**3. Results**

## *3.1. Field Experiment*

The presence and development of *M. anisopliae* was detected both under compost and soil cover managemen<sup>t</sup> systems. While no differences in crop weight (an average of 1600 gr/plant) were detected between treatments (Table 1), there were differences in tuber damage, and significantly higher damage (only at level 2—deep damage) was detected when the crop was covered with agro-foil, and not treated with compost and fungus. These damage figures, however, were not different from those obtained by agro-foil and compost cover, with or without fungal treatment, and textile cover without compost and *M. anisopliae* (Figure 2, Table 2). Altogether, a tendency of lower damage of Melolontha larvae was detected when sweet potato was treated with *M. anisopliae* strain NCAIM 362 and covered by agro-textile (Figure 2, Table 2).

**Table 1.** Statistical analyses of tuber weight in the open field experiment (F values below and *p* values above line), data were compared using average tuber weight/plant/compos<sup>t</sup> application/soil cover/block (*n* = 22).



**Table 2.** Statistical analyses of tuber damages in the open field experiment (*F* values below and *p* values above line), the effect of *M. anisopliae* on deep tuber damages (level 2) were compared using average damage/tuber/plant/compos<sup>t</sup> application/soil cover/block (*n* = 22). Bold numbers represent statistically significant *p* values.

## *3.2. Greenhouse Experiment*

Again, the presence and development of *M. anisopliae* was detected in all containers M<sup>+</sup>. While no differences in crop weight (an average of 1700 gr/plants) were observed (Table 3, Figure 3A), variations in *M. melolontha* larvae survival and damage were detected between treatments (Figure 3B,C). Significantly lower numbers of survived larvae were detected in plots treated with α-cypermethrin (CipermP + ControlP+ *U* = 3.2, *p* < 0.01; CipermP + MetarhP+ *U* = 3.0, *p* < 0.01) and no differences were detected between *Metarhizium* treatment and control (MetarhP + ControlP+ *U* = 0.67, *p* < 0.23), where generally half of the larvae died before the end of the experiment. The numbers of dead larvae were higher in plots treated with α-cypermethrin (CipermP + ControlP+ *U* = 4.1, *p* < 0.01; CipermP + MetarhP+ *U* = 3.9, *p* < 0.01) and again no differences were detected between *Metarhizium* treatment and control (MetarhP + ControlP+ *U* = 0.88, *p* < 0.56) (Figure 3B). Signs of fungal infection among larvae were hardly observed at all (an average of one infected larva was found in 10 *Metarhizium* treated pots) at the end of the experiment in M<sup>+</sup> treatments, and this did not make statistical analysis possible (Figure 3B). The damage rate of tubers also varied between treatments. Significantly lower damage rates were detected in pots with α-cypermethrin (CipermP + ControlP+ *U* = 5.2, *p* < 0.01; CipermP + MetarhP+ *U* = 3.9, *p* < 0.01), while no differences between *Metarhizium* treatment and control were observed (MetarhP + ControlP+ *U* = 0.66, *p* < 0.45) (Figure 3C).

**Figure 3.** (**A**–**C**) Crop weight (g) withα-cypermethrin (red bars), control (blue bars), and fungal treatment (grey bars) (**A**); average number of alive, dead, and infected *Melolontha* larvae (**B**); and damages under different treatments and control (**C**). Crop weight data was analyzed using ANOVA, followed by Tukey's HSD test using average tuber weight (g)/plant/treatment/block (*n* = 35). For bulb damage and *M. melolontha* larval survival data Kruskal-Wallis test was used, followed by a Mann-Whitney using average values for plants/treatments/blocks (*n* = 35)) and average values of alive, dead, and infected larvae (average number/plant/treatment/block (*n* = 35)). Bars represent standard errors. Different letters (a,b) means statistical significant differences.


**Table 3.** Statistical analyses of tuber weight in the greenhouse. \* Significance: same letters indicate no significant differences (Tukey HSD test).

#### *3.3. Chemical Composition Assay of Sweet Potato Soil*

Representative elemental composition of sweet potato soil results was averaged out from five measurement each. Since a small amount of sample was used, results below 0.5 are considered as qualitative information, given that these elements only appear in trace amount. No differences in the chemical composition of sweet potato soil were detected between treatments (Table 4).

**Table 4.** Representative elemental composition of sweet potato soil. Results were obtained by calculating the average of five measurements. \* Significance: same letters indicate no significant differences (Tukey HSD test).


#### *3.4. Microbial Community and Biological Activity in Sweet Potato Soil*

A total of 697,221 high-quality bacterial 16S rRNA gene sequences were obtained from the samples (38,734 ± 9 336 reads per sample). Good's coverage values were higher than 0.94 in all cases, whic indicated that sequencing depth was sufficient to recover all major bacterial taxa (Figures S5–S7, and Table S1). The average length of sequences was ~450 nt, which allowed genus-level taxon identification. No significant differences between bacterial community were detected when the soil was treated with insecticide or the fungus, and control with and without *M. melolontha* larvae (Welch *F* test *F* = 0.0006, *df* = 22.37, *p* < 0.9) (Figure 4). Also, no significant differences in soil biological

activity were detected when treatments and control were compared Welch *F* test *F* = 0.03, *df* = 6, *p* < 0.76 (Figure 5).

**Figure 4.** Soil bacterial community analysis performed on amplicon sequencing of the 16S rRNA gene. Total genomic DNA was extracted using the DNeasy PowerSoil Kit (Qiagen), a part of the 16S rRNA gene was amplified with primers containing the Bacteria-specific sequences Bakt\_341F (5-CCTACGGGNGGCWGCAG-3; Herlemann et al., 2011) and Bakt\_805NR (5-GACTACNVGGGTATCTAATCC-3). DNA sequencing was performed on an Illumina MiSeq platform using MiSeq standard v2 chemistry as a service provided by the Genomics Core Facility RTSF, Michigan State University, USA. Cont.II represents control P<sup>−</sup>, Cont.I represents control P<sup>+</sup>, Meth.II represents Metarhizium P<sup>−</sup>, Meth.I represents Metarhizium P+, Cyp.II represents insecticide P− treatments, Cyp.I represents insecticide P+. Only data of 3 sample/treatment are presented.

**Figure 5.** Soil biological activity with insecticide, *M. anisopliae* treatment, and control. The concentration of fluorescein was determined with spectrophotometer (*PG Instruments T60 UV*/*VIS Spectrophotometer*) on 490 nm. The FDA enzyme activities of the soil probes performed in μg/g soil/h. Same letters indicate no significant differences (Tukey HSD test).
