Sky radiance and spectral gradient are orienting cues for the sandhopper Talitrus saltator (Crustacea, Amphipoda) Free

The supralittoral sandhopper Talitrus saltator (Montagu 1808) is one of the best-known arthropod models for behavioural studies on celestial orientation. To return to the belt of damp sand by the sea, where it lives buried during the day, this species can rely on both the sun and the moon as cues for compass orientation (Papi and Pardi, 1953; Pardi and Papi, 1953; Pardi and Ercolini, 1986; Ugolini et al., 1999a,b). However, the sun and moon are not the only celestial cues used by this species. Beside astronomical cues, sandhoppers can use local cues, such as the slope of the substratum and the vision of the landscape as references to return to their belt of damp sand (Ercolini and Scapini, 1974; Craig, 1973; Hartwick, 1976; Ugolini et al., 1986, 2006; Ugolini, 2014; Ugolini and Ciofini, 2016). In addition to the sun, sunlight scattering on atmospheric particles generates additional skylight cues, such as a polarization pattern as well as profiles of radiance and spectral distributions (e.g. see Rossel and Wehner, 1984; Wehner, 1989, 1997; see also Horváth and Varjú, 2004; Cronin et al., 2014, for reviews). The pattern of skylight polarization provides directional information concordant with that provided by the sun, and helps many arthropods to orient themselves during the day even when the sun is not directly visible (e.g. under cloudy sky conditions) (see Waterman, 1981; Horváth and Varjú, 2004; Cronin et al., 2014, for reviews). Although the celestial polarization pattern server as a reliable orienting reference for many arthropods, sandhoppers do not seem to use it as a compass cue in their zonal recovery (Ugolini et al., 2013).

Only T. saltator and the dung beetle Scarabaeus lamarcki are known to be able to obtain compass information from the skylight radiance gradient (Ugolini et al., 2009, 2012; el Jundi et al., 2014a). Even though these two species face different ecological problems, they are both under pressure to orientate in the most efficient way: T. saltator must return to the damp belt of sand as quickly as possible by a type of chronometric orientation defined as unidirectional not-vector orientation to avoid death due to dehydration. In contrast, S. lamarcki needs to maintain a straight route while rolling its ball away from a dung pile in a randomly chosen compass direction to avoid intraspecific competition (e.g. Dacke et al., 2013, 2014; Khaldy et al., 2019).

Although the use of the skylight radiance gradient as an orientation cue was first described in various species of animals, including T. saltator, by Verheijen (1978), since then only a few investigations have been conducted to study its role in orientation in arthropods (Wehner, 1997; Ugolini et al., 2009; el Jundi et al., 2014a). In insects, the relevance of the skylight radiance as a compass reference seems to be relatively low and used only in combination with other cues. For instance, the orienting system of the desert ant Cataglyphis bicolor is disrupted when it is allowed to orientate using only this cue (individuals lose their ability to maintain a correct homeward direction but perform phototactic responses toward the solar hemidome instead; Wehner, 1997). The dung beetle S. lamarki (probably not performing time-compensated orientation) can orientate with high precision only by using intensity information in combination with additional skylight cues, such as skylight polarization (el Jundi et al., 2014a).

Although sandhoppers use the skylight radiance gradient for orientation (Ugolini et al., 2009), how the degree of cloudiness of the sky affects this ability has never been tested. An additional celestial cue used by some insects to achieve directional information is the spectral gradient. This is characterized by the predominance of longer (460–700 nm) wavelengths in the celestial hemicycle including the sun (‘solar’ hemidome) and shorter (293–460 nm) wavelengths in the opposite, ‘antisolar’ hemidome (Coemans et al., 1994). Some arthropods, like the honey bee Apismellifera (Brines and Gould, 1979; Edrich et al., 1979; Rossel and Wehner, 1984), the desert ant C. bicolor (Wehner, 1997) and the dung beetle S. lamarcki (el Jundi et al., 2015b, 2016), are known to be able to discriminate between the solar and the antisolar direction based on spectral information and to use them for spatial orientation. Interestingly, it has been demonstrated (Ugolini et al., 2010) that T. saltator perceives UV–blue (390–450 nm) and green (500–550 nm) wavelengths, which opens up the possibility that these animals may also derive compass information from the spectral pattern of the sky. The aim of this work was therefore to evaluate the ability of T. saltator (1) to detect and use skylight radiance profiles attenuated with respect to that occurring in conditions of clear sky and (2) to investigate the use of the celestial spectral gradient in its zonal recovery.

Adult individuals of T. saltator were collected in spring–summer 2013–2017 on a sandy beach in the Natural Park of Migliarino, San Rossore, Massaciuccoli (Pisa, Italy, 43°40′03″N, 10°20′29″E, sea–land axis 265–85 deg). Tests were conducted within 2 weeks of collection. In the laboratory, animals were kept in Plexiglas boxes containing wet sand at room temperature (25°C), under a light:dark cycle of 12 h:12 h in phase with the natural photoperiod. Food (universal dried food for fish, SERA^® Vipan, Heinsberg, Germany) was constantly available.

We tested groups of 10–12 individuals within one trial, as it has been shown that the directional choice of individual sandhoppers is not influenced by the heading directions of their conspecifics (Scapini et al., 1981). At the end of the experiments, the sandhoppers were released at the collection site.

We performed experiments to test the use of radiance gradients and spectral gradients, releasing sandhoppers under artificial light conditions. To increase the motivation to rapidly orientate seaward, the animals were dehydrated for 2–3 min before the experiments. The sandhoppers were released at the centre of a transparent Plexiglas bowl (diameter 18 cm) placed on a horizontal transparent plate (diameter 30 cm) at local noon (Fig. 1A). The sandhoppers can jump or walk, slipping, trying to climb the wall of the bowl. A single direction for each radially oriented individual (with the head pointed toward the outside of the bowl and the longitudinal axis of the body oriented no more than ±45 deg from the radius of the bowl) was recorded after 2 min from each release from freeze-framed images taken by a camera placed below the bowl. Sandhoppers belonging to each group were released only once. A goniometer was set below the bowl to measure the directions taken by the individuals with respect to 0 deg (north). The bowl was surrounded by an opaline Plexiglas cylinder (diameter 30 cm) closed at the top by an opaline Plexiglas plate (Fig. 1A). The artificial sky was projected (once reflected by a mirror) onto the circular opaline Plexiglas plate placed over the bowl, using a video-projector provided with a 200 W halogen lamp (LC4341, Philips).

In addition, we designed images attenuated to the gradient present in natural conditions of clear sky by reducing light radiance variation along the solar meridian. In particular, we elaborated patterns attenuated by 40% (Fig. 2D,E) and 75% (Fig. 2G,H).

Sandhoppers were released in a confined environment using the device described previously (see Ugolini et al., 1998, 2005). A white opaline Plexiglas dome (diameter 80 cm) was placed on top of the bowl. The internal surface of the dome (the artificial sky) was illuminated by a fibre-optic illuminator (Schott KL1500). The end of the light guide (diameter 8 mm) was accurately placed in a tube at the centre of the bowl (Fig. 1B) which allowed presentation of a wide stimulus across the inner surface of the dome (dome illumination). The end of a second light guide (diameter 4.5 mm) was mounted at the outer surface of the dome at an elevation of 45 deg and allowed presentation of a light spot to the animals (simulated sun). Tests were carried out with and without the simulated sun switched on (Fig. 1B). Some releases were carried out with the only the dome illuminated (i.e. with the simulated sun switched off) to exclude any possible use of additional orientation cues, such as the Earth's magnetic field. Artificial celestial blue profiles were reproduced by positioning a blue gelatine filter (no. 118 Light Blue, λ_max=450 nm, transmittance=22.16%, SpotLight, Milan, Italy) on the half of the tip of the fibre light guide illuminating the dome. To exclude the possible use of intensity information as orientation cues (radiance gradient), differences in light radiance between the two halves of the dome were reduced by placing two neutral density filters (no. 298, transmittance=69.3% and no. 209, transmittance=51.2%, Illuminoservice, Bologna, Italy) on the other half of the tip of the fibre, superimposed. This arrangement produced a spectral gradient across the dome's inner surface but not any absolute intensity differences: thus, the only orientating cue could be a spectral gradient across the artificial sky. To compare the artificial spectral gradient with the natural one, disregarding the natural radiance gradient, spectra achieved with outdoor measurement of the open sky were normalized to the light radiance pattern and elaborated according to blue and green sensitivity as explained below. The experiments with the spectral gradient stimulus were carried out in the morning, around noon, and in the afternoon. Thus, if the sandhoppers use spectral information as an orientation cue, they should orientate toward the blue-rich hemisphere in the morning, along the theoretical separation line between the two hemidomes at noon, and away from the blue-rich hemisphere in the afternoon. Prior to each experiment, the position of the blue filter and the azimuth of the simulated sun were adjusted according to the azimuthal position of the ‘real’ sun at the respective time of day (Fig. 3B–D).

Fig. 1C shows the artificial radiance profile recorded with respect to the natural one measured under clear sky conditions. The curve is relatively flat in the range of about 70–100 deg (it varies by about 20%) compared with the natural one: this radiance difference (20%) is too weak to serve as a compass cue for the sandhoppers (see Fig. 2E,F,H,I: a difference of about 40% and 75% is not sufficient to allow compass information with respect to the radiance gradient). The flat radiance profile in the artificial environment is crucial for the evaluation of the sandhoppers’ response to differences in sky spectral characteristics. To compare the colour (as perceived by sandhoppers) of the two halves of the artificial sky, we must balance the part of the spectrum, of each half, that falls in the blue sensitivity and the one which falls in the green sensitivity of T. saltator under a constant sky radiance.

If we consider a constant-radiance sky, from the point of view of a sandhopper, the blue-rich hemidome is much brighter than the green-rich hemidome both in nature and under our artificial spectral gradient. To consider this, we calculated the green/blue ratio (G/B) (Fig. 1D). The curves for the G/B ratio measured in nature and under our artificially generated spectral gradient show a similar shape with relatively more green light in the solar hemidome and relatively more blue light in the antisolar hemidome. Also, in this case, natural measurements were normalized in order to obtain a constant radiance.

Data were analysed following the methods proposed by Batschelet (1981) for circular distributions. For each distribution, the length of the mean resultant vector and the mean angle were calculated. We used the V-test to establish whether distributions differed from uniformity (P<0.05 at least) (Aneshansley and Larkin, 1981). The confidential limits for the mean angles (95% confidence; Batschelet, 1981) were calculated to assess the orientation towards the seaward (expected) direction. Comparisons among the number of radially orientated sandhoppers were made by the G-test (Zar, 1984).

Sandhoppers tested around local noon, under the artificial radiance gradient corresponding to the profile of the clear natural sky, orientated significantly toward the seaward direction of their home beach. The mean resultant vector was deflected from the expected direction by only 23 deg (Fig. 2C).

When the sandhoppers’ orientation behaviour was tested under weaker radiance gradients, attenuated by 40% (Fig. 2D–F) or 75% (Fig. 2G–I), individuals did not exhibit directional choices in accordance with the expected direction; in both cases, they showed a positive phototactic orientation behaviour toward the solar hemidome (Fig. 2F,I).

The number of the radially orientated sandhoppers out of the total number of individuals released was significantly higher in tests performed under the non-attenuated artificial gradient of radiance (corresponding to natural radiance) than in tests carried out under attenuated gradients (Fig. 4A, G=13.100, d.f.=2, P<0.01, G-test).

In the control experiment (Fig. 3A) under an illuminated dome that did not provide any visual cue as an orientation reference (i.e. a simulated sun or intensity radiance or spectral gradient), the sandhoppers were not statistically orientated (the distribution was not statistically dissimilar from uniformity). Thus, only 27% (N=8/30) of the released individuals were radially orientated (G=6.683, d.f.=1, P<0.01, G-test).

Sandhoppers tested under the artificial spectral gradient, combined with a simulated sun that was switched on were significantly orientated. Irrespective of the time of day tested, the mean directions were in agreement with the seaward direction of their home beach (Fig. 3B–D), deviating about 2–10 deg from the expected direction. In addition, most of the sandhoppers (80%, N=210) made a clear directional choice (Fig. 4B, white bars): the ratio of orientated animals released at noon and afternoon was about 95% and 94%, respectively. However, only about 50% of the sandhoppers released in the morning were radially orientated.

In experiments carried out under the artificial spectral gradient without the simulated sun being on (Fig. 3E–G), the sandhoppers kept mean directions in good agreement with the expected seaward direction at any time of the day: the deviation of the mean direction from the expected direction was 11 deg in the test performed in the morning, 25 deg at noon and 15 deg in the experiment conducted in the afternoon (Fig. 3E–G). In all cases, the ratio of orientated animals was about 45% out of the total number of individuals released (n=203, Fig. 4B, black bars).

In the noon and afternoon releases, the frequency of radially oriented individuals was significantly higher in the test conducted with the simulated sun switched on with respect to those carried out in conditions with it switched off (noon tests: G=31.192, d.f.=1, P<0.001; afternoon tests: G=41.779, d.f.=1, P<0.001, G-test). This was not the case in the morning, where the proportion of orientated animals was similar irrespective of whether the spectral gradient stimulus was presented with the simulated sun switched on or off (G=0.450, d.f.=1, P-value not significant, G-test).

Our results confirm the use of the skylight radiance gradient as a compass cue in the zonal orientation of T. saltator. Sandhoppers released under the artificial radiance gradient that matched the natural intensity gradient on a cloudless day orientated correctly toward the seaward direction of their home beach (Fig. 2A–C). In contrast, they lost their ability to use this celestial cue as an orientation reference when released under artificial radiance profiles that exhibited a weaker intensity gradient than the natural gradient (Fig. 2D–I). The significant reduction of the proportion of radially orientated sandhoppers further confirmed the inability of the sandhoppers to orientate according to attenuated radiance patterns (Fig. 4A). Indeed, as previously stated, a skylight radiance gradient could be used by T. saltator to orientate along the sea–land axis of the beach even when the sun is not directly visible (Ugolini et al., 2009, 2012). We hypothesize that in natural conditions this can occur in the presence of fog or cloud cover in the sky, the spectral gradient being absent or very reduced.

The results obtained in experiments carried out under the artificial spectral gradient (Fig. 3) represent the first evidence on the use of the celestial spectral gradient as a compass cue in T. saltator. In fact, sandhoppers performed time-compensated directional choices and kept the correct seaward directions of their home beach independent of the time of release. The uniform distribution obtained by releasing sandhoppers with only the dome illuminated, in the absence of the simulated sun and radiance and/or spectral gradients (Fig. 3A), demonstrates that T. saltator does not use the natural magnetic field as a compass reference under these conditions at least.

It is important to note that the spectral gradient was the only available cue in the experiments where the simulated sun was switched off. Thus, the reduction of the number of radially orientated individuals could be caused by the absence of any further reliable cues, such as the substrate slope and the surrounding landscape (Ercolini and Scapini, 1974; Craig, 1973; Hartwick, 1976; Ugolini et al., 1986, 2006; Ugolini, 2014; Ugolini and Ciofini, 2016). It remains an open question why the sandhoppers exhibited a relatively low number of radially oriented individuals (50%) registered from the morning releases with the artificial sun switched on (Fig. 4B), which was not statistically different from the frequency registered in the morning releases with the artificial sun switched off (43%). We hypothesize that it could be due to the conflict between the use of the sun compass and a positive phototactic tendency: in fact, in contrast to the other test conditions, in the morning, the azimuth of the sun (105–110 deg) and the seaward direction (265 deg) are almost in opposite directions.

Based on ethological and neuroanatomical investigations, it is known that the polarization pattern is detected by a specialized region of the compound eye (Labhart and Meyer, 1999), while the main retina perceives the spectral and intensity information of the sky in insects (el Jundi et al., 2014b). Information provided by these skylight references is then integrated into the same network in the central nervous system to generate a consistent compass signal (Brines and Gould, 1979; Edrich et al., 1979; Rossel and Wehner, 1984; Wehner, 1989; el Jundi et al., 2014b, 2015a,b, 2019). Previous investigations (Pardi and Papi, 1953, Ugolini et al., 2013) revealed that T. saltator perceives the skylight polarization but it does not use the pattern of skylight polarization as a compass reference. Moreover, it is not yet clear what advantage there is for the sandhoppers to perceive wavelengths in the UV–blue range.

As hypothesized by Coemans et al. (1994), it is possible to speculate that the UV sensitivity is part of a colour-processing system well adapted to derive the position of the sun from the use of skylight spectral contrast. According to this hypothesis, we reasoned that the ability of T. saltator to detect light in the UV range may be ecologically advantageous in that it renders its overall compass system more reliable. However, T. saltator perceives UV–blue wavelengths mainly with the lower part of the eye, while the upper part presents photopigments both for UV–blue and for green wavelengths (Ciofini et al., 2020). Therefore, we can imagine that the radiance and spectral gradients are perceived with the upper (dorsal) part of the eye while the UV–blue sensitivity of the lower part is more related to the identification of the sea hemicycle, given the importance of the perception of the coloured landscape (Ugolini et al., 2006), and not only of the contrast between the sea and land hemicycle in the zonal recovery of T. saltator.

Some of the results represented in Fig. 2 of this paper are reproduced, modified, from the PhD thesis of A. Ciofini (Ciofini, 2018). We wish to thank the Ente Parco Naturale Regionale Migliarino S. Rossore Massaciuccoli (Pisa) for permission to collect the sandhoppers. Many thanks are also due to Dr el Jundi (Würzburg University) for critical reading of a draft version of our paper.