*3.2. Codon Usage Analysis of PCGs*

In the *P. vulgaris* mitogenome, most of the PCGs use ATG as the start codon, while *mttB* and *nad1* start with ACG (C to U RNA editing on the second site is presumed) as the start codon (Table 3). Four types of stop codons were found in the PCGs: (1) TAA (15 genes; *atp4*, *atp8*, *atp9*, *cox1*, *nad1*, *nad2*, *nad3*, *nad4L*, *nad5*, *nad6*, *nad9*, *rpl5*, *rpl16*, *rps1*, and *rps4*), (2) TGA (10 genes; *atp1*, *ccmB*, *ccmC*, *ccmFN*, *cox3*, *matR*, *mttB*, *nad4*, *rps10*, and *rps12*), (3) TAG (5 genes; *atp6*, *cob*, *nad7*, *rps3*, and *rps14*), and (4) CGA (*ccmFC*; C to U RNA editing on the first site is presumed). As shown in Figure 3, the codon usage analysis revealed that leucine (Leu) and serine (Ser) are the most frequently-used amino acid residues, while cysteine (Cys) and tryptophan (Trp) are the least-used amino acid residues in the plant mitochondrial proteins. By comparison of the composition of *P. vulgaris* with other angiosperms plants, we found that the distribution of amino acid residues across the mitochondrial proteins are very similar in angiosperms (Figure 3). In addition, most of the amino acid residues were found to be very conserved between angiosperms (*P. vulgaris*, *G. max*, *L. japonicus*, *V. radiata*, *V. faba*, *A. thaliana*, *C. lanatus*, and *T. aestivum*) and gymnosperms (*Ginkgo biloba* and *Cycas taitungensis*), except for five of them (Phe, Leu, Pro, Arg, and Ser).

The relative synonymous codon usage (RSCU) analysis for the *P. vulgaris* mitogenome is shown in Figure 4, which indicates that all codons are present in the PCGs. Excluding the termination codons, the 31 PCGs in the *P. vulgaris* mitogenome consist of 9545 codons in total. Additionally, the codon usage showed that the RSCU values of the NNT and NNA codons are higher than 1.0 except for Ile (AUA) and Leu (CUA; Figure 4), suggesting a strong As or Ts bias in the third codon position of *P. vulgaris* mitochondrial PCGs, which is a very common phenomenon observed in all studied mitogenomes (Table S3). The codon usage pattern of *P. vulgaris* mitogenome is highly consistent with two other papilionoid legumes. The distributions of some codons encoding Pro (CCU, CCA, and CCG) differ between dicotyledons (*P. vulgaris*, *G. max*, *V. angularis*, *C. lanatus*, and *A. thaliana*) and monocotyledons (*T. aestivum*), and some codons (UCG, AGU, AGC, CCU, CCG, ACG, CGG, and AGA) are distributed differently between angiosperms and gymnosperms.


**Table 3.** Gene profile and organization of PCGs in *P. vulgaris* mitogenome.
