. 2021 Dec 16;25(1):103529. doi: 10.1016/j.isci.2021.103529

Table 3.

Results from the fixed-factor structure of the linear-mixed models for mosquitofish, with life-history (body condition and shape) and fertility traits (sperm number and velocity in males, and egg number and weight in females) as dependent variables

Model	Mean sq.	df	F	p
Body condition (K; g mm⁻³ 10⁴)

Treatment	0.019	1,57	4.070	0.048
Trial	0.004	1,81	0.791	0.376
Sex	0.031	1,66	6.465	0.013
Tank	0.007	10,57	1.502	0.162
Body weight	0.051	1,94	1.075 10¹	0.001
Treatment × trial	0.001	1,67	0.268	0.607
Treatment × sex	0.002	1,57	0.347	0.558
Trial × sex	0.257	1,74	5.373 10¹	<0.001
Treatment × trial × sex	0.008	1.67	1.668	0.201

Body shape ♂ (RW2)

Treatment	0.001	1,23	4.527	0.035
Trial	0.002	1,38	1.446 10¹	0.003
Tank	0.001	10,24	4.257	0.003
Body size	<0.001	1,36	0.006	0.593
Treatment × trial	0.001	1,34	5.197	0.030

Body shape ♀ (RW2)

Treatment	0.001	1,57	2.973	0.126
Trial	0.004	1,57	1.858 10¹	0.004
Tank	<0.001	10,57	2.259	0.055
Body size	<0.001	1,57	<0.001	0.838
Treatment × trial	<0.001	1,57	0.051	0.838

Sperm number ♂ (counts; 10⁶)

Treatment	4.880 10¹	1,12	5.653	0.035
Body size	0.265	1,12	0.031	0.864
Tank	0.588	1,12	0.068	0.799

Sperm velocity ♂ (μm s⁻¹)

Treatment	0.050	1,13	<0.001	0.990
Body size	6.924 10¹	1,13	0.244	0.630
Tank	2.903 10³	1,13	1.022 10¹	0.007

Egg number ♀ (counts)

Treatment	1.816 10¹	1,7	5.456	0.052
Body size	6.044 10¹	1,7	1.816 10¹	0.004
Tank	6.738	7,7	2.024	0.186

Egg weight ♀ (mg)

Treatment	6.738 10¹	1,21	7.458	0.012
Body size	2.190 10²	1,21	2.424 10¹	<0.001
Tank	1.518 10¹	10,21	1.680	0.152

Treatment (non-exposed and robot-exposed), body size/weight, and mesocosm tank are included as fixed effects in each model. Models on body shape (Relative Warps score 2, RW2) also included trial and treatment × trial as covariates, because body shape was measured twice for each individual during the study—before and after exposure to the robot—and false discovery rate-adjusted p values are presented. Body condition was tested on males and females together, and the model also included sex and its interaction with treatment and trial. Random intercepts (individual ID) are included in body condition and shape models, and they are absent in models for sperm and eggs as fertility traits were measured once per individual, on sexually mature individuals only. Analysis of variance was performed with Satterthwaite's method. Significant results are in bold.