Neural and sensory basis of homing behavior in the invasive cane toad, Rhinella marina

The behavioral, sensory, and neural bases of vertebrate navigation are primarily described in mammals and birds. However, we know much less about the navigational abilities and mechanisms of vertebrates that move on smaller scales, such as amphibians. To address this knowledge gap, we conducted an extensive field study on navigation in the cane toad, Rhinella marina. First, we performed a translocation experiment to describe how invasive toads in Hawaiʻi navigate home following displacements of up to one kilometer. Next, we tested the effect of olfactory and magnetosensory manipulations on homing, as these senses are most commonly associated with amphibian navigation. We found that neither ablation alone prevents homing, suggesting that toad navigation is multimodal. Finally, we tested the hypothesis that the medial pallium, the amphibian homolog to the hippocampus, is involved in homing. By comparing neural activity across homing and non-homing toads, we found evidence supporting the involvement of the medial pallium, lateral pallium, and septum in navigation, suggesting the conservation of neural structures supporting navigation across vertebrates. Our study lays the foundation to understand the behavioral, sensory, and neural bases of navigation in amphibians and to further characterize the evolution of behavior and neural structures in vertebrates.

Methods
This manuscript includes datasets related to cane toad (Rhinella marina) spatial movements, the effects of sensory manipulations on toad spatial movements, and brain activity during toad navigation acitvities. The following types of data are presented:

1) Toad metadata: HI 2022 Toad List.csv, TransToadInfo.csv, and Toads.csv contain information about individuals in the experiment. This includes ID, sex, size (mass and snout-vent-length), tagging date and times, sensory ablation information (whether an ablation happened, what type, when it was performed, and information about how many coordinate points post-ablation fall into the area the toad occupied pre-ablation), the toad experimental group, whether the toad homed or not, the field site where the toad was monitored, the date the toad was translocated, and the day the toad was euthanized. HI 2022 Toad List.csv is used in analysis of baseline spatial data, TransToadInfo.csv is used in analysis of trajectories post translocation, and Toads.csv is used in analysis of brain activity related to homing. 

2) Baseline spatial data (Baseline.csv): This dataset includes geographic coordinates and timestamps of toad positions collected in the wild on the island of O‘ahu in the state of Hawaiʻi. This data was gathered by using radio telemetry to track individual toads over several nights and record their positions with handheld GPS devices. Coordinates were uploaded to ArcGIS Online and ArcGIS Pro for visualization and to convert coordinates to UTM projection, and then processed in R Studio (v 2023.09.1+494, Posit Software, PBC, Sunnyvale, CA) running R (v 4.3.1, R Foundation for Statistical Computing, Vienna, Austria). Trajectories were generated for each toad using the package ‘adehabitatLT’ in R; "ltraj" objects generated in ‘adehabitatLT’ were used to calculate spatial and temporal attributes, including cumulative movement during baseline tracking, daily movement during this period, and total movement range.

3) Ablation related baseline spatial data (Ablation_Baseline.csv): This dataset includes the baseline coordinates and timestamps found in Baseline.csv, but only for toads involved in olfaction and magnetroeception manipulation experiments. There is a column that specifies whether the baseline point was recorded pre-ablation or post-ablation. Separate trajectories were made for pre- and post- periods using the package ‘adehabitatLT’ in R, and baseline metrics were compared to see if ablations or sham manipulations disrupted space baseline space use. 

4) Coordinates from translocation experiment (Translocation_Coordinates.csv): This dataset includes coordinates and timestamps of tods during translocation experiments where toads were moved ~500 m or ~1000 m away from their outermost baseline coordinate point and tracked for three days to observe whether they would return home. Toads were tracked until they moved within 200 meters of their nearest baseline point or 3 days had passed post-translocation. Trajectories were generated for each toad using the package ‘adehabitatLT’ in R; "ltraj" objects generated in ‘adehabitatLT’ were used to calculate cumulative movement, straightness index, homing duration/speed, and active homing duration/speed. The dataset includes a column "Grp" which specifies whether the toads were unablated (in which case "500" or "1000" is listed, corresponding to the translocation distance) or had an ablation or sham ablation (OAE for olfactory ablation, OAC for olfactory control/sham, MAE for magnetoreception ablation, and MAC for magnetoreception control/sham).

5) Angle correction for normalization of trajectories (Transect_Correction.csv): This dataset contains the coordinates of the translocation release site and the outermost  baseline reference point used to calculate translocation distance for each toad. The package ‘adehabitatLT’ in R was used to create trajectories from these points for each toad to extract the angle of travel needed to reach home, which was used to normalize trajectories for plotting. The timestamps are arbitrary time points picked to separate the two points for each toad: translocation release point is 2022-02-22 and baseline reference is 2022-05-13 or 2022-05-14). 

6) Metadata for analyzed brains (Brains.csv): This dataset contains information about brains extracted from toads, including the toad, aproximate trannslocation distane (0, 500, or 1000m), condition (either Homing, Lost, Baseline, or Homing* if brain collected at late time point after toad homed), ablation (ablation information if applicable), extraction (date that brain was collected), sodaz (0 if brain was not put in PBS with sodium azide, 1 if it was, depending on the time the brain was collected in relation to transport to the laboratory). This dataset was used to assign identifiers to count data described below.

7) Region specific pS6 count data for analysis of brain activity differences between homing conditions (BranCounts_02.csv): This dataset inclusdes the count of pS6-positive cells in six different brain regions across 9 sections of brain in every toad considered in analysis. Count data was generated by completing pS6-immunohistochemistry on cryosectioned brain sections, imaging sections at 20x magnification using brightfield microscopy, and counting positive cells in each region manually with ImageJ. Count data was used to build models in R. We used generalized linear mixed models (‘glmmTMB’ package in R) to test for differences in brain activity and its relation with behavior.  Models used a negative binomial distribution and best model fit was confirmed using the ‘DHARma’ package in R . Homing condition, brain region, and their interaction were the main effects predicting the number of pS6-positive cells. Toad ID was included as a random factor and brain region area was included as an offset variable to account for brain area size. We modeled differences in brain activity due to ablations separately using the same model parameters as above except that ablation was also included as a main effect. Post hoc pairwise contrasts (between homing condition groups and ablation treatments) were calculated with estimated marginal means using the “emmeans” package in R.

In addition, to test if brain activity was related to either straightness or duration of homing, we performed linear mixed effects models using the “lme4” package, where pS6-count was the response variable and the homing descriptor (straightness or duration), brain region, and their interaction were fixed effects. Toad ID was a random effect and brain region area was an offset variable in the model. Linear mixed effects models were also performed within count data specific to each brain region, in which the response variable was pS6 counts within a given region, the homing descriptor was the fixed effect, Toad ID was a random effect, and region area was an offset variable. 

8) Count of pS6-positive cells in the olfactory bulbs of olfaction ablated and sham ablated toads (olf_ct.csv): This dataset includes pS6 count data from olfactory bulbs of toads in olfactory ablation experiment. Three sections of olfactory bulb were considered for each animal and pS6-postive cells were counted in a 90,000 micrometer squared box on each section, in order to test whether there was an effect of olfactory ablation on the number of pS6-positive cells in olfactory bulb. The comparison was made between sham and ablated animals using a Wilcoxon Rank Sum Test. 

9) Weather data from translocation experiments (Weather.csv): This dataset includes daily weather data taken from NOAA for any day a translocation experiment was happening at a particular field site. This data includes precipitation amoutn, max temperature, and minmum temperature. These measurements were compared between field sites using Kruskal-wallis Tests and post hoc Dunn's Tests to identify differences in eather conditions between locations during tracking experiments.