Non-Mendelian assortment of homologous autosomes of different sizes in males is the ancestral state in the Caenorhabditis lineage

Organismal genome sizes vary by six orders of magnitude and appear positively correlated with organismal size and complexity. Neutral models have been proposed to explain the broad patterns of genome size variation based on organism population sizes. In the Caenorhabditis genus, hermaphrodite genomes are smaller than those of gonochoristic species. One possible driving force for this genome size difference could be non-random chromosome segregation. In Caenorhabditis elegans, chromosome assortment is non-independent and violates Mendel’s second law. In males, the shorter homologue of a heterozygous autosome pair preferentially co-segregates with the X chromosome while the longer one preferentially co-segregates with the nullo-X (O) chromosome in a process we call “skew”. Since hermaphrodites preferentially receive the shorter chromosomes and can start populations independently, their genome size would be predicted to decrease over evolutionary time. If skew is an important driver for genome size reduction in hermaphroditic Caenorhabditis species, then it should be present in all congeneric species. In this study, we tested this hypothesis and found that skew is present in all eight examined species. Our results suggest that skew is likely the ancestral state in this genus. More speculatively, skew may drive genome size patterns in hermaphroditic species in other nematodes.

Supplementary Text S1. The skew transmission bias ratio difference for mIs10 between the WHR10 and PD4793 strains is likely due to genetic modifiers.
The transmission bias ratios (TBRs) for mIs10 in strains WHR10 (3.86, this study) and PD4793 (6.55, previous study 1 ) differed. PD4793 was the original strain obtained from the CGC and is 3x outcrossed. We outcrossed the PD4793 strain to N2 an additional 10 times to obtain WHR10.
We considered four possible explanations for the TBR differences. First, skew TBR is variable, perhaps affected by subtle environmental factors. In this case, the two sets of crosses were conducted in different countries (Taiwan and Switzerland). Second, the tester hermaphrodite strains were different; this study used BRC189 (unc-119(ed9) III; ttTi5605 II) while the previous study used CB184(dpy-13(e184)). A third possibility is that the mIs10 insertion allele changed in size. Because there is a correlation between insert size and TBR 1 , the presumption would be that WHR10 carries a shorter allele. Finally, there were genetic modifiers that were removed or introduced during outcrossing.
We tested the first two possibilities by reordering the original PD4793 strain from the CGC and then re-assaying skew for both PD4793 and WHR10. Specifically, we crossed heterozygous mIs10/+ males to Unc hermaphrodites (BRC189) and scored their progeny. We found that PD4793(new) had a TBR of 6.01 which was not different from the PD4793(old, TBR: 6.55, P = 0.29, χ 2 value = 1.09, 1 df). Similarly, WHR10(new, TBR: 4.33) was not different from the previous WHR10(old, TBR: 3.86, P = 0.19, χ 2 value = 1.65, 1 df). These results indicate that skew TBR is stable and not dependent on the tester strain. For the third case, we estimated insert sizes using qPCR 1 in both strains (4 technical replicates each) and found no difference (P = 0.12, Welch's two-sample t-test = 1.88, df=4.575).
Based on these results, we conclude that one or more skew modifier mutations were removed or introduced during our outcrossing of mIs10.

Supplementary Text S2. Insertion size and transmission bias for the interspecies comparisons.
For the interspecies comparisons associated with Supplementary Figure 1, we could not use the same set of reference primers for DNA real-time qPCR due to sequence divergence among the species. Instead, we first identified putative highly conserved elements using the iHCE software package 2 (v4.34; --nompi, all other settings default). For the input dataset, we downloaded the genomes for C. elegans (PRJNA13758), C. briggsae (PRJNA10731), C. remanei (PRJNA53967), and C. sinica (PRJNA194557) from Wormbase (all release WS260). From the output, we selected the 9 longest single copy autosomal loci (all >80 bp) for additional sequence inspection across Caenorhabditis species using blastn 3,4 . We chose two sequences for preliminary tests based on high nucleotide identity across most species and found that only LG2_6028 produced acceptable qPCR performance (i.e., >95% PCR efficiency and single peak melting curves).
The sequence for LG2_6028 is: TCAACGGCGCGAGAGATCGGGATAAAGATCGCGTCCGGAACCCATGGTCTGAGTTGAGTGTGTGGTGGTGG AGGACGGAACCAGGCCAT We conducted genomic DNA real-time qPCR assays on bla 5 (targets the ampicillin resistance marker on plasmids) and LG2_6028 for six strains from four species (C. elegans, C. briggsae, C. tropicalis, and C. brenneri). We could not test C. portoensis because qPCR tests of the reference locus were not reliable. We did not test C. remanei because of a repeated bacterial contamination issue.
We calculated copy number using the Pfaffl method that adjusts for PCR efficiency 6 and simplified for single 'ΔCt'. Then, we estimated transgene sizes by multiplying the 'unit' insert size by the approximate copy number. A 'unit' was defined as the average of the injected plasmid sizes, except for mfIs42, which was the sum of the plasmid and the associated PCR product sizes. Transgene sizes were then normalized to genome size. Finally, we tested for a correlation between the TBRs and normalized transgene sizes using the Spearman's rank correlation test.

Supplementary Figure S1. Scatter plot of transmission bias ratio and relative insert size.
There is no correlation between TBR and insert size across the six samples (P = 0.41, Spearman's rank correlation test) possibly indicating that species-specific factors modify TBR. The three C. briggsae data points (green) may be consistent with a within species positive correlation. Transgene names are indicated below in italics.

Supplementary Figure S2. Full transmission patterns of extrachromosomal transgenes.
Percentages of individuals carrying (gray bars) or not carrying (white bars) extrachromosomal arrays. Species names are indicated above the plots; transgene names are indicated below in italics. Error bars are 95% confidence intervals. P-values, χ 2 tests assuming random segregation of the extrachromosomal transgenes by sex; ***, P < 0.001; **, P < 0.01; n.s., not significant (P > 0.05). TBR, transmission bias ratio; N_total, total number of individuals scored; N_males, number of males tested. See Figure 2 for plots of only individuals inheriting the extrachromosomal array. P-values are identical between these two figures because the χ 2 tests were conducted on this dataset.

Supplementary Figure S3. Transmission patterns of the antEx50 and antEx51 extrachromosomal transgenes.
Percentages of hermaphrodites (white) and males (gray) inheriting the extrachromosomal array. Unc progeny were not scored because self-progeny of the tester unc-119 strain could not be excluded. Transgene names are indicated above barplots in italics. Error bars are 95% confidence intervals. P-values, binomial tests assuming equal segregation of the extrachromosomal arrays by sex; ***, P < 0.001. TBR, transmission bias ratio; Total_GFP, total number of GFP individuals scored; N_males, number of males tested. Figure S4. Schematic of fem-3 and her-1 oocyte crosses. fem-3 and her-1 XO individuals are phenotypically females and hermaphrodites, respectively. Crosses with males carrying an X-linked RFP will produce six viable and two dead progeny classes. Three viable classes correspond to the "preferred" gamete combinations (red box) and the other three classes correspond to the "anti" gamete combinations (blue boxes).  Figure 3B focus on gfp/gfp individuals. In both models, chromosomes are presumed to segregate randomly in oocytes (see Fig. 3A). The bottom Punnett square can be generalized for any skew TBR (≥1). The relationship between TBR and the XO:XX ratio is: TBR = XO:XX ratio -1. b, Photograph taken through a fluorescence stereomicroscope of dpy-11 her-1 individuals carrying the mIs10 transgene either as a homozygote (gfp/gfp) or heterozygote (gfp/+).