Acacia longifolia : A Host of Many Guests Even after Fire

: Acacia longifolia is a worldwide invader that cause damage in ecosystems, expanding largely after wildﬁres, which promote germination of a massive seed bank. As a legume, symbiosis is determinant for adaptation. Our study aims to isolate a wider consortium of bacteria harboured in nodules, including both nitrogen and non-nitrogen ﬁxers. Furthermore, we aim to evaluate the e ﬀ ects of ﬁre in nodulation and bacterial diversity on young acacias growing in unburnt and burnt zones, one year after the ﬁre. For this, we used molecular approaches, M13 ﬁngerprinting and 16S rRNA partial sequencing, to identify species / genera involved and δ 15 N isotopic composition in leaves and plant nodules. Nitrogen isotopic analyses in leaves suggest that in unburnt zones, nitrogen ﬁxation contributes more to plant nitrogen content. Overall, A. longifolia seems to be promiscuous and despite Bradyrhizobium spp. dominance, Paraburkholderia spp. followed by Pseudomonas spp. was also found. Several species not previously reported as nitrogen-ﬁxers were identiﬁed, proposing other functions besides ammonia acquisition. Our study shows that bacterial communities are di ﬀ erent in nodules after ﬁre. Fire seems to potentiate nodulation and drives symbiosis towards nitrogen-ﬁxers. Taken together, a multifunctional community inside nodules is pointed out which potentiate A. longifolia invasiveness and adaptation. Bradyrhizobium cytisi is the most represented one (Supplementary Table S1). In fact, through the diversity and evenness indexes, we found a higher diversity in the unburnt zone and a dominance of a species, Bradyrhizobium cytisi , in both zones, as already mentioned. As shown by Shannon–Wiener diversity index, there was a higher diversity in isolates from the unburnt zone (H’ = 1.0) than from the burnt zone (H’ = 0.74). The Pielou evenness index revealed that in both zones some isolates were dominant (J’ = 0.75 for unburnt zone and 0.67 for burnt zone). The Simpson diversity index also showed a high diversity for both the unburnt zone (D’ = 0.97) and burnt zone (D’ = 0.98).


Introduction
In a fast-changing planet and under a climate change scenario, biological invasions have become a serious problem. To overcome them, understanding species mechanisms to adapt to a new environment is crucial for biodiversity protection [1], conservation ecology and management strategies [2]. Exotic species introduction and outcompeting with natives can lead to invasion [3], due to their ability to easily adapt to new environments. One of the largest and widespread families of flowering plants is Fabaceae, which includes species that are becoming major threats for biodiversity [4].
Acacia is one of such genera and it constitutes a polyphyletic group comprising over 1350 species [5], the majority native to Australia [6]. Several acacias have been introduced outside Australia and have resulted in invasive populations worldwide, present at a higher frequency in Mediterranean climates like California, the Iberian Peninsula, and South Africa [7]. Acacia longifolia (Andrews) Willd. (Sidney golden wattle) is considered nowadays one of the most aggressive species worldwide as well as one of the most interesting invaders. In Portugal, this species was introduced for dune stabilization, preservation of sand erosion and with ornamental purposes during the late nineteenth century and the

Soil Characteristics
Soil samples were collected from a depth of 0-20 cm after removing litter layer. A mixed sample was made through the collection of soil from three spots in each zone according to Sankhla et al. [28], and each sample was approximately 1.5 kg. For soil analysis, three subsamples were collected and mixed in a composite sample; these samples were analysed for basic characteristics such as texture and particle size through gravimetric essays, pH (water and KCl 1 M) through a suspension method and potentiometry, organic matter (OM) with thermic decomposition, P2O5 (phosphorus) and K2O (potassium oxide) and the amount of total and mineral N (N-NH₄⁺ and N-NO₃⁻), all through molecular absorption in a segmented flux analyzer. Soil was characterized by a coarse texture in both studied zones, with a pH of 5.4-5.5 (Table 1). Analysis was performed by Plants and Soils Laboratory from Universidade de Trás-os-Montes e Alto Douro (UTAD), Portugal. Table 1. Mean values of soil parameters: texture, water pH, soil organic matter (OM) (%), amount of P₂O₅ (mg•kg −1 ), total amount of nitrogen (N) (g•kg −1 ) and amount of mineral nitrogen (mg•N•kg −1 ), including NO₃ − and NH₄ + . Statistically significant differences, according to T-test with an α = 0.05, are represented by *.

Isotopic Analysis
Leaves and nodules from young plants were collected from the six sampled sites and were dried during 48 h in a drying kiln at 60 °C. Each sample was ground using a ball mill and 2-2.5 mg was weighted for isotopic analysis. 13 C/ 12 C and 15 N/ 14 N ratios in the samples were determined using a continuous flow isotope mass spectrometry on a Sercon Hydra 20-22 (Sercon, Crewe, UK) stable isotope ratio mass spectrometer, coupled to a EuroEA (EuroVector, Pavia, Italy) elemental analyser

Soil Characteristics
Soil samples were collected from a depth of 0-20 cm after removing litter layer. A mixed sample was made through the collection of soil from three spots in each zone according to Sankhla et al. [28], and each sample was approximately 1.5 kg. For soil analysis, three subsamples were collected and mixed in a composite sample; these samples were analysed for basic characteristics such as texture and particle size through gravimetric essays, pH (water and KCl 1 M) through a suspension method and potentiometry, organic matter (OM) with thermic decomposition, P 2 O 5 (phosphorus) and K 2 O (potassium oxide) and the amount of total and mineral N (N-NH 4 + and N-NO 3 − ), all through molecular absorption in a segmented flux analyzer. Soil was characterized by a coarse texture in both studied zones, with a pH of 5.4-5.5 (Table 1). Analysis was performed by Plants and Soils Laboratory from Universidade de Trás-os-Montes e Alto Douro (UTAD), Portugal.

Isotopic Analysis
Leaves and nodules from young plants were collected from the six sampled sites and were dried during 48 h in a drying kiln at 60 • C. Each sample was ground using a ball mill and 2-2.5 mg was weighted for isotopic analysis. 13 C/ 12 C and 15 N/ 14 N ratios in the samples were determined using Diversity 2020, 12, 250 4 of 15 a continuous flow isotope mass spectrometry on a Sercon Hydra 20-22 (Sercon, Crewe, UK) stable isotope ratio mass spectrometer, coupled to a EuroEA (EuroVector, Pavia, Italy) elemental analyser for online sample preparation by Dumas-combustion. Delta (δ) calculation was performed according to δ = [(Rsample − Rstandard)/Rstandard] × 1000, where R is the ratio between the heavier and lighter isotopes. δ 15 N air values are referred to air and δ 13 C VPDB values are referred to PDB (Pee Dee Belemnite). The (secondary) reference materials used were Sorghum Flour Standard OAS/Isotope and Wheat Flour Standard OAS/Isotope (Elemental Microanalysis, UK) for nitrogen and carbon isotope ratio (with, respectively, δ 15 N air (Sorghum Flour OAS) = 1.58 ± 0.15% , δ 15 N air(Wheat Flour OAS) = 2.85 ± 0.17% , δ 13 C VPDB (Sorghum Flour OAS) = −13.68 ± 0.19% , δ 13 C VPDB(Wheat Flour OAS) = −27.21 ± 0.13% ), regularly checked against certified reference materials. Uncertainty of the isotope ratio analysis, calculated using values from six to nine replicates of secondary isotopic reference material interspersed among samples in every batch analysis, was ≤0.1% . The major mass signals of N and C were used to calculate total N and C abundances, using Sorghum and Wheat Flour Standard OAS (Elemental Microanalysis, UK, with 1.47% N, 46.26% C and 1.47% N, 39.53% C respectively) as elemental composition reference materials.
All the analyses were performed at the Stable Isotopes and Instrumental Analysis Facility (SIIAF), Faculty of Sciences of the University of Lisbon, Portugal.

Isolation and Phenotypic Characterization of Nodule Bacteria
For bacterial isolation, nodules were rehydrated in water during 12 h and surface-disinfected in 70% EtOH for 1 min, then transferred to commercial bleach for 6 min and 1 min in 70% EtOH, followed by six washes in sterile distilled water (adapted from Vincent [29]). For disinfection control, nodules were dried with sterile filter paper and rolled (surface printing) in YMA (Yeast Mannitol Agar) plates, incubated at 28 • C for four days. YMA was chosen, once it is selective to nitrogen-fixers, allowing pre-selection when isolation. Only in the absence of growth the nodules were processed, preventing the growth of microorganisms from the soil or nodule surface.
Macroscopic appearance of bacterial growth was analysed, through patterns of growth described for nutrient agar plates (days of growth, size, pigmentation, form, margin, and elevation), according to Cappuccino and Sherman [30]. Routine tests, namely Gram staining and potassium hydroxide (KOH) test, catalase test and oxidase test were performed to cluster the colonies according to the results. Each of these tests has a dichotomic response, positive (+) or negative (-). All these tests were performed after 24 h of visible colonies growth.

DNA Fingerprinting of Bacterial Isolates
Genomic DNA was extracted using GES (Guanidium thiocyanate, Ethylenediamine tetraacetic acid (EDTA) and Sarkosyl) modified protocol [32]. One loop of colonies from each isolate was used. For mucous colonies, five washes in sterile water were performed through suspension in 1 mL of autoclaved water before lysis. The protocol was then followed, and DNA was resuspended in 100 µL of 1× TE (Tris-EDTA).
Polymerase chain reaction (PCR )amplification for molecular fingerprinting was performed on a final volume of 25 µL, containing 50 ng of template DNA, 1 U of Taq DNA polymerase (Invitrogen), 25 pmol of the primer csM13 (5' GAGGGTGGCGGTTCT 3'; [33]), 3 mM of MgCl 2 , 0.2 mM of each dNTPs and 1× PCR buffer. The PCR temperature profiles were 5 min followed by 40 cycles of 95 • C for 1 min, 50 • C for 2 min, 72 • C for 2 min and a final extension at 72 • C for 5 min. Amplification products were resolved through electrophoresis on 1% (w/v) molecular biology agarose gel dissolved in 0.5× TBE buffer. Gels ran at 85 V for 5 h. After running, they were stained in 0.5 µg·mL −1 ethidium M13-PCR fingerprinting profiles were compared using BioNumerics software (Applied Maths, Sint-Martens-Latern, Belgium) and a dendrogram was performed using the Pearson correlation coefficient as association measure and the unweighted pair-group method with arithmetic mean (UPGMA) as clustering algorithm. Through this analysis, it was possible to cluster similar isolates, reducing our sample size, facilitating the selection to identification. An inherent limitation is ascribed to culture-dependent methods, which restricts our diversity analysis to cultivable microorganisms.
Reproducibility was analysed based on a random sample of 10% replicates of the total isolates, in order to establish the cut-off level. This represents the maximum level of similarity between two isolates to support their difference.
Shannon-Wiener and Simpson diversity indexes and Pielou evenness index [34] were used to calculate the diversity and evenness of the bacterial isolates of unburnt and burnt zones. A global approach was performed through cluster analysis, considering all the isolates obtained from unburnt zone and all from burnt zones. Indexes were then compared for further analysis.

Identification of Bacterial Isolates by 16S rRNA Gene Sequencing
After discriminating isolates by DNA fingerprinting, the 16S rRNA gene was amplified for a sub-set of isolates representative of almost all clusters, using two different primers combinations: PA(8f) with 907r or 104f with 1392r based on the Escherichia coli numbering system [35], depending on the success of amplification. SurfTaq (StabVida, Portugal) was the DNA polymerase used. The final volume of the PCR reaction was 20 µL, and the same master mix reagent concentration and temperature profiles were used, as previously described.
Following confirmation of a unique amplicon with the correct size, PCR products were purified using ExoSAP-IT™ PCR Product Cleanup Reagent (Thermo Fisher Scientific), according to the manufacturer's protocol. After purification, the samples were sequenced through Sanger sequencing (StabVida, Portugal).
DNA sequences were analysed with the software Geneious [36], performing alignments among each other and later with data available from GenBank through BLAST. Each sequence was considered individually and aligned with sequences available in the GenBank. The alignments with high similarity were considered and taxonomic identification was achieved according to maximum pairwise identity. Through this identification, it was possible to putatively identify other non-sequenced isolates by comparing similarities through dendrogram analysis. Sequences were submitted to GenBank with accession numbers MT465339 to MT465388 (Supplementary Table S1).

Statistical Analysis
Principal component analysis (PCA) was performed with normalized values of the number of nodules, the soil properties and isotopic analysis, by subtracting the mean of each variable to each value and dividing by the standard deviation. This approach was used to explore differences between the 6 sampled zones in order to identify the main discriminatory variables. Mean number of nodules collected in all unburnt and burnt zones were analysed by t-test at a 95% confidence level (α = 0.05), as well as soil and isotopic data. All data collected were statistically analysed using packages FactoMiner and stats in R studio (v.3.6.1).

Nodulation of Young Plants
In this study, 242 nodules were collected from unburnt zones and 337 nodules from burnt zones, for a total of 579 nodules. However, some variability was present in young acacias growing in sampled areas from both fire conditions. In unburnt zones 1, 2 and 3, 83, 128 and 31 nodules were counted, Diversity 2020, 12, 250 6 of 15 respectively, while in burnt zones 1, 2 and 3 these values were 168, 95 and 74, respectively. Although no relevant differences in size and morphologies were detected between nodules from zones with and without fire (data not shown), a higher and more diverse nodulation index was found in burnt zones, with an average number of 14.0 nodules per young plant, against 10.1 nodules per young plant in unburnt zones. The differences found between treatments, however, were not statistically significant (p > 0.05).

Isotopic Analysis
Nitrogen fixation efficiency in leaves was analysed through δ 15 N, revealing values close to 0% in both zones, in accordance with the occurrence of atmospheric nitrogen fixation through symbiosis. Despite no statistically significant differences between fire treatments, the values −1.0% for unburnt and 0.8% for burnt ones were obtained from leaves, suggesting a higher nitrogen fixation in acacias growing in unburnt zones. In the nodules, no differences were found in δ 15 N values from unburnt and burnt zones (7.9% and 7.4% respectively). δ 15 N in the nodules remained extremely enriched (very positive values). Regarding δ 13 C, there were no major differences between plants growing after fire conditions, both in leaves (−30.2 for unburnt and −29.5 for burnt zones) and nodules (−29.5 for unburnt and −28.8 for burnt zones) (see Table 2). Table 2. Nitrogen and Carbon isotopic composition from leaves and nodules of A. longifolia young plants from unburnt (UBZ) and burnt (BZ) zones. The N and C isotopic compositions, δ 15 N and δ 13 C, respectively, and the percentage of these elements in leaves and nodules by % N and % C, as well as the ratio C/N are also shown. Differences were not statistically significant, according to t-test with an α = 0.05.

Leaves
Nodules

PCA Analysis
PCA biplots showed that three main components explained 76.9% of the variance. The unburnt and burnt zones are clearly separated in PC_1 (Dim1). This Dim1 explained 32.6% of the total variance, mainly due to differences in δ 15 N_L, P 2 O 5 , K 2 O and N-NO 3 . PC_2 (Dim2) explained 25.6% of the variance, considering the OM, total_N and N-NH 4, while PC_3 (Dim3) explained 18.7% of the variance, mainly due to differences in the number of nodules. The bidimensional representation of both PC1_PC2 and PC1_PC3 (Figure 2a,b) reveals a clear separation between the burnt and unburnt zones in both biplots, given the differences in the variables P 2 O 5 , K 2 O, N-NO 3 , OM, total_N, N-NH 4 and δ 15 N_L. All the variables increased with fire, except δ 15 N_L, which was lower in plants growing without fire, revealing a potentially higher nitrogen fixation. The difference in the number of nodules, higher after fire occurrence, is particularly relevant for PC3 (Figure 2b).

Bacterial Fingerprinting and Identification
A total of 153 isolates were obtained, 94 from the unburnt zone and 59 from the burnt zone. After phenotypic analysis, genomic fingerprinting based on csM13 was performed and the results are presented in the dendrogram of the Supplementary Figure S1. This dendrogram showed that isolates obtained from UBZ and BZ were scattered, and there were no clustering grouped accessions according to fire treatment. The isolates were mainly grouped according to its genera, although some genera with more representatives (e.g., Bradyrhizobium sp.) were clustered in different groups. A total of 19 clusters were identified. Some of the isolates were clustered independently, forming a group with a unique representative. 16S rRNA gene sequencing allowed for the preliminarily identification of 50 isolates, with up to 94.2% pairwise identity (See Supplementary Table S1). Bradyrhizobium and Paraburkholderia were the most represented genera with 23 and 10 isolates, respectively, followed by Pseudomonas, represented by seven isolates. Caballeronia, Duganella, Micrococcus, Moraxella, Paenibacillus, and Rhizobium were also identified genera (Supplementary Table S1). These data were considered together with the dendrogram analysis, allowing the inference of the identification of other isolates belonging to the same cluster, following a previous similarity evaluation of the fingerprinting profile. As a result, more genera are represented in UBZ comparing to BZ. Considering the 153 isolates obtained, the 94 isolates from unburnt zone were distributed in five different classes: Alphaproteobacteria (39.4%), Betaproteobacteria (26.6%) and Gammaproteobacteria (16%) from phylum Proteobacteria; Actinobacteria (3.2%) from phylum Actinobacteria and Bacilli (2.1%) from phylum Firmicutes. Part of the collection remained unclassified, accounting for 12.8% of the isolates. The 59 isolates from burnt zone were distributed in four different classes: Alphaproteobacteria (45%), Betaproteobacteria (13.3%) and Gammaproteobacteria (10%) from phylum Proteobacteria; Actinobacteria (8.3%) from phylum Actinobacteria. Additionally, 23.3% of the isolates remained unclassified (Figure 3 and Table 3). A curious and unexpected result was the presence of only one cluster with three isolates identified as Rhizobium sp. and the absence of isolates from Sinorhizobium and Mesorhizobium genera. Regarding species identification, Bradyrhizobium cytisi is the most represented one (Supplementary Table S1). In fact, through the diversity and evenness indexes, we found a higher diversity in the unburnt zone and a dominance of a species, Bradyrhizobium cytisi, in both zones, as already mentioned. As shown by Shannon-Wiener diversity index, there was a higher diversity in isolates from the unburnt zone (H' = 1.0) than from the burnt zone (H' = 0.74). The Pielou evenness index revealed that in both zones some isolates were dominant (J' = 0.75 for unburnt zone and 0.67 for burnt zone). The Simpson diversity index also showed a high diversity for both the unburnt zone (D' = 0.97) and burnt zone (D' = 0.98).
with a unique representative.
16S rRNA gene sequencing allowed for the preliminarily identification of 50 isolates, with up to 94.2% pairwise identity (See Supplementary Table S1). Bradyrhizobium and Paraburkholderia were the most represented genera with 23 and 10 isolates, respectively, followed by Pseudomonas, represented by seven isolates. Caballeronia, Duganella, Micrococcus, Moraxella, Paenibacillus, and Rhizobium were also identified genera (Supplementary Table S1). These data were considered together with the dendrogram analysis, allowing the inference of the identification of other isolates belonging to the same cluster, following a previous similarity evaluation of the fingerprinting profile. As a result, more genera are represented in UBZ comparing to BZ. Considering the 153 isolates obtained, the 94 isolates from unburnt zone were distributed in five different classes: Alphaproteobacteria (39.4%), Betaproteobacteria (26.6%) and Gammaproteobacteria (16%) from phylum Proteobacteria; Actinobacteria (3.2%) from phylum Actinobacteria and Bacilli (2.1%) from phylum Firmicutes. Part of the collection remained unclassified, accounting for 12.8% of the isolates. The 59 isolates from burnt zone were distributed in four different classes: Alphaproteobacteria (45%), Betaproteobacteria (13.3%) and Gammaproteobacteria (10%) from phylum Proteobacteria; Actinobacteria (8.3%) from phylum Actinobacteria. Additionally, 23.3% of the isolates remained unclassified (Figure 3 and Table  3). A curious and unexpected result was the presence of only one cluster with three isolates identified as Rhizobium sp. and the absence of isolates from Sinorhizobium and Mesorhizobium genera. Regarding species identification, Bradyrhizobium cytisi is the most represented one (Supplementary Table S1). In fact, through the diversity and evenness indexes, we found a higher diversity in the unburnt zone and a dominance of a species, Bradyrhizobium cytisi, in both zones, as already mentioned. As shown by Shannon-Wiener diversity index, there was a higher diversity in isolates from the unburnt zone (H' = 1.0) than from the burnt zone (H' = 0.74). The Pielou evenness index revealed that in both zones some isolates were dominant (J' = 0.75 for unburnt zone and 0.67 for burnt zone). The Simpson diversity index also showed a high diversity for both the unburnt zone (D' = 0.97) and burnt zone (D' = 0.98).

Nodulation: Does Fire Play a Role?
After fire, soils are enriched in ammonia and nitrates, as expected considering they are the inorganic forms of nitrogen produced [37], which is shown by values of N-NH 4 + and N-NO 3 − (mineral N forms) that almost doubled for burnt soils (see Table 1). Previous studies proposed that nodulation is downregulated through environmental feedback within the presence of ammonia, ultimately saving resources [38,39]. Additionally, Streeter [40] and Gordon et al. [41] showed that the presence of nitrates on the soil could delay nodule development. Notwithstanding, in the present study, nodulation seems to be potentiated somehow after the fire, as observed by the higher average number of nodules found in young plants growing in burnt study sites. Interestingly, through PCA biplot analysis, it can be seen that the isotopic signature of nitrogen in leaves, represented by δ 15 N_L suggests that nitrogen fixation in post-fire conditions is not correlated with the number of nodules (Figure 2b). Even considering the lower number of nodules in unburnt zones, an increased symbiotic nitrogen fixation occurred. On the other hand, there is a negative correlation between δ 15 N in leaves and mineral N forms, pointing out symbiotic nitrogen fixation dependency in the absence of mineral N forms, as it is occurring in unburnt zones (Figure 2a). With this in mind, why does A. longifolia invest in nodulation? We may hypothesize that these young plants may respond to fire events, showing a different behaviour in this "new" environment, which may be particularly relevant for plant fitness. Additionally, studies developed by Harper [42], showed that a supply of both soil and symbiotic nitrogen is required for a more favourable production of soybean. For this reason, we can possibly extend this hypothesis of partial contribution of both soil and symbiotic nitrogen to A. longifolia too, especially in the after-fire scenario that requires a faster adaptation. Interestingly, δ 15 N in nodules is extremely enriched in both zones, indicating absence of atmospheric nitrogen fixation. One potential explanation could be ascribed to a fractionation leading to an enrichment associated with nodule development and compounds' synthesis to export, with the second one contributing more [43]. Moreover, and in the same direction of an additional fractionation towards a higher enrichment in nodule tissue, Michelsen et al. [44] showed that 15 N abundance could be influenced by the presence of mycorrhizal fungi, leading to an enriched value of the host plant up to 8% . Considering this, we can hypothesize that δ 15 N signature and the respective 15 N enrichment in collected nodules could be due to an association with fungi, along with bacteria presence, which was not explored in the present study. Further studies should focus on exploring this possibility of tripartite symbiosis already described for other Acacia species [45], along with nitrogen compounds exported to the plant.
The nodulation process involves the acquisition of symbiotic partners engaged in nitrogen fixation that consequently allows A. longifolia to access a different pool of nitrogen, facilitating recolonization. For this reason, nodulation seems to be a good process to allocate energy in, leading to its successful dominance. The higher quantity of P 2 O 5 present in burnt soils (five times higher compared with unburnt soils, Table 1), is also an important factor to consider and previous studies have related it with nodules development, quantity, and function [46,47]. In the present study, we can consider it a facilitator for nodule formation in burnt zones in comparison to unburnt, due to a possible limitation in the second.
PCA biplots show a clear separation between unburnt and burnt zones, but besides this natural variability among three unburnt and three burnt zones, there is a tendency for a similar response due to ecosystem rebalance capacity after a disturbance, like a fire.

Nodule Bacteriome: Who Is Taking Part?
A. longifolia seems to establish symbiosis with different bacteria, beyond the commonly described rhizobia and, for the first time, our study was focused on studying bacteriome diversity in this plant species beyond nitrogen-fixers. Its "bacteriome" seems to include α-Proteobacteria, β-Proteobacteria, γ-Proteobacteria, Firmicutes and Actinobacteria. Thus, we can hypothesize that within an invasive range, A. longifolia can take advantage of its promiscuity, outcompeting native species, and investing energy in nodulation, considering that its ability to obtain symbiotic partners is facilitated. Furthermore, A. longifolia could be an example of Taylor et al.'s [48] studies who suggested that legumes, as individuals, could establish symbiosis with multiple rhizobia species simultaneously, again a direct consequence of promiscuity, making A. longifolia a generalist mutualist. Besides this, several studies [19,[49][50][51] showed that Bradyrhizobium is the most common symbiont genus in both native and invasive range of Acacia species, A. longifolia included, which is confirmed in our study both by dominance and intrageneric diversity. Bradyrhizobium cytisi is the main partner among this genus and it is described here for the first time as being involved in symbiosis with A. longifolia. On the other hand, B. japonicum, previously described as the major partner by Rodríguez-Echeverría [23], in both native and non-native ranges [52], was not present among our isolates. Besides other Bradyrhizobium species including isolates only identified to genus level, B. canariense, B. ganzhouense/B. rifense and B. pachyrhizi were also present highlighting the intrageneric diversity and genus dominance, which is in accordance with Rodríguez-Echeverría [18], who already mentions B. ganzhouense as present in Acacia nodules.
Surprisingly, we only observe three isolates belonging to Rhizobium genus, namely Rhizobium rhizogenes at species level (Supplementary Figure S1, Supplementary Table S1), which was a genus already described as one of the main symbionts among legumes [53], and particularly among Acacia genus in Australia (native range) [49,54]; In our study, i.e., within an invasive range, it was not so dominant. Rhizobia obtained from nodules, isolated in nitrogen-free medium as performed in our work, has been described as functional in nitrogen fixation in Leguminosae by several authors [55,56]. These authors showed that Sinorhizobium and/or Mesorhizobium related strains, isolated from Medicago and Acacia, were highly effective in nitrogen fixation (as assessed by its N 2 fixation effectiveness index) and induction of nodulation.
Paraburkholderia and Pseudomonas were two genera also present in our collection. Paraburkholderia caledonica and Pseudomonas moorei were already described as plant growth promoting bacteria (PGPB), playing the role of nodule inducers, presenting similarities to rhizobial species, regarding nif genes and nod factors [57]. Additionally, Saїdi et al. [58] showed that Pseudomonas spp. could have a role as P-solubilizer and in siderophore production. Furthermore, Martínez-Hidalgo and Hirsch [26], in their review, also highlighted the role of Micrococcus strains, a genus to which some isolates in our collection belong, as a plant-development "helper". With this in mind, the question that remains is what could be the role of these non-fixing bacteria in A. longifolia nodulation?
In this context, recent studies postulate that the more diverse a bacterial community is within a symbiosis, the more likely it contains an effective symbiont. Such diverse symbiotic partnerships were explained by Mårtensson et al. [59] that showed that legumes cannot predict the nitrogen fixation efficiency before nodules are established and fixation is in progress; if so, we may hypothesize that A. longifolia emits signals that can be received by several soil bacteria. Other authors [60] have proposed that legumes can control nodulation through oxygenation of nodule microenvironment, leading to bacteria death and nodule senescence, showing that "the host controls the party" [61]. Considering this, diversity is easy to be under control. This could be A. longifolia's strategy, supported by Bradyrhizobium spp. dominance and diversity. We can hypothesize that a process of specialization is present between A. longifolia and Bradyrhizobium spp., by comparison of unburnt and burnt zones. Thus, this symbiotic partner ensures efficient nitrogen fixation, as an obligatory partner. This great representation of Bradyrhizobium spp. can occur considering that different strains of the same rhizobia may differ in their effectiveness [62].
It is also known that some bacteria have functional traits that could complement each other in a way to facilitate a third functional trait [63], which is also potentiated within a wide-range community, with Paraburkholderia spp., Pseudomonas spp. and Micrococcus spp. presence, as possibly occurred in the present study. For this reason, further investigation should rely on the functions that could be performed by bacteria hosted in nodules, along with nitrogen fixation. In other words, nodulation would be much more than just a way to get ammonia. An interesting comparison could be carried out using Next Generation Sequencing techniques to assess a much greater diversity present inside nodules and that will complement this culture-dependent approach.
While Richardson et al. [64] suggested that mutualisms render plant species less prone to invade, our study shows that A. longifolia symbiosis seems to contribute for plant growth and colonization that can be due to an unspecific plant-bacteria interaction already mentioned. In fact, one of the main reasons why A. longifolia is such an aggressive invader, described as ecosystem-engineer, is its capacity to be eventually infected and establish relationships with a wide-range diversity of bacteria available in soils. In addition, belowground microbial diversity is substantially different after fires and this non-specific partnership can as such be useful. Besides the higher diversity in unburnt zones, some selection seems to occur in burnt zones. As reported by Franche et al. [65], diversity found within nodules' "bacteriome" is exclusively between nif genes carrying bacteria and/or already described nitrogen-fixers. This specialization is also according to the nitrogen fixation pathways corroborated by the δ 15 N in leaves ascribed to atmospheric nitrogen fixation through symbiosis (Table 2).

Bacteriome: What Could be Occurring?
As a costly process, nodulation has implicit a complex signal exchange, with the plant responsible for attracting bacteria possessing nif genes to fix nitrogen [61]. This is, as far as we know, the main goal of nodulation. This nodule "bacteriome" diversity can be explained by the inherent highly functional-and taxonomically diverse soil microbiome, along with an absence of restriction on entry into A. longifolia root system. Of equal importance is the process of horizontal gene transfer (HGT) that can be a determinant mechanism to facilitate this entry. In short, genes involved in signal exchange and nodulation are part of symbiotic plasmids or highly mobile "symbiotic islands", which can be transferred easily between different bacterial species, and even genera [66]. For this reason, "bacteriome" diversity could lead more easily to effective nodulation, once beneficial bacteria can take part in nodulation, allowing A. longifolia to grow and spread, faster than other species, underlining the absence of competition after fires. Of course, among potential efficient nitrogen-fixers, some hitchhikers could take a ride and take advantage of nodule environment, stressing why bacteriome functionality should be explored.
Plant-bacteria interactions might, in fact, be highly regulated by environment. The observed highly efficient bacterial community inside nodules makes A. longifolia a top invasive species. Future studies should rely on nodule functionality, activity, and regulation, once which host "controls" the party is known, with guests having their own behaviour. The extent of this approach to native range could be a great contribution to understanding the invasive behaviour of A. longifolia and ultimately leading to its control.

Conclusions
In the present study, fire influenced bacterial diversity inside nodules, maintaining its nitrogen fixation functionality. After the fire disturbance, A. longifolia apparently "selects" nitrogen-fixing bacteria, culminating in Bradyrhizobium spp. dominance and intrageneric diversity. B. cytisi and other species in this genus seem to have a determinant role in symbiosis with A. longifolia, revealing a close relation and a putative facilitation role. However, besides this straight relation with A. longifolia-Bradyrhizobium spp., a considerable bacterial diversity was reported in our study, that could be functionally diverse and render nodules a highly complex structure.
A. longifolia is a typical invader that easily adapts to disturbances, and environmental changes seem to cause a different response in unburnt and burnt zones. This highlights the mutual contribution of ammonia/nitrates and symbiotic nitrogen fixation to plant development, albeit with fast ecosystem rebalance capacity.
Thus, regarding its major impacts, A. longifolia is not only an "ecosystem-engineer" in the aboveground environment, but also, due to its efficiency in selecting bacterial guests, it behaves as an "engineer" of the belowground environment, too.
Supplementary Materials: The following are available online at http://www.mdpi.com/1424-2818/12/6/250/s1. Figure S1: Dendrogram based on cluster analysis of fingerprinting PCR profiles of the isolates from A. longifolia nodules, using the Pearson correlation coefficient and the unweighted pair-group method with arithmetic mean algorithm (UPGMA). 84% was the cut-off level below which isolates could be considered different. On the right are represented: isolate identification (CJJ xxx), zone from where it was isolated (UBZ/BZ x), Gram test result, morphology (rods (B) or cocci (CC)), catalase test result and oxidase test result, both (+) or (−). Colours are according to the phylum/class into each genus belong to: Proteobacteria/α-proteobacteria (blue), Proteobacteria/β-proteobacteria (orange), Proteobacteria/γ-proteobacteria (green), firmicutes/Bacilli (purple) and actinobacteria/Actinobacteria (yellow). Roman numbering identifies clusters. Isolates are identified up to genus level, Table S1: Identification of bacterial isolates obtained from unburnt and burnt zones by BLAST analysis of the 16S rRNA gene sequences. GenBank accession numbers are also indicated.
Author Contributions: J.G.d.J. did the data collection. J.G.d.J. and H.T. outlined the original draft. All authors contribute to the conceptualization, methodology, data analysis and interpretation, as well as the review and edition of the manuscript. All authors have read and agreed to the published version of the manuscript.