The quality and completeness of the reference database have a direct impact on the accuracy of forage plant identification, thereby influencing the level of conservation and management of wildlife resources. In our research, target amplification was subjected to first-generation sequencing to assemble a local reference database using chloroplast trnL metabarcoding. We found that the primers c-h outperformed g-h as a universal DNA metabarcoding and 162 valid choloroplast trnL sequences were submitted (GenBank ID: PP081756 - PP081917), which exhibited an obvious preference for A and T nucleotides (60.49%). The haplotype diversity (Hd), nucleotide diversity (Pi) and average number of nucleotide differences (K) of these trnL sequences were 0.978, 0.0484 and 4.743, respectively. To assess the availability of the local database in identifying the diet of South China sika deer (Cervus nippon kopschi), high-throughput metabarcoding sequencing and BLAST analysis were performed. Ultimately, 25 forage plant species were identified, belonging to 19 families and 25 genera. Shrubs and herbaceous plants, such as Potentilla freyniana, Persicaria perfoliata, Rosa laevigata and Ardisia japonica etc, dominated the forage plants. This study established a local trnL reference database that holds immense value for the forage plant identification and nutritional evaluation for sika deer and other sympatric herbivores, as well as the conservation and management of biodiversity.