Background matching can reduce responsiveness of jumping spiders to stimuli in motion

ABSTRACT Motion and camouflage were previously considered to be mutually exclusive, as sudden movements can be easily detected. Background matching, for instance, is a well-known, effective camouflage strategy where the colour and pattern of a stationary animal match its surrounding background. However, background matching may lose its efficacy when the animal moves, as the boundaries of the animal become more defined against its background. Recent evidence shows otherwise, as camouflaged objects can be less detectable than uncamouflaged objects even while in motion. Here, we explored whether the detectability of computer-generated stimuli varies with the speed of motion, background (matching and unmatching) and size of stimuli in six species of jumping spiders (Araneae: Salticidae). Our results showed that, in general, the responsiveness of all six salticid species tested decreased with increasing stimulus speed regardless of whether the stimuli were conspicuous or camouflaged. Importantly, salticid responses to camouflaged stimuli were significantly lower compared with those to conspicuous stimuli. There were significant differences in motion detectability across species when the stimuli were conspicuous, suggesting differences in visual acuity in closely related species of jumping spiders. Furthermore, small stimuli elicited significantly lower responses than large stimuli across species and speeds. Our results thus suggest that background matching is effective even when stimuli are in motion, reducing the detectability of moving stimuli.

Recent evidence shows otherwise, as camouflaged objects can be less detectable than 23 uncamouflaged objects even while in motion. Here, we explored if the detectability of 24 computer-generated stimuli varies with the speed of motion, background (matching and 25 unmatching) and size of stimuli in six species of jumping spiders (Araneae: Salticidae). Our 26 results showed that in general, the responsiveness of all six salticid species tested decreased 27 with increasing stimulus speed regardless of whether the stimuli were conspicuousness or 28 camouflaged. Importantly, salticid responses to camouflaged stimuli were significantly lower 29 compared to conspicuous stimuli. There were significant differences in motion detectability 30 across species when the stimuli were conspicuous, suggesting differences in visual acuity in 31 closely related species of jumping spiders. Furthermore, small stimuli elicited significantly 32 lower responses than large stimuli across species and speeds. Our results thus suggest that 33 background matching is effective even when stimuli are in motion, reducing the detectability 34 of moving stimuli.  Tree trunks were photographed at breast height (~1.5 m), at a 0.5 m distance using a tripod-148 mounted Nikon D800 digital SLR camera (Nikon Corp., Tokyo, Japan). The aperture, shutter 149 speed and ISO were kept at f16, 1/60 and 100 respectively. A total of 52 rubber tree trunks 150 were imaged at Pulau Ubin, Singapore. Images were stored in RAW format to avoid loss of 151 information due to compression (Stevens et al., 2007). We used the software ImageJ v. In both the visual responsiveness and background matching assays, the stimulus 166 comprised of a rectangle moving horizontally across the screen. In the visual responsiveness 167 assay, we used a black rectangle stimulus (< 1 cd m -2 ; Fig. 2B small stimuli are comparable to the body lengths of a salticid and its potential prey, 172 respectively. In the background matching assay, we used a background matching stimulus 173 presented against the tree bark background (Fig. 2C). The background matching stimulus was 174 extracted from the middle section of the background and the dimensions were the same as 175 that used for the large stimulus in the visual responsiveness assay. 176 Before each trial, we placed the test salticid in a cylindrical, viewing chamber 177 positioned on an elevated platform 7 cm from the centre of a tablet screen (iPad 8 th 178 generation, Apple Inc, California, US), where the stimulus was displayed ( Fig. 2A). Although 179 salticids have been reported to discriminate prey from distances of up to 32 cm, some species, 180 such as Menemerus sp., have a maximum discrimination distance of up to 12 cm (Harland et 181 al., 1999). Thus, we deemed 7 cm to be an acceptable distance for our assays. The chamber 182 consisted of a white base (diameter: 1.5 cm), a thin, transparent wall made of clear acetate 183 (thickness: 0.1 cm, height: 0.7 cm) and a transparent lid (adhered to the wall using white 184 tack). As Po. labiata individuals were generally larger than the rest of the species used, we 185 placed them in a larger viewing chamber (diameter: 3 cm; height: 0.7 cm). The tablet screen 186 was 20.8 × 15.6 cm, with a pixel resolution of 2160 × 1620 pixels (264 pixels per inch (ppi)) 187 and a refresh rate of 60 Hz, which was deemed to be acceptable for this experiment as The test salticid was first acclimatised in the chamber for 10 min. Next, we waved a 196 thin brush in front of the test salticid to redirect its attention to the screen or gently rotated the 197 chamber using a brush such that the salticid faced the screen. When the salticid turned 198 towards the screen, the stimulus (Fig. 2B, C) was animated to move horizontally across the 199 screen, with the direction of movement of stimuli randomised for each trial. The test ended 200 after the stimulus animation moved across the screen. The stimuli and background varied 201 depending on the assay. 202

Visual responsiveness assay 203
To determine how salticids respond to an uncamouflaged stimulus, each test salticid was 204 exposed to a black stimulus presented against the white background at different speeds. Each 205 trial consisted of eight presentations, where the stimulus was presented in two sizes -large 206 and small -at four moving speeds (duration of animation on screen: 0.25, 0.5, 1, and 1.5 s, 207 corresponding to the speeds: high (83.2 cm s -1 ), medium (41.6 cm s -1 ), low (20.8 cm s -1 ), and 208 very low (13.9 cm s -1 )). The mean luminance of the white screen was approximately 300 209 cd/m 2 , measured using an illumination meter (Topcon IM-2D, Tokyo, Japan) at the start of 210 each trial. Each presentation was separated by an inter-stimulus black screen interval of at 211 least 50 s, and the stimuli were presented in a random order. During the assay, we observed 212 that, unlike the other species tested, the behavior of Po. labiata was affected by the transition 213 from a black to white screen, as Po. labiata would exhibit freezing behavior for a long time 214 upon the screen transition. Thus, for Po. labiata, we used a white screen instead of a black 215 screen during the inter-stimulus interval to minimise potential distractions for this species. 216 Each trial was repeated four times, with at least a one-day interval between each trial. Thus, a 217 total of 32 stimulus presentations were showed to each salticid -each stimulus type (with two 218 different sizes) presented at four speeds for four times. 219

Background matching assay 220
To determine how salticids respond to a camouflaged stimulus moving at different speeds, 221 salticids were tested using the similar procedures as in the visual responsiveness assay except 222 that only the large, background matching stimuli (Fig. 2C) were presented against the tree 223 bark background in the background matching assay. Thus, only essential details are described 224 here. To ensure that the salticids were sufficiently responsive to the camouflaged stimulus, 225 only individuals with an average response rate of at least 50 % to large, black stimuli in the 226 visual responsiveness assay were included in this assay (Fig. S1). Portia labiata had very low 227 response rate (only 25 % of Po. labiata had a response rate of at least 50 %) and were thus 228 excluded from this assay. The mean luminance of the screen when the tree bark background 229 was displayed was 45 cd m -2 , measured using an illumination meter at the start of each trial. 230 Each salticid was exposed to a trial consisting of four stimuli presentations, where the 231 background matching stimulus was presented at four speeds (corresponding to the speeds 232 tested in the visual responsiveness assay) in a random order. We repeated each experimental 233 trial four times, with at least a one-day interval between each trial. Thus, a total of 16 stimuli 234 were presented to each salticid. 235

Behavioral responses 236
To determine whether the stimulus was detected by the salticids, we scored the responses of 237 the test salticids following each stimulus presentation (

Statistical analysis 240
We performed two analyses to determine -i) the effectiveness of background matching for To determine the effectiveness of background matching for moving stimuli, we 252 compared the salticids' responses across both assays and included salticid species, stimulus 253 moving speed, salticid sex, and background type as fixed effects. We proposed a total of 14 254 models, which comprised of i) a null model (1 model); ii) each fixed effect alone (4 models); 255 iii) salticid species interacting with the other fixed effects (3 models); iv) stimulus speed 256 interacting with background/stimulus type (1 model); v) full models containing all fixed 257 effects with and without interactions (5 models). A comparison of the models can be found in 258 Table A1. 259 To determine if there are variations in responses among salticid species, we examined 260 the salticids' responses in the visual responsiveness assay only. We included the same fixed 261 effects and models as above but replaced background type with stimulus size. The proposed 262 models are ranked in Table A2. Additionally, we determined if there were differences in 263 responses among species at each of the four stimulus speeds tested by using four separate 264 CLMMs to compare the responses among species at each speed in the visual responsiveness 265 assay. For each CLMM, salticid species, stimulus size and sex were included as fixed effects. 266 We then conducted pairwise comparison tests for significant effects using emmeans 267 (https://CRAN.R-project.org/package=emmeans) for each CLMM. 268

269
The effectiveness of background matching 270 The model containing species, stimulus moving speed, sex, background type, as well as the 271 interaction between species and stimulus moving speed best predicted the responses of 272 salticids between camouflaged and uncamouflaged stimuli (AICc = 3971.4, weight = 1; Table  273 A1). Salticids generally showed a significantly higher response to uncamouflaged stimuli 274 when compared to camouflaged stimuli regardless of species (Table 2; Fig. 3), indicating that 275 background matching effectively reduced the detectability of moving stimuli for salticids. 276 Salticids responded less with increasing stimulus moving speed across species, sexes, and 277 backgrounds; salticids generally responded differently among species across stimulus moving 278 speeds, sexes and backgrounds. However, females had no significant differences in 279 responsiveness compared to males across species, stimulus moving speeds and backgrounds 280 (Table 2). 281

A B
moving speed (AICc = 9872.6, weight = 1, Table A2). In general, the conspicuous stimuli 293 elicited different levels of responses among the salticids species tested (Fig. 4A). Salticids 294 responded significantly less with increasing moving speed, and small stimuli elicited 295 significantly lower responses than large stimuli (Table 3 and Fig. 4B). However, sex did not 296 significantly predict salticid responses though it is included in the best-fitting model (Table  297 3). Species interacting with stimulus moving speed significantly predicted salticid responses 298 to conspicuous, moving stimuli: salticids that responded differently to the different stimulus 299 moving speeds depended on salticid species (Table 3). 300   (Table 4). Furthermore, analyses at specific speeds showed significant differences in the 312 salticids' level of responsiveness (Fig. 4). Portia labiata was less responsive, and displayed 313 milder responses compared to the rest of the species at most stimulus moving speeds (Fig. 5). 314 Interestingly, T. bhamoensis displayed stronger responses compared to the rest of the species 315 at lower speeds (Fig. 5). At the highest speed, C. umbratica and M. bivittatus displayed 316 significantly stronger responses than T. bhamoensis (Fig. 5). No significant difference in 317 level of responsiveness was found at the medium speed (Fig. 5). 318

329
Our study demonstrates that salticids displayed lower levels of response to camouflaged than 330 uncamouflaged stimuli, suggesting that background matching may be an effective 331 camouflage strategy for moving stimuli, especially at higher speeds, in the eyes of salticids. 332 Salticids displayed higher levels of response when the stimulus was larger and moving 333 slower, indicating that the speed of moving stimuli inversely correspond to detectability. 334 Importantly, we observed significant differences in motion detectability across species when 335 the stimuli were conspicuous, thus suggesting differences in visual acuity across closely 336 related salticid species. 337 Our findings support our hypothesis that camouflaged moving stimuli, compared to 338 uncamouflaged moving stimuli, were less detectable across the salticid species tested in this 339 study. One possible reason can be that the optical performance of these salticid species tested 340 may not be well equipped in discriminating stimuli moving against a similarly patterned 341 background. Although little is known regarding the visual acuity of the tested species, our 342 results suggest that the salticids may have difficulty resolving complex textures such as the 343 bark of rubber trees. Understanding the effectiveness of background matching for camouflage 344 can provide insights to how moving animals exploit the visual constraints of the receiver to 345 escape detection and capture. This is relevant as animals need to move to forage, find mates 346 or escape from potential predators. Previous studies indicate that when a background 347 matching prey moves, its boundaries become more defined, rendering them more detectable 348 Thus, a well-camouflaged prey would be able to escape detection more easily than a 355 conspicuous one. Although our results indicate that background matching stimuli in motion 356 experience reduced detection, it remains to be shown if these stimuli evade capture more 357 successfully than non-background matching stimuli when detected. In addition, the effect of 358 different motion types (e.g., irregular bursts of speed or unpredictable motion trajectory) 359 could further improve the camouflaging effectiveness of background matching in moving 360

prey. 361
As predicted, the responses of the salticids were negatively correlated with the speed 362 of moving stimuli regardless of species, sex, and background type. Our findings are in line 363 with prior studies, which found that prey moving at higher speeds tend to avoid capture 364 (Smart et al., 2020;Stevens et al., 2008). At high speeds, the stimuli could have moved so 365 quickly that it exceeded the motion processing capabilities of the salticids, thus resulting in 366 lower response levels. Additionally, fast moving stimuli have an added advantage in avoiding 367 detection if it is moving against a similarly patterned background. Thus, in our study, 368 camouflaged stimuli elicited lower responses compared to conspicuous stimuli, especially at 369 high speeds. compared to the other species tested in our study. Variations in luminosity could have also 396 influenced stimulus detectability against different background types. In our study, the 397 textured background had lower mean luminosity compared to the more conspicuous white 398 background. Thus, it is possible that the camouflaged stimuli appear to be less conspicuous 399 than the uncamouflaged stimuli due to differences in luminosity. 400 Low levels of response may not represent low motion processing capabilities but 401 could be due to species-specific responses. Furthermore, large targets can appear to move more slowly than small targets (Brown, 1931). 418 Prey body size is shown to influence a predator's foraging decision, especially since the 419 predator needs to evaluate if the rewards obtained from prey capture outweigh the costs/ risks 420 involved (Juanes, 1992; Nakazawa et al., 2013). Oftentimes, predators must balance potential 421 risks and rewards before pursing prey -a relatively large prey may be energetically costly to 422 predators while a small prey may not yield sufficient energetic returns (Schmitz, 2017