(I) = −2 and showed a sharp drop to background levels between log(I) = −3 and −4 (Fig. 5A). The I/R function of TuTu1 neurons in solitarious animals was intensity independent between log(I) = 0 and −3, but at a logarithmic step of −4, the response broke down to background levels (Fig. 5A). Statistically, no differences were observed at each intensity step of the I/R function of TuTu1 neurons between gregarious and solitarious locusts.
I/R functions of LoTu1 neurons were based on 24 recordings from gregarious locusts and 18 recordings from solitarious animals. Eighteen LoTu1 neurons from gregarious locusts and five LoTu1 neurons from solitarious animals that were recorded during the day (Fig. 5B, left) showed similar I/R curves that gradually decreased to background levels between log(I) = 0 and log(I) = −4. In addition, recordings from seven LoTu1 neurons from gregarious animals and 13 LoTu1 neurons from solitarious animals during the night revealed similar sensitivity curves (Fig. 5B, right). As in TuTu1 neurons, I/R curves from LoTu1 neurons did not significantly differ between solitarious and gregarious locusts.
As mentioned above, the R value is a measure for the modulation strength of firing activity during stimulation but does not give information about the directedness of the response. Therefore, we tested whether the length of the mean vector r, which serves as a measure for the directedness of the response to polarized light (Pfeiffer et al. 2011), differed between both locust phases (Fig. 5, C and D). No significant differences in the directedness of TuTu1 (Fig. 5C) and LoTu1 (Fig. 5D) neurons between solitarious and gregarious locusts were found. Taken together, the data suggest that there are no differences in the neural network of the AOTu underlying the processing of polarized light between both locust phases.
Differences in neural responses between AOTu neurons.
Because no differences between solitarious and gregarious locusts were found in general tuning characteristics and light intensity dependence, we pooled data from both forms to compare the neural response properties of the different AOTu cell types in detail. Interestingly, I/R curves between both types of intertubercle neurons differed substantially. Whereas the R value to polarized light of TuTu1 neurons remained relatively constant between log(I) = 0 and log(I) = −3 but declined to background levels within the final log unit (Fig. 6A), the R value in the LoTu1 neuron decreased gradually from one logarithmic intensity step to the next (Fig. 6B). This was also reflected in the statistical analysis: in TuTu1 neurons, the response at log(I) = 0 differed only from the response at the lowest light intensity step [log(I) = −4; Fig. 6A], whereas in the LoTu1 neuron, the response to the highest analyzed light intensity differed significantly from all other light intensities (Fig. 6B). Furthermore, the responses at several light intensity steps in LoTu1 neurons differed significantly among each other. I/R curves of TuLAL1a and TuLAL1b neurons were similar to the I/R function of TuTu1 neurons but showed a slightly more shallow decline between log(I) = −2 and log(I) = −4 to baseline levels (Fig. 6, C and D). Thus, in contrast to TuTu1 and TuLAL1 neurons, which may signal E-vector orientation above threshold levels independent of light intensity, the response in LoTu1 neurons is strongly dependent on the intensity of the polarized light throughout all intensities tested.
We compared the response properties of the neurons in greater detail to further characterize the distinct roles of the different cell types in the processing of polarized light. LoTu1 neurons showed a significantly lower background firing rate than TuTu1 and TuLAL1a neurons (Fig. 7A). Furthermore, the background spiking rate in darkness was significantly lower in TuLAL1b cells than in TuLAL1a neurons. Whereas LoTu1 neurons showed low background variability and R values to polarized light stimulation, TuTu1 neurons showed significantly higher background firing variability and higher R values (Fig. 7, B and C). No statistical differences were observed between TuLAL1a and TuLAL1b neurons in background variability and R values and between intertubercle cells and TuLAL1 neurons. Finally, the directedness of the response showed no differences between all AOTu neuron types (Fig. 7D).
In the next step, we analyzed possible correlations between the tuning characteristics. Not surprisingly, the R value of all cell types correlated significantly with the length of r (data not shown). However, in all other tuning properties, no significant correlations were found except for a correlation between the background activity and the length of r in LoTu1 neurons. Whereas in TuTu1 neurons no correlation between the background spiking activity and the length of r was found (Fig. 7E), a linear correlation was present in LoTu1 neurons (t-test for slope = 0, Rcorr: −0.5, P = 0.0001; Fig. 7F). Owing to the small number of recorded neurons, a conclusion for TuLAL1 neurons was not possible, but, statistically, no correlation was found in TuLAL1a cells (t-test for slope = 0, Rcorr: −0.54, P = 0.212) or in TuLAL1b neurons (t test for slope = 0, Rcorr: −0.55, P = 0.46).
In the next analysis, the distributions of Φmax within the receptive field were investigated (Fig. 8). In contrast to TuTu1 cells, which did not show systematic changes in the preferred E-vector orientation within the receptive field (Fig. 8A), the preferred E-vector angle of LoTu1 neurons increased within the receptive field from ipsi- to contralateral positions (t-test for slope = 0, Rcorr: 0.23, P = 0.005; Fig. 8B). Likewise, in TuLAL1a neurons, deviations from zenithal Φmax values depended on the hemispheric side of the stimulus (t-test for slope = 0, Rcorr: 0.61, P = 0.015; Fig. 8C), whereas in TuLAL1b cells, no changes of Φmax values within the field of view were found (Fig. 8D).
To elucidate further differences between AOTu neurons, we next analyzed the correlation between tuning characteristics and the time of day of the recording (Fig. 9). TuTu1 neurons did not show any systematic changes in background activity (t-test for slope = 0, Rcorr: 0.16, P = 0.39) or variability (t-test for slope = 0, Rcorr: 0.31, P = 0.09) depending on the time of day (data not shown). We also did not observe a correlation between the directedness of the response to polarized light in the center of the receptive field and the time of day (t-test for slope = 0, Rcorr: 0.15, P = 0.39; data not shown) or between R values in the center of the receptive field and ZT (t-test against slope = 0, Rcorr: 0.09, P = 0.63; Fig. 9A). Neurons responded to polarized blue light during the subjective day and subjective night with similar R values (P = 0.68 by Mann-Whitney U-test; Fig. 9D). Also, the response directedness did not differ between TuTu1 neurons recorded at night and during the day (P = 0.94 by Mann-Whitney U-test). Similar to the conditions observed in TuTu1 neurons, no correlation between the general tuning characteristics (background activity and background variability), directionality, or R value (Fig. 9C) and the time of day was observed in TuLAL1 cells. LoTu1 cells showed no correlation between background activity or background variability and the time of day (t-test against slope = 0, Rcorr: 0.05 and −0.03, respectively, P = 0.64 and P = 0.83, respectively; data not shown). Although the length of r in the center of the receptive field did not change systematically with daytime (t-test against slope = 0, Rcorr: 0.24, P = 0.07; data not shown), LoTu1 neurons showed an increased directedness in neurons that were recorded at night compared with neurons that were recorded during the day (P = 0.042 by Mann-Whitney U-test; Fig. 9E). Although no significant differences were found between the I/R functions of LoTu1 neurons recorded during the day and LoTu1 cells analyzed at night (P > 0.05 by ANOVA combined with Games-Howell post hoc or with Tukey's HSD post hoc test; Fig. 5B), we found a correlation between the R values in the center of the receptive field and the time of day of the recordings (t-test for slope = 0, Rcorr: 0.41, P = 0.002; Fig. 9B). LoTu1 neurons also showed a significantly higher R value to polarized blue light at night than during the day (P = 0.0004 by two-tailed t-test; Fig. 9E).
Responses to high intensities of polarized light.
After analyzing responses under low-light conditions, we were interested in how the interneurons of the AOTu responded to stimulation with polarized light at higher light intensities as occur, e.g., at noon on a cloudless sky. Therefore, we stimulated LoTu1 and TuTu1 neurons with high-intensity polarized white light (1.68 × 1016 photons·cm−2·s−1) that ranged in a similar order of magnitude as illumination from the sun at noon (Elvegård and Sjöstedt 1940). As shown by Kinoshita et al. (2007), both intertubercle neurons have broad spectral sensitivities and are excited at Φmax during presentation with different wavelengths of polarized light. Whereas the LoTu1 neuron was activated during stimulation with polarized blue light, high-intensity polarized white light resulted in strong inhibition of spiking activity without any action potentials in 3 of 19 recorded LoTu1 neurons. The other neurons responded with reduced modulation of spiking activity (Fig. 10, A–D). Opposite lights on-responses to the different light intensities were particularly typical for LoTu1 neurons: whereas the cells showed an increase in firing rate at low light intensities (Fig. 10A), they were inhibited when the light was turned on at high intensities of light (Fig. 10C). Accordingly, the R value of the LoTu1 neuron was significantly higher during stimulation with polarized blue light compared with stimulation with polarized bright white light (Fig. 10E).
TuTu1 cells were analyzed in seven experiments. In contrast to LoTu1 cells, TuTu1 cells showed no significant difference in the responses between blue light stimulation and bright white light stimulation (Fig. 10G). To further exclude the possibility that the effect of the reduced R value in LoTu1 neurons is the result of the different spectral composition of the stimuli, we determined I/R relationships for different intensities of white light stimulation (Fig. 10, F and H). Five LoTu1 neurons showed a bell-shaped sensitivity curve with a maximum R value at log(I) = −2 and reduced R values at lower and higher intensities (Fig. 10F). Statistically, the R value at log(I) = 0 differed significantly from the maximum R value at log(I) = −2 (P = 0.04 by ANOVA combined with Games-Howell post hoc test). In individual LoTu1 neurons, the R value at log(I) = 0 was reduced by 29–70% compared with their maximum responses at lower light levels. In addition, we also analyzed the difference between the R values at each logarithmic step and the background variability. Because the background variability was calculated analogous to the R value, however, from a section of the spike train without stimulation, it is a measure for the modulation of a neuron at darkness. The R values between log(I) = −1 and −3 were significantly higher than the background variability, whereas the R values at log(I) = 0 and −4 were not different from background variability. In one investigated TuTu1 neuron, the R value at log(I) = 0 was similar to the R value at intensities between log(I) = −1 to −3 [6% reduction of the R value at log(I) = 0], again exhibiting an intensity-independent response (Fig. 10H). Taken together, these data confirm that the reduced R value in LoTu1 neurons at bright white light is an effect of light intensity.
We analyzed general neural activities, E-vector tuning, receptive field properties, and I/R functions in four classes of polarization-sensitive neurons of the AOTu in the locust brain. To our surprise, we found no differences in these physiological parameters between gregarious and solitarious locusts despite their different lifestyles and activity patterns. Consistent differences in physiological parameters were, however, found when we compared the different neuronal cell types. The LoTu1 neuron was exclusively activated by polarized light at moderate light intensities, whereas TuTu1 and TuLAL1 neurons showed polarization opponency. Above a certain threshold of light intensity, TuTu1 and TuLAL1 cells were largely invariant to changing light intensities and daytime, whereas the E-vector response of LoTu1 neurons was clearly dependent on light levels as well as on the time of day. In addition, stimulation with polarized light intensities mimicking illuminance levels of midday sunlight led to a strong reduction of E-vector-dependent responses in LoTu1 but not TuTu1 neurons. Taken together, these observations suggest that TuTu1 and TuLAL1 neurons provide a robust compass signal throughout the day, whereas the LoTu1 neuron is tuned to signal polarization information at low light conditions during sunset or sunrise.
General tuning properties.
While the R values between TuTu1 and LoTu1 neurons differed significantly, no differences were found in the averaged directedness of the response between any of the AOTu cell types. This indicates that the tuning width is similar in all AOTu neurons. The higher background variability of TuTu1 neurons compared with the LoTu1 cell could mean that TuTu1 neurons receive synaptic input from a larger number of neurons than the LoTu1 cell. Furthermore, in the LoTu1 neuron, a correlation between background activity and directedness was found. This implies that in this cell type, the background activity has an effect on the tuning width around Φmax. For the LoTu1 neuron, this is not surprising because its activity at Φmin is similar to background levels. Experiments in flies have shown that neural activities of visual neurons in the brain are modified during flight or walking (Rosner et al. 2010; Maimon et al. 2010; Chiappe et al. 2010). In visual neurons of locusts, a change of firing activity dependent on the behavioral state has also been observed (Homberg 1994). If the spiking rates of the LoTu1 neuron is, likewise, modified during walking or flight, the directedness of the response to polarized light should also be modified in a behavior-dependent context.
The E-vector tuning of the LoTu1 neuron in gregarious animals was significantly different from a random distribution. It had a mean value of ∼128.4°, which is similar to the mean Φmax of 134° reported for the LoTu1 neuron by Pfeiffer et al. (2005). In contrast, the distribution of Φmax orientations of TuTu1 neurons was not significantly different from randomness, unlike the two peaks around 135° and 175° reported by Pfeiffer et al. (2005), corresponding to the presence of 2 TuTu1 neurons/brain hemisphere. Therefore, we have to assume considerably larger interindividual differences in E-vector tuning of TuTu1 neurons than in the population of animals studied by Pfeiffer et al. (2005). Like in TuTu1 neurons, the preferred E-vector orientations of TuLAL1 neurons varied considerably. This might be explained by the existence of 40–50 TuLAL1 neurons, as suggested from fiber counts (Homberg et al. 2003).
The receptive fields of intertubercle neurons are directed to the contralateral visual hemisphere, with a maximum R value at an elevation of 60° along the left-right meridian. This fits very well to data from Pfeiffer et al. (2005), who showed that LoTu1 and TuTu1 neurons receive polarization input from the ipsilateral eye, and to anatomic data showing contralaterally pointing visual axes of DRA ommatidia (Homberg and Paech 2002). The extent of the receptive fields of TuTu1 (110–120°) and LoTu1 (130–135°) neurons along the left-right meridian was considerably wider than the receptive fields of DRA photoreceptor neurons [∼30° (Eggers and Gewecke 1993)]. This suggests that photoreceptor neurons with different spatial tuning are recruited by the intertubercle cells. In contrast to intertubercle neurons, TuLAL1 neurons differed widely in receptive field properties, including receptive field orientation and bilateral expansion. However, this may not be surprising in view of the high number of TuLAL1 cells per brain hemisphere (Homberg et al. 2003).
One of the main functions of the intertubercle neurons is probably to provide contralateral visual input to postsynaptic TuLAL1 neurons. As shown here in one example, TuLAL1a neurons do receive binocular input, and their postsynaptic partners, type TL2 tangential neurons with projections to the lower division of the central body (Träger et al. 2008), are likewise binocular (Heinze et al. 2009). TuLAL1b neurons, in contrast, might receive ipsilateral visual input only, because their likely postsynaptic partners, TL3 tangential neurons of the lower division of the central body (Träger et al. 2008), are also dominated by ipsilateral visual input (Heinze et al. 2009).
In contrast to TuTu1 and TuLAL1b neurons, E-vector tuning in LoTu1 and TuLAL1a neurons changed systematically along the left-right meridian. In both types of neurons, an increase in E-vector tuning from the ipsilateral to contralateral hemisphere was observed. As already illustrated for neurons of the protocerebral bridge, this shift of Φmax values within the receptive field suggests that neurons of the right AOTu respond more strongly when the sun is behind the animal and neurons of the left AOTu respond more strongly when the locust faces the sun (Heinze et al. 2009). Together with input from the sky chromatic contrast, as proposed by Pfeiffer and Homberg (2007), the E-vector tuning shift may further aid in distinguishing between the solar and antisolar hemispheres of the sky.
Responses of intertubercle neurons to different light intensities.
I/R functions of the intertubercle cells were determined in the center of their receptive fields. The response threshold at which the neurons showed no differences to firing activity at darkness was at a light intensity of ∼6.9 × 109 photons·cm−2·s−1. This is similar to the sensitivity of the intertubercle neurons analyzed during zenithal stimulation (Kinoshita et al. 2007). Clear differences were observed in I/R functions between LoTu1 neurons and the remaining AOTu neurons. In TuTu1, TuLAL1a, and TuLAL1b neurons, R values remained relatively constant and showed a sharp drop within the final 2 log intensity steps. This is in accordance with data from cricket POL1 neurons, which were also intensity independent above a certain threshold level of polarized light intensity (Labhart and Petzold 1993; Labhart et al. 2001; Petzold 2001). These neurons should, therefore, not be susceptible to changes in sky clouding conditions and are thus ideally suited to process polarized light signals for spatial orientation. In contrast, the response of the LoTu1 neuron was intensity dependent over at least 4 log units of light levels.
LoTu1 but not TuTu1 neurons showed higher R values and increased directedness at night than during the day. No differences, however, were observed in the absolute sensitivity measured by the I/R curves of LoTu1 neurons during the day and at night. As shown by Pfeiffer et al. (2005), Kinoshita et al. (2007), and the present study, the LoTu1 neuron receives at least two visual inputs from dorsal direction: 1) a low threshold excitatory input that is polarization sensitive and 2) a higher threshold inhibitory input that is insensitive to E-vector orientation. The lack of inhibition by unpolarized light at low light levels leads to stronger responses to polarized light at night than during the day and to a steeper I/R function of the LoTu1 neuron at night without affecting its absolute threshold for polarized light. When the intensity of polarized light is increased, however, the increasing contribution of the higher threshold inhibitory input eventually dominates the response of the LoTu1 cell. It leads to a reduction and finally to complete elimination of the E-vector-dependent response (Fig. 10F). Based on these characteristics, the LoTu1 neuron is probably adapted to signal polarized light at low light conditions during sunset or sunrise [1 × 1010 photons·cm−2·s−1 (Johnsen et al. 2006)], wherease high light intensities at noon should inhibit the LoTu1 neuron and strongly reduce its polarization sensitivity. A model proposed by Pfeiffer et al. (2011) suggests that the temporal dynamics of neural responses might largely account for the reversal of responses of the LoTu1 neuron with increasing light intensities. According to that model, stimulus-dependent release of histamine by DRA photoreceptors combined with dynamic membrane properties of lamina neurons could explain the opposing effects of high- and low-intensity polarized light on the LoTu1 neuron.
In contrast, the responses of TuTu1 neurons to high- and low-intensity polarized light did not significantly differ. As shown by Kinoshita et al. (2007), an inhibition of TuTu1 neurons by dorsal unpolarized light was observed in only one of a total of four recordings, and TuTu1 neurons did not show significant responses to UV or green unpolarized light presented from dorsal direction (Kinoshita et al. 2007). This correlates well with our data indicating intensity-independent responses of TuTu1 neurons to zenithal polarized light.
Comparison between gregarious and solitarious locusts.
The transformation in locusts from the solitarious to gregarious form strongly depends on population density and is mediated by the level of serotonin in the thoracic ganglia (Anstey et al. 2009). Field experiments have suggested that solitarious locusts prefer to migrate as individuals during the night (Waloff 1963; Roffey 1963). Whereas Roffey (1963) observed flight activity at night of locusts ranging from solitarious to transiens and gregarious, Waloff (1963) reported that flight of solitarious locusts occurred exclusively at night. He observed spontaneous flight activity of solitarious locusts from sunset to ∼5 h after disappearance of the sun. In both field studies, however, migrating solitarious locusts were also observed during the day, but their flight activity was interpreted as forced flights or as exceptions.
Substantial differences in the size and proportion of brain areas involved in visual processing have been observed between gregarious and solitarious locusts (Ott and Rogers 2010). Thus, gregarious locusts have larger brains than solitarious locusts, a noticeable larger optic lobe and central complex, and, in addition, a smaller optic lobe-to-midbrain ratio than solitarious locusts (Ott and Rogers 2010).
We did not find significant differences in the physiological responses of polarization-sensitive neurons of the AOTu between both forms. Receptive field structures as well as absolute sensitivities were analyzed extensively in the intertubercle cells, and, in both, no correlation between physiological responses and locust phase was found. In addition, no differences in E-vector tunings or in I/R functions were found between the two phases. The data therefore suggest that polarized light signals may be of similar significance as navigational cues in both forms. The presence of a high (TuTu1)- and low (LoTu1)-intensity polarization channel in both phases might, therefore, allow solitarious and gregarious locusts to use the same polarization coding system despite their different lifestyles. Although solitarious animals have larger eyes than gregarious locusts (Rogers et al. 2010), this difference is not reflected in the width and alignment of the receptive fields of AOTu intertubercle cells. The size and orientation of intertubercle receptive fields suggest that the size of the DRA as well as the number of DRA photoreceptors may not be different between both locust phases, and, thus, the increased eye size of solitarious locusts might be restricted to the main retina. It is, however, conceivable that specific adaptations of AOTu neurons in the two locust phases may be present in the neurons' responses to unpolarized chromatic stimuli, which were not tested here.
Possible functional role of AOTu neurons.
LoTu1 and TuTu1 neurons receive polarized light signals mainly from the ipsilateral eye (Pfeiffer et al. 2005); thus, an interconnection between the intertubercle cells of both hemispheres can be excluded. Both intertubercle neurons likely transmit signals to TuLAL1 neurons, which provide input to the sky compass in the central complex. Physiologically, TuTu1 neurons are similar to TuLAL1 cells and could implement an early comparison of E-vector signals of both eyes in TuLAL1a neurons. Owing to the strong dependence of the firing activity of the LoTu1 neuron to changes of the light intensity and, thus, on solar elevation, the LoTu1 neuron might act as a gain modulator that controls the R value of TuLAL1 cells in a daytime-dependent manner. Due to the higher absolute sensitivity of the LoTu1 neuron to polarized light than to unpolarized light (Kinoshita et al. 2007), light intensity conditions might be crucial in the LoTu1 neuron for the impact of the detected sky compass cue. As previously discussed by Kinoshita et al. (2007), high light intensities at noon will probably reduce the contribution of polarized light input and simultaneously increase the significance of unpolarized light input for detection of the azimuthal direction of the sun. Thus, the LoTu1 neuron could control the balance in TuLAL1 neurons between input from the sky polarization pattern and direct azimuthal input from unpolarized sunlight.
This work was supported by Deutsche Forschungsgemeinschaft Grant HO 950/16-2. Parts of the effort were sponsored by the Air Force Office of Scientific Research, Air Force Material Command, United States Air Force, under Grant FA8655-08-1-3021.
The United States Government is authorized to reproduce and distribute reprints for Government purpose notwithstanding any copyright notation thereon. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the Air Force Office of Scientific Research or the United States Government.
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: B.e.J. and U.H. conception and design of research; B.e.J. performed experiments; B.e.J. analyzed data; B.e.J. and U.H. interpreted results of experiments; B.e.J. prepared figures; B.e.J. drafted manuscript; U.H. edited and revised manuscript; U.H. approved final version of manuscript.
The authors thank Dr. Keram Pfeiffer for providing the Spike2 script for data analysis and Sebastian Richter and Manfred Peil for constructing the stimulation device and control equipment. The authors further thank Tim-Henning Humberg for rehydrating and scanning neurons, and the authors are grateful to Dr. Carsten Heuer for suggestions on the manuscript and to Martina Kern and Karl-Heinz Herklotz for maintaining the gregarious locust cultures. The authors thank Martin Kollmann for establishment of the solitarious culture and Evelyn Rieber and Tim-Henning Humberg for rearing the solitarious locusts.
Present address of B. el Jundi: Dept. of Biology, Lund Vision Group, Lund Univ., Lund 22362, Sweden.
- Copyright © 2012 the American Physiological Society