Giant Australian cuttlefish use mutual assessment to resolve male-male contests. 1 Alexandra K ...

43 Pages·2015·0.76 MB·English

1
  Giant Australian cuttlefish use mutual assessment to resolve male-male contests.
2
  Alexandra K. Schnell*a, Carolynn L. Smitha, Roger T. Hanlonb, Robert Harcourta
3
  a Department of Biological Sciences, Macquarie University, Sydney, NSW, Australia
4
  b Program in Sensory Physiology and Behavior, Marine Biological Laboratory, Woods Hole,
5
  MA, U.S.A.
6
  Received 12 January 2015
7
  Initial acceptance 11 February 2015
8
  Final acceptance 15 May 2015
9
  MS. number: 15-00033R
10
  *Correspondence: A. K. Schnell, Macquarie University, Department of Biological Sciences,
11
  North Ryde, Sydney, NSW 2109, Australia.
12
  E-mail address: [email protected] (A. K. Schnell).
13

14

15

16

17

18

19

20

21
22
  Game theory models provide a useful framework for investigating strategies of conflict
23
  resolution in animal contests. Model predictions are based on estimates of resource-holding
24
  potential (RHP) and vary in their assumptions about how opponents gather information about
25
  RHP. Models can be divided into self-assessment strategies (energetic war-of-attrition, E-
26
  WOA; cumulative assessment model, CAM) and mutual assessment strategies (sequential
27
  assessment model, SAM). We used laboratory-staged contests between male giant Australian
28
  cuttlefish, Sepia apama, to evaluate RHP traits and to test game theory models. Mantle length
29
  was a key indicator of RHP because it predicted contest outcome, whereby larger individuals
30
  were more likely to win a contest. Winners and losers did not match behaviours, ruling out
31
  the E-WOA. There was no relationship between contest outcome, duration and escalation
32
  rates, arguing against the CAM. Persistence to continue a contest was based on RHP
33
  asymmetry, rather than loser and/or winner RHP, providing support for the SAM. Motivation
34
  to fight was determined from a male’s latency to resume a contest following the introduction
35
  of a female during a contest. The latency to resume a contest was negatively related to the size
36
  of the focal male and positively related to the size of their opponent. These results show that
37
  competing males are able to gather information concerning RHP asymmetries, providing
38
  support for mutual assessment. Furthermore, males showed significant behavioural
39
  differences in their responses to relatively larger than to relatively smaller opponents. Using
40
  an integrative approach, our study provides a well-substantiated example of mutual
41
  assessment.
42
  Keywords: contest competition, fighting ability, game theory, resource-holding potential,
43
  sequential assessment model, visual signalling
44

45
46
  Animals competing over limited resources are likely to incur costs, including increased
47
  energy expenditure and risk of predation, injuries or fatal attacks (Maynard Smith, 1974;
48
  Maynard Smith & Price, 1973). During contests, animals may gather information from
49
  multiple sources to assess the potential costs and benefits of continued conflict, in turn
50
  facilitating economic and tactical decision making (Maynard Smith & Parker, 1976; Parker,
51
  1974). The decision to withdraw from a contest is usually influenced by the fighting ability of
52
  a contestant, termed resource-holding potential (RHP; Maynard Smith, 1974; Parker, 1974;
53
  Parker & Stuart, 1976). The information that facilitates these decisions will be dictated by the
54
  assessment capabilities of the species (Taylor & Elwood, 2003).
55
  Game theoretical approaches serve as an analytical tool for understanding the patterns
56
  of behaviour observed in contests across many taxa. Currently, three major game theory
57
  models may be applied to animal contests to determine the assessment strategy used for
58
  decision making (Table 1). The models can be divided into self-assessment and mutual
59
  assessment strategies. The self-assessment models include the energetic war of attrition (E-
60
  WOA; Payne & Pagel, 1996; 1997) and the cumulative assessment model (CAM; Payne,
61
  1998). These models assume that contestants evaluate their own RHP, but fail to assess their
62
  opponent’s RHP. Contestants differ in rates of escalation within phases (i.e. periods defined
63
  by behaviours of similar aggressive intensity). The decision point to withdraw is determined
64
  by the weaker individual’s threshold for costs. For the E-WOA model, the threshold is based
65
  on self-imposed energetic costs. For the CAM, the threshold is determined by combined costs
66
  that accumulate as a function of time and energy expenditure, as well as the damage inflicted
67
  by the opponent. Mutual assessment is modelled through the sequential assessment model
68
  (SAM), which assumes that contestants evaluate their own RHP relative to their opponent’s
69
  RHP (Enquist & Leimar, 1983). In this model, contests progress through a series of
70
  successive phases, which are thought to provide increasingly accurate information about the
71
  RHP asymmetry between contestants. Predictions for these three models are based on
72
  estimates of RHP and vary in their assumptions about how opponents gather information
73
  about RHP (Table 1).
74
  Mutual assessment is assumed to be a more efficient strategy than self-assessment
75
  because animals can minimize costly and futile persistence by gathering information about
76
  relative RHP (Enquist & Leimar, 1983). However, studies on a wide range of animal contests
77
  that have shown mutual assessment (e.g. Englund & Olsson, 1990; Junior & Peixoto, 2013;
78
  Kemp, Alcock, & Allen, 2006; Pratt, McLain, & Lathrop, 2003) have recently been called
79
  into question (Briffa & Elwood, 2009; Elwood & Arnott, 2012; Taylor & Elwood, 2003).
80
  Taylor and Elwood (2003) contended that such studies may have actually presented artefacts
81
  of alternative mechanisms. For example, a negative association between contest duration and
82
  RHP asymmetry, which is thought to be indicative of the SAM (i.e. mutual assessment),
83
  could also arise if the weaker contestant accrued costs faster than its opponent (i.e. self-
84
  assessment, E-WOA). Taylor and Elwood (2003) recommended a statistical framework to
85
  distinguish between mutual and self-assessment strategies. This framework has been
86
  implemented in many studies, revealing that self-assessment is more prevalent than
87
  previously thought (e.g. Brandt & Swallow, 2009; Prenter, Elwood, & Taylor, 2006; Stuart-
88
  Fox, 2006). However, distinguishing between assessment strategies remains a challenge, and
89
  consequently several recent studies report inconclusive results (e.g. Batista, Zubizarreta,
90
  Perrone, & Silva, 2012; Egge, Brandt, & Swallow, 2011; Jennings, Gammell, Carlin, &
91
  Hayden, 2004; Kelly, 2006).
92
  Recently, there has been renewed debate about whether mutual assessment is more
93
  cognitively complex than self-assessment because of its apparent requirement for comparative
94
  decision making (Elwood & Arnott, 2012; Elwood & Arnott, 2013; Fawcett & Mowles,
95
  2013). Elwood and Arnott (2012) and Fawcett and Mowles (2013) argued that mutual
96
  assessment could entail cognitively simple threshold decision making. They noted that the
97
  original SAM model (i.e. mutual assessment) does not require an explicit comparison of RHP;
98
  rather, information about RHP is directly transmitted as a relative measure (i.e. as error-prone
99
  estimates of relative fighting ability). Moreover, Elwood and Arnott (2012) argued that many
100
  studies provide insufficient evidence of individuals comparatively assessing RHP, and that
101
  many claims of comparison of body size, claw size or dewlap size still need to be
102
  substantiated. One experimental approach to substantiate such claims involves assessing the
103
  motivational state of an animal in a contest by using a novel stimulus that causes a contestant
104
  to temporarily cease fighting (see Arnott & Elwood, 2009a; Elwood, Wood, Gallagher, &
105
  Dick, 1998). The latency to resume the contest provides a measure of the individual’s
106
  motivation to fight (see Table 1 for predictions). Another approach is to test the ability of a
107
  contestant to assess relative values (e.g. body size or claw size) in the context of aggression
108
  (see e.g. dogs, Canis familiaris, Taylor, Reby, & McComb, 2010). Testing such capabilities
109
  during a contest may validate claims of mutual assessment.
110
  This study investigated the assessment strategy used to resolve conflict in male giant
111
  Australian cuttlefish, Sepia apama. These cuttlefish engage in dynamic signalling during
112
  agonistic contests, similar to other species in which game theory models have been tested
113
  (e.g. hermit crabs, Briffa & Elwood, 2001; chameleons, Stuart-Fox, 2006; wasps, Tibbetts,
114
  Mettler, & Levy, 2010). Contests typically occur in the presence of females during their
115
  reproductive season (austral winter months: May–August). However, even in the absence of
116
  females, males engage in contests in both field (Hanlon, 1999) and laboratory settings
117
  (Schnell, 2014). The fighting tactics of males are influenced by body size, which varies
118
  widely at maturity. Small males tend to reduce direct aggression by being surreptitious or
119
  through deceptive signalling (i.e. female mimicry; Hanlon, Naud, Shaw, & Havenhand,
120
  2005). Large males regularly engage in agonistic contests, which are typically mediated
121
  through visual displays but can escalate to physical pushing and grappling (Hall & Hanlon,
122
  2002). Variation in body size and its effect on agonistic behaviours suggest that this species
123
  has evolved the ability to assess the size of its opponents and alter its behaviour accordingly.
124
  However, the assessment strategy used during these contests has not been tested. The
125
  application of game theory models to cuttlefish contests may be an effective tool for
126
  determining patterns of fighting behaviour (i.e. self-assessment or mutual assessment
127
  strategy) in this particular system.
128
  The central aim of our study was to determine the fighting strategy used by giant
129
  Australian cuttlefish during male–male contests. First, we assessed the male traits that may be
130
  associated with RHP. Second, we used specific predictions of the three major game theory
131
  models (E-WOA, CAM, and SAM; see Table 1 for predictions) to determine whether the
132
  decision to withdraw from a contest was based on the absolute RHP of the loser (self-
133
  assessment) or on the RHP of the loser relative to the winner (mutual assessment). Third, the
134
  contestant’s assessment of RHP was substantiated by measures of motivation and aggression.
135

136
  <H1>METHODS
137
  <H2>Study species, collection and husbandry
138
  Thirty-four male and four female adult giant Australian cuttlefish were caught via scuba in
139
  coastal areas of Sydney, Australia (34°50’S, 151°22’E) between April and May 2012. They
140
  were transported (< 50 min) to the aquarium facility at Cronulla Fisheries Research Centre in
141
  a custom-made transport tank (9.0 x 8.0 cm and 8.0 cm high, maximum capacity = 3
142
  subjects). Water in the transport tank was oxygenated and maintained at a natural ambient sea
143
  temperature (15–17 °C). Sex was determined by coloration and the dimorphic state of the
144
  fourth arm. Subjects were housed individually in open-air tanks that received a constant flow
145
  (approximately 10 litres/min) of filtered ambient sea water. Cuttlefish were fed a mixed diet
146
  of food items including live Australian ghost shrimp, Trypaea australiensis, and thawed
147
  frozen prawn, squid or pilchard every evening.
148
  <H2>Male traits
149
  We measured mantle length and dimorphic arm length (ventral-most pair; Fig. 1) to the
150
  nearest 0.1 mm and weighed cuttlefish to the nearest 1 g using a Precisa electronic balance
151
  (30000D IP65 Wedderburn scales TYP 480-9580, Switzerland). We also measured ‘passing
152
  cloud’ behaviour which is a chromatic signal typically expressed by males during the lateral
153
  display (Fig. 1b). It involves the expansion and contraction of chromatophores to produce the
154
  appearance of light and dark bands flowing unidirectionally over the mantle. The number of
155
  clouds and speed of travel are relatively consistent; however, the expression of this behaviour
156
  varies in intensity. To measure changes in intensity, we recorded the duration and the contrast
157
  of the bands. There was no variation in the duration of bands; however, the contrast of bands
158
  changed throughout the agonistic interactions. We therefore measured the intensity using
159
  contrast differences between light and dark bands. Contrast differences are likely to be
160
  visually conspicuous because the visual systems of cuttlefish are sensitive to polarized light
161
  (Mäthger, Shashar, & Hanlon, 2009; Shashar, Rutledge, & Cronin, 1996). We calculated the
162
  brightness of the bands on the mantle of each male displaying passing cloud. This was
163
  measured on a laptop (Apple Macintosh, OS X 10.9.2) broadcasting the video recordings and
164
  using Apple Macintosh Digital Color Meter. We recorded the RGB values at 10 random
165
  locations on both light and dark bands. Brightness values were calculated from RGB values (0
166
  = black and 255 = white) using the luminance formula from Poynton (2003):
167
  Y = 0.2126 × R + 0.7152 × G + 0.0722 × B
168
  The means of each set of brightness values was then used to approximate contrast differences
169
  between the bands. Correlations of all four male attributes (mantle length, arm length, mass
170
  and passing cloud intensity) were then determined using Pearson correlation coefficients (see
171
  Appendix Table A1).
172
  <H2>Male contests
173
  Laboratory-staged contests were carried out in June and July 2012. Twenty-two males that
174
  varied in size (mean mantle length = 415.1 mm; range 295–509 mm; mean body weight =
175
  6348 g; range 4015–9324 g) were used. We used a repeated measures design, in which each
176
  male was assigned to (1) a size-symmetric opponent (within 7% of mantle length of each
177
  other) and (2) a size-asymmetric opponent (at least 20% mantle length difference). Subjects
178
  were assigned a random sequence of the treatments (i.e. size-symmetric and size-asymmetric)
179
  to control for order effects. Trials were staged 4 weeks apart to minimize experience effects,
180
  since different agonistic experiences can sometimes affect the outcome of a contest (i.e.
181
  winner–loser effect; Hsu & Wolf, 1999).
182
  The contest arena was a circular 5000-litre (height 12.0 cm, diameter 23.4 cm; 2.38 m2
183
  per cuttlefish) tank divided in half. A clear partition was fixed in place to physically isolate
184
  subjects and prevent injury during the contests. An opaque partition slid loosely next to the
185
  clear partition to visually isolate subjects during the first phase of the tests. Water was able to
186
  flow between both compartments to facilitate chemical exchange between the male subjects.
187
  A high-definition video camera (SONY HDR-SR11E) fitted with a wide-angle lens (Raynox
188
  HD-5050PRO 0.5 x) was placed directly over the test arena to record (MTS format 1920 ×
189
  1080 lines) behavioural responses.
190
  To acclimate subjects to the apparatus, individuals were placed in the test arena for
191
  three 65 min periods over 3 consecutive days. During acclimation sessions the subjects were
192
  tested separately, so that males did not come into visual contact or chemical exposure with
193
  one another until the experimental phase. Subjects were placed gently into one of the two
194
  compartments in the test arena. Following a 5 min period, the opaque partition was removed
195
  by sliding it out of the water. After 60 min subjects were recaptured and transferred back to
196
  their home tank. After each acclimation session, the test arena was drained, cleaned and
197
  refilled using fresh filtered sea water to ensure subjects were not responding to any chemical
198
  cues left in the water from previous test subjects. By the third acclimation session subjects
199
  exhibited no signs of disturbance, such as inking or jetting, from the transfer procedure. The
200
  experimental trials followed the same procedure used in the acclimation sessions. Subjects
201
  typically engaged in more than one contest during the 60 min staged experimental trials and
202
  therefore the repeated measures design was unbalanced. We observed 75 contests in 21
203
  experimental trials (mean number of contests = 3.43; range 1–7); one trial without mutual
204
  displaying between contestants was excluded.
205
  <H2>Probing motivational state in response to a female stimulus
206
  The motivational state experiments were carried out in June 2012. Twelve males that varied
207
  in size (mean mantle length = 411.76 mm; range 350–464 mm; mean body weight = 5113 g;
208
  range 4278–6324 g) and had not participated in any previous experiments were used. We used
209
  a repeated measures design, in which each test male was assigned a size-asymmetric opponent
210
  (at least 20% mantle length difference) and tested twice using the same opponent. Probing
211
  motivational state is typically tested using a startle stimulus (see Arnott & Elwood, 2009a;
212
  Elwood et al., 1998). However, initial pilot observations revealed that male subjects did not
213
  respond to small startle stimuli whereas larger startle stimuli caused focal males to ink,
214
  making it impossible to record the behaviour of the test subjects. We used a female to distract
215
  the focal male instead. This consistently caused the focal male to cease fighting and inspect
216
  the female before resuming the contest. The test arena was the same as the contest arena
217
  except it was divided into three compartments that did not allow chemical exchange. The
218
  edges of the clear Perspex partition were sealed with a silicon sealant. Two compartments
219
  were equal in size and the third compartment was smaller. Test males were placed in the
220
  equal-sized compartments and a female was placed in the smaller compartment. Following a
221
  30 min acclimation period the opaque partition between the males’ compartments was
222
  removed. One minute after the onset of display by both male cuttlefish, the opaque partition
223
  to the female compartment was removed on one side only, so that one male contestant (i.e. the
224
  focal male) could see the female. The female compartment in the test arena was designed so
225
  that when only one opaque partition was removed her presence was undetected by the male
226
  opponent (i.e. stimulus male). The specific timing of introducing the female was chosen
227
  following initial laboratory-staged contest experiments, and selected to give sufficient periods
228
  of display and escalation. Following each interaction, the subjects were returned to their home
229
  tanks and 48 h later the original pairs were retested in the same manner, but with the
230
  previously nondistracted stimulus male now being designated as the focal male and exposed
231
  to the female. Four females of similar size (mean mantle length = 299.71 mm; range 293–304;
232
  mean body weight = 3266.34 g; range 3104–3326) were used as a stimulus to reduce
233
  pseudoreplication. Males had not encountered the females prior to this experiment.
234
  <H2>Decision making in response to aggressive rivals with variable RHP
235
  The decision-making experiments were carried out in July 2012. Eight medium-sized males
236
  (mean mantle length = 415.88 mm; range 390–431 mm; mean body weight = 5424 g; range
237
  4515–5834 g) that were used in the laboratory-staged contest experiments were subsequently
238
  used as focal subjects. A repeated measures design was used, in which each focal male was
239
  assigned a smaller live-male stimulus (mantle length approximately < 20%) and a larger live-
240
  male stimulus (mantle length approximately > 20%). Focal males were assigned a random
241
  sequence of the treatments (i.e. small and large) to control for order effects. Focal males were