Compass Cue Integration and Its Relation to the Visual Ecology of Three Tribes of Ball-Rolling Dung Beetles

Simple Summary To escape competition for food at the dung pat, ball-rolling dung beetles shape a piece of dung into a sphere and roll it away. To maintain their bearing, these beetles integrate directional information from a range of celestial cues. For the majority of diurnal dung beetles living in open habitats, the most dominant of these cues is the sun. It has recently been demonstrated that beetles living in closed habitats, with closely spaced trees and tall grass, rely predominantly on directional information provided by polarised skylight rather than the sun. Taken together, these findings suggests that the orientation strategy of the beetle is influenced by the animal’s visual ecology. To further investigate the relative weighting of cues in the orientation system of beetles, and its relation to their visual ecology, we investigated the orientation strategy of ball-rollers from three different dung beetle tribes, all present within the same savanna biome. We find that species within a tribe share the same orientation strategy, but that this strategy differs across tribes. We conclude that, despite dramatic intertribal differences in body size and external eye design, the dynamic heading direction network of the South African ball-rolling dung beetles is well adapted to guide the foraging insect in the habitat that it normally traverses. Abstract To guide their characteristic straight-line orientation away from the dung pile, ball-rolling dung beetles steer according to directional information provided by celestial cues, which, among the most relevant are the sun and polarised skylight. Most studies regarding the use of celestial cues and their influence on the orientation system of the diurnal ball-rolling beetle have been performed on beetles of the tribe Scarabaeini living in open habitats. These beetles steer primarily according to the directional information provided by the sun. In contrast, Sisyphus fasciculatus, a species from a different dung-beetle tribe (the Sisyphini) that lives in habitats with closely spaced trees and tall grass, relies predominantly on directional information from the celestial pattern of polarised light. To investigate the influence of visual ecology on the relative weight of these cues, we studied the orientation strategy of three different tribes of dung beetles (Scarabaeini, Sisyphini and Gymnopleurini) living within the same biome, but in different habitat types. We found that species within a tribe share the same orientation strategy, but that this strategy differs across the tribes; Scarabaeini, living in open habitats, attribute the greatest relative weight to the directional information from the sun; Sisyphini, living in closed habitats, mainly relies on directional information from polarised skylight; and Gymnopleurini, also living in open habitats, appear to weight both cues equally. We conclude that, despite exhibiting different body size, eye size and morphology, dung beetles nevertheless manage to solve the challenge of straight-line orientation by weighting visual cues that are particular to the habitat in which they are found. This system is however dynamic, allowing them to operate equally well even in the absence of the cue given the greatest relative weight by the particular species.


Introduction
To successfully navigate the world, animals often rely on directional information from more than one sensory channel [1][2][3]; fruit flies require visual feedback to localize an odour source [4], bees are better at pin-pointing their nest entrance with the addition of olfactory cues [5] and Bogong moths use the Earth's magnetic field in combination with landmarks to localize Alpine caves [6]. In addition, orienting insects seem to afford the greatest weight to the directional information that conveys the highest certainty at a given moment [7]. Monarch butterflies primarily rely on the sun to find their route across the North American continent [8] but will refer to polarised skylight for directional guidance as soon as this bright solar cue is obstructed [9][10][11], and as the sun climbs high in the sky, becoming less reliable for directional input, dung beetles and ants rely more heavily on directional information provided by wind [1,12]. Homing ants, which find their way back to their nest by path integration and landmarks, also employ a dynamic strategy for reliable navigation; as these foragers are displaced further and further from their nest and the visual scenery around them becomes increasingly unfamiliar, they shift the relative directional weight of their path integrator and landmark guidance in favour of the former to lead them back home [13].
Not surprisingly, the visual ecology of an animal influences what directional cues to follow where and when. Rodent and fish species living in spatially complex environments will rely more on egocentric cues to find their way compared to species inhabiting more open habitats [14][15][16]. Another example can be found among ants, where species inhabiting cluttered, landmark-rich spaces, rely more on landmark guidance compared to desert ants that forage in open, featureless habitats [13,[17][18][19][20]. While these differences in directional guidance appear to be species specific, and strictly tuned to the visual environment in which the animal lives, a dynamic influence of the visual ecology of the navigator can be observed in the strictly nocturnal, savanna-living dung beetle, Scarabaeus satyrus (Fabricius). During a moon-lit night, this beetle orients using polarised lunar skylight in preference to the moon, but if coerced to roll during the day, the heading direction network of the beetle shifts the relative weight of these two types of celestial cues in favour of directional information provided by the sun [21].
Diurnal ball-rolling dung beetles steer their characteristic straight-line escapes from a dung pile [22][23][24][25][26][27] by directional information provided by the sun [21,23,24,27], the polarised skylight [26,28], the gradients of intensity [28] and colour that form across the daytime sky [29], as well as the prevailing winds [1]. Our understanding of how these insects roll straight over the sun-lit savanna is largely based on behavioural, morphological and neurobiological studies of beetles from the tribe Scarabaeini [1,21,23,27,28,30] (but see [24,26,31] and below). In these studies, we repeatedly found that if the directional information from the sun is set in conflict with other celestial cues, the beetles change their bearings according to the position of the sun [23,30]. These experiments clearly demonstrate that directional information from the sun is given the greatest relative weight during straight-line orientation in these large and iconic dung beetle species. In addition, it seems that, contrary to homing ants, which choose an intermediate route when directional cues are set in conflict [12,[32][33][34][35][36], dung beetles do not average the dictates of the directional sources, but instead predominantly rely on the directional information given the greatest weight at that time. Consequently, only when the sun is hidden from view will the Scarabaeini beetles turn in response to the rotation of an overhead pattern of polarised light [28].
It was recently shown that a savanna woodland-living species of the tribe Sisyphini, Sisyphus fasciculatus, displays a different behaviour; when rolling under a polarising filter under a sun-lit sky outdoors, this beetle changes its bearing in accordance with the turn of the polariser [26]. This suggests that these small beetles, which traverse litter strewn terrain under closely spaced trees or through tall grass, predominantly rely on directional information from the celestial pattern of polarised light [26]. Due to this contrasting behaviour, Khaldy et al. [26] suggested that the visual ecology of the orientation system of the different species of dung beetle, just as in ants, is influenced by their distinct habitat associations. Here, we continue to explore the relative weight of directional information in Figure 1. Dung beetles from three tribes of ball-rollers and the bioregions they inhabit. Beetles from three tribes of ball-rollers (blue-bordered image: Scarabaeini; yellow-bordered image: Gymnopleurini; red-bordered image: Sisyphini) were collected within the same savanna biome. K. nigroaeneus and G. unicolor were predominantly found actively foraging in the open habitat (a) and S. fasciculatus predominantly foraged within the closed habitat (b) of the same bioregion. A histogram, illustrating the percentage of individuals found in the open and closed habitat over three consecutive sampling days, is presented below each respective beetle image. A 180° view of the sky from the ground perspective of the beetle is included at the bottom of each panel.   emptied and re-baited with fresh dung every 3 h during daylight hours. For more details regarding the trapping method, see Khaldy et al., 2020 [26]. The habitat preference for G. nitens, that shares the same savanna biome, was not defined.

Statistical Analysis of Habitat Preference
To test for differences in species abundance between habitat types, generalized linear mixed models (GLMMs) [38] in R (R Core Team 2020, Vienna, Austria, https://www.Rproject.org/, accessed on 31 May 2021), used with lme4 [39], were fitted. Each trap of each sampling event was used as a sampling unit, with a total of 155 sampling units. The Shapiro-Wilk test was used to test for normality in the residual distribution of the species abundance. The abundance of each species was non-normal count data and Poisson error distribution was specified in each model [39]. In all GLMMs, habitat type was treated as a fixed factor and sampling day as a random factor to block the layout of the sampling design.

Eye Size
To measure the eye surface area, the right eye of ten individuals of each species was covered with a thin layer of transparent nail polish. Once dried, the coat of nail polish was peeled off from the eye, cut and mounted flat on a microscope slide. The images of the flattened impression of the eyes were taken with a stereo microscope (Zeiss Stereo Discovery V12) and the absolute area was measured using ImageJ (Rasband, W.S., ImageJ, U. S. National Institutes of Health, MD, USA, https://imagej.nih.gov/ij/, 1997-2018, accessed on 20 May 2021). As K. nigroaeneus possesses a complete canthus, the absolute eye area for this species was calculated as the sum of the dorsal and ventral eye area.

Behavioural Experiments
Outdoors, experiments were performed under clear skies, at solar elevations between 45 • and 60 • , at Bersig Eco Estate and Pullen nature reserve between March 2018 and November 2019. In Lund University, Sweden, the beetles were presented with a green unpolarised light spot (Adafruit DotStar Digital LED Strip; emission peak 530 nm, Adafruit Industries, New York, NY, USA), a previously documented replacement for the sun in the heading direction network of the beetle [21], at an elevation of 45 • , in an otherwise completely darkened indoor room. An overhead Sony Handycam HDR-CX730E (fitted with a 0.42× wide angle lens), mounted from above with the lens facing downwards, was used to record exit bearings.

Orientation Performance of Dung Beetles
To determine the beetle's orientation performance under an open sky, each individual was repeatedly placed beside its ball in the centre of a circular, flat, sand-coated arena, where the effective radius was set to a distance equivalent to the length of 20 steps for the species tested (K. nigroaeneus; 59 cm, G. unicolor; 32 cm, S. fasciculatus; 32 cm) (for detailed data see Supplementary Table S1). Each beetle was allowed to roll its ball to the arena perimeter ten times. Ten individuals per species were tested.

Relative Weighting of Directional Cues in the Orientation System of Dung Beetles
For each experimental treatment, the beetle was placed alongside its dung ball, in the centre of a 50 cm radius circular arena and allowed to roll its ball to the perimeter where its exit bearing was noted. For conditions requiring a polarising filter, a circular 30 cm radius, UV/Visible light-transparent polarisation filter (BVO UV Polarizer, Bolder Vision Optik©, Boulder, CO, USA) was positioned over the centre of the arena. The filter was mounted on four legs (10 cm in height) and fitted with a black cloth curtain around its perimeter to prevent the entry of light from outside the filter. The exit bearing was recorded when the beetle reached the filter perimeter. Upon completion of the beetle's first roll, the position of the test cue(s) was rotated by either 90 • or 180 • (see Section 2.4.3. below). The beetle was allowed to exit the arena and its second exit bearing was noted. A third exit, presenting the same visual parameters as in the first trial, was performed as a control to test whether the beetle could follow approximately the same bearing throughout the experiment. Angular change was calculated as the difference in bearing between the first and second exit (test), or first and third exit (control). In total, each individual rolled from the centre to the edge of the arena (or filter perimeter) three times. In all outdoor experiments, 20 individuals per species were tested. For the indoor experiments, 10 individuals were tested for each species.

Manipulation of Directional Input
Sun (ersatz or real): In the field, the sun's apparent position was changed by 180 • using a mirror (30 × 30 cm), while simultaneously concealing the real sun from the beetle's field of view using a wooden shade board (100 × 75 cm). Indoors, the azimuth of the ersatz sun was changed by 180 • between trials by switching off and on the green light spot at different relative positions.
Polarised light: In the field, the UV/Visible light-transparent polarisation filter was turned by 90 • , between consecutive rolls, either under a full view of the sun or with the sun shielded from the beetle's field of view by the shade board. The initial orientation of the filter was alternated for each beetle, with every second beetle starting with the polarisation filter aligned to the natural polarisation band of the sky, and every other beetle with the filter aligned perpendicular to the natural polarisation band of the sky.
Sun and polarised light: In these experiments, the polarising filter was turned by 90 • in combination with a 180 • change in the solar position, as described above.

Circular Statistics
Circular statistics on measured data was performed using Oriana 4.0 (Kovach Computing Services, Anglesey, UK). All circular data are reported as mean ± one circular standard deviation. Distributions of exit angles were analysed using Rayleigh's uniformity test for circular data [40]. Changes in direction between treatments were calculated by measuring the angular difference in exit bearing between two exits from the arena and analysed using a v-test with an expected mean of 0 • for the control experiments and 180 • for the mirrored sun/ersatz sun experiments. To test for homogeneity of two or more samples, a Mardia-Watson-Wheeler test was used.

Differences in Eye Size and Shape
The relatively big eye (1.60 ± 0.57 mm 2 ) of K. nigroaeneus is completely divided into a dorsal (0.59 ± 0.22 mm 2 ) and a ventral part (Figure 2a), while the smaller eyes of G. unicolor (0.21 ± 0.03 mm 2 ) and S. fasciculatus (0.15 ± 0.03 mm 2 ) rather have a more oval-shaped dorsal eye (0.05 ± 0.01 and 0.02 ± 0.01 mm 2 , respectively) which connects to the ventral part of the eye (N = 10) (Figure 2b,c) (for detailed data see Supplementary Table S1).

Orientation Performance under the Natural Sky Is Equal for All Species
The outdoor orientation performance of the three species, as determined from the mean resultant vector length (R) of 10 exit bearings per beetle from the centre of the circular arena (the closer to 1, the better oriented the beetle) did not differ between the species  Supplementary Table S2). We also found that within a species, the first bearing chosen by each individual was not biased towards any particular heading (K. nigroaeneus: p = 0.06, Z = 2.8; G. unicolor: p = 0.54, Z = 0.64; S. fasciculatus: p = 0.22, Z = 1.53, Rayleigh uniformity test, N = 10). Although the evidence for this was weaker in K. nigroaeneus (p = 0.06), previous work on closely related Scarabaeini species [22,30] suggests that this is most likely an effect of the small sample size. Taken together, this indicates that, under an open sky, our test species, from three different tribes, are able to travel along any given bearing with the same angular precision.

Differences in Eye Size and Shape
The relatively big eye (1.60 ± 0.57 mm 2 ) of K. nigroaeneus is completely divided into a dorsal (0.59 ± 0.22 mm 2 ) and a ventral part (Figure 2a), while the smaller eyes of G. unicolor (0.21 ± 0.03 mm 2 ) and S. fasciculatus (0.15 ± 0.03 mm 2 ) rather have a more oval-shaped dorsal eye (0.05 ± 0.01 and 0.02 ± 0.01 mm 2 , respectively) which connects to the ventral part of the eye (N = 10) (Figure 2b, c) (for detailed data see Supplementary Table S1).

Figure 2.
Habitus, eye design and dominant orientation cue in three tribes of ball-rolling dung beetles. Diurnal species within the tribe Scarabaeini (a) attribute greatest relative weight to directional information provided by the sun during straight-line orientation. In contrast, two nocturnal species from the same tribe, as well as the smaller, diurnal Sisyphus fasciculatus, from the tribe Sisyphini (b), rely predominantly on polarised skylight for directional information. The underlying weighting strategy for straight-line orientation within the tribe Gymnopleurini (c) differs from that previously mentioned, where neither directional information from the sun nor the polarisation pattern dominates the output from its compass network. As can also be noted from our test species, Sisyphini are generally much smaller than Gymnopleurini, which in turn are smaller than Scarabaeini. The relative eye sizes across the three tribes follow the same pattern, but they differ in shape. The canthus (c) completely separates the roughly equal sized dorsal (d) and ventral (v) eyes of the Scarabaeini, while the dorsal portion of the eye of the Gymnopleurini and Sisyphini is only partially separated and much smaller than the ventral part. Species tested in this study are indicated by an asterisk (*). Habitus, eye design and dominant orientation cue in three tribes of ball-rolling dung beetles. Diurnal species within the tribe Scarabaeini (a) attribute greatest relative weight to directional information provided by the sun during straight-line orientation. In contrast, two nocturnal species from the same tribe, as well as the smaller, diurnal Sisyphus fasciculatus, from the tribe Sisyphini (b), rely predominantly on polarised skylight for directional information. The underlying weighting strategy for straight-line orientation within the tribe Gymnopleurini (c) differs from that previously mentioned, where neither directional information from the sun nor the polarisation pattern dominates the output from its compass network. As can also be noted from our test species, Sisyphini are generally much smaller than Gymnopleurini, which in turn are smaller than Scarabaeini. The relative eye sizes across the three tribes follow the same pattern, but they differ in shape. The canthus (c) completely separates the roughly equal sized dorsal (d) and ventral (v) eyes of the Scarabaeini, while the dorsal portion of the eye of the Gymnopleurini and Sisyphini is only partially separated and much smaller than the ventral part. Species tested in this study are indicated by an asterisk (*).

Ball-Rolling Dung Beetles Can Orient to a Single Green Light Spot
Beetles presented with a green light spot (indoors) as an ersatz sun in the same azimuthal position between two consecutive exits from the centre of the arena (control), and showed no significant change in direction in any of the three species, (K. nigroaeneus: µ = 28.   , while the changes in exit bearings for G. unicolor were randomly distributed within the population (graph 2). Beetles were then allowed to roll a third time, with the manipulated cue(s) moved back to its/their initial position. The mean angular difference between the first and second exit (test), and the first and third exit (control), is represented by a red solid vector and a grey dotted vector, respectively, in each graph. Error bars represent one circular standard deviation. The data presented for S. fasciculatus in (a) and (c) (graph 3, respectively) were adapted from Khaldy et al., 2020 [26]. For detailed data see Supplementary Table S3.
When the position of the ersatz sun was changed by 180 • between two exits from the centre of the arena (test), all species changed their headings accordingly (K. nigroaeneus: µ = 219.59 • ± 37.73 • , p < 0.01, V = 2.77; G. unicolor: µ = 186.27 • ± 49.40 • , p < 0.001, V = 3.47; S. fasciculatus: µ = 185.9 • ± 41.67 • , p < 0.001, V = 3.41, v-test (with the expected mean of 180 • ), N = 10) (Figure 3a). These changes in headings showed that the species tested can steer with reference to a single point-light source and with no significant difference in performance between species (p = 0.139, W = 6.95, Mardia-Watson-Wheeler test, N = 10). As for the control for the experimental treatment (including our handling of the beetles), the changes in bearing between two exits under an unmanipulated sky was also calculated (control); the average change of bearings was clustered around 0 • for all species To further investigate the role of polarised skylight on the orientation system of K. nigroaeneus and G. unicolor, the polarising filter was again placed above the arena, but now with the sun obstructed from view. The changes in headings recorded for the two species in response to a 90 • rotation of the filter now clustered around 67.36 • ± 35.45 • for K. nigroaeneus, and around 72.49 • ± 36.57 • for G. unicolor (N = 20) (Figure 3d), demonstrating that when the sun is obstructed from view, directional information from the overhead polarised light pattern is now attributed a relatively greater weight in the orientation system of these two species.

The Combined Role of Sun and Polarised Skylight in the Orientation System of Garreta Unicolor and G. nitens
Given that G. unicolor did not turn despite a displacement of the sun or rotation of the pattern of polarised light under the open sky but did orient to an ersatz sun indoors and to a polarised light pattern in the shade, we then rotated the polariser by 90 • while simultaneously mirroring the sun by 180 • and shielding the real sun from the beetle's view (Figure 3e, graph 2). To our surprise, the angular changes in bearing recorded for G. unicolor in response to this manipulation were not different from a random distribution (p = 0.70, Z = 0.37, Rayleigh uniformity test, N = 20) (Figure 3e, graph 2). It is important to note that the beetles still maintained a straight trajectory when rolling. As soon as the real sun was revealed and the polarising filter was turned back to its original position, the beetles resumed their initial direction of travel (µ = 344.67 • ± 49.03 • , p < 0.001, V = 4.23, v-test (with an expected mean of 0 • ), N = 20).
To further evaluate this somewhat surprising observation, we repeated this experiment on K. nigroaeneus and the close relative G. nitens (due to their experimentally frailer nature, the tiny S. fasciculatus would not perform under this condition, but rather flew away from the setup at any given chance). While K. nigroaeneus altered its heading towards a 180

Discussion
In this study, we demonstrated how the orientation system of ball-rolling dung beetles, belonging to three different tribes that co-occur within the same savanna biome, attribute different relative weights to directional information during straight-line orientation.

Diurnal Scarabaeini Attribute the Greatest Relative Weight to the Directional Information Provided by the Sun
As with sandhoppers, monarch butterflies and birds [41][42][43], ball-rolling dung beetles can direct their straight-line movements according to directional input from a single source of light in an indoor setting (Figure 3a and [21,24,27]). Outdoors, however, the beetles are exposed to a range of celestial directional cues, including the sun, polarised skylight [44][45][46], as well as the gradients of intensity [47] and colour [48][49][50] that form across the natural sky. Therefore, if the apparent position of the real sun is changed by 180 • with the aid of a mirror and a shading board, the directional information from the sun is set in conflict with that of the rest of the sky. Nevertheless, Kheper nigroaeneus changed its roll bearing in accordance with such an experimental displacement of the sun (Figure 3b, graph 1). A comparable response to this manipulation has also been documented for three other members of the Scarabaeini; K. lamarcki [21,23], Scarabaeus ambiguus (Boheman) [30] and Pachysoma femoralis Kirby [24], suggesting that the orientation system of these species attributes the greatest relative weight to the directional information provided by the sun.
Consistent with this observation, K. nigroaeneus did not respond to a 90 • turn of an artificial, highly polarised pattern of polarised light, when presented from above in full view of the unmanipulated sun (Figure 3c, graph 1). However, as soon the sun was hidden from view, the beetles showed a clear 90 • turn in response to the rotated polariser (Figure 3d,  graph 1). It appears that once the sun is absent, which also naturally happens when it is obscured by a passing cloud, the distribution of the relative weight between the directional cues that remained can shift in favour of the polarised light input (Figure 3d, graph 1). The same holds true also for the close relative, K. lamarcki [28], where a behavioural response to the directional input from the gradients of colour and intensity can be seen when presented in isolation [28,29].

Sisyphus fasciculatus Attributes Greatest Relative Weight to the Directional Information Provided by the Celestial Polarisation Pattern
Neither Garreta unicolor nor S. fasciculatus changed their bearings according to the displacement of the sun (Figure 3b, graph 2, 3), indicating that the relative weight attributed to this directional cue in their orientation system is somewhat lower. This supports the results of a recent study [26], where, in contrast to Kheper nigroaeneus, S. fasciculatus turns in accordance with the turn of the polariser under a natural sky. Together, these findings clearly demonstrate that the smaller S. fasciculatus attributes the greatest relative weight to the directional information provided by the (artificial) linear pattern of polarised light. This sky-wide celestial cue is also known to play a significant role in the orientation system of other insects (locusts [51], honeybees [52] and bull ants [32]), and in some cases, even plays a dominant role (nocturnal dung beetles [21,53,54], flies [55] and desert ants [56]).

A Different Weighting of Directional Reference Cues in Garreta Species
To our surprise, G. unicolor kept to its original direction of travel both in the presence of a mirrored sun (Figure 3b, graph 2), and under a turned polariser (Figure 3c, graph 2). Only when these two cues were rotated in combination did this species demonstrate a behavioural response, which was an angular change in bearing which appeared to be randomly distributed within the population (Figure 3e, graph 2). A similar response could be confirmed in its congeneric, G. nitens (), suggesting that this is a tribe-specific orientation strategy.
Due to experimental constraints, this combined manipulation of directional information from the sun and the over-head pattern of polarisation was achieved by a 180 • shift in the apparent position of the sun in combination with a 90 • rotation of the polarisation pattern. The outcome of these manipulations was that the position of the two cues were not only changed in relation to the unmanipulated gradients of intensity and colour that spans the sky, but also in relation to each other. This drastic and multi-angular change in directional input could potentially cause the beetles to simply re-set their roll bearings, effectively contributing to the random changes in bearings displayed by G. unicolor (Figure 3e). However, this conjecture can be refuted, as the beetles faithfully returned to their initial bearings as soon as the cues were returned to their initial positions (Figure 3e, grey dotted vector, graph 2). Additionally, when tested under the same multi-conflict paradigm, K. nigroaeneus showed a clear and directed response. Attributing the greatest relative weight to the sun, these beetles simply continued to follow the angular displacement of this cue also under this experimental condition (compare Figure 3b,e). One possibility is of course that in our experiments with G. unicolor, each beetle followed an individual strategy; some turned 180 • according to the sun, some 90 • according to the overhead polarisation and some followed the stable gradients of intensity and colour. This is, however, unlikely, as we would then have expected to see a different and much more varied response when these cues were manipulated on their own (see Figure 3b,c). The random spread of changes in bearings observed for the Garreta sp. rather points to a more even weighting of directional information where the combined directional information in this artificial cue conflict experiment, results in a weak directional signal. While the beetles were still able to exit from the centre of the arena along straight paths, small, individual differences in the weighting of cues could now be seen in large differences in angular change. It would have been interesting to evaluate this theory further by testing the same beetle repeatedly before rotating all cues back to their initial positions again, but this was unfortunately not within the scope of this study.
While our results do not reveal the precise nature of the orientation strategy of the Garreta species, we can still conclude that the heading direction networks of our three test species process the directional information provided by the sky somewhat differently; K. nigroaeneus preferentially steers according to the sun, S. fasciculatus with the pattern of polarised light, and G. unicolor (and G. nitens) does not attribute a greater relative weight to either of these cues.

Compass Cue Integration and Its Relation to the Visual Ecology of Ball-Rolling Dung Beetles
Given that a navigator can reliably perceive and analyse directional information provided by the sun and its pattern of polarised light, neither of these cues should be inherently more reliable for orientation than the other. We previously showed that ballrolling beetles that attribute the greatest weight to directional information provided by the sun, are equally well directed in its absence when an alternative cue is available [23,28]. This holds true also for ants [56], monarch butterflies [9] and fruit flies [55]. It is further important to note, that the three tribes of dung beetles tested in this study-each attributing a different relative weight to the sun and the celestial polarised light pattern-all orient with the same precision under a clear, open sky (Figure 2b, grey mean vector (control)). Taken together, this indicates that the directional information provided by the sun or polarised light in the photon-rich African sky can (i) support orientation with the same precision, and (ii) be processed with comparable accuracy by the visual system and heading direction network of the Scarabaeini beetles. This is most likely also the case for the Gymnopleurini (species G. unicolor and G. nitens), that do not seem to employ differential weighting to any of the celestial cues tested.
Even though our test species are active within the same bioregion, K. nigroaeneus and G. unicolor were found actively foraging in the open habitat (Figure 1a), while the smaller S. fasciculatus rather foraged for dung within the closed habitat (Figure 1b). In this habitat, with tall grass and a high density of trees, the sun will be frequently obstructed from view, while a wide-field cue, such as the celestial polarised light pattern, will remain visible through any overhead vegetation [45,57,58]. This is also the cue attributed the highest directional weight in the orientation system of S. fasciculatus [26]. While it would have been preferable to explore the orientation strategy of additional species within the tribe Sisyphini from a different visual habitat, this unfortunately proved impossible as the species available to us (Sisyphus manni [59] and Sisyphus seminulum [60]) are so small (pronotum width: 3-5 mm) and timid, that not even the most experienced beetle experimentalist could coerce them into performing in our experiments. Still, our limited results from this tribe again suggest that they afford the greatest weight to the most consistent source of celestial directional information in their cluttered habitat [7], a strategy also found in ants [18,19].
As day turns into night, the visual world changes drastically, most notably in the decrease in light intensity [61]. At this time of the day, visually driven orientation systems need to capture as much light as possible. One common way to meet this challenge is by an increase in eye size [62,63], but it is also interesting to note that some neurons within the heading direction network of the desert locust have a higher absolute sensitivity to polarised than to unpolarised light [64]. The larger eyes of the nocturnal, open habitat Scarabaeini beetles (Scarabaeus satyrus and S. zambesianus), possess a large dorsal rim area (DRA) (the polarisation sensitive region known to detect polarised light in insects [65][66][67]) and rely on directional information from the polarised skylight above that provided by the moon itself [25,53,54]. In contrast, the diurnal K. lamarcki, which is active in the same habitat, only possess a single dorsal row of polarisation-sensitive ommatidia (Dacke unpublished data). The orientation systems of beetles active under more challenging light conditions-in the dark or under vegetation canopies-thus seem well adapted to their respective visual ecologies.
It is interesting to note that differences in external eye morphology between the ball-rolling beetles are more pronounced between the three tribes, than within the tribes themselves [68] (Figure 2). As representatives of their respective tribes ( [69][70][71]), S. fasciculatus and G. unicolor possess a more oval-shaped dorsal eye compared to that of K. nigroaeneus, where the dorsal eye of the medium-sized G. unicolor is proportionally smaller than that of K. nigroaeneus, while the small and spindly S. fasciculatus has the smallest dorsal eye of the three (Figure 2b). If we would assume that the small and narrow dorsal eyes of the smaller species also have a smaller visual field [72][73][74][75][76], the heading direction networks of the narrow-eyed Sisyphini and Gymnopleurini could possibly benefit from a sky-wide orientation signal, such as the celestial polarisation pattern, rather than using the position of a single light source. These inter-tribal differences might be an additional influence on how species within each tribe weight the sources of directional information they can reliably use.
In conclusion, despite exhibiting different body size, eye size and morphology, dung beetles nevertheless manage to solve the challenge of straight-line orientation by weighting visual cues that are particular to the habitat in which they are found. This system is however dynamic, allowing them to operate equally well even in the absence of the cue given the greatest relative weight by the particular species.