Table 2

Results of generalized linear model of host mortality after exposure to N.whitei using binomial error distribution. A)

Factor

Df

Deviance

Resid. Df

Resid. Dev

P(> |Chi|)


Null deviance

1125

1276.171

Host line

4

91.71

1121

1184.459

< 0.001

Parasite isolate

4

45.98

1117

1138.484

< 0.001

Selection regime

1

40.12

1116

1098.366

< 0.001

Line:Parasite

16

8.25

1100

1090.116

0.941

Line:Selection

4

4.73

1096

1085.386

0.316

Parasite:Selection

4

15.41

1092

1069.976

0.004

Line:Parasite:Selection

16

27.58

1076

1042.396

0.035

Results of post hoc contrast analysis of mortality

Within host (coevolved lines)

Within parasite (coevolved lines)

Within host (control lines)

Within parasite (control lines)

Line/Isolate

Z value

P

Z value

P

Z value

P

Z value

P

1

-2.54

0.049

-0.65

0.974

-2.65

0.039

2.03

0.194

3

1.81

0.272

-0.55

0.988

1.64

0.378

-1.96

0.225

4

0.12

0.999

2.14

0.153

2.93

0.016

2.38

0.084

5

0.56

0.969

0.58

0.948

-1.19

0.686

-0.03

1.000

6

2.75

0.028

-0.08

1.000

-1.38

0.551

-1.72

0.361


Levels of significance for the GLM-model fits were tested using analysis of deviance with chi-square distribution. Post hoc test results for contrasts between sympatric and allopatric combinations are shown below. Positive Z values indicate combinations where local antagonists show higher mortality than foreign antagonists.

Bérénos et al. BMC Evolutionary Biology 2012 12:11   doi:10.1186/1471-2148-12-11

Open Data