Landmark navigation in a mantis shrimp

Abstract

Mantis shrimp commonly occupy burrows in shallow, tropical waters. These habitats are often structurally complex where many potential landmarks are available. Mantis shrimp of the species Neogonodactylus oerstedii return to their burrows between foraging excursions using path integration, a vector-based navigational strategy that is prone to accumulated error. Here, we show that N. oerstedii can navigate using landmarks in parallel with their path integration system, correcting for positional uncertainty generated when navigating using solely path integration. We also report that when the path integration and landmark navigation systems are placed in conflict, N. oerstedii will orientate using either system or even switch systems enroute. How they make the decision to trust one navigational system over another is unclear. These findings add to our understanding of the refined navigational toolkit N. oerstedii relies upon to efficiently navigate back to its burrow, complementing its robust, yet error prone, path integration system with landmark guidance.

1. Introduction

Stomatopods, better known as mantis shrimp, are benthic crustaceans renowned for their ballistic strikes and complex visual systems. As adults, most mantis shrimp species reside in shallow tropical marine waters, environments that are often structurally varied and therefore contain many potential visual landmarks [1]. In these environments, mantis shrimp typically occupy small holes or crevices for use as burrows, where they reside concealed for most of the day. During foraging, many stomatopod species leave the safety of their burrows for extended excursions, where they become vulnerable to predation [2–5]. Returning to the burrow efficiently is critical to minimize predation risk and to also reduce the chance that the vacated burrow will be claimed by another animal.

Mantis shrimp of the species Neogonodactylus oerstedii employ path integration to efficiently navigate back to their burrows between foraging bouts [5]. During path integration, an animal monitors the distances it travels in various directions from a reference point (usually home) using a biological compass and odometer. From this information, a home vector (the most direct path back to the reference point) is continuously calculated, allowing the animal to return to its original location [6–8]. As animals update their home vectors during excursions, small errors in odometric and orientation measurements are made. Over the course of an animal's travel, these small errors accumulate in its path integrator. Therefore, with longer outward paths, increased positional errors of home vectors are expected [7,9]. As theory suggests, path integration in N. oerstedii is prone to this accumulated error [10]. To overcome this error, many path-integrating animals use landmarks to accurately pinpoint their goal [9,11–14]. We hypothesized that in addition to path integration, N. oerstedii uses landmarks when available during navigation. The benthic habitats N. oerstedii occupy are structurally complex with an abundance of sponges, coral, rock and seagrass to serve as potential visual landmarks (figure 1). Using landmarks during navigation would allow N. oerstedii to accurately pinpoint the location of its burrow, irrespective of error accumulated during path integration. While some path-integrating animals do not appear to use landmarks while homing (like fiddler crabs [9,15]), many others use the panoramic image of distant landmarks in their environment for homeward navigation (such as bees and ants [12,13,16–19]) or for orientation behaviours (as soldier crabs do [20]). Owing to extreme scattering of light in the water column, an extended landmark panorama is an often absent underwater, likely influencing how landmarks are used for underwater navigation.

Figure 1.

Neogonodactylus oerstedii inhabits shallow waters that offer an abundance of potential landmarks. Burrows are indicated by orange arrows. Note the abundance of environmental features, including marine vegetation, sponges, coral fragments and rock rubble, available in the scenes. (Online version in colour.)

2. Material and methods

(a). Animal care

Individual N. oerstedii collected in the Florida Keys, USA were shipped to the University of Maryland, Baltimore County (UMBC). Animals were housed individually in 30 parts per thousand (ppt) sea water at room temperature under a 12 : 12 light : dark cycle. Animals were fed whiteleg shrimp, Litopenaeus vannamei, once per week. Data were collected from 13 individuals (five male and eight female). All individuals were between 30 and 50 mm long from the rostrum to the tip of the telson.

(b). Experimental apparatuses

Four relatively featureless, circular navigation arenas were constructed from 1.5 m diameter plastic wading pools that were filled with pool filter sand and artificial seawater (30 ppt; figure 2a). Arenas were placed in a glass-roofed greenhouse on the UMBC campus (same as reported in Patel & Cronin [5]). The spectral transmittance of light through the greenhouse glass was nearly constant for all wavelengths, excluding the deep-ultraviolet-wavelength range [5]. Celestial polarization information was transmitted through the glass roof of the greenhouse [5]. Vertical burrows created from 2 cm outer diameter PVC pipes were buried in the sand 30 cm from the periphery of the arena so that they were hidden from view when experimental animals were foraging. Vertical 2 cm diameter, 8 cm high PVC columns with alternating 1 cm thick black and white horizontal stripes were placed adjacent to the burrows to function as removable landmarks. Stripe cycle widths of the landmarks were approximately twice the visual resolving limit of Gonodactylus chiragra (0.8 cycles degree⁻¹ [21]), a closely related mantis shrimp that can be slightly larger than N. oerstedii, when viewed from the food location in the arena (a distance of 70 cm). Trials were recorded from above using C1 Security Cameras (Foscam Digital Technologies LLC) mounted to tripods placed above the arenas. During landmark displacement experiments, a thin 11 × 82 cm acrylic track with a movable platform was placed adjacent to the burrow (figure 2b). A landmark identical to the one used in trials in which the landmark was fixed in place, was mounted to the movable platform.

Figure 2.

Neogonodactylus oerstedii uses a landmark to navigate back to its burrow while foraging. (a) Arena design. A vertical burrow was set into the base of the arena so it was invisible at range (empty circle). A landmark was placed adjacent to the burrow during some experiments (gold-filled star). Food was placed in one of two locations near the centre of the arena (filled circles). (b) Landmark displacement experimental design. Homeward paths were observed when a landmark adjacent to the burrow was displaced to a new location in the arena (gold arrow), while experimental individuals were away foraging. (c) Examples of foraging paths from and to the burrow during the three experimental conditions. Blue lines represent outward paths from the burrow while red lines represent homeward paths before search behaviours were initiated. Grey lines represent homeward paths after search behaviours were initiated. Empty and filled circles represent the location of the burrow and food, respectively. Gold-filled stars represent the location of the landmark. Arrows represent paths of landmark displacements. (d) Data from all homeward paths. Lines and filled circles represent the same as in (c). The grey rectangle represents the track along which the landmark was displaced. The gold rectangle marks the range of locations to which the landmark was displaced during landmark displacement trials. The black tracing in the ‘landmark displaced' group marks the homeward path of an individual on its second run which, after orientating its initial homeward path towards the displaced landmark (in red), it turned back (marked by an asterisk, *) returning to the food location and orientated towards the burrow (in black). (e) Orientations of homeward paths at one-third the beeline distance from the location of the food to the burrow (initial orientations). Each point along the circumference of the circular plot represents the orientation of the homeward path of one individual with respect to the position of the burrow (empty triangle). Grey arcs in the ‘landmark displaced' orientation plots represent the range of the directions of the displaced landmark from at the location of the food. Arrows in each plot represent mean vectors, where arrow angles represent vector angles and arrow lengths represents the strength of orientation $(\overline{R})$. Dashed lines represent 95% confidence intervals. Asterisks (*) denote a significant difference between groups (p < 0.05). ‘Landmark absent' data were obtained from Patel & Cronin [5]. (f) Homeward path orientations of groups same as in (e) measured immediately before search behaviours were initiated (terminal home vector orientations). (g) Straightness of homeward paths from the location of food to the burrow. Larger path straightness values indicate straighter paths with a value of one being a completely straight path from the food location to the burrow (a beeline path). Bars represent medians, boxes indicate lower and upper quartiles, and whiskers show sample minima and maxima. Asterisks indicate significant differences in path straightness between groups (p ≤ 0.05; landmark absent: n = 13, landmark fixed: n = 13, landmark displaced: n = 10).

(i). Experimental procedures

Individual N. oerstedii were placed in each arena and were allowed to familiarize themselves to the arena for 24 h. During familiarization, the striped landmark was placed adjacent to the burrow, marking it during the animals' initial explorations of the arena.

After familiarization, the landmark was either removed for trials in which the landmark was absent, or left in place for trials in which the landmark was present. Empty Margarites sp. snail shells stuffed with pieces of food (whiteleg shrimp) were placed at one of two locations 50 cm from the periphery of the burrow. Each animal was allowed three successful foraging excursions (i.e. food placed in the arena was found) before foraging paths were used for analyses. If an individual did not successfully locate food within one week in the arena, it was replaced with a new individual.

During landmark displacement experiments, the landmark was carefully displaced along the track to a new location in the arena by the pulling of a thin fishing line tethered to the platform when animals were foraging away from their burrows. The distance from the food location to the landmark remained relatively constant while the landmark was displaced.

Experiments reported in Patel & Cronin [5], in which animals were displaced in identical arenas to those used in the current study, demonstrated that N. oerstedii do not rely on odour cues to find their burrows (at least in the navigation arenas in which behaviours were tested). Therefore, odour cues were not controlled for in the current study.

(c). Data and statistical analyses

Foraging paths to food locations and from them to the burrow were video recorded from above. In order to differentiate homeward paths from continued arena exploration, paths from the food locations were considered to be homeward paths when they did not deviate more than 90° from their initial trajectories for at least one-third of the beeline distance (the length of the straightest path) from the food location to the burrow. From these homeward paths, search behaviours were determined to be initiated when an animal turned more than 90° from its initial trajectory.

Paths were traced at a sampling interval of 0.2 s using the MTrackJ plugin [22] in ImageJ v. 1.49 (Broken Symmetry Software), from which the output is given as Cartesian coordinates. From these data, the inbound and outbound path lengths, beeline distances from food to burrow, and inbound and outbound indices of path straightness were calculated, where

$$\begin{array}{rl}\text{path straightness}& =\hspace{0.17em}\frac{\text{beeline distance}}{\text{path length}}\hspace{0.17em}\\ & =\hspace{0.17em}\frac{\sqrt{{(x_n-x_1)}^2+{(y_n-y_1)}^2}}{{\sum }_{k=1}^n(\sqrt{{(x_{k+1}-x_k)}^2+{(y_{k+1}-y_k)}^2})},\end{array}$$

n = the last coordinate of the path.

Additionally, the orientations of homeward paths when animals were both at one-third of the beeline distance from the food source to the burrow (at which point the orientation of the home vector was usually observed) and at the end of the home vector (when search behaviours were initiated) were recorded using ImageJ. The orientations of animals after they had initiated their searches were also recorded for experiments when the landmark had been displaced. These measurements were made in respect to the burrow's location when animals were approximately 10 cm from the point where they initiated their searches.

We also measured the orientations of the body axes of all animals in respect to the landmark while it was displaced. These body axis orientations were sampled at a rate of 0.2 s. From these body axis orientations, a mean body axis orientation was calculated for each individual.

Data from the ‘landmark absent' group in this study were taken from the ‘not manipulated' trials of the greenhouse experiments published in Patel & Cronin [5].

All statistical analyses were run on R (v. 3.3.1, R Core Development Team 2016) with the ‘CircStats', ‘circular', ‘Hmisc’ and ‘boot' plugins. Orientation data were analysed using the following procedures for circular statistics [23]. All reported mean values for orientation data are circular means. All circular 95% confidence intervals were calculated by bootstrapping with replacement over 1000 iterations.

As reported in Patel & Cronin [5], no significant difference was observed between homeward orientations of males and females during experiments in the absence of a landmark (p > 0.5), so data from both sexes were pooled for all experiments.

Rayleigh tests of uniformity were used to determine if homeward paths were orientated within a group for all trials. Parametric Watson–Williams tests for homogeneity of means were used to determine if those group orientations were significantly different from one another. The orientations of groups which did not fit the assumptions of the Watson–Williams test were instead compared using the non-parametric Watson's two sample test of homogeneity. These tests were also used to compare differences between initial homeward path orientations (orientations at one-third the beeline distance from the food to the burrow) and final homeward path orientations (orientations at the initiation of search behaviours) for each group.

Homeward path lengths of trials in which the landmark was fixed in place were compared to those in which the landmark was absent using a paired t-test. A paired Wilcoxon signed-rank test was used to compare homeward path lengths of trials in which the landmark was fixed to those in which the landmark was displaced.

Pearson's correlation tests were used for all correlative analyses.

Holm–Bonferroni multiple testing corrections were used for all tests when applicable.

3. Results and discussion

(a). Neogonodactylus oerstedii uses landmarks during navigation

We placed N. oerstedii individuals in relatively featureless circular arenas filled with sand and sea water in a glass-roofed greenhouse. Vertical burrows were buried in the sand so that they were hidden from view when experimental animals were away. Snail shells stuffed with small pieces of shrimp were placed at one of two fixed locations approximately 70 cm from the location of the burrow in the arena (figure 2a). Foraging paths to and from the location of the food were video recorded from above.

As described by Patel & Cronin [5,10], we observed that animals would make tortuous paths away from the burrow until they located the food placed in the arena. After animals located the food, they would usually execute a fairly direct home vector towards the burrow. Animals would locomote by both walking and swimming. If the burrow was not found using the home vector, animals would initiate a stereotyped search behaviour (figure 2c and electronic supplementary material, video S1).

To determine if N. oerstedii use landmarks during homeward navigation when available, a vertical cylinder with alternating horizontal black and white stripes was placed adjacent to the burrow to serve as a landmark. Stripe cycles of the landmark would appear to span approximately 0.8 cycles degree⁻¹ at the location of the food, approximately twice the visual resolving limit of G. chiragra [21], a closely related mantis shrimp that can be slightly larger than N. oerstedii. Trials with the landmark present were compared to the results of previous experiments in which the landmark was absent [5].

Return trips in the presence of the landmark were more direct than trips in the landmark's absence (p < 0.05; figures 2c,d and 3, and the electronic supplementary material, videos S1 and S2), supporting the hypothesis that N. oerstedii uses landmarks during navigation. This was primarily owing to the virtual elimination of stereotyped search behaviours at the ends of homeward paths in the presence of the landmark. Instead, short-directed searches for the burrow around the landmark were observed. Return trips were initially orientated similarly between the two groups (groups were orientated: p < 0.001 for both groups; orientations were not significantly different between groups: p > 0.5; all statistical outcomes are presented in the electronic supplementary material, table S1). However, during trials in the presence of the landmark, individuals appeared to correct for their initial homeward error over the course of the homeward path (p < 0.05), in contrast with what we observed in the absence of the landmark (p > 0.5; figure 2d–f). These results indicate that in the presence of a landmark, N. oerstedii uses both path integration and landmark navigation to navigate back to its burrow.

Figure 3.

Mantis shrimp orientated towards the location of their burrows after deviating from their home vectors when the landmark was displaced. (a) Initial home vector orientations (Φ₁) were measured at one-third the beeline distance from the location of the food to the burrow while the orientations of homeward paths following home vectors (Φ₂) were measured approximately 10 cm from the point where the home vector was terminated. (b) Initial search path orientations when the landmark was displaced were significantly orientated towards the burrow (p < 0.001; $\overline{R}$ = 0.87, n = 9). One individual immediately found its burrow following its home vector and was therefore not included in the analysis. (Online version in colour.)

(b). Mantis shrimp exhibit varied homeward paths when landmark navigation and path integration are placed in conflict

In the light of the above results, we were interested in the confidence N. oerstedii places in its landmark navigation system when it is in conflict with its path integrator. In order to create this situation, homeward paths were observed when a landmark adjacent to the burrow was displaced to a new location in the arena while experimental individuals were away foraging. The landmark remained at roughly the same distance from the food location both before and after displacement. If N. oerstedii navigates using landmarks and trusts a landmark's location over the location designated by its path integrator when homing, animals should orientate towards the displaced landmark rather than the burrow's location (figure 2b).

Homeward paths were less direct (p < 0.05; figure 2g) and were differently orientated (p < 0.05; figure 2d–f) when landmarks were displaced compared to when they were left in place, further supporting the hypothesis that N. oerstedii navigate using landmarks. Some individuals orientated towards the displaced landmark while others ignored the displaced landmark, orientating towards the burrow (figure 2c and electronic supplementary material, videos S3 and S4). Many individuals initially orientated towards the displaced landmark, but broke away from their initial trajectories during their homeward paths, orientating towards the burrow instead (figures 2d and 3; p < 0.001). One individual initially orientated its homeward path towards the landmark, only to turn around and return to the food location before adopting a revised homeward path orientated towards the burrow (figure 2d). These observations suggest that the path integrator of N. oerstedii continually updates its home vector throughout foraging excursions, even during homeward travel.

The varied results observed during the landmark displacement experiments demonstrate that N. oerstedii must make decisions when the navigational strategies it relies on are in conflict. What governs this decision making? Owing to errors inherent in path integration, home vector errors grow with increased outward path lengths [10]. When the landmark was displaced, individuals may have evaluated this accumulated error during foraging, choosing to trust the position of the landmark when the accumulated error of the path integrator was high (i.e. confidence in the path integrator was low). However, we found that the orientations of homeward paths during landmark displacement experiments were not significantly correlated with the outward path lengths from the burrow to the food location (p = 0.16; electronic supplementary material, figure S1A). On the other hand, the effect size of this relationship was fairly strong (R = −0.48), so the hypothesis should not be completely discounted.

Cataglyphid desert ants are model terrestrial species for studying navigation using path integration and visual landmarks. In experiments with these ants, when their path integrators are placed in conflict with the surrounding landmark panorama, displaced desert ants will orientate either more towards the location indicated by their path integrator or towards a local landmark array depending on their distance from their nest, not on the error accumulated in their path integrators [24,25]. Similarly, N. oerstedii may weigh the contributions of path integration and landmark guidance during homeward travel depending on the length of its home vector. While lengths of home vectors in our experiments were similar (about 0.7 mm), in nature N. oerstedii may use home vector length to weigh its confidence in path integration versus landmark guidance when returning home.

Alternatively, during experiments when the landmark was displaced, the deviation between the home vector and the landmark's perceived position may have been at a preference threshold for either of the two navigation systems. For example, if the landmark was displaced further from the burrow, the majority of animals may have trusted their home vector, while if the landmark was moved less far from the burrow, the animals may have been more likely to trust the landmark's position. However, when homeward path orientations during landmark displacement experiments were compared to the distances of landmark displacement, no correlation was observed (p = 0.92, R = −0.04; electronic supplementary material, figure S1B). This suggests that the degree of landmark displacement did not influence the decision to orientate towards the home vector or the displaced landmark during these trials.

It was also possible that animals which may have observed the landmark's displacement were more likely to disregard its location than those that may not have noticed the displacement. To investigate this hypothesis, we measured the orientations of all animals' body axes with respect to the landmark while it was displaced. We compared the means of these body axis orientations to the orientations of homeward paths and found no correlation (p = 0.604, R = 0.19; electronic supplementary material, figure S1C). Thus, either the animal did not notice the landmark's displacement, or observing its displacement did not influence an animal's decision to favour using the displaced landmark's position or using its home vector.

4. Conclusion

Our results demonstrate that N. oerstedii uses landmark navigation together with path integration while navigating back to its burrow while foraging. Landmarks are reliable references which can be used to correct for error accumulated by path integration; this is especially important during path integration when using idiothetic orientation cues, which N. oerstedii uses when allothetic cues become unreliable [5].

Landmarks were used in a very basic situation during our experiments— as a beacon to home towards. Many other questions about how landmarks may be used by mantis shrimp arise from this work: do mantis shrimp learn to recognize landmarks encountered during foraging routes, exhibiting ‘trapline foraging'? Can mantis shrimp estimate the relative position of a goal to multiple landmarks? Do stomatopods use a snapshot mechanism like that employed by some insects to learn landmark arrays [13,16]? Stomatopod visual systems incorporate several types of eye movements, often with each eye moving independently of one another [26,27]. The extreme mobility of stomatopod eyes complicates spatial orientation, especially if stomatopods use a snapshot-based mechanism to locate a goal in relation to landmarks. Further, contrary to terrestrial environments, a distinct, distant landmark panorama would often not be available to an underwater animal owing to veiling light in the water column. While insects are largely thought to use a panoramic image matching mechanism to locate goals in relation to landmarks [12,13,16–19], the absence of a reliable submarine landmark panorama may influence how mantis shrimp recognize and interact with landmarks in their environment. Alternatively, might stomatopods possess cognitive maps akin to those thought to exist in mammals [28]? Finally, mantis shrimp are famed for possessing complex colour vision, linear polarization vision in two spectral channels and circular polarization vision [29]. Besides spatial vision alone, do stomatopods use these visual channels to identify landmarks? If so, how?

Mantis shrimp occupy a wide variety of marine habitats and depths, from structurally complex reefs to nearly featureless mud flats. Stomatopod species that occupy landmark-rich environments may weigh the importance of landmarks more heavily during navigation than stomatopods which occupy benthic environments relatively void of landmarks. Also, because visual information rapidly attenuates with distance underwater owing to the extreme scattering of light in water, the relative importance of landmark navigation over path integration may differ for mantis shrimp species occupying waters of different depths and turbidities.

Taken together with our previous work on mantis shrimp navigation [5,10], this work offers an opportunity to study the neural basis of navigation, learning, memory and decision making in stomatopods. Mushroom bodies, centres for arthropod learning and memory, are thought to play a prominent role in landmark learning in insects [30–33]. Prominent hemiellipsoid bodies, brain regions thought to be similar in function to insect mushroom bodies, exist in stomatopod eyestalks [34]. As in insects, these neuropils may be crucial for navigation and landmark learning in mantis shrimp. A separate brain region, the central complex, plays a role in landmark orientation in Drosophila melanogaster. Here, orientation to landmarks is neurally based in the ellipsoid body of the central complex [35]. Stomatopods themselves possess a highly developed central complex composed of a collection of neuropils anatomically very similar to those found in insects [36]. Investigation of the function of stomatopod brain regions in light of our work may have implications for the evolutionary origins of navigational strategies and the neural architecture of the brain within the ancient Pancrustacean clade, a taxon which includes all insects and crustaceans [37], as well as in other arthropods.

In summary, N. oerstedii possesses a robust navigational toolkit on which it relies to efficiently navigate back to its burrow. First, N. oerstedii relies on path integration using multiple redundant compass cues to navigate back to its home [5]. If path integration does not lead N. oerstedii directly to its burrow, it relies on a stereotyped search behaviour which is scaled to the amount of error it accumulates during its outbound foraging path to locate its nearby lost target [10]. Finally, the stomatopod will use landmarks, if available, to quickly pinpoint its target, correcting for positional uncertainty accumulated during path integration.

Data accessibility

The data that support the findings of this study are available on Mendeley Data at: http://dx.doi.org/10.17632/wfngvp2fd7.1.

Authors' contributions

R.N.P. designed and conducted all research, analysed all data and prepared the manuscript. T.W.C. provided guidance and research support.

Competing interests

We declare we have no competing interests.

Funding

This work was supported by grants from the Air Force Office of Scientific Research under grant no. FA9550-18-1-0278 and the University of Maryland Baltimore County (SR18CRON).