This study investigated the interactive effects of different light conditions and weed control methods on the medicinal compound composition and rhizosphere bacterial community structure of Epimedium sagittatum. A completely randomized block design was employed, incorporating four treatments: full light with manual weeding (LN), shade with manual weeding (SN), full light with weed-control fabric mulch (LG), and shade with mulch (SG). Active compound levels in two-year-old plants were quantified using HPLC, and rhizobacterial diversity was assessed via high-throughput sequencing. The results indicated that the SG treatment significantly enhanced the photosynthetic efficiency and yielded the highest levels of epimedin C and total active compounds. In contrast, the SN treatment fostered a beneficial rhizosphere environment—characterized by increased pH, ammonium nitrogen (NH₄⁺-N), bacterial diversity, and the abundance of Flavobacterium—which supported the highest production of epimedin B and icariin. Redundancy analysis confirmed that these microbial shifts were primarily driven by soil pH, nitrate nitrogen (NO₃⁻-N), and shading. Furthermore, while stochastic processes governed bacterial community assembly, deterministic selection intensified across the treatments from LN to SG. Collectively, our findings demonstrate that light and mulching can be strategically tailored to manipulate the plant–soil-microbe system, thereby enabling precise modulation of the medicinal quality of E. sagittatum.

This study aims to clarify the codon usage bias and influencing factors of protein-coding genes in the chloroplast genome of the medicinal plant Corydalis yanhusuo. The chloroplast genome sequence of C. yanhusuo was obtained by resequencing, approximately 50 protein-coding genes were screened, and the nucleotide composition and codon usage patterns were calculated and analyzed using CodonW 1.4.2 and EMBOSS software. The results showed that the total guanine and cytosine (GC) content of codons in the chloroplast genome of C. yanhusuo was 40.06%, and the GC contents at the third, second, and first codon positions (GC3, GC2, and GC1) were 32.12%, 40.21%, and 47.84%, respectively, indicating that codons in the chloroplast genome of C. yanhusuo preferentially used adenine (A) or uracil (U). The effective number of codons (ENC) ranged from 42.87 to 61.00, with an average value of 50.54, indicating weak codon usage bias. A significant positive correlation existed between the GC content at the third codon position (GC3) and ENC, showing that codon bias was mainly affected by the third base. Neutral plot, ENC-plot, and PR2-plot analyses showed that the codon bias of the chloroplast genome of C. yanhusuo was mainly influenced by natural selection. Sixteen optimal codons—UUA, AUU, GUU, GUA, UCU, AGU, CCU, ACU, GCU, CAA, AAA, GAU, UGU, CGU, CGA, and GGU—were finally determined based on the relative synonymous codon usage analysis of high-frequency and highly expressed codons, all of which preferentially ended with A/U. Overall, this study reveals the codon usage bias of the chloroplast genome of C. yanhusuo and its influencing factors, and provides a theoretical basis for chloroplast genetic engineering and phylogenetic research.