1. Introduction
The central role of the plant-associated microbiome in maintaining host’s fitness is being recognized more and more, and plants are now regarded as “holobionts”, which include the plant itself and the entire community of associated microbes, seen as a single unit of evolution [
1,
2]. The plant microbiome is known to be species- and cultivar-specific [
3,
4]; moreover, each plant habitat harbors its own specific microbiome [
5]. Microorganisms interact with their host plants in several ways, from specific symbioses to relatively nonspecific beneficial effects, including plant growth promotion and protection from phytopathogens [
6,
7]. Like most plant species studied so far [
8], olive (
Olea europaea L.) has also an associated microbiome, which was shown to be cultivar-specific [
9,
10,
11]. Being a tree species growing in arid and semiarid regions, olive might have established positive relationships with microbes to profit from several beneficial functions, including disease alleviation [
12]. Indeed, microbes isolated from olive plants showed ability to inhibit phytopathogens, at least in in vitro assays [
13]. In contrast with animals and humans, very little is known about the role of microbiome in plants and their response to plant diseases. Passos da Silva et al. [
14] analyzed ten olive knots infected by
Pseudomonas savastanoi pv.
savastanoi and showed the presence of a highly diverse bacterial microbiota. However, they did not analyze
P. savastanoi-uninfected plants as control, which makes it impossible to derive conclusions about the response of the native olive microbiome to the disease.
One of the most dramatic diseases for olive trees is the Olive Quick Decline Syndrome (OQDS), which resulted in devastation of thousands of hectares of plants in the Salento region, in Southern Italy), including very old trees [
15,
16], and infected more than twenty other plant species in the region [
17]. Future projections suggested that the pathogen will persist in Europe [
18]. It is caused by a strain (named “De Donno”, previously known as “CoDiRO”) of
Xylella fastidiosa Wells et al., subsp.
pauca, a xylem-limited bacterial pathogen transmitted in olive trees by sap-feeding insect vectors, i.e., the meadow spittlebug (
Philaenus spumarius L.) [
18].
X. fastidiosa can infect more than 550 plant species, including trees of major importance in forest ecosystems and urban greening plantations, such as oak, elm, sycamore, and maple, representing one of the major threats to agro-forest-ecosystems all over the world [
19,
20,
21,
22]. The pathogen invades the xylem, resulting in the occlusion of the vessels with subsequent restriction of water movement, inducing the related parts of tree crown to dry out [
23,
24,
25,
26]. To date, the most widespread olive cultivars in the Salento region, “Cellina di Nardò” and Ogliarola di Lecce, showed high sensitivity to
X. fastidiosa, while a notable resistance was observed in the less common cultivar “Leccino” [
27,
28,
29].
The mechanisms of resistance of the “Leccino” trees are still unclear. It was previously shown that “Leccino” resistance is probably influenced by the lignin amount in the xylem vessels, which can limit the bacteria movement and the host invasion by slowing down the disease progression [
30], or by the constitutive amount of secondary metabolites such as hydroxytyrosol glucoside [
29]. However, in a citrus plant or grapevine affected by
X. fastidiosa, it was shown that the nature of the endophytic bacterial community is able to downregulate the pathogen growth or plant symptoms, either because they compete with the pathogen or because they secrete substances able to modulate its virulence [
31,
32]. The first results on olive tree microbiome were recently released, and they indicated that a very high proportion of the detected fungi occurs in the resistant cultivar FS17 [
33].
Microbial diversity is associated with plant health and productivity [
34,
35], similarly to the gut microbiome in humans [
36]. It is possible that the autochthonous microbiome associated with “Leccino” trees contributes to its resistance to
X. fastidiosa, for example, by inhibiting the pathogen via microbe–microbe interactions or by triggering the reaction of the plant immune system. This interaction could synergistically enforce the natural resistance of “Leccino” trees to this pathogen.
In this study, we explored and compared the autochthonous fungal and bacterial microbiota associated with both X. fastidiosa-infected (Xf-infected) and -uninfected (Xf-uninfected) “Leccino” trees (X. fastidiosa-resistant), using the X. fastidiosa-susceptible “Cellina di Nardò” trees as a control. Our aims were as follows: (i) to assess differences in terms of assemblage, diversity, and structure between the two cultivars; (ii) to understand the response of the native microbiota to the X. fastidiosa infection; and (iii) to identify taxa specifically associated to the cultivar “Leccino”. Moreover, we explored the microbial co-occurrence/co-exclusion network within the endophytes of Xf-infected “Leccino”, to gain insights into the multispecific interactions involving X. fastidiosa. We hypothesized the following: (i) the cultivar “Leccino” harbors specific microbiota, with a different assemblage and a higher diversity compared to “Cellina di Nardò”; (ii) the response of the olive microbiota to X. fastidiosa infection is different between the two cultivars; (iii) microbial taxa exist that are maintained, or that appear only, during X. fastidiosa infection in “Leccino” but not in “Cellina di Nardò”; and (iv) these “Leccino”-specific taxa show potential multispecific interactions with X. fastidiosa.
3. Discussion
In this work, we aimed to characterize the microbiota associated with the
X. fastidiosa-resistant olive cultivar “Leccino”, a highly promising cultivar able to survive and grow in Salento despite the severe
X. fastidiosa outbreak that affects this region of Italy since several years [
38]. So far, no effective method was found to control the pathogen [
22,
39,
40], but only several conventional and innovative diagnostic approaches were tested for the De Donno strain [
28,
41,
42]. It is of primary importance to understand the basic resistance mechanisms of “Leccino”, in order to implement successful agronomical strategies to overcome the current critical situation of the disease. The working hypothesis of this study was that the autochthonous microbiota associated with the cultivar “Leccino” plays a role in the cultivar’s resistance to
X. fastidiosa, perhaps acting synergistically with the plant’s own resistance mechanisms, as suggested for the olive cultivar FS17 [
33]. Plant microbiomes are species- and cultivar-specific, and therefore the final desired goal of our research is to identify microbial species or consortia, specifically associated to “Leccino” that could be used for the biological control of the Olive Quick Decline Syndrome. First, we characterized the microbiome associated with branches and leaves of the cultivar “Leccino”. To do this, we applied a cultivation-independent approach (Illumina sequencing) and used an
X. fastidiosa-susceptible cultivar (“Cellina di Nardò”) as control for comparison. Our sampling strategy followed a full factorial scheme, including the separation of endo + epiphytic microbiota and endophytic only (by surface-sterilizing a subset of samples), the differentiation between branches and leaves, and the analysis of
Xf-infected and -uninfected plants. Both Prokaryotic and Fungal microbiota were analyzed. To the best of our knowledge, no data have been available so far in the literature on the total microbiome associated with
Xf-infected and
Xf-uninfected olive trees in Salento.
The
Xf-infected and -uninfected samples of our study were collected from two areas within the same natural park, at 3 km distance from each other. This was a strategical choice, due to the higher risk to select false-negative samples in orchards in which positive plants were previously detected compared to sampling in orchards in which the pathogen was not yet ascertained. This event is not uncommon for an erratically distributed pathogen with a long period of latency [
22]. However, the sampling area is flat, without surrounding or dividing orographic elements, which might influence notably the average weather conditions (
Table S5). Although some micro-climatic conditions of the sampling sites (difficult to track) might have influenced the olive-associated microbiome, soil analysis showed largely similar pedological parameters (
Table S6). Thus, pedo-climatic conditions can be reliably considered as homogeneous within the whole sampling area.
Our samples showed an erratic distribution of
X. fastidiosa, especially in the branches, confirming previous reports [
25,
26,
41]. The taxonomic composition of the microbiota showed hundreds of families and genera, which is in agreement with the data available from other tree species [
43,
44]. Dominant organisms were typical plant-associated taxa, also known to be endophytic, such as
Pseudomonas,
Sphingomonas, and
Methylobacterium among bacteria, and
Acremonium,
Aureobasidium, and
Sarocladium, among fungi. Interestingly, Archaea were not detected, although the primers used in this work are able to amplify archaeal 16S rRNA genes [
44]. Müller et al. [
9] analyzed the leaf microbiota of 10
O. europaea cultivars from the Mediterranean basin, including “Leccino”, and found abundant Archaea in all of them. However, these plants were all grown at a single agricultural site in Spain, and therefore this abundance might have been determined by the local soil and environmental conditions.
In our study,
Xf-infected “Cellina di Nardò” samples appeared severely dysbiotic, especially for the leaf endophytic bacteria. Here the endophytic bacteria were dominated by
Ammoniphilus, an obligate oxalotrophic bacterium that requires high concentration of ammonium to grow [
45]. To the best of our knowledge, its presence in diseased plants has never been shown until now. Likely, its abundance in the
Xf-infected “Cellina di Nardò” is linked to the advanced status of the disease in our samples, where high amount of ammonium might have accumulated due to tissue decay. The nitrogen level in the soil where
Xf-infected plants grew is good (1.92 g kg
−1); however, it is unlikely that a high concentration of ammonium in plant tissues could have derived directly from soil nitrogen, otherwise we would have observed a prevalence of
Ammoniphilus also in the
Xf-infected “Leccino” samples.
Xf-infected “Leccino”, instead, showed a bacterial assemblage and structure similar to that of the
Xf-uninfected samples, and this was particularly evident for the leaf endosphere. Considering that this is the primary site of
X. fastidiosa infection, the stability of the endophytic microbiota in the leaves of
Xf-infected “Leccino” appears very relevant and promising for the biological control of
X. fastidiosa. Dysbiosis is the unbalanced microbial status associated to several diseases in humans and animals [
46,
47]. However, little is known about the equivalent situation in plants, and this is one of the first reports showing dysbiosis associated to a plant disease. We argue that the stability of the endophytic microbiota of “Leccino” during
X. fastidiosa infection contributes to the maintenance of a good healthy status of the plant. In fact, it is known that the plant microbiome provides several ecological services important for the maintenance of the host’s fitness and health [
48], within the concept of the “plant holobiont” [
2,
49], as it was shown, for example, in the case of the tomato var. Hawaii 7996’s resistance to
Ralstonia solanacearum [
50]. Moreover, the endophytes are expected to establish a more intimate relationship with the host than the epiphytes [
8,
51]. As such, the bacterial species forming the leaf endophytic microbiota of “Leccino” might be a promising source of strains with potential biocontrol activity against
X. fastidiosa.
The microbiota of “Leccino” also showed a higher diversity and equitability with respect to those of “Cellina di Nardò”. Again, this difference was especially evident for bacteria of both, endo + epiphytic and endophytic microbiota (
Figure 2). Microbial diversity was linked to the health and fitness of ecosystems in general [
52,
53] and plants in particular [
54,
55]. A well-balanced microbiome supports a series of functions that turns beneficial to the host, including the production of inhibitory compounds and growth-promoting factors, and can provide a “barrier effect” that limits both space and nutrients for potential alien species’ and pathogens’ growth [
6]. We suggest that the more diverse “Leccino” microbiota reduces the effects of
X. fastidiosa infection or modifies its output effect for the plant.
The total fungal microbiota appeared very diverse, but no clear difference could be seen in the
Xf-infected samples (
Figure 3D). The endophytic fungal community did not show clear distinguishable pattern between the two cultivars, and therefore we concentrated our analysis on the bacterial microbiota. A statistical comparison between the microbiota of the two cultivars indicated that some bacterial species occurred only in “Leccino”, either in both,
Xf-infected and -uninfected plants, or only in the
Xf-infected ones. These “Leccino”-specific taxa include species known to exert beneficial effects on the host plants, such as
Rhizobium [
56],
Burkholderiaceae [
57],
Sphingomonas [
58], and
Enterobacter [
59].
Massilia is a genus recently shown to be involved in plant-microbe interactions at root and rhizosphere level [
60,
61]; however, it was never shown so far in the plant endosphere or phyllosphere.
Pseudomonas and
Sediminibacterium were reduced in
Xf-infected “Cellina di Nardò”, while the former is a very well-known plant beneficial genus [
62], the latter is a typical sediment bacterium that was never shown associated to plants or beneficial. Although the genus
Pseudomonas includes some phytopathogenic species (such as the mentioned
P. savastanoi pv.
savastanoi), in our study the detected
Pseudomonas OTUs were taxonomically related mainly to different species, such as
P. aeruginosa,
P. stutzeri and, at a less extent, other
Pseudomonas spp. (BLAST analysis of representative sequences). We hypothesize that these “Leccino”-specific taxa could interact with
X. fastidiosa in “Leccino”, and therefore the next step was to perform a co-occurrence analysis to assess potential interactions in the endophytic microbial network.
Microbe–microbe interactions can change the net effect of a microbiome on the host, including plants [
63] and animals [
64]. Therefore, it is of primary importance to investigate microbial interactions also between uncultivated species, which represent the majority [
65,
66]. Inference of co-occurrence networks is a computational method able to detect potential interactions between microbes based on their relative abundances in the samples [
36]. This method is increasingly used to characterize the plant-associated microbiota, for example, in the rhizosphere [
61,
67,
68,
69] or in the pollen habitat [
70]. Here we showed that
X. fastidiosa is potentially interacting with several species of bacteria and fungi in the leaf endosphere of
Xf-infected “Leccino”. Although we were particularly interested in negative correlations, which could indicate a direct inhibition of
X. fastidiosa, the positive correlations could point to beneficial species that grow together with the pathogen (therefore being positively correlated) but at the same time modify its metabolism and reduce its pathogenicity. To perform this analysis we used only the endophytic samples from “Leccino” (and excluded ones of “Cellina di Nardò”) for two reasons: First,
X. fastidiosa was consistently detected only in “Leccino” (
Figure 1A;
Figure S2); second, the microbiota of
Xf-infected “Cellina di Nardò” was so different, due to the dysbiosis, that the network would have been strongly biased by the “habitat filtering” effect [
36], which would make network interpretation impossible [
71]. On the other hand, as a consequence, the number of samples used for the co-occurrence analysis in our work was relatively low, which is a critical point in co-occurrence network inference [
71]. In the future, a wider analysis of the endophytic leaf microbiota from
Xf-infected “Leccino” trees should be done, to perform a more robust co-occurrence correlation analysis and confirm our results.
“Cellina di Nardò”
Xf-infected samples in our study showed a low presence of
X. fastidiosa. This was probably due to the fact that we sampled plants at a very advanced disease state, when the pathogen could become very erratic [
72]. The putative lower amount of
X. fastidiosa observed in 2018 in susceptible cultivar compared to resistant one should be likely due to the progression of symptoms, which became very severe in “Cellina di Nardò” trees in 2018, causing an adverse habitat for the pathogen. Moreover, the advanced state of disease in the
Xf-infected “Cellina di Nardò” plants might have had an effect on the whole microbiome. However, to analyze such plants was a strategic decision because we aimed to have certainly
Xf-infected olive trees as control to be compared to the resistant “Leccino” trees, which are definitively the subject of this work. To assess the actual effect of the disease status on the whole microbiome, an analysis of susceptible olive cultivars at different disease stages will be necessary, in order to investigate the dynamic of the dysbiosis and to link it to the general infection status of the tree.