Physicochemical properties

In all sampled environments, Typha angustifolia was the main emergent macrophyte and formed almost monospecific communities covering large surface areas. Moreover, sampled environments were located in a relatively small geographical area, thus minimizing weather effects (temperature, average rain, and solar radiation). Nevertheless, physicochemical conditions of the chosen environments were essentially different in terms of nutrient concentrations, salinity and redox potential. Unfortunately, nutrient concentrations in water were not recorded at the moment of sampling. Nitrate and ammonia concentrations in the two studied systems are relatively variable (ranging from undetectable to above 10 mg/L for NO3 and NH4+, NO2 is not detected above 0.2 mg/L) and highly influenced by the performance of the WWTP in the Empuriabrava area, or nutrient discharges due to agricultural activities in the Baix Ter area10, 35, 36. Water temperatures were around 26 °C, being slightly lower in Bassa de les Tortugues and higher in the Daró river mouth (Table 1). When samples were grouped according to the location (i.e. FWS-CW and Baix Ter), no significant differences of temperature were observed (U Mann–Whitney test, p > 0.05). Sampled environments spanned along a salinity gradient, ranging from slightly saline (conductivity values of 11.95 mS/cm2, estimated salinity 6.84 ppt), such as Bassa de les Tortugues, to low salinity fresh water, such as Rec Coll (0.823 mS/cm2, salinity 0.40 ppt). Water in the Empuriabrava FWS-CW showed typical conductivity values for the system in summer37.

Table 1 Main water physicochemical parameters in the studied sites.

Water oxygen concentration, pH and Redox values were significantly different between the two geographical locations (U Mann–Whitney test, p < 0.05). The samples from the Empuriabrava FWS-CW were characterized by lower oxygen concentration (0.36 ± 0.06 ppm), higher pH (8.85 ± 0.21), and variable Redox values (from − 92 to 70 mV). In contrast, samples from the Baix Ter exhibited a higher oxygen concentration (on average 11.64 ± 1.89 ppm O2), relatively lower pH (7.34 ± 0.17), and high positive Redox values (from 90 to 170 mV) (Table 1). Lower oxygen concentrations in the Empuriabrava FWS-CW can be due to the relatively lower oxygen concentration at the Empuriabrava WWTP effluent and to the fact that planted soils lay on a impermeabilized clay layer minimizing diffusion effects compared to natural systems in the area38.

Radial oxygen loss

Oxygen concentrations were measured orthogonally to the longitudinal axis of the root at different positions, from the tip to the base and were used to estimate potential radial oxygen loss (ROL). Ten different roots from Typha angustifolia, five from each of the two geographical sites, were analysed. In all cases, oxygen concentration increased at the root surface confirming diffusion from the root (Fig. 1). For most of the roots analysed, oxygen loss was higher in the middle section (from 1.5 to 4.5 cm from the tip) although a high variability of estimated diffusion rates was observed between roots, from 0.003 to 0.316 μmol O2/L/μm. Potential oxygen diffusion at the root tip ranged from non-detectable to values similar to those obtained in the middle section (from 3 × 10–5 to 0.196 μmol O2/L/μm). Basal portions (next to the plant junction) always showed the lower estimated diffusion values, ranging from 0.002 to 0.038 μmol O2/L/μm (Table 2). For the two geographical areas analysed, significant differences of potential oxygen diffusion rates were found between the middle section compared to either the tip or the basal sections (paired sample Wilcoxon test, p < 0.05), thus confirming that only a small section of Typha angustifolia roots is likely participating in the oxygenation of the sediment.

Figure 1


Contour plots showing the relative concentration of oxygen (% of maximum measured concentration per root, inset value in each graph) of 10 different roots at different points along the longitudinal axis. Left- Empuriabrava FWS-CW. Right- River Daró estuary. Cell 1 and 2- Sampling points 1 and 2 and the treatment cells of the Empuriabrava FWS-CW, Europa Europa Lagoon, BT Bassa de les Tortugues, DD Daró river mouth, RC Rec Coll.

Table 2 Estimated oxygen diffusion rates and relative values at different root sections grouped according to the geographical area.

Plant aerenchyma is the main tissue by which oxygen can be exchanged between shoots and roots, ensuring an even distribution within plant tissues39. Aerenchyma is especially important in aquatic and wetland plants that grow in hypoxic sediments. Effective radial oxygen loss is proportional to the intensity of the concentration gradient at the root surface, the physical resistance of the dermis and cuticle to O2 diffusion in the radial direction, and the consumption of O2 by cells along this radial diffusion path39, 40. According to the equilibrium of production-consumption, actively growing root areas, such as the root tip, consume abundant oxygen and diffusion to the rhizosphere is diminished6. Similarly, older parts exhibiting a well formed and impermeable root barrier will also prevent oxygen diffusion into the adjacent sediment. Adventitious roots of many wetland species contain a barrier to ROL in the basal zones12, 32, 41,42,43. A barrier to ROL in basal root zones can enhance longitudinal O2 diffusion towards the apex, by diminishing losses to the rhizosphere6. However, diffusion from different portions of the root system can differ markedly but appears to be closely related to aerenchyma formation and maturation of the exodermal bands and suberin lamellae (ROL barriers). Particularly, Typha sp. has extensive casparian bands and suberin lamellae in all layers of the exodermis. Those differentiated cell structures are produced after extensive material deposition on the cell wall strengthening the tissue impermeability. Casparian bands and suberin provide resistance to gas exchange, minimizing losses in less active tissues32. As a result, O2 diffusion in Typha angustifolia roots is restricted to a narrow band (circa 4 cm in length) probably limiting the effect to the surrounding microbiota. To test this hypothesis, extensive microbiological analyses of the rhizoplane were performed using ammonia oxidizers as model organisms.

Abundance of ammonia oxidizers

Bacterial cells were clearly visible in all sections of the root surface, albeit at relatively low densities (Supporting Figure S1). In order to determine changes on cell distribution along the root longitudinal axis, qPCR analyses of 16S rRNA and amoA genes for bacteria and archaea were used. Unfortunately, due to the low number of cells attached to the root surface, root sections from different root hairs had to be pooled in composite samples, precluding the use of paired-tests for this data set, which would have led to more conclusive results. Bacterial and Archaeal 16S rRNA genes ranged from 5.79 × 105 to 1.09 × 109 and 1.91 × 103 to 2.49 × 108 copies/g of fresh weight (FW) in roots, respectively (Supporting Figure S2). amoA genes were detected at much lower gene densities, and varied between 1.26 × 104 and 8.06 × 105 copies/g FW for AOB, and from 0.32 × 103 to 4.03 × 105 copies/g FW for AOA. Abundance of amoA genes, as well as for 16S rRNA bacterial an archaeal genes, was low in comparison to what have been previously found in Typha sp. colonizing environments of similar characteristics44, 45.

In order to determine differences in gene abundances between root sections, each sampling location was analysed separately. In the Empuriabrava FWS-CW, abundances of 16S rRNA Bacteria and amoA Archaea were significantly higher in the middle section compared to the root tip, and 16S rRNA Bacteria was significantly more abundant in middle compared to the basal section (Kruskal–Wallis and Dunn’s test with Bonferroni correction, p < 0.05). In the temporary lagoon, Bassa de les Tortugues, 16S rRNA Bacteria and amoA Archaea were found at significantly higher abundances in the middle than compared to the basal section of the root (Kruskal–Wallis and Dunn’s test with Bonferroni correction, p < 0.05) (Supporting Figure S2). In the Bassa de les Tortugues, a higher abundance of archaeal 16S rRNA and amoA genes was observed in the root tip section, despite no significant differences could be measured compared to the middle section of the roots. In agreement, two out of the ten intact roots studied to estimate ROL potential rates in laboratory conditions (one of them collected in the Bassa de les Tortugues, Fig. 1), exhibited a high diffusion rate at the very tip thus suggesting a potential effect on microbial abundance. However, a higher number of tips samples should have been analyzed in order to confirm these results. Moreover, in Daró River mouth, a higher abundance of amoA Archaea were found in the middle section compared to the basal section (Kruskal–Wallis and Dunn’s test with Bonferroni correction, p < 0.05). Finally, for Rec Coll samples, significantly higher abundances of all four genes were found in the middle section compared to the basal section. Generally speaking, the highest abundance of all genes in the rhizoplane was observed in the middle section, and the lowest in the basal section, although differences occurred between locations, suggesting that general physicochemical conditions of lagoons have an impact on the abundance of bacteria that remain attached to root at the low-scale level.

In previous works, higher abundances of Bacteria over Archaea were found in the Typha rhizosphere and rhizoplane thriving in similar environments as those analysed here45, 46. However, a direct analysis of ammonia oxidizers resulted in non-conclusive effects of plant roots on the relative abundance of archaea when compared to the surrounding sediment. For instance, archaeal amoA was more abundant compared to the bacterial counterpart in the rhizosphere of emergent aquatic macrophytes46, 47, agricultural crops, such as Ipomoea batatas and Zea mays48, and grassland vegetation49. In contrast, Sterngren et al. showed a higher nitrification activity of AOB, and Huang et al. found higher AOB densities compared to AOA in sediments and rhizospheres from Ceratophyllum demersum and Potamogeton malainus in Lake Taihu.

We checked for a selective effect of oxygen diffusion on either archaeal or bacterial ammonia oxidizers by comparing the abundance of amoA Archaea/abundance of amoA Bacteria (AOA/AOB) ratio of different root sections with the surrounding sediment (Fig. 2). Taking all data collectively, higher values of AOA/AOB were found in roots compared to adjacent sediments (Mann–Whitney test, p < 0.05; ranging from 0.001 to 0.747 and 0.001 to 0.119 for roots and sediments, respectively), being the differences more pronounced in the Empuriabrava FWS-CW and the Rec Coll, two environments of contrasting salinity and redox potential. Previous works have highlighted the importance of pH in the dominance of AOB over AOA in alkaline media, which holds also true for coastal lagoons46. Despite having relatively high pH values, we observed a higher AOA/AOB ratio in the Empuriabrava FWS-CW, in disagreement with the previous statement. Probably, sediment had different physicochemical variables compared to water, which could affect microbial communities in different way than roots, favouring AOB over AOA in sediment. AOA were described to be more sensible to salinity and to pH than AOB in microcosms and estuarine sediments31, 50; though it depends on the ecotype, archaea are able to survive in a wider range of salinity and oxygen concentration51. Considering the variability of gene abundances between samples of the same location, more analyses should be done to define a clear trend in this sense. In relation to changes according to root sections, significant differences of the AOA/AOB ratio were only found in Rec Coll and the Daró River mouth, in which archaea were significantly enriched in the middle compared to the basal section (Kruskal–Wallis and Dunn’s test with Bonferroni correction, p < 0.05). Rec Coll and Daró River mouth were the two environments showing the lowest water conductivity and no drastic seasonal changes in water flow are expected in the area. Responses of AOA/AOB ratios to various environmental factors have highlighted the importance of temperature, ammonia and oxygen level in addition to salinity and pH. AOA were suspected to be more sensitive to salinity and pH fluctuations than AOB according to the results of microcosms incubations and analyses of estuarine sediments31, 50. Although AOA are generally considered to have higher affinities for oxygen compared to AOB51, 52, some exceptions to this rule exist. For instance, Santoro et al. retrieved almost constant archaeal amoA gene copies in aerobic subterranean aquifer sediments with pore water at dissolved oxygen levels of 0.1–0.2 mM. Qin et al. showed that not all strains and isolates respond uniformly to increased oxygen concentration and some of them were stimulated at higher O2 concentrations. Könneke et al. reported the fully aerobic growth of Nitrosopumilus maritimus during cultivation and near-stoichiometric conversion of ammonium to nitrite, showing a limited effect of O2 concentration.

Figure 2


Box plot showing the ratio between ammonia oxidizing archaea and bacteria (qamoA Archaea/qamoA Bacteria) in the four studied environments according to the root section. Ratios in sediment samples are included for comparison. Distances from the root apex: Tips 0 to 15 mm, Middle 15 to 45 mm, Base- > 45 mm.

Microbial community structure

Microbial communities at different root sections were studied on the basis of the 16S rRNA gene sequence. A total of 3,836,951 sequences passed quality filtering. On average, 66,154 sequences were obtained per sample (ranging from 23,001 to 142,783). Four samples (i.e. two replicates of the middle section, Cell 2 and Europa Lagoon, one replicate from the basal section, Bassa de les Tortugues, and one from sediment, Rec Coll) were discarded from the analysis due to poor sequencing depth. For comparisons between samples, a subset of 22,500 sequences per sample was randomly selected and used for alpha and beta diversity analyses. Observed richness (Sobs, number of OTUs) varied between 158 and 2,916 (Supporting Table S1). Shannon’s and Phylodiversity diversity indices varied from 3.65 to 6.54 and from 10.85 to 100.33, respectively. Despite this variation, no significant differences were found across root sections and not between roots and sediment in alpha diversity indices of the analysed sites (Kruskal–Wallis and Dunn’s test with Bonferroni correction, p > 0.05).

The phylum Proteobacteria was the most abundant in all samples and accounted for more than 50% of all sequences (Supporting Figure S3). Results agree with what has been previously found in natural systems and in wastewater treatment plants53, 54. Bacteroidetes, Chloroflexi (mainly in CW) and Firmicutes were also abundant. Bacteria were dominant over Archaea in all samples, although differences did not match completely qPCR results. Archaea accounted from less than 0.01% of the total sequences reads in the sediment of the Daró River Mouth to up to 8% of sequences in the sediment of the Empuriabrava CW. According to qPCR, archaea to bacteria 16S gene ratio was significantly higher in the Empuriabrava sediment and rhizoplane in agreement with sequencing data. However, this higher proportion of archaea could not be solely attributable to a higher abundance of AOA. Only for the sake of comparison, AOA/16S rRNA archaeal genes was 1.9 × 10–3 (average value) in the Empuriabrava samples, and up to 1.4 × 10–1 for Rec Coll.

In order to specifically analyse changes in the composition of nitrifiers, sequences were classified down to the genus level and putative ammonia oxidizers were selected and analysed separately (Supporting Table S2). AOB were generally more represented than AOA, according to the relative abundance of sequences, which was in concordance to qPCR results (Fig. 3). Indicator species analyses did not result in conclusive results at the large-scale level (root vs sediment samples). Nevertheless, “Candidatus Nitrososphaera”, Thaumarchaeota Group C3, and “Candidatus Nitrosoarchaeum” and Thaumarchaeota Marine Group I were more prevalent in the sediment samples of the Empuriabrava FWS-CW, Bassa de les Tortugues, and in the Daró River Mouth, respectively. Those differences in potential indicator species among systems revealed a clear effect of the prevailing physicochemical conditions on the nitrifier community. In contrast, unclassified Nitrosomonadaceae and Nitrospira spp. were found to be related to root samples in the Daró River Mouth and Rec Coll, environments with the lowest salinity. Nitrospira spp. were detected in all samples (0.1 to 2.7% of sequences) being more abundant in the Daró River Mouth. Comammox Nitrospira spp. is supposed to be widely distributed in environments of diverging characteristics24, suggesting the possibility of the presence of microorganisms harbouring a complete nitrification gene set, especially in the Daró River Mouth.

Figure 3


Relative abundance (per total number of sequences in samples) of identified ammonia oxidizers (groups as in Supporting Table S2). Blue colours—Ammonia Oxidizing Archaea. Orange colours—Ammonia Oxidizing Bacteria.

Dissolved oxygen potentially has an advantageous effect on ammonia oxidizers. Due to differences in the affinity of nitrifying microbes for oxygen (AOA > AOB > NOB, including Nitrospira)55, environment selection on different populations was expected. High oxygen affinity makes AOA more competitive than AOB in hypoxic environments51. In fact, Nitrosomonadaceae and Nitrospiraceae, which had a lower affinity for oxygen, were the most represented taxa in the Daró River Mouth and Rec Coll, where the highest oxygen concentrations were found, especially in root samples. Common estuarine AOB, such as Nitrosospira spp., could not be unequivocally determined at the genus level in the samples analysed, although the proportion of unassigned Nitrosomonadaceae sequences was relatively high within nitrifiers’ related sequences. According to sequences assigned to genera, Nitrosomonas spp. were dominant on those systems. Nitrosomonas is commonly found in freshwater environments with relatively high C/N ratios or low total N conditions24, 56, conditions similar to those described for the studied environments.

Root vs sediment differences in nitrifiers community structure

Differences in the structure of microbial communities were analysed with a Principal Coordinates Analysis (PCoA), based on the Unifrac weighted distance matrix. Two clusters were observed in PCoA samples distribution according to the two studied geographical sites (Supporting Figure S4), which was confirmed by PERMANOVA (pseudo-F = 10.716, p = 0.001). Redox and pH were correlated to the distribution of samples (Spearman’s test, r2 > 0.8). pH and Redox potential has been described as having an opposite effect on the structure of sediment microbial communities of constructed and natural wetlands57,58,59.

In order to infer changes in the ammonia oxidizing community in the sampled sites, ordination plots (PCoA based on Unifrac weighted distances) were constructed considering only those sequences assigned to any of the taxa depicted in Fig. 3. Similar to what has been observed for the analysis of the total microbial community, samples grouped according to the geographical area (Fig. 4). Two groups could be clearly separated, the Baix Ter, and the Empuriabrava FWS-CW areas at a significant level (PERMANOVA test, pseudo-F = 15.727, p = 0.001). As a whole, samples distribution appeared to be correlated to redox potential, oxygen and temperature (Spearman’s test, r2 > 0.6). These results suggested an important effect of the environmental physicochemical properties in shaping the ammonia oxidizers community. Water oxygenation has been shown earlier to have a direct advantageous impact on the ammonia oxidizing community45, and differences in the affinity of nitrifying microbes for oxygen have been confirmed55, thus having a direct effect on the community composition. However, no clear effects of redox potential on ammonia oxidizers have been described, though an increase of redox potential (e.g. by oxygen release from the root or by low pH environment) could stimulate ammonia oxidation24, 60.

Figure 4


PCoA distribution of samples according to putative ammonia oxidizers community composition determined by Unifrac weighted matrix. Vectors show the correlation of environmental variables to PCoA axis 1 and 2. Sample labelling, Cell 1 and 2- Sampling points 1 and 2 and the Treatment cells of the Empuriabrava FWS-CW, EE Europa Lagoon at the Empuriabrava FWS-CW, BT Bassa de les Tortugues, DD Daró river mouth, RC Rec Coll. Inset showing clusters for root (green) and sediment (brown) samples in the four sampled environments.

For all sampled sites, communities in the rhizoplane appeared to be different compared to the surrounding sediment, although significant differences were only found for the Empuriabrava FWS-CW, probably due to higher number of samples analysed. More specifically, middle and basal root sections appeared to have distinct ammonia oxidizing communities compared to sediments (PERMANOVA pseudo-F = 3.44, p < 0.01), whereas the two root tip samples that could be correctly analysed appeared at a long distance from each other. Despite the low number of samples, a similar distribution was observed in all sampling sites, that is basal and middle root sections harbouring a distinct nitrifier community. We can speculate on a progressive succession of nitrifier phylotypes eventually coming in contact with the root system at sections were radial oxygen loss occurs (middle section and tips). As root growth progresses, previously colonized areas develop stronger barriers to diffusion, thus slowing down the selective pressure of oxygen on the root surface (basal sections). At this point, little changes in relation to oxygen diffusion are supposed to occur in close contact to the root and nitrifying communities are potentially receiving a higher influence from the surrounding sediment. In this sense, the presence of microhabitats with evolving properties, sediment or different parts on the root, may have an impact on the community composition5, 61. However, to what extent is this occurring at the root surface during development should be further investigated using molecular proxies for activity of ammonia oxidizers.