*2.4. Determination of* alp *and* nifH *Genes Copy Numbers*

The abundance of phosphate mineralizing and nitrogen fixing bacteria were quantified by targeting the *alp* and *nifH* genes, respectively, in a quantitative real-time PCR (qPCR). The reactions were performed in an ABI Prism 7300 Cycler (Applied Biosystems, Germany) in 25 μL reaction mixtures containing 1 <sup>×</sup> SYBRTM Select Master Mix (Applied Biosystems, Germany), 0.4 <sup>μ</sup>g/μL BSA, 1 μL of target DNA (approximately 50 ng) and 0.5 μM of primers. For the quantification of *alp* gene, primers ALPS-F730/ALPS-R1101 [27] were used. The amplification conditions were 2 min at 50 ◦C, 10 min at 95 ◦C, and 40 cycles consisting of 30 s at 95 ◦C, 60 s at 57 ◦C, and 30 s at 72 ◦C. Standard curves were prepared by serial dilutions (from 108 to 10<sup>1</sup> gene copy numbers/μL) of plasmid DNA (pTZ57R/T; CloneJET™ PCR Cloning Kit - Thermo Fisher Scientific, Waltham, MA, USA) containing the cloned *alp* gene PCR products from bacterial strain 006.30 [34]. For the quantification of the *nifH* gene, primers FGPH19 [28] and PolR [29] were chosen as described by Taketani, et al. [35]. The amplification conditions were an initial denaturation step at 94 ◦C for 10 min, followed by 40 cycles of 1 min at 94 ◦C, 27 s at 57 ◦C, and 1 min at 72 ◦C. Standard curves were obtained using serial dilutions of the *Escherichia coli*-derived vector plasmid JM109 (Promega, Madison, WI, USA) containing a cloned *nifH* gene from *Bradyrhizobium liaoningense*, using 102 to 10<sup>7</sup> gene copies/μL. Significant differences between samples were tested in pairwise comparisons using the Tukey test (*p* ≤ 0.05; SAS 9.3; SAS Institute Inc., Cary, NC, USA).

#### **3. Results**

### *3.1. Structure of Phosphate Mineralizing and Nitrogen Fixing Bacterial Communities Analyzed by DGGE*

The effects of plant growth stage and genotypes on the structure of the phosphate mineralizing and nitrogen fixing bacterial communities present in the rhizosphere of tuberous roots of the genotypes IPB-149, IPB-137, and IPB-052 were analyzed by DGGE based on *alp* gene and *nifH* gene fragments, respectively, which were amplified from TC-DNA.

For the DGGE profiles based on the *alp* gene, principal component analysis (PCA) showed the formation of two separate groups: IPB-137\_6M and IPB-149\_3M (Figure 1A). The fingerprints of the phosphate mineralizing bacterial community in the rhizosphere were statistically analyzed using the permutation test (*p* ≤ 0.05). The sampling time had a significant effect on the structure of the bacterial community in the IPB-137 genotype (Table 1). Moreover, the results showed that the genotypes also significantly influenced the structure of the phosphate mineralizing bacterial community in IPB-137 and IPB-149 (Figure 1B). These communities were statistically different for both sampling times (Table 1). Therefore, the different genotypes and, to a minor extent, the time influenced the phosphate mineralizing bacterial community present in the tuberous roots of sweet potato studied here. The DGGE fingerprint analyses showed that the structure of both bacterial communities were complex with high variability among the replicates (Figure S1A,B). The samplings of IPB-137 and of IPB-149 were grouped separately (per genotype) with more than 60% similarity after UPGMA cluster analysis (Figure S1A). A high similarity (more than 80%) was observed within IPB-137\_6M samples, suggesting a slight influence of the plant age, as was observed in PCA.

**Figure 1.** Principal component analyses (PCA) was conducted using Denaturing Gradient Gel Electrophoresis (DGGE) patterns of phosphate mineralizing bacteria based on the *alp* gene from the rhizosphere of three different sweet potato genotypes (IPB-149, IPB-137, and IPB-052) sampled after three and six months after planting (t1, \_3M and t2, \_6M, respectively). (**A**) and (**B**) highlight the grouping observed within the sampling time (t1 × t2) and within the different genotypes, respectively.

For the DGGE profiles based on *nif* gene, the influence of the sampling time was not evident in PCA, except for some of IPB-137\_6M replicates (Figure 2A). The results also showed that the genotypes affected the structure of the nitrogen fixing bacterial community (Figure 2B). The fingerprints of the nitrogen fixing bacterial community in the rhizosphere were statistically analyzed using the permutation test (*p* ≤ 0.05). No significant effect of the sampling time on the structure of the nitrogen fixing bacterial community within the different genotypes was observed (Table 1). In contrast, the structure of the nitrogen fixing bacterial communities in sweet potato genotypes IPB-149 and IPB-137 were statistically different from those in IPB-052 after six months of planting (Table 1). The visual interpretation of the DGGE profiles corroborated the PCA. UPGMA cluster analysis of the DGGE profiles based on

*nifH* gene showed a high similarity within the replicates of IPB-137\_6M. The replicates of IPB-052\_6M formed a group that was separate from the other genotypes (Figure S1B).

**Table 1.** Dissimilarity (d value in %) of rhizosphere bacterial fingerprints (DGGE) based on *alp* and *nifH* genes with comparisons between samplings (t1, \_3M and t2, \_6M) or among IPB-149, IPB-137, and IPB-052 genotypes.


**Figure 2.** Principal component analyses (PCA) were conducted using DGGE patterns of nitrogen fixing bacteria based on the *nifH* gene from the rhizosphere of three different sweet potato genotypes (IPB-149, IPB-137, and IPB-052) sampled after three and six months after planting (t1, \_3M and t2, \_6M, respectively). (A) and (B) highlight the grouping observed within the sampling time (t1 x t2) and within the different genotypes, respectively.
