Table 1. Predicted relationships between various contest behaviors and the sizes of the two opponents from the assessment models and the non-assessment models.

(1)

Table 1. Predicted relationships between various contest behaviors and the sizes of the two opponents from the assessment models and the non-assessment models.

Indices Assessment Models Non-Assessment Models Smaller

¹

Bigger

²

SDiff

³

Smaller Bigger

contest duration ＋

⁴

－－＋＋ / × contest intensity

＋

－－ NA NA display frequency

＋

－－＋＋ / × attack frequency

＋

－－＋＋ / ×

1

：size of the smaller opponents in this study, measured in body length

2

：size of the bigger opponents in this study, measured in body length

3

：Size Difference between opponents, the relative size difference between the two

opponents in this study=( 100

fish smaller of

size

fish smaller the

of size - fish bigger the

of

size × %)

4

：predicted relationships between sizes and contest behaviors.

＋ positive relationship.

－ negative relationship.

× no relationship

NA

no prediction from the models

(2)

Table 2. Factors examined in this study.

name of variables definition

C

¹

_HON9

I used four R. marmoratus clones for this study. By using the clone RHL

C_VOL

as the the baseline group, I set up 3 dummy variables representing clone

C_SLC8E

HON9, VOL, and SLC8E for the regression model.

LO

²

_W

³

I took the outcome of the pair’s last dyadic contest into consideration.

By using the loser pairs the baseline group, I set up a dummy variable representing the contests pairs having won their last dyadic contests.

S

⁴

_BL

⁶

The body length of the smaller opponent

B

⁵

_BL

The body length of the bigger opponent

1

：

C stands for Clone

2：

LO stands for Last Outcome of fighting.

3：

types of last fighting outcome, Winning or Losing

4：

S stands for Smaller fish

5：

B stands for Bigger fish

6：

BL stands for Body Length as the size measurement in the experiment.

(3)

Table 3. Results of multiple linear regression analysis on contest duration

¹

.

X factors

R_adj²

DF b SE F ratio p>F

Full Model

Overall test

²

0.03 6 1.41 0.22 C_HON9 0.57 0.44 1.73 0.19 C_VOL

-0.21 0.45 0.21 0.65 C_SLC8E

-0.05 0.43 0.01 0.91 LO_W 0.61 0.31 3.91 0.05 S_BL 0.19 0.39 0.23 0.64 B_BL -0.10 0.09 1.33 0.25

1

： contest duration is the period between 1

^st

display and retreat of one contestant in

the dyadic contest

2

：the whole model test including all independent variables listed in Table 2

Table 4. Results of survival regression analysis on contest duration.

X factors DF b SE L-R χ

²

p> χ

²

Full Model

Overall test 6 9.83

0.13 C_HON9

-0.48 0.34 2.03 0.15 C_VOL -0.73 0.35 4.37 0.04 * C_SLC8E

-0.03 0.33 0.01 0.92 LO_W -0.25 0.25 1.02 0.31 S_BL -0.19 0.27 0.49 0.49 B_BL

0.13 0.07 3.30 0.07

* p<0.05

(4)

Table 5. Results of ordinal logistic regression on contest intensity.

X factors R DF b SE L-R

²

χ

²

p> χ

²

Full Model

Overall test 0.10 6 18.71

0.005 C_HON9 0.12 0.62 0.04 0.84 C_VOL

0.17 0.64 0.07 0.80 C_SLC8E

0.05 0.62 0.01 0.94 LO_W

-1.31 0.46 8.38 0.004 **

S_BL

-1.24 0.58 4.54 0.03 * B_BL

0.35 0.13 7.00 0.008 **

Reduced Model I

Overall test 0.10 3 18.63 0.0003 LO_W

-1.32 0.46 8.70 0.003 **

S_BL

-1.24 0.56 4.78 0.03 * B_BL 0.36 0.13 8.01 0.005 **

Reduced Model II

Overall test 0.09 2 16.71 0.0002 LO_W

-1.32 0.46 8.75 0.003 **

RD

¹

_BL

0.09 0.03 11.37 0.0007 ***

1

：the relative difference of size between 2 contestants：

( 100

fish smaller of

BL

fish smaller the

of BL - fish bigger the

of

BL × %)

*** p<0.001

** p<0.01

* p<0.05

(5)

Table 6. Relationships between the sizes of the two contestants and the frequency of displays and attacks in the contests.

Response X factors r

_s ¹

p S_BL 0.08 0.48

B_BL

0.04 0.72

S_BL

0.22 0.05 B_BL

-0.18 0.12

1

：Spearman Rank correlation

Table 7. Results of multiple logistic regression on the probability of the bigger opponents winning contests.

X factors R DF b SE L-R

²

χ

²

p> χ

²

Full Model

Overall test 0.20 6 18.71 0.005 C_HON9

-0.75 0.77 0.96 0.33 C_VOL

0.33 0.29 0.16 0.69 C_SLC8E

0.71 0.85 0.71 0.40 LO_W 1.68 0.65 7.87 0.005 **

S_BL

-1.37 0.78 3.27 0.07 B_BL

0.28 0.18 2.48 0.12

Reduced Model

Overall test 0.09 1 8.25 0.004 LO_W 1.55 0.58 8.25 0.004 **

** p<0.01

Display frequency

Attack frequency

(6)

Table 8. Multiple logistic regression on the probability of the bigger contestants initiated displays.

X factors R DF b SE L-R

²

χ

²

p> χ

²

Full Model

Overall test 0.10 6 10.23 0.11 C_HON9

0.80 0.70 1.34 0.25 C_VOL 0.50 0.73 0.48 0.49 C_SLC8E

0.75 0.71 1.15 0.28 LO_W 0.44 0.51 0.76 0.38 S_BL

-0.06 0.65 0.01 0.93 B_BL -0.43 0.15 9.09 0.003 **

Reduced Model

Overall test 0.07 1 7.83 0.005 B_BL -0.38 0.15 7.83 0.005 **

** p<0.01

(7)

Table 9. Multiple logistic regression on the probability of the bigger contestants initiated attacks.

X factors R DF b SE L-R

²

χ

²

p> χ

²

Full Model

Overall test 0.19 6 18.39 0.005 C_HON9

-0.24 0.72 0.11 0.74 C_VOL

0.96 0.78 1.57 0.21 C_SLC8E

1.51 0.85 3.43 0.06 LO_W 0.52 0.57 0.85 0.36 S_BL

-2.30 0.77 10.59 0.001 **

B_BL

0.02 0.17 0.01 0.93

Reduced Model

Overall test 0.10 1 10.58 0.001 S_BL

-2.11 0.71 10.58 0.001 **

** p<0.01

(8)

Table 10. Multiple logistic regression on the probability of escalation.

X factors DF b SE L-R χ

²

p> χ

²

Full Model

Overall test 6

15.80 0.01 C_HON9 -1.10 0.73 2.34 0.13 C_VOL

-0.74 0.74 1.02 0.31 C_SLC8E

0.31 0.73 0.18 0.67 LO_W 1.00 0.53 3.72 0.05 S_BL 1.09 0.71 2.51 0.11 B_BL -0.30 0.16 3.73 0.05

Reduced Model

Overall test 2 8.67

0.01 LO_W

1.13 0.50 5.30 0.02 * B_BL

-0.30 0.15 4.61 0.03 *

* p<0.05

(9)

Table 11. Multiple logistic regression on the probability of the bigger contestants winning the escalations.

X factors R DF b SE L-R

²

χ

²

p> χ

²

Full Model

Overall test 0.29

6 10.81

0.09 C_HON9 -3.30 2.01

3.06 0.08 C_VOL

-0.38 1.62 0.05 0.82 C_SLC8E

-0.64 1.29 0.25 0.62 LO_W

2.48 1.21 5.54 0.02 * S_BL

-1.91 1.48 1.78

0.18 B_BL

0.76 0.42 4.30

0.04 *

Reduced Model

Overall test 0.21

2 10.56

0.03 LO_W

2.33 1.14 5.37 0.02 * B_BL

0.71 0.40 4.30

0.04 *

* p<0.05

.

(10)

Table 12. Results of multiple linear regression on the latency to first attack

¹

(Ln transformed).

X factors

R_adj²

DF b SE F ratio p>F

Full Model

Overall test 0.03 6

1.33 0.26 C_HON9 -0.15 0.37 0.16 0.69 C_VOL

-0.19 0.38 0.23 0.63 C_SLC8E

-0.31 0.38 0.66 0.42 LO_W

-0.50 0.27 3.41 0.07 A

²

_BL

0.11 0.08 2.05 0.16 R

³

_BL

0.19 0.09 4.26 0.04 *

Reduced Model

Overall test 0.02 1

2.85 0.10 R_BL

0.12 0.07 2.85 0.10

1

：the latency to first attack is the period between 1

^st

display of either one contestants and 1

^st

attack of the attacker in the dyadic contest

2

：A stands for Attacker

3

：R stands for Attack Receiver

* p<0.05

Table 13. Relationships between sizes and the attack rate by winner.

X factors r p

_s

(11)

Table 14. Effects of the body sizes of the two contestants on different assessment indices. Size effects on 3 indices(contest duration, contest intensity and attack frequency) have the same pattern as predicted in assessment models;

although some of them were insignificant.

Indices S_BL

¹

B_BL RD_BL correlation

²

p correlation p correlation p

contest duration ＋ 0.64 － 0.25 － 0.21 contest intensity

＋ 0.03 － 0.005 － 0.0007 display frequency

＋ 0.48 ＋ 0.72 ＋ 0.94 attack frequency

＋ 0.05 － 0.12 － 0.05

1

：S stands for Smaller contestant, B stands for Bigger contestant and RD stands for Relative Size Difference between two contestants. BL stands for Body Length as size measurement.

2

Table 1. Predicted relationships between various contest behaviors and the sizes of the two opponents from the assessment models and the non-assessment models.