*3.3. Estimation of the Stability of the Reference Genes under Different Experimental Conditions*

To identify the optimal reference genes for the normalization of gene expression analysis in *L. megaphylla*, the stability of the 14 candidate genes was assessed by four different algorithms. The RefFinder software was used for overall ranking.

#### *3.4. Delta Ct Method Analysis*

The delta Ct (ΔCt) method ranks the stability of candidate reference genes based on the relative expression levels of "gene pairs" in each group of sample comparisons, while the mean standard deviation of gene expression differences (STDEV) is inversely proportional to its stability using the raw Ct value [34]. The stability of the transcript levels of each candidate reference gene was evaluated based on the STDEV value. The gene with the minimum STDEV value was regarded as the most stably expressed gene. The results demonstrated that the optimal reference genes were different in the different experimental sets (Figure 2). In seedling samples, *helicase-15* and *PAB2* showed the lowest ΔCt values (0.43), indicating the most stability (Figure 2A), while *UBC28* (0.31) was the most stably expressed gene in adult trees (Figure 2B). In other tissues, *helicase-15* (0.51) and *UBC28* (0.57) were the most stable reference genes (Figure 2C), which is consistent with the results for

seedlings and adult trees. *ACT7* (0.58) and *UBC36* (0.59) were more stable across different leaf developmental stages (Figure 2D). Data analyses from the entire growth cycle indicated that *ubiquinone* (0.67) and *UBC36* (0.68) were the most stable (Figure 2E). *UBC36* showed good stability in all three cold stress sets (Figure 2F–H). *PAB2* (0.69) had the highest stability under heat treatment (Figure 2I). Across all temperature stresses, *PAB2* (0.6) was the most stably expressed gene (Figure 2J). For all samples, ubiquinone (0.74), *EF2* (0.75) and *PAB2* (0.76) showed the most stability (Figure 2K). *TUA* had relatively higher Ct values, indicating that it was the least stable reference gene in most of the experimental sets (Figure 2A–K).

**Figure 2.** Ranking of expression stability of the 14 candidate reference genes in *L. megaphylla* using ΔCt analysis. Genes are listed across bottom of each plot in order of increasing stability from left to right. Results from (**A**) different seedling tissues; (**B**) different adult tree tissues; (**C**) all seedling and adult tree tissues; (**D**) 16 leaf developmental stages; (**E**) entire growth cycle including all different seedlings and adult tree tissues as well as 16 leaf developmental stages; (**F**) cold treatment for 7 days; (**G**) cold treatment for 24 h; (**H**) cold treatments including 7 days and 24 h; (**I**) heat treatment for 24 h; (**J**) different temperature treatments including cold and heat; (**K**) total samples.

#### *3.5. geNorm Analysis*

The stability of the reference genes was ranked by calculating the average expression stability values (M value) using the geNorm program, taking into account only similar intergroup variation [35].

Genes with an M value less than 1.5 were considered stably expressed, with smaller M values indicating a more stable gene [28]. The geNorm analysis results for the 14 candidate genes in the different experimental sets are shown in Figure 3A–K. The M values of the 14 candidate reference genes were less than 1.5 under the different experimental conditions (Figure 3). *EF2* and *PAB2* had the highest stability in seedlings, with M values of 0.012 (Figure 3A), while *CYP26-2* (0.076) and *helicase-15* (0.076) were most stably expressed in the adult tree (Figure 3B). Between the different tissue sets, *PAB2* and *helicase15* were the optimal candidate genes, which is similar to the results in seedlings and adult trees (Figure 3C). The genes *PAB2* and *helicase-15* had the lowest stability values (0.285), which is consistent with the results of the ΔCt analysis. The two most stably expressed genes among the different leaf developmental stages were similar to those for the entire growth cycle, namely, *UBC36* and *UBC7* (Figure 3D,E). These results are also similar to those of the ΔCt analysis. *UBC36* was more stable than the other candidate reference genes under the cold treatments (Figure 3F–H) and was the same regardless of whether the cold treatment lasted for 7 days or 24 h. *GAPDH* (0.269), *CYP20-2* (0.269), and *PPR* (0.306) were more stable than the other candidate genes under heat treatment (Figure 3I). However, *PPR* and *PAB2* (0.343) exhibited the strongest stability under temperature stress (Figure 3J). These results are consistent with the results of the ΔCt analysis. *UBC36* and *UBC7* (0.406) showed the strongest stability in all samples (Figure 3K). In contrast, *TUA* and *PPR* were the least stable across most sets.

Best practices include using multiple reference genes as internal controls for standardization to improve the accuracy of RT-qPCR data [28,49]. The number of optimal genes for standardization of the different datasets from *L. megaphylla* was calculated using the Vn/n+1 function of geNorm, with a threshold of 0.15 (Figure 4). Interestingly, the values of V2/3 were less than 0.15 for most experimental groups (0.033, 0.045, 0.111, 0.112, 0.126, 0.034, 0.044, 0.052, 0.098, and 0.129) except for the 'all samples' group, as shown in Figure 4. This suggested that two was the optimal number of reference genes for each type of samples. However, for the 'all samples' set, the V2/3 and V3/4 values were greater than 0.15, and it is not until four reference genes are used (V4/5 value of 0.117) that the values is less than 0.15. Thus, at least four genes are required to obtain accurate results across many tissues and treatments.
