bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1Title

2Look before you jump: jumping discriminate different ants by visual cues

3Sajesh Vijayan1,3, Chethana Casiker2,3 and Divya Uma3*

4

51. School of Biology, Indian Institute of Science Education and Research,

6Thiruvananthapuram, Kerala

72. Ashoka Trust for Research in Ecology and the Environment, Bangalore

83. School of Liberal Studies, Azim Premji University, Bangalore

9*Correspondence: [email protected]

10

11

12

13

14

15

16

17

18

19

20

21

22

23

1 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

24Abstract

25

26Ants, being ubiquitous, aggressive, and top predators, play a predominant role in terrestrial

27ecosystems. Jumping spiders are another prominent invertebrate predator that are present in

28similar habitats as that of ants. Most jumping spiders are thought to avoid ants, yet little is known

29if they discriminate among them. In this study we examined the response of jumping

30genus towards three different ant species (Oecophylla smaragdina, the weaver ants;

31Camponotus sericeus the golden-back carpenter ants, and Leptogenys processionalis, the

32procession ants). In a behavioral assay that excluded tactile and chemical cues, we tested if

33spiders distinguish the three ants by visual cues alone. We recorded and analysed behaviors such

34as ‘look’, ‘approach’, 'stalk', 'attack', and ‘avoidance’ by spiders towards ants. Our results show

35that the three ants differ in their color, movement and aggressive behavior. Spiders gave 'short

36looks' to live ants, suggesting movement is important in detecting ants. Furthermore, spiders

37gave significantly more ‘long looks’ to procession and golden-back ants compared to weaver

38ants. Spiders approached, stalked and attacked procession ants more compared to weaver ants.

39Numerous jumping spiders and ants overlap in their habitat, and it is advantageous to selectively

40avoid some ants over others. Our results suggests that jumping spiders can indeed distinguish

41ants that co-occur in their habitat by visual cues alone, however, the precise nature of visual cues

42warrants further studies.

43

44Running title: Look before you jump

45Key Words: Jumping spiders, Weaver ant, Procession ant, Carpenter ant, Visual cues bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

46Introduction

47Ants play a prominent role in ecological communities. They are widely distributed, and top

48invertebrate predators in tropical ecosystems (Hölldobler, & Wilson, 1990; Roslin et al., 2017).

49They compete for resources (including prey or nesting areas) with other , leading to

50non-consumptive effect on communities (Halaj et al., 1997; Ibarra-Isassi, & Oliveira,

512018). Ants also serve as partners in mutualistic interactions, and their colonies sometimes serve

52as exploitable resources. Because of their ubiquitous presence, aggressive nature and communal

53defense, ants also serve as suitable models for numerous arthropod mimics (McIver, &

54Stonedahl, 1993; Cushing, 1998, 2012)

55

56Since many invertebrate and vertebrate species have overlapping habitats with ants, and many

57even engage in various kinds of interactions with them, it may be advantageous to identify and

58distinguish among different ants. For example, coastal horned lizards selectively consume some

59ants over others and their preferences seem to be influenced by ant size (Suarez et al., 2000).

60Butterflies avoid egg laying on plants frequented by predaceous ants, and this is influenced by

61ant size and shape (Sendoya et al., 2009). Praying mantids avoid ant mimicking spiders because

62of their close resemblance to ants (Ramesh et al., 2016).

63

64Jumping spiders are another top arthropod predator in the terrestrial ecosystem, often found in

65similar habitat as that of ants. Earlier research suggests that many jumping spiders have an innate

66avoidance of ants or subsequently learn to avoid them (Edwards, & Jackson, 1994; Nelson et al.,

672006). However, it is not clear if jumping spiders can distinguish among different kinds of ants.

3 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

68Why might such a discrimination be important? Ants and jumping spiders overlap in their

69habitats, and are often intraguild predators (Okuyama, 2002; Sanders, & Platner, 2006; Sanders

70et al., 2008). Furthermore, some ants may be more aggressive than others: For example, the

71weaver ant (Oecophylla smaragdina) is territorial and highly aggressive towards intruders. In

72such cases, it makes sense for spiders to employ a ‘safety first’ approach, and stay away from

73these ants. However, other ants may not be as aggressive, and jumping spiders could afford to

74forage closer to ant colonies, and sometimes even prey on insects tended by ants (Del-Claro, &

75Oliveira, 2000). In such cases, it may be beneficial for spiders to discriminate one ant from

76another. Jumping spiders may use various subcomponents of the visual signal such as movement,

77color and/or shape to distinguish between ants. Yet, the proximate visual cues these spiders use

78to discriminate different ants are not known. Multimodal cues used by these spiders in the visual,

79chemical or vibrational domain will have important implication for ecology and evolution of ant-

80spider interactions (Clark et al., 2000; Allan, & Elgar, 2001; Nelson, & Jackson, 2006a), &, &.

81The vision of jumping spiders is unique in the world, as they have achieved a high degree

82of spatial resolution due to their ‘principal eyes’ (Harland et al., 2012), as well as a wide field of

83view due to ‘secondary eyes’ (Land, 1972; Forster, 1985; Jackson, & Pollard, 1996). Jumping

84spiders use shape, movement, and color to recognize a potential prey: for example, Yllenus

85arenarius rely on the direction of movement, and head-indicating features such as eye spots to

86strike a potential prey (Bartos, & Minias, 2016). Evarcha culicivora, which preys on female

87Anopheles mosquitoes appear to make discriminations based on relative orientation of body parts

88(Dolev, & Nelson, 2014). While movement is more essential than shape for predatory response

89in Phidippus audax (Bednarski et al., 2012) the ant-eating Habrocestum pulex

90seems to distinguish its prey even in the absence of motion cues (Li et al., 1996). Both naïve and bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

91field-caught Habronattus pyrrithrix avoided red colored crickets (Taylor et al., 2014), but they

92could also be trained to prefer them (Taylor et al., 2016), demonstrating flexibility in color

93learning during foraging. Although these studies indicate the factors that influence prey

94recognition, cues that salticids use to identify a potential enemy/threat such as an ant have not

95been explicitly examined. Visual predators such as praying mantids avoided red colored ant

96mimicking spiders more than black colored ant mimics (Ramesh et al., 2016). But, it is not

97known if salticids generalize all ants as dangerous or if they can distinguish between different

98species of ants.

99

100The objective of the present study is to examine if jumping spiders can distinguish between three

101species of ants that co-occur in their habitat. We choose three ant species of comparable sizes:

102weaver ants (Oecophylla smaragdina Fabricius 1775), carpenter ants (Camponotus sericeus

103Fabricius 1798) and procession ants (Leptogenys processionalis Jerdon 1851) that differ in color,

104movement and aggression (Table 1). We first establish aggressive behavior, and movement of

105the three ants were indeed different. We then examined if jumping spiders differentiate these ants

106using vision alone by performing behavioral assays where tactile and chemical cues were

107controlled. Finally, by controlling for ant movement, we discuss the importance of color and

108movement cues influencing spider responses.

109

110Methods

111Study species

5 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

112Subadult and adult jumping spiders were collected from in and around Bengaluru, Karnataka

113from Oct 2016 to July 2017. After capture, the spiders were individually housed in plastic boxes

114with holes for ventilation, and a moistened piece of cotton for hydration. Spiders were captured

115two days prior to the behavioral assays. For the assays spiders in the genus Plexippus (Plexippus

116petersi Karsch 1878 and Plexippus paykulli Audouin 1826) were used, as they were

117predominantly found in the same habitat as that of ants. Foragers of all three species of ants and

118houseflies (used as control) were collected on the day of the experiment. Ant and representative

119spider pictures are shown in Fig. 1. All test were maintained and released back into their

120natural habitat in compliance with the laws of the country.

121

122Ant aggression towards spiders

123Ant aggression was calculated by restricting an ant (weaver, carpenter, or procession ants n=16

124for each ants) and a spider in a 5 cm3 arena and recording the latency of the ant to bite the spider.

125Ants which took more time to bite (increased latency) were considered to be less aggressive.

126

127Response of spider towards ants

128We examined the response of spiders towards ants, where trials were carried out in an 8cm

129diameter Petri dish. The dish was modified to include an inner transparent ring made with a strip

130of transparent sheet into which the ant was released. The transparent ring neither allowed ants

131nor chemical cues to get into the outer compartment, thus allowing only visual cues to be seen

132from either side of the ring. An individual ant was introduced first, followed by a spider into the

133outer portion, upon which the trial commenced, and was video recorded using a Sony handycam bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

134(HDR-PJ600VE). At the end of 5 minutes, the trial was stopped and the test subjects were

135released back into the storage boxes. Trials were carried out from 9 am to 4 pm, well illuminated

136with a 4.5watt LED lamp. After each trial, the image of the spider was recorded using a Leica

137EZ4E microscope connected to a host computer, later used for identification. Each spider was

138presented with a weaver ant, carpenter ant or a procession ant (n = 29, n = 28 and n = 30 trials

139respectively). Using the same experimental setup, a different set of trials were carried out pairing

140a spider with a housefly to make sure spiders exhibited a typical hunting behavior (n = 25). This

141was to ensure that the set up did not hamper vision-mediated behavior of the spiders. Spiders or

142ants were used only once for all the trials.

143

144Spider responses towards ants/flies were coded as follows: a) Short look: spider orients its

145cephalothorax towards the stimulus for a short amount of time (0-7 seconds) and then leaves b)

146Long look: spider orients its cephalothorax towards the stimulus for longer than 7 seconds and

147then leaves c) Approach: spider moving towards the stimulus, d) Stalk: spider moves slowly

148towards the stimulus, while visually oriented towards it e) Attack: spider pouncing on the wall of

149the outer ring at the stimulus f) Avoid: spider moving away from the stimulus. Short and long

150looks were used by Huang et. al. (2011) to classify jumping spider behavior towards different

151prey. Short looks in our study, on average, lasted for 1.5 secs (median 2.29 secs, range 0 to 7

152secs) across all trials, and followed a beta distribution (Suppl Fig S1). Anything more than 7

153seconds was considered long look, and it ranged from 7 to 112 seconds (average 20.3 secs,

154median 14.1seconds). The proportion of time spent and the frequency of these behaviors in 5

155minutes was measured.

156

7 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

157Difference in ants' movements

158Ant movement was calculated as the amount of time the ants spent moving inside the inner

159transparent ring. The movement scores were obtained from the same trials where spider

160responses (see above) were scored. The presence of spiders outside the ring did not influence ant

161movement.

162

163Responses of spiders in the absence of ant movement

164To control for the movement of ants, in a separate experiment, we created ‘lures’ of ants where

165freshly killed ants were positioned in a life-like manner inside the transparent barrier. Except

166movement cues, color and textural cues were intact. Spider responses (short and long looks)

167towards weaver, procession and carpenter ant lures were compared.

168

169Statistical analyses

170We compared the aggression levels of three ants towards spiders, and movement of ants inside

171the ring by Dunn's-test for multiple comparisons of independent samples with Bonferroni

172correction (Pohlert, 2014). We compared spider responses towards 1) weaver ant and carpenter

173ant 2) weaver ant and procession ant and, 3) carpenter ant and procession ant. The frequency of

174spider responses (approach, attack and avoid) and the proportion of time spent stalking and

175looking towards different species of ants were compared using Kruskal-Wallis test followed by

176post-hoc analysis with Dunn's-test. Behavioral responses of the jumping spiders towards live ants bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

177and ant-lures; and house flies and ants were compared using Mann-Whitney U test. All statistical

178tests were carried out on R version 3.3.2 (R Core Team, 2018).

179

180Results

181Ants differ in their aggression towards spiders

182Our results show that weaver ants, carpenter ants and procession ants differ in their aggression

183towards jumping spiders. Weaver and procession ants were very aggressive, and bit spiders in

184less than a minute after they were introduced as indicated by their average latency time to bite

185the spiders (Table 2). However, carpenter ants were not aggressive: 12 out of 16 trials did not

186elicit any aggressive response from the ants. Carpenter ants are observed to feed on extra floral

187nectaries and tend plant hoppers in the wild, and are known to be less aggressive (Mody, &

188Linsenmair, 2003).

189

190Ants differ in their movement

191Weaver ants and procession ants moved significantly more than carpenter ants when placed

192inside the ring (Table 3). While weaver and procession ants moved continuously, carpenter ants

193moved in a staggered manner, where they would stop and move multiple times. This behavior is

194similar to what is observed in the wild (pers. obs.).

195

196Spiders differentially respond to three ants

9 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

197When controlled for aggression (by putting the ants inside a ring), the jumping spiders

198differentiated the ants by looks. Spiders gave quick short looks (median 2.29 sec) to all the three

199ants (Fig 2. Kruskal-Wallis χ2 = 1.337, df = 2, p-value = 0.5). On the other hand, spiders gave

200significantly longer looks (median 14.1 sec) to golden-backed carpenter and procession ants

201compared to weaver ants (Fig 3. Kruskal-Wallis χ2= 20.029, df= 2, p < 0.001. p values for post

202hoc Dunn’s test: carpenter ant vs weaver ant = 0.001, carpenter ant vs procession ant = 1 and

203procession ant vs weaver ant < 0.001). Furthermore, spiders approached, stalked and even

204attacked procession ants more than weaver ants. But these behaviors were not significantly

205different between carpenter ants and procession ants or carpenter ants and weaver ants (Table 4).

206Since ants were enclosed inside a barrier, they did not impose any threat towards spiders, and

207hence the spider sometimes approached and stalked these ants. When barrier was removed, the

208same spiders avoided these ants (unpublished results).

209Not surprisingly, jumping spiders, readily differentiated ants from the flies, corroborating that

210spiders can indeed differentiate prey from non-prey within our setup. When placed with a fly, all

211spiders exhibited a typical hunting behavior where they stalked (Mann-Whitney U = 665.5, p <

2120.001) and attacked (Mann-Whitney U = 539.5, p< 0.001) the fly significantly more than they

213attacked any of the ants.

214

215 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

216Spiders responses in the absence of ant movement

217In the absence of ant movement, spiders still looked at the ants, but did not differentiate them

218based on looks, suggesting that movement plays a crucial role in differentiating the three ants.

219Dunn's test for looks by spiders towards lures gave non-significant results across all

220comparisons.

221

222Discussion

223Most of the earlier studies suggest that jumping spiders avoid ants, but few have examined the

224response of spiders towards different ants. Our findings indicate that jumping spiders can

225distinguish and respond differentially to different species of ants. Weaver ants, procession ants,

226and carpenter ants differed in their color, aggression and movement, and spiders visually

227discriminated them. In this study we show that both movement and color of the ants influence

228the response of spiders.

229Spiders when placed inside the Petri dish stayed predominantly at the periphery, and responded

230towards ants around 10-20 % of the time during which they differentiated between the ants. Our

231results suggest that movement of the ants is crucial for spiders to detect them. An earlier study

232had shown that jumping spiders innately avoided motionless dead lures of ants by visual cues

233alone (Nelson, & Jackson, 2006). Researchers found that spiders, at the end of 10 hrs, chose to

234stay away from the lures of ants devoid of chemical, movement, and textural cues. Although our

235study is not directly comparable to theirs, in the absence of movement cues the spiders failed to

236differentiate the three ants. Jackson, & Tarsitano (1993) found that out of 11 species of salticids,

237only one ant-specialist spider (Corythalia canosa) attacked motionless ant lures. Movement of an

11 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

238object, be it an enemy, prey or a mate is one of the primary cues that elicit spiders to orient

239towards the signal (Forster, 1977; Jackson, & Pollard, 1996; Bednarski et al., 2012; Bartos, &

240Minias, 2016).

241Our study suggests that while movement of ants elicit detection, other sub-components of visual

242cues (color, brightness, texture and/or behavior) enables spiders to distinguish between them.

243Spiders barely gave long looks to weaver ants compared to golden-back and procession ants.

244Additionally, our preliminary studies (not shown here) with seven other species (Hyllus sp.,

245Menemerus sp., and from five other unidentified species) suggest a similar pattern. We have not

246explicitly tested whether spiders in our study perceive the red, black or black-gold colors of these

247ants, but recent studies have shown that spiders can see red (Zurek et al., 2015), and avoid red

248colored prey (Taylor et al., 2016). Regardless of Plexippus sp can see red, black or black-gold, it

249can nevertheless discriminate the ants.

250While spiders seem to detect ants using short looks, they discriminate ants by giving long looks.

251What may be the significance of such a distinction? First, by giving frequent short looks spiders

252constantly scan their surrounding environment, keeping the target in its field of view. All three

253ants were moving (either continuously or in a staggered manner) inside the transparent ring, thus

254eliciting repeated short looks from spiders. Jumping spiders are known to continuously scan

255novel objects by looking for features like ‘legs’ or ‘head spots’ (Land, 1999; Bartos, & Minias,

2562016). Moreover, jumping spiders in Huang et al.'s (2011) study gave more number of short

257looks to ants and ant mimicking spiders than non-mimetic spiders. Second, quick decisions have

258to be made when an enemy or a potential prey is dangerous or fast moving. For example,

259bumblebees make speedy decisions to avoid potentially dangerous foraging patches (Chittka et bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

260al., 2009). Spiders may give frequent short looks to all ants and make a decision early on if it is a

261potential threat.

262In addition to movement, spiders could use color to distinguish ants. Spiders gave more number

263of long looks to black colored ants; black color could belong to a potential prey or a predator.

264Spiders also approached, stalked and even attacked the black colored procession ants

265significantly more than the red colored weaver ants. This suggests that red colored ants could

266pose a stronger threat than black colored ones. Spiders' response toward black-gold colored

267carpenter ants fell in between their response towards red and black ants. A recent study has

268suggested that gold/yellow color is also aposematic to a wide range of invertebrate and vertebrate

269predators (Pekár et al., 2017). Whether color could play a significant component in

270distinguishing ants remains to be tested.

271If predators with high visual acuity can differentiate ants , do they also respond differently

272towards different ant-mimicking spiders? Interestingly, praying mantids approached spiders that

273mimic black colored carpenter ants significantly more than the spiders that mimic red colored

274weaver ants (Ramesh et al., 2016). That walking like an ant is indeed protective has been shown

275in recent studies (Pekár, & Jarab, 2011; Nelson, & Card, 2016; Shamble et al., 2017). Whether

276visual predators respond to the mimics in the same way, by giving longer looks to some and

277short looks to others needs to be further tested.

278There are over 5000 species of jumping spiders and most of them share habitats with different

279kinds of ants, especially in the tropics. Clearly it is advantageous to differentiate among ants

280rather than to avoid all of them and their mimics. How a spider perceives some ants as

281potentially dangerous is governed by dynamic interaction of multiple sensory modalities and

282contexts. We have shown the importance of visual cues in distinguishing among ants.

13 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

283

284

285Acknowledgement

286This research was funded by a DBT BT/Bio-CARe/04/9809/2013-14, and APU grant to DU.

287Funding for CC and SV came from the DBT grant. We would like to thank Aranya Bagchi who

288helped collect spiders, and Dr. Deepak Deshpande for taking ant pictures.

289Contribution of authors

290CC, SV and DU were involved in project design, data collection and analysis. SV and DU wrote

291the paper.

292

293References

294Allan, R.A., & Elgar, M.A. (2001) Exploitation of the green tree ant, Oecophylla smaragdina, by

295 the salticid spider Cosmophasis bitaeniata. Australian Journal of Zoology, 49, 129–137.

296Bartos, M., & Minias, P. (2016) Visual cues used in directing predatory strikes by

297 thejumping spider Yllenus arenarius (Araneae, Salticidae). Animal Behaviour, 120,

298 51–59.

299Bednarski, J.V., Taylor, P., & Jakob, E.M. (2012) Optical cues used in predation by jumping

300 spiders, Phidippus audax (Araneae, Salticidae). Animal Behaviour, 84, 1221–1227.

301Chittka, L., Skorupski, P., & Raine, N.E. (2009) Speed–accuracy tradeoffs in animal decision

302 making. Trends in Ecology & Evolution, 24, 400–407. bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

303Clark, R.J., Jackson, R.R., & Cutler, B. (2000) Chemical cues from ants influence predatory

304 behavior in Habrocestum pulex, an ant-eating jumping spider (Araneae, Salticidae).

305 Journal of Arachnology, 28, 309–318.

306Cushing, P.E. (1998) Population Structure of the Ant Nest Symbiont Masoncus

307 pogonophilus (Araneae: Linyphiidae). Annals of the Entomological Society of America,

308 91, 626–631.

309Cushing, P.E. (2012) Spider-Ant Associations: An Updated Review of Myrmecomorphy,

310 Myrmecophily, and Myrmecophagy in Spiders. Psyche: A Journal of Entomology, 2012,

311 1–23.

312Del-Claro, K., & Oliveira, P.S. (2000) Conditional outcomes in a neotropical treehopper-ant

313 association: temporal and species-specific variation in ant protection and

314 homopteran fecundity. Oecologia, 124, 156–165.

315Dolev, Y., & Nelson, X.J. (2014) Innate Pattern Recognition and Categorization in a Jumping

316 Spider. PLoS ONE, 9, e97819.

317Edwards, G.B., & Jackson, R.R. (1994) The role of experience in the development of

318 predatory behaviour in Phidippus regius, a jumping spider (Araneae, Salticidae) from

319 Florida. New Zealand Journal of Zoology, 21, 269–277.

320Forster, L. (1985) Target Discrimination in Jumping Spiders (Araneae: Salticidae). In

321 Neurobiology of (ed. by Barth, F.G.). Springer Berlin Heidelberg, Berlin,

322 Heidelberg, pp. 249–274.

15 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

323Forster, L.M. (1977) A qualitative analysis of hunting behaviour in jumping spiders

324 (Araneae: Salticidae). New Zealand Journal of Zoology, 4, 51–62.

325Halaj, J., Ross, D.W., & Moldenke, A.R. (1997) Negative effects of ant foraging on spiders in

326 Douglas-fir canopies. Oecologia, 109, 313–322.

327Harland, D., Li, D., & Jackson, R. (2012) How jumping spiders see the world. How animals

328 see the world: comparative behavior, biology, and evolution of vision, 133–164.

329Hölldobler, B., & Wilson, E.O. (1990) The Ants. Harvard University Press.

330Huang, J.-N., Cheng, R.-C., Li, D., & Tso, I.-M. (2011) Salticid predation as one potential

331 driving force of ant mimicry in jumping spiders. Proceedings of the Royal Society B:

332 Biological Sciences, 278, 1356–1364.

333Ibarra-Isassi, J., & Oliveira, P.S. (2018) Indirect effects of mutualism: ant–treehopper

334 associations deter pollinators and reduce reproduction in a tropical shrub. Oecologia,

335 186, 691–701.

336Jackson, R.R., & Pollard, S.D. (1996) Predatory Behavior of Jumping Spiders. Annual Review

337 of Entomology, 41, 287–308.

338Jackson, R.R., & Tarsitano, M.S. (1993) Responses of jumping spiders to motionless prey.

339 Bull. Br. Arachnol. Soc, 9, 105–109.

340Land, M. (1972) Mechanisms of orientation and pattern recognition by jumping spiders

341 (Salticidae). Information processing in the visual systems of arthropods, 231–247. bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

342Land, M.F. (1999) Motion and vision: why animals move their eyes. Journal of Comparative

343 Physiology A: Sensory, Neural, and Behavioral Physiology, 185, 341–352.

344Li, D., Jackson, R.R., & Cutler, B. (1996) Prey-capture techniques and prey preferences of

345 Habrocestum pulex , an ant-eating jumping spider (Araneae, Salticidae) from North

346 America. Journal of Zoology, 240, 551–562.

347McIver, J.D., & Stonedahl, G. (1993) Myrmecomorphy: Morphological and Behavioral

348 Mimicry of Ants. Annual Review of Entomology, 38, 351–377.

349Mody, K., & Linsenmair, K.E. (2003) Finding its place in a competitive ant community: leaf

350 fidelity of Camponotus sericeus. Insectes Sociaux, 50, 191–198.

351Nelson, X.J., & Card, A. (2016) Locomotory mimicry in ant-like spiders. Behavioral Ecology,

352 27, 700–707.

353Nelson, X.J., & Jackson, R.R. (2006) Vision-based innate aversion to ants and ant mimics.

354 Behavioral Ecology, 17, 676–681.

355Nelson, X.J., Jackson, R.R., Li, D., Barrion, A.T., & Edwards, G.B. (2006) Innate aversion to

356 ants (Hymenoptera: Formicidae) and ant mimics: experimental findings from

357 mantises (Mantodea): ANT MIMICRY. Biological Journal of the Linnean Society, 88, 23–

358 32.

359Okuyama, T. (2002) The role of antipredator behavior in an experimental community of

360 jumping spiders with intraguild predation. Population Ecology, 44, 121–125.

17 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

361Pekár, S., & Jarab, M. (2011) Assessment of color and behavioral resemblance to models by

362 inaccurate myrmecomorphic spiders (Araneae): Inaccurate myrmecomorphic

363 spiders. Invertebrate Biology, 130, 83–90.

364Pekár, S., Petráková, L., Bulbert, M.W., Whiting, M.J., & Herberstein, M.E. (2017) The golden

365 mimicry complex uses a wide spectrum of defence to deter a community of predators.

366 ELife, 6.

367Pohlert, T. (2014) The Pairwise Multiple Comparison of Mean Ranks Package (PMCMR).

368R Core Team. (2018) R: A Language and Environment for Statistical Computing. R

369 Foundation for Statistical Computing, Vienna, Austria.

370Ramesh, A., Vijayan, S., Sreedharan, S., Somanathan, H., & Uma, D. (2016) Similar yet

371 different: differential response of a praying mantis to ant-mimicking spiders.

372 Biological Journal of the Linnean Society, 119, 158–165.

373Roslin, T., Hardwick, B., Novotny, V., Petry, W.K., Andrew, N.R., Asmus, A., et al. (2017)

374 Higher predation risk for insect prey at low latitudes and elevations. Science, 356,

375 742–744.

376Sanders, D., Nickel, H., Grützner, T., & Platner, C. (2008) Habitat structure mediates top–

377 down effects of spiders and ants on herbivores. Basic and Applied Ecology, 9, 152–

378 160.

379Sanders, D., & Platner, C. (2006) Intraguild interactions between spiders and ants and top-

380 down control in a grassland food web. Oecologia, 150, 611. bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

381Sendoya, S.F., Freitas, A.V.L., & Oliveira, P.S. (2009) Egg‐Laying LayingButterfliesDistinguish Butterflies Distinguish

382 Predaceous Ants by Sight. The American Naturalist, 174, 134–140.

383Shamble, P.S., Hoy, R.R., Cohen, I., & Beatus, T. (2017) Walking like an ant: a quantitative

384 and experimental approach to understanding locomotor mimicry in the jumping

385 spider Myrmarachne formicaria. Proceedings of the Royal Society B: Biological Sciences,

386 284, 20170308.

387Suarez, A.V., Richmond, J.Q., & Case, T.J. (2000) PREY SELECTION IN HORNED LIZARDS

388 FOLLOWING THE INVASION OF ARGENTINE ANTS IN SOUTHERN CALIFORNIA.

389 Ecological Applications, 10, 711–725.

390Taylor, L.A., Amin, Z., Maier, E.B., Byrne, K.J., & Morehouse, N.I. (2016) Flexible color

391 learning in an invertebrate predator: Habronattus jumping spiders can learn to prefer

392 or avoid red during foraging. Behavioral Ecology, 27, 520–529.

393Taylor, L.A., Maier, E.B., Byrne, K.J., Amin, Z., & Morehouse, N.I. (2014) Colour use by tiny

394 predators: jumping spiders show colour biases during foraging. Animal Behaviour, 90,

395 149–157.

396Zurek, D.B., Cronin, T.W., Taylor, L.A., Byrne, K., Sullivan, M.L.G., & Morehouse, N.I. (2015)

397 Spectral filtering enables trichromatic vision in colorful jumping spiders. Current

398 Biology, 25, R403–R404.

399

400

401

19 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

402

403Figure legends

404

405Figure 1: Three ant species and the jumping spider used in the study: A. Aggressive fast-

406moving weaver ants B. Aggressive, fast-moving procession ants C. Docile, staggered-moving

407carpenter ants D. Plexippus petersi, jumping spider (Photographs not according to scale)

408

409Figure 2: Proportion of time spider spent giving short looks: Proportion of time spent by

410spiders giving short looks to ants did not vary across the different ants. Kruskal-Wallis χ2 =

4111.3377, df = 2, p = 0.5.

412

413Figure 3: Proportion of time spider spent giving long looks: More time was spent giving long

414looks to procession ant and carpenter ants compared to the weaver ant. Kruskal-Wallis χ2 =

41520.029, df = 2, p < 0.001. Post hoc Dunn's-test: Weaver ant vs Carpenter ant p = 0.001, Weaver

416ant vs Procession ant p < 0.001, Procession ant vs Carpenter ant p = 1df = 2.

417

418

419

420

421 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

422

423

424

425

426

427

428

429

430Table 1: Ants differed in color, aggressive behavior and movement.

Ant Color Aggression Movement Weaver ant Red High Continuous Procession ant Black High Continuous Carpenter ant Black-gold Low Sporadic

431

432Table 2: Aggression of ants, measured as latency to bite a spider: The more aggressive ants

433take less time to bite a spider. Both procession ants and weaver ants are aggressive compared to

434carpenter ants. Dunn's test for ant aggression: Weaver ant vs Carpenter ant p = 0.007, Weaver ant

435vs Procession ant p = 1 and Procession ant vs Carpenter ant p = 0.001

Ant Latency to attack (sec) Weaver ant (n = 16) 63.32 ± 68.67 Procession ant (n = 16) 59.8 ± 77.6 Carpenter ant (n = 16) 156.9 ± 52.1

21 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

436

437Table 3: Percentage of time ants spent moving in experimental arena: Both weaver ants and

438procession ants move continuously, whereas carpenter ants moved sporadically. Dunn's test for

439percentage of time spent moving by the ants: Weaver ant vs Carpenter ant p < 0.001, Weaver ant

440vs Procession ant p = 1 and Procession ant vs Carpenter ant p < 0.001

Ant Percentage of time spent moving Weaver ant (n = 30) 93.48 ± 9.69 Procession ant (n = 29) 86.41 ± 22.79 Carpenter ant (n = 26) 47.69 ± 38.54 441

442Table 4: Spider responses towards three ants: Spiders differed in their approach, stalk and

443attack towards the three ants (Kruskal-Wallis test: Approach: χ2: 7.45, P=0.02; Stalk: χ2: 10.13,

444P=0.006; Attacks: χ2: 6.35, P= 0.04; Avoid χ2: 0.51, P=0.7). Dunn Post hoc P values are

445presented below.

Weaver ant vs Weaver ant vs Procession ant vs

Behavior Carpenter ant Procession ant Carpenter ant Approach 0.29 0.02 0.20 Stalk 0.185 0.005 0.131 Attack 0.47 0.043 0.167 Avoid 1.000 1.000 1.000

446

447

448

449 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

450

451

452

453

454

455

456

457

458

459

460

461

462 A B 463

464

465

466 C D 467Figure 1

23 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

468

469

470

471

472Figure 2

473

474 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

475

476Figure 3.

477

478

479

480

481

482

483

25 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

484

485Supplementary information

486

487Figure S1: Histogram of duration of looks by spiders towards ants: The cut-off for

488differentiating between long and short looks has been established as 7 seconds. The normalized

489durations of short looks (duration less than 7 seconds) follow a beta distribution, with a median

490value of 0.02 (equivalent to 2.29 seconds).

491

492

493

494 bioRxiv preprint doi: https://doi.org/10.1101/349696; this version posted June 18, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

495Table S1: Spider responses towards live and dead ants:

496Spiders spent more time giving short looks to the live ants than their lure. Mann-Whitney U test

497scores comparing long and short looks of spiders towards ants and their lures are given below.

Weaver ant vs Weaver Procession ant vs Carpenter ant vs

Behavior ant lure Procession ant lure Carpenter ant lure Short look U = 390, p<0.001 U = 411, p <0.001 U = 262, p = 0.12 Long look U = 298.5, p = 0.062 U = 313, p = 0.084 U = 296, p = 0.012 498

27