1. Introduction
The olive oil extraction requires a considerable amount of water and produces vast quantities of oil mill wastes (OMW) in a limited period with a significant environmental impact [
1,
2,
3,
4,
5]. The type of OMW produced depends on the extraction system adopted. Indeed, the OMW from the three-phase centrifugation is solid pomace and olive mill wastewater (OMWW). In contrast, only a wet pomace is generated from the two-phase centrifugation oil extraction system [
6,
7,
8]. The OMWW has a low oil content (about 3%). A high water content, mainly deriving from the olive fruits (depending on their stage of ripeness, water availability, and cultivars they contain 40–50% of water), the extraction plants during the washing phase (10–15% of the weight of the processed olives), or, in continuous systems, during the extraction phase, where a possible aqueous dilution of the paste (40–60% of the weight of the processed olives) facilitates the extraction of oil. In the Mediterranean countries, more than 30 × 106 m
3 of OMWW are produced during the harvest and processing season [
8]. The chemical-physical characteristics of the OMWW depend on the climatic conditions of the cultivation area, the cult, the state of ripeness of the olives, and especially the type of processing adopted [
9,
10,
11,
12]. In general, OMWW contains many organic compounds in solution in colloidal state and suspension (i.e., sugars, amino acids, tannins, organic acids, lipids, alcohols, pectins, carotenoids, and phenols) [
11,
12], but also a significant amounts of inorganic salts (i.e., potassium phosphates, sulphates, and chlorides) both as a soluble and insoluble fraction (carbonates and silicates, about 20%), and mineral elements (i.e., potassium, magnesium, calcium, sodium, and iron) [
13,
14]. The amendment with OMWW could supply macronutrients (P, K, and N available) and organic matter to the soil, thus limiting the use of chemical fertilizers, with related economic and environmental advantages [
15,
16,
17]. In fact, since OMWW derives exclusively from the mechanical extraction process, no synthetic chemicals, additives, pathogenic microorganisms, and viruses are contained or could occur for the urban sewage [
18].
On the other hand, OMWW has a high potential for pollution due to the high content of salt, organic matter, phenolic compounds (catechol, hydroxytyrosol, tyrosol, and oleuropein), oxygen with a high biological (BOD) and chemical (COD) content, and high acidity. Considering salt content, the direct diffusion of the OMWW could increase the soil’s electrical conductivity (EC), with different effects on the soil depending on the quantity applied [
8,
13]. Chaari et al. [
19] observed that the application of OMWW over a long period (9 years) at high doses (200 m
3 ha
−1) caused soil salinization. Furthermore, the increase of the C/N ratio in the soil after the application of the OMWW affects microbial activity, and then the mineralization and humification [
8,
17], as well as the organic substance extractable from the water (WEOM), which is well-known to be affected quantitatively and qualitatively after the addition of differently stabilized organic materials [
2,
20,
21,
22]. Phenolic compounds distributed in the upper soil layers by OMWW application are degraded by specific bacteria and yeasts [
13,
17]. Still, a high-dose application is a stressful condition that could reduce their activity [
23]. The acidity of the OMWW temporarily affects the pH of the soil for a high application rate and repeated treatments, showing a buffer capacity after soil disturbances [
8,
23]. Even the long-term repeated application of OMWWs can lead to relevant changes in the chemical and biochemical parameters of the soil, but only in the first period of application [
24]. However, despite the presence of potential phytotoxic compounds, the OMWWs spreading could positively affect the growth performance of wheat, maize, and olive crops [
15,
16,
25,
26].
In many olive-growing countries, i.e., Italy, the agronomic use of OMWW by directly spreading in the field is a widespread method of legal disposal [
1,
27,
28]. Italian Law n. 574/96 (Nuove Norme in Materia di Utilizzazione Agronomica delle Acque di Vegetazione e di Scarichi dei Frantoi Oleari. Gazzetta Ufficiale n. 265, 12 November 1996) integrated with Legislative decree of 6 July 2005 allows the spreading of up to 50 or 80 m
3 ha
−1 year
−1 olive mill wastewater generated by the press or continuous centrifugation systems, respectively [
28]. Such use has taken great interest not only because it is relatively inexpensive and easily feasible but also because it allows giving organic matter to the soil. Intensive farming is a source of soil depletion that worsens the physical, chemical, and biological soil properties, leading to degenerative phenomena, of which erosion and loss of fertility are the most prominent and worrying aspects [
7]. However, the spreading in the field of OMWW should be implemented with rationality since the little stabilized organic materials may inhibit or reduce the development of crops due to (I) the presence of tannins, fatty acids, and phenols, (II) the competition for the nitrogen among the microorganisms of the soil and the roots due to a high C/N ratio, and (III) anoxia of the roots caused by the microorganism consumption of oxygen [
8,
29,
30,
31,
32,
33]. In particular, the high concentration of polymeric phenols could have a bacteriostatic effect on microorganisms and some phytotoxic effects on cultivation. Moreover, the high content of salts and the slightly acidic reaction suggest the need to manage this waste carefully. Information on the influence of OMWW spreading in the biological activity of the soil is lacking [
34]. To the best of our knowledge, few studies deal with the molecular characterization of the microbial communities in OMWW and their effect on the soil microbiome [
5,
35]. Some studies characterized microbial communities with polymerase chain reaction denaturing gradient gel electrophoresis (PCR-DGGE) and polymerase chain reaction temperature gradient gel electrophoresis (PCR-TGGE) from olive mill wastes produced by an anaerobic process of two-phase centrifugation [
35,
36,
37] and from OMWW generated by two olive varieties (
Olea europaea var.
mastoiditis and
O. europaea var.
koroneiki), studied by 16S rRNA clone libraries [
35]. Tsiamis et al. [
38] were also able to identify a cultivar-specific profile of bacterial communities in OMWW with a culture-dependent and -independent approach. Nowadays, next-generation sequencing (NGS), coupled with bioinformatics tools, has made it easier to analyze microbial communities on any matrix, including soils [
39,
40]. To date, few studies characterized microbial communities that inhabit the olive grove’s soil with NGS technologies [
41,
42]. On the other hand, many studies investigated the effect of OMW on soil microorganisms using cultivation-enumeration methods and enzymatic activity [
27,
43,
44,
45,
46,
47]. According to our knowledge, only Federici et al. [
2] provided evidence on the short-term modifications of olive grove soil microbial communities after OMW spreading combining chemical and molecular data.
The present work aimed to evaluate the effects of OMWW spreading on soil chemical characteristics and soil microbial structures to study the evolution and dynamics of the soil microbiome of an olive orchard over time. Moreover, the effect of the OMWW spreading on the vegetative and productive activities of the olive trees was evaluated. In particular, in this study, the OMWW was directly applied in the grooves, and the soil analysis was carried out at different depths (0–20 cm and 20–40 cm) to investigate the potential risk of leaching. This unique OMWW spreading method was adopted to facilitate its management in the study area.
3. Results
3.1. Effect of OMWs Spreading on Soil Chemical Properties
The results of soil chemical characteristics of unamended control, 14 days after spreading, one year and two years from the end of the OMWW spreading are reported in
Table 2 and
Table 3 for 0–20 and 20–40 cm, respectively. Fourteen days after spreading, the pH showed significantly lower values than the control at both depths due to the acidic pH of OMWW added to the soil. No changes in soil pH were observed in all other sampling times. The addition of OMWW also caused a significant increase of soil EC at 14 days after spreading at both depths. Afterward, a gradual reduction of salinity was observed. Concerning the TOC content, all soils did not show significant differences with mean values of 2 and 0.71% at 0–20 and 20–40 cm, respectively, except for a decrease in 2 years treated soils in the upper soil layers. Even the soluble organic C in alkaline solution (TEC) did not show significant change after OMWW application in the short period, except for a slight increase after one year. The WEOC concentration increased significantly, especially at 14 days after spreading at both depths, and then decreased over time, reaching values similar to the unamended control after two years. The addition of OMWW resulted in a slight significant increase of total N only after one year. In contrast, the available P was higher than the unamended control even after two years of treatment at both depths. In the short term (14 days), the exchangeable K increased in the upper soil layers, showing values similar to the unamended control after one year. The CEC did not show any change in the soils amended with OMWW and the concentration of exchangeable Ca and Mg.
3.2. Leaf Net Photosynthesis (Pn)
Soil amendment with OMWW did not significantly influence Pn. The highest photosynthesis values were recorded in June (15.16 ± 0.38 and 12.24 ± 0.95 mmol CO2 m−2 s−1 for the control and treated trees, respectively) and July (10.71 ± 2.29 and 13.45 ± 0.62 mmol CO2 m−2 s−1 for the control and treated trees, respectively), without significant variations between control and treated trees. The lowest Pn values were recorded in August (9.02 ± 1.09 and 6.87 ± 0.81 mmol CO2 m−2 s−1 for the control and treated trees, respectively) in correspondence with a period characterized by the water deficiency and the high temperatures.
3.3. Vegetative and Productive Activity, Characteristics of Fruit
The canopy volume was not significantly different between control and treated trees (14.55 ± 0.32 and 14.73 ± 0.78 m3 for the control and treated trees, respectively). The pruning weight recorded did not show statistical differences between control and treated trees (18.85 ± 1.4 and 17.74 ± 2.5 kg for the control and treated trees, respectively). The amendment with OMWW did not influence the production per tree and the fruits’ fresh and dry weight, color, and oil content.
3.4. High-Throughput Sequencing Results
After quality filtering, the total reads were 289,299 in soil and 75,831 in OMWW, with a total of 5190 and 80 ASVs, respectively. An average number of approximately 12,000 and 19,000 reads/samples were reported. OMWW samples were investigated only in their bacterial composition, and no further analysis was carried out.
3.5. Bacterial Composition of Soil and OMWW
Relative abundances at the phylum level (Relative abundance > 1%) for each analyzed soil sample were calculated and reported in
Figure 1. The bacterial communities in the olive grove soil were dominated overall by Proteobacteria (23.6%), Firmicutes (22.1%), and Actinobacteria (20.6%). Considering the soil treatment period, it was possible to determine that Firmicutes (33.9%) dominate the soil not treated with OMWW, whereas Proteobacteria was the most represented phylum in 14-days (37.9%) and 1-year (26.79%) samples. The soil where OMWW spreading stopped two years before presented a high presence of Actinobacteria (25.82%).
Relative abundance (relative abundance > 1%) was also calculated for OMWW. The bacterial communities were characterized only by Cyanobacteria (91.2%) and Proteobacteria (8.8%) in these samples. Whereas among Proteobacteria, the Gammaproteobacteria and Alphaproteobacteria classes were described, in Cyanobacteria, the only Chloroplast class was recognized. In OMWW, four ASVs represented 93.4% of the total reads (
Supplementary Table S1). Therefore, to find a more precise taxonomic identification, their sequences were used as queries against the GenBank [
75] database using the Basic Local Alignment Search Tool (BLAST) algorithm [
76]. Two of them were associated with the chloroplast genome; the other two showed a BLAST match ≥98% to
Cronobacter sakazakii and mitochondrion genome of
Olea europaea, respectively.
3.6. Microbial Richness
Supplementary Table S2 shows the observed richness, Shannon’s diversity index, and Simpson’s dominance index for the four treatment periods. Despite Shannon and Simpson indexes reported a lower diversity in 14-days samples, no significant differences (
p > 0.05) in ASVs richness of the bacterial community were found between the periods of treatment (data not shown).
3.7. Comparison of Soil Bacterial Communities over Time (β-Diversity)
PERMANOVA analysis applied to the Bray-Curtis dissimilarity distance matrix confirmed a significant effect of the treatment with OMWW on the bacterial communities (p ≤ 0.05). On the other hand, the same analysis showed no significant impact of the sampling depth on the prokaryotic communities (p > 0.05).
Principal coordinate analysis (PCoA) ordination graph (
Supplementary Figure S1), performed on the 16S rDNA data, shows that the difference between prokaryotic communities of the no-treated soil and the 2-Years samples with the more recently treated samples (14-days and 1-year) explained the variance obtained in the first axis (19%). On the other hand, the second axis (11.3%) does not show a persistent pattern according to treatment time.
The CAP ordination explains approximately 25% (17.7% in the first and 7.6% in the second dimension, respectively) of the total variance observed in bacterial communities (
Figure 2). This ordination graph separated data into three groups, 2-Years+no-treatment, 1-year, and 14-days. This evidence showed a similar structure of bacterial communities in 2-years and no-treated samples.
3.8. Differential Abundance Analysis (ANCOM)
We used ANCOM differential abundance analysis using QIIME2 to visualize the bacterial taxa that significantly differed among the soil samples by treatment period. Overall, ANCOM statistical results reported that 12 ASVs significantly differed (null hypothesis rejected) among samples, according to the treatment period (
Table 4).
Table 5 shows also the ANCOM percentile abundance of features by the treatment period for the significantly differing ASVs. All the significantly different ASVs were associated with the 14-days sample category; in particular, two of them (c98061c433607fe4b2f06fceb61c7458 and 67465099c7e7995ae38f287944b6094f) were clear outliers in the ANCOM volcano plot (
Supplementary Figure S2), with high values for both W (the number of null hypotheses rejected) and clr (effect size). These two had high sequence similarity using NCBI blastn to
Kosakonia arachidis and
Mangrovi bacterplantisponsor species, both belonging to
the Enterobacteriaceae family.
To better elucidate the dynamics and changes in bacterial communities over time, we carried out ANCOM analysis on pairs of treatment period categories.
Table 6 shows as significant differences in ASVs abundance was detectable comparing no-treated soil vs. 14-days (6 ASVs), 14-days vs.1-year (2 ASVs), and 14-days vs. 2-years (5 ASVs). ANCOM-returned ASVs were all associated with the 14-days samples (data not shown). We did not find any ASV that significantly differed comparing no-treated soil vs. 1-year and 2-years and 1-year vs. 2-years.
3.9. Groves Characterization
The first two PCA axes accounted for 74.77% of total data variability, and the biplot highlighted some notable differences among groves and depths (
Figure 3). In detail, the PCA showed two clusters: samples named Unamended_40, two years_40, and one year_40 belonged to the first one, while samples named 14 days_40, Unamended_20, and two years_20 belonged to the second one. In particular, these Unamended_20 and two years_20 appeared very similar. Samples 14 days_20 and one year_20 were discriminated with respect to other groves on the PC1. In contrast, they discriminated each other to a slightly lower extent on this same PC1 and discriminated each other on PC2.
Chemical soil variables seem to differentiate the samples collected at different depths very well. All the chemical soil variables had significant effects on the ordination of groves (as indicated by the lengths of their vectors). The variables were clustered into three groups: the first one included TKN, CEC, exchangeable Mg, TOC, and TEC, the second one had exchangeable K, available P, WEOC, and EC, while the third group included Ca and pH. Within each group, variables were highly positively correlated. Still, the variables belonging to the first group were generally negatively correlated to pH, and they did not correlate with those belonging to the other remaining group and Ca. The variables belonging to the second group were negatively correlated to both pH and Ca. Both Unamended samples, one year_40, two years_20, and 40 samples, were mainly characterized by high (above-average) values of pH and Ca. Samples 1 year_20 were characterized primarily by high (above-average) CEC, TKN, Mg, TEC, and TOC values. In contrast, sample 14 days_20 was mainly characterized by high (above-average) values of other chemicals in the soil. Finally, sample 14 days_40 was characterized by high values, on average, of all chemical soil variables.
3.10. Redundancy Analysis
The chemical variables selected by the “forward” stepwise factorial regression procedure were: EC, TEC, exchangeable Ca, available P, and exchangeable Mg. These variables explained 49.56% of the variability in the species abundance matrix.
The biplot derived from RDA for species abundance shows a clear representation of the relationships between chemical soil variables, bacterial communities scores, and Phyla scores (
Figure 4). The first axis accounted for 22.11% of the variation, while the second axis accounted for 17.28% (for 39.39% of total abundance variation). All fitted variables were located at about the same distance from the center of the biplot (
Figure 4), indicating that they had the same effect on the abundance matrix, except Mg, which had the lowest impact. TEC, available P, and EC were positively correlated; in particular, EC and P had opposite effects with respect to Ca and Mg along Axis 1; while TEC had opposite effects with respect to Ca and Mg along Axis 2 (
Figure 4). At the same time, Mg and Ca were highly positively correlated.
The bacterial communities named one year_40, 14 days_20, and 14 days_40 were characterized by phyla mainly positively influenced by high (above-average) values of EC, available P, and TEC, and low values of Ca and Mg. In contrast, the bacterial communities named two years_20, two years_40 and which found unamended control were constituted by phyla which were positively influenced by high (above-average) values of Ca and Mg and consequently by low (below-average) values of EC, P, and TEC. The bacterial communities named one year_20 were characterized by phyla mainly positively influenced by high (above-average) TEC and available p values; they were not affected by EC and were negatively impacted by high (above-average) values of exchangeable Mg and exchangeable Ca. Only the replicates two year_20a, two year_40a, and Unamended_20c showed opposite behavior with respect to the other replicates of the same samples.
Regarding phyla, Firmicutes (second quadrant) and Proteobacteria (third quadrant), which were the most abundant phyla, were very distant from each other in terms of their score on Axis 2 and from all the other phyla along Axis 1. Other major bacterial phyla, such as Actinobacteria, Chloroflexi, and Gemmatimonadetes, were spread in Quadrant I, while Acidobacteria, Bacteroidetes, Planctomycetes, and Verrucomicrobia were spread in Quadrant IV. In addition, these phyla could be considered indicators of high or low values of the variables fitted in this model. Proteobacteria were favored by high (above-average) EC, available P, and TEC values and low (below-average) values of Ca and Mg. Firmicutes were selected by high (above-average) EC, Ca, and Mg values, and it was not influenced by P, while it negatively affected TEC. Acidobacteria and Bacteroidetes were favored by high (above-average) values of TEC and available P and by low (above-average) values of Ca and Mg, while EC did not influence them. Finally, the other most abundant phyla: Actinobacteria, Chloroflexi, Verrucomicrobia, Planctomycetes, and Gemmatimonadetes, took advantage of high (above-average) values of Ca and Mg, while they were negatively correlated to EC, available P, and TEC.
5. Conclusions
To the best of our knowledge, this is the one of few studies where bacterial communities of an olive grove after OMMW treatment were characterized and their evolution over time investigated. Moreover, the changes of main chemical parameters of soil were investigated and correlated to the variations of soil microbial community. The metagenomic approach allowed a wide description of the bacterial communities present in the soil providing both a quality and quantitative characterization. This work has demonstrated that the application of OMWW affected in the first two weeks the pH, salinity, available P, WEOC parameters at both depths. Afterwards all changes were restored, except for the available P. Gram-positive bacteria, particularly the Firmicutes, were the phyla with the lowest tolerance to the addition of OMWW. On the other hand, among the Gram-negative bacteria, Protobacteria were positively correlated to EC, available P, and TEC, suggesting their rapid growth respect to the Gram-positive bacteria and their important role in the C dynamics. Even the microbial analysis suggested that the treatment with OMWW modified in few days the abundances of bacterial taxa in the soil, which are restored along time. The ability of soil to well retain different components could allow to some phyla to try benefit against others for longer periods. As a consequence, the microbial communities could be directionally changed, promoting bacterial phyla that take advantages from components present in OMWW and disadvantaging others, when the treatments are repeated for a long time. Moreover, the application of OMWW did not negatively affect leaf net photosynthesis, the olive tree vegetative activity, yield, and fruits characteristics.
In conclusion, the data reported in the present paper provide positive indications on the long-term use of OMWW as soil amendment. Although the OMWW can be potentially rich in phytotoxic compounds, its long-term use did not affect negatively the chemical-microbiological characteristics of the soil, and on the olive tree vegetal-productive activities. The other disposal methods (incineration, ultrafiltration, concentration, etc.,) commonly used for the OMWW are generally not capable of reducing the pollutants to the levels set by the law and are expensive for most of the oil mills which are mostly of small size and produce sludge that is difficult to dispose of. Even lagooning, while not requiring large investments, is difficult in practical because it has a very slow biodegradation activity, producing bad smell that spread over large areas.
On the contrary the soil spreading of OMWW can be considered an environmentally friendly practice to recycle this organic material and improve soil fertility, permitting a reduction in the use of chemical fertilizers, which are expensive and often can determine pollution phenomena and energy consumption. Although with the dosage imposed by the law (80 m3 ha−1 year−1) no toxicity phenomena have been observed, it is prudent not to increase this quantity because in other contexts (e.g., sandy soils) it could have fewer positive results.