Key bacterial groups maintain stream multifunctionality in response to episodic drying

Microbial biodiversity is fundamental to maintain ecosystem functioning in seasonally variable ecosystems. However, it remains unclear how alterations in water availability caused by episodic drying compromise the ability of stream microbes to maintain multiple functions simultaneously (e.g., primary production and carbon cycling). Using data from 32 streams, we investigated how the phenology of annual drying influences stream sediment microbial biodiversity and their capacity to sustain multifunctionality. Our results showed that stream multifunctionality and most bacteria did not respond to changes in drying phenology. Only two bacterial groups, the drying‐resistant Sphingobacteriia and the drying‐sensitive Acidobacteria_Gp7, exhibited positive associations with multifunctionality; whereas, bacterial diversity showed a negative correlation with functions. Among these biodiversity aspects, Sphingobacteriia showed the strongest capacity to maintain multifunctionality at low and moderate performance levels. Our findings will help to better understand the mechanisms through which biodiversity sustains the functioning of seasonally variable streams and their responses to global change.

at low and moderate performance levels.Our findings will help to better understand the mechanisms through which biodiversity sustains the functioning of seasonally variable streams and their responses to global change.
The availability of water plays a crucial role in determining the biodiversity and biogeochemistry in running waters (Dudgeon et al. 2006;Sofi et al. 2020).Stream discharge dynamics exhibit distinct seasonal patterns that influence critical aspects of organisms' life cycles and biogeochemical activity (Rolls et al. 2017;Zimmer et al. 2022).Owing to the seasonal fluctuations in water availability, many streams experience recurring dry periods that differ in their phenology, such as the timing, frequency and duration of drying events (Döll and Schmied 2012).The spatial extent of these drying watercourses and the duration of their dry periods are on the rise owing to global change (Messager et al. 2021;Sabater et al. 2023).This trend poses a threat to aquatic biodiversity and biogeochemical cycles.Despite the global distribution of drying watercourses, it remains unclear how changes in drying phenology affect the ability of stream biota to maintain multiple functions simultaneously (e.g., maintaining moderate to high levels of primary production, secondary production, and organic matter cycling at the same place and time).
Microbes are the most abundant and diverse life forms on Earth, playing a vital role in stream biogeochemistry (Battin et al. 2016;del Campo et al. 2021).Among them, bacteria inhabiting stream biofilms and sediments have a significant impact on primary production, organic matter decomposition, nutrient cycling, and energy transfer to higher trophic levels (Gessner et al. 2010;Rüegg et al. 2021).In terrestrial ecosystems, the importance of microbial biodiversity in driving multifunctionality has been demonstrated under various environmental contexts and spatial scales (Delgado-Baquerizo 2016;Delgado-Baquerizo et al. 2017a,b).However, in freshwater systems, previous studies have focused on individual functions, which have limited our capacity to understand how bacterial biodiversity sustain multifunctionality.
Previous evidence suggests that drying phenology can cause contrasting patterns across functions, depending on how much they rely on water availability (Duarte et al. 2017;Baert et al. 2018).For instance, organic matter decomposition and sediment respiration can be maintained during long dry periods thanks to streambed moisture and the rapid recovery of the microbial activity after flash storms (Gionchetta et al. 2019a(Gionchetta et al. ,b, 2020a;;Arias-Real et al. 2020).In contrast, functions such as primary production undergo significant declines when faced with drying conditions (Timoner et al. 2012;Sabater et al. 2016).However, it remains unclear to what extent multifunctionality is impacted by the changes and trade-offs that arise across individual functions in response to drying phenology.
Another critical factor in predicting how stream multifunctionality responds to drying understands the role of complementarity and selection mechanisms in maintaining ecosystem multifunctionality (van der Plas et al. 2016;Gamfeldt and Roger 2017).Prior studies on soils have demonstrated that complementarity mechanisms, such as resource partitioning and facilitation, seem to be fundamental for maintaining multifunctionality (Violle et al. 2011;Delgado-Baquerizo et al. 2016, 2017b).The rationale is that diverse microbial communities tend to occupy complementary niches, promoting resource-use efficiency and the flux of energy and matter through the ecosystem (Crump and Hobbie 2005;Arias-Real et al. 2023a,b).However, dry sediments might exhibit reduced multifunctionality because their limited water content can constrain microbial diversity and activity.Alternatively, multifunctionality can be sustained under reduced water availability if drying tolerant taxa are able to maintain or even increase some functions under drying conditions (Allison and Martiny 2008).These expectations are highly dependent on how bacteria's ability to withstand drying conditions balances with their contributions to ecosystem function.Therefore, to predict biogeochemical responses to altered drying phenology, it is crucial to understand how the complementarity and selection mechanisms sustain multifunctionality throughout drying periods.
Here, we investigate how the phenology of annual drying influences stream multifunctionality by evaluating the effects of drying stress and bacterial biodiversity on the capacity of streams to sustain multiple ecosystem functions.First, we examine how stream multifunctionality (an index that integrates eight functional indicators, see Table 1) and bacterial biodiversity (diversity and composition) responded to drying phenology (duration and frequency) and water content in the sediment.Second, we identified which aspects of bacterial diversity and composition best explain stream multifunctionality.Finally, we assess whether drying phenology altered the capacity of bacterial biodiversity to sustain stream multifunctionality.

Study area and sampling details
This study was conducted in the northeastern Iberian Peninsula at 32 independent streams located in nine river basins along a wide drying gradient.The upstream catchments are dominated by low impact land-uses (natural vegetation and occasional extensive farmland) and showed different phenological patterns in drying duration and frequency (Supporting Information Table S1).In autumn 2016, at each stream, we collected five sediment replicates, which were stored in sterile plastic bags and transported to the laboratory at 4 C. Once in the laboratory, we prepared 1 mL-sediment subsamples to characterize their biodiversity and multifunctionality by analyzing bacterial DNA in sediments and three associated functional dimensions (photosynthetic activity, organic matter decomposition, and microbial activity).See details in Supporting Information Appendix S1.

Drying phenology characterization
Based on the daily variation in the streambed temperature and water levels, we characterized the drying duration (number of dry days) and frequency (number of dry periods) of each stream site.To do that, 245 d before our sampling, we placed temperature and water level data loggers at each sampling site (Gionchetta et al. 2020b;Arias de Real et al., 2021).Sediment water content was calculated as the ratio between the weight of fresh and dry sediment.

Bacterial biodiversity
DNA was extracted using the FastDNA™ Spin Kit for soils.The extracted DNA was assessed for quality and quantity with a NanoDrop spectrophotometer and stored at À20 C. Bacterial 16S rRNA genes were sequenced using the V4_515F/ V4_806R61 primer pair on Illumina MiSeq technology.The DADA2 algorithm in QIIME2 was used for sequence analysis, resulting in 2,785,940 high-quality reads.These reads were distributed across 32 samples, with an average Q35 score of 80%.The forward and reverse reads were merged, yielding 3353 amplicon-sequence variants (ASVs).Taxonomic assignment of the ASVs was performed using the SILVA reference database v.138 and a feature-classifier script in QIIME2.
To assess bacterial biodiversity and selection mechanisms, we utilized three diversity measures (ASV richness, Shannon diversity and phylogenetic diversity) and analyzed the relative abundance of major taxa.Shannon diversity index was transformed into effective numbers by calculating its exponential, following Jost (2006).We calculated the phylogenetic bacterial diversity through the Rao's quadratic entropy index (Rao 1982), as the sum of phylogenetic distances between each pair of species in the communities weighted by the species relative abundance.The phylogenetic distance (i.e., difference in base pairs) between each pair of species was calculated with the function cophenetic() in the picante package (Kembel et al. 2010), using a phylogenetic tree that was built in QIIME2 using the SEPP plugin (Bolyen et al. 2019).The sequencing data for this study are available in NCBI archive with accession number PRJNA557375 (see details in Supporting Information Appendix S2).

Stream multifunctionality
We measured eight functional variables as proxies of three main functional dimensions of the stream ecosystem (Table 1): photosynthetic activity (dim1), organic matter decomposition (dim2) and microbial activity (dim3).Photosynthetic activity is represented by chlorophyll a (Chl a) and carotenoid content.Chl a characterizes the photosynthetically active biofilm biomass and carotenoids are essential and ubiquitous pigments in photosynthesis that enhance energy transport and provide protection against photo-oxidation (Gong and Bassi 2016).Pigments were extracted using standard methods (Jeffrey and Humphrey 1975) based on spectrophotometric measurements at 665 nm and 480 nm wavelengths for Chl a and Car concentrations, respectively (Jodłowska and Latała 2011).
Organic matter decomposition is characterized by the activity of two hydrolytic enzymes (β-D-1,4-glucosidase and β-xylosidase) and one oxidative enzyme (phenol-oxidase).These enzymes degrade glucose, cellulose, and lignin, respectively.The enzymatic activity was obtained from fresh sediment samples (1 mL) following the same procedure as in Gionchetta et al. (2019b).See Supporting Information Appendix S3 for more information.Microbial activity is represented by bacterial cell density, sediment respiration, and the concentration of extracellular polymeric substances.Bacterial cell density was determined through flow cytometry (FACSCalibur, Becton Dickinson), following a protocol adapted from Amalfitano and Fazi (2008).Sediment respiration was determined through the resazurin assay modified version by Haggerty et al. (2008).The content of extracellular polymeric substances was quantified from frozen sediments using cation exchange resin (CER, Dowex Marathon C sodium form, Sigma-Aldrich) (Romaní et al. 2008).We estimated two missing values for community respiration using the remaining seven functions as predictors through the mice package (Van Buuren and Groothuis-Oudshoorn 2011).We provided expanded methodological details of functional measurements in Supporting Information Appendix S3.
Based on these functional indicators, we used four independent approaches to quantify multifunctionality: (i) the weighted-averaging approach (Manning et al. 2018), (ii) the unweighted-averaging approach (Delgado-Baquerizo et al. 2016), (iii) principal component analysis approach (Meyer et al. 2018), and (iv) the multiple threshold approach (Byrnes et al. 2014).Details of each approach are available in Supporting Information Appendix S3.To reduce redundancy, we show the results of the weighted-averaging index (i) and the multiple-threshold (iv) methods, as they show complementary perspectives.Nonetheless, our results are robust to the choice of the multifunctionality measure (Supporting Information Tables S2  and S3).

Data analysis
We applied Spearman correlations to explore the effects of annual drying phenology (including duration, frequency, and sediment water content) on stream multifunctionality and bacterial biodiversity (diversity and composition), as well as to assess the connections between bacterial biodiversity and multifunctionality.We used Spearman correlations to avoid problems with non-normally distributed data.
To identify the relative importance of bacterial biodiversity in explaining stream multifunctionality, we used a multimodel inference approach (Burnham and Anderson 2002).To do this, we fitted models including all potential combinations of biodiversity (bacterial diversity and relative abundance of key bacterial groups) and drying phenology predictors (drying duration and sediment water content).We removed some potential predictors (i.e., phylogenetic diversity and drying frequency) to ensure a low collinearity degree (variance inflation factor <2).We ranked models using the Akaike information criterion for small sample sizes (AICc) and calculated model weights using the function dredge() from the MuMIn R package (Barto n 2020), retaining those with the greater explanatory capacity (ΔAICc ≤4).We calculated predictor importance as the weighted-mean explained variance across retained models (Hoffman and Schadt 2016).Some variables were log-or squared-root-transformed to reduce distribution skewness.Predictors were standardized (mean = 0, SD = 1) to allow the comparison of model coefficients.Model residuals were visually assessed to verify linear model assumptions.Statistical analyses were performed using R statistical software version 3.4.1 (R Development Core Team 2011).Data are available in the Dryad repository (Gionchetta et al. 2024).

Results
The studied streams showed a contrasting drying phenology, ranging from sites with permanent flowing conditions (no drying events) to ephemeral streams showing up to 245 dry days per year.Intermittent and ephemeral steams also showed a high variability in their drying patterns with 1-17 drying events per year (median = 4 drying events).Average water content in sediments was 15%, ranging from 1.2% to 49.6%.Bacterial richness in these sites ranged from 365 to 2666 amplicon-variant sequences (ASV).The five dominant bacterial groups were Alphaproteobacteria (mean relative abundance = 21.9%),Actinobacteria (13.8%),Betaproteobacteria (13.0%),Gammaproteobacteria (9.2%), and Deltaproteobacteria (5.2%).

Effects of drying on multifunctionality and sediment biodiversity
Stream multifunctionality (obtained through the weighted-averaging index), functional dimensions and individual functions did not significantly respond to drying duration and frequency nor to the percentage of water content in sediments (Fig. 1A; Supporting Information Table S4).Similarly, bacterial taxonomic and phylogenetic diversity did not significantly vary over these gradients of annual drying and water content (Fig. 1B; Supporting Information Table S5).Bacterial taxa showed a variable response to drying duration, with the majority of groups showing a negative response to drying duration and frequency (Fig. 1B).However, only the relative abundance of Acidobacteria_Gp7 (r S = À0.41,p = 0.019) and Bacteroidia (r S = À0.39,p = 0.028) significantly decreased with drying duration and frequency, respectively (Supporting Information Table S5).Water content in sediments had a significant positive effect on the relative abundance of Gemmatimonadetes (r S = 0.53, p = 0.002) and Acidobacteria_Gp10 (r S = 0.36, p = 0.041) (Supporting Information Table S5).

Relationship between sediment biodiversity and stream multifunctionality
Bacterial phylogenetic diversity showed a significant negative relationship with ecosystem multifunctionality (r S = À0.47,p = 0.007) (Fig. 2A), which was obtained through the weighted-averaging index.Bacterial ASV richness and diversity also showed negative correlations with ecosystem multifunctionality, but both were non-significant (Shannon diversity-multifunctionality plot: Fig. 2A).Bacterial groups showed both positive and negative effects on ecosystem multifunctionality (Fig. 2B).Only Sphingobacteriia (r S = 0.43, p = 0.014) and Acidobacteria_Gp7 (r S = 0.37, p = 0.037) showed a significant positive association with ecosystem multifunctionality (Fig. 2A).These two bacterial groups tended also to have positive relationships with the three functional dimensions.Acidobacteria_Gp7 showed the highest positive correlations with photosynthetic activity (functional dimension 1, r S = 0.47, p = 0.007) and organic matter decomposition (functional dimension 2, r S = 0.53, p = 0.002), whereas Sphingobacteriia was the most explanatory group for the microbial activity (functional dimension 3, r S = 0.43, p = 0.013).Thermoleophilia and Actinobacteria also tended to have a positive association with ecosystem multifunctionality, functional dimensions 1 and 2 (Fig. 2A), but these correlations were generally weaker compared with those observed for Sphingobacteriia and Acidobacteria_Gp7.Remarkably, Epsilonproteobacteria (r S = À0.57,p < 0.001), Thermomicrobia (r S = À0.37,p = 0.036), Alphaproteobacteria (r S = À0.37,p = 0.038), and Planctomycetia (r S = À0.37,p = 0.039) had a significant negative relationship with ecosystem multifunctionality (Fig. 2A).These patterns reflect pervasive negative associations of these bacterial groups with functional dimensions and individual functions (Supporting Information Table S3).
The relative abundance of Sphingobacteriia was positively and significantly correlated with the number of functions exceeding low and moderate performance levels (10%, 25%, and 50%) (Fig. 3A).The highest relative abundances of Sphingobacteriia maintained up to 7 functions at ≥10% of their potential, 6 functions at ≥25% of their potential and 4 functions at ≥50% of their potential, reflecting strong functional trade-offs.However, neither Shannon diversity nor the relative abundance of Acidobacteria_Gp7 had a significant effect on the number of functions exceeding performance thresholds (Fig. 3B,C).

Drivers of stream multifunctionality over a gradient of annual drying
The relative abundance of Sphingobacteriia had consistent positive effects on ecosystem multifunctionality across competing models that also considered bacterial Shannon diversity, drying duration and water content in sediments as predictors (Supporting Information Table S6).Acidobacteria_Gp7 tended to have positive effects on ecosystem multifunctionality, but with the mean effect size overlapped with zero (Supporting Information Table S6).The relative abundances of Sphingobacteriia (14.3% of explained variance) and Acidobacteria_Gp7 (7.4%) were the most important drivers of ecosystem multifunctionality, with the rest of variables having a very limited explanatory capacity (<1%) (Supporting Information Fig. S1).

Discussion
Our study shows that multifunctionality, bacterial diversity and the majority of bacterial groups did not respond to annual drying.However, we identify two bacterial groups, the drying-resistant Sphingobacteriia and the drying-sensitive Acidobacteria_Gp7, which can provide functional resistance to annual drying, by sustaining ecosystem functioning under both wet and dry conditions.In addition, our results suggest that drying duration and frequency are relevant facets of a stream's phenology, as they integrate variations in the timing of water availability throughout the hydrological year.For example, annual drying duration exceeding 90 d or drying frequency surpassing one dry event per year indicate occurrences of dry events during spring and/or autumn.Collectively, our results support the hypothesis of functional compensation over the drying gradient and reinforce the role of selection effects in driving stream ecosystem functioning.
Previous studies investigating the impact of drying on individual ecosystem functions mediated by microbes have shown Fig. 2. Relationship between stream multifunctionality and multiple bacterial biodiversity aspects.Multifunctionality here was obtained through the weighted-averaging method.The regression fit (solid line), Spearman correlation coefficient (r S ) and p-value are shown for relationships between stream multifunctionality and Shannon diversity, phylogenetic diversity and the relative abundance of Sphingobacteriia and Acidobacteria_Gp7 (A).We also show the relationship between the relative abundance of bacterial taxa and stream multifunctionality and the three functional dimensions (B).MF indicates ecosystem multifunctionality, dim1 refers to photosynthetic activity, dim2 to organic matter decomposition and dim3 to microbial activity.Bacterial taxa are ranked following a decreasing sensitivity to drying.n = 32 sites.
contrasting responses, thus reinforcing the use of multifunctionality.These studies show that desiccation affects processes occurring in the surface biofilm, such as primary production and organic matter decomposition (Timoner et al. 2012;Gionchetta et al. 2019a;Arias-Real et al. 2020).However, our proxies of autotrophic production did not respond to drying, probably, because they were measured in sediments, which seem to be more resistant to drying events (Marxsen et al. 2010;Gionchetta et al. 2019a;Coulson et al. 2021;Arias-Real et al. 2022).For example, sediment organic matter decomposition can be maintained during drying events in patches with less exposure to solar radiation that preserve sufficient moisture (Arias-Real et al. 2020;Gionchetta et al. 2020a).In addition, episodic flash storms can resume microbial activity and increase the availability of labile carbon and nitrogen (Gionchetta et al. 2019a;Shumilova et al. 2019;von Schiller et al. 2019).However, although other studies have found that heterotrophic processes tend to increase in sediments exposed to recurrent drying (Acuña et al. 2015;Coulson et al. 2022;Zhang et al. 2023), our results suggest limited drying effects on proxies of autotrophic production and heterotrophic functions.Consequently, sediment biogeochemistry could be more resistant to drying than previously thought.Nonetheless, these findings should be verified by future studies exploring whether microbes can sustain stream multifunctionility under longer and more frequent drying conditions (Fu and Feng 2014).
Our findings highlight that drying causes few changes in bacterial diversity and community composition.Episodic rainfall events can maintain sufficient sediment moisture and labile carbon to enhance microbial activity, thus allowing most bacterial groups to persist during the dry period.This process might have been favored by the functional adaptations of drying specialists and the potential rapid colonization of early-soil bacteria (Allison and Martiny 2008;Shade et al. 2012).As occurs in soil ecosystems exposed to aridity, drying phenology might have driven the evolution of specialized bacterial traits, such as resting cysts, dormancy, osmolytes production, and protective pigments that promote resistance or quick recovery to drying stress (Coulson et al. 2022).
Previous studies in soil ecosystems have revealed positive associations between microbial diversity and multifunctionality (Delgado-Baquerizo et al. 2016, 2020;Fan et al. 2023).In contrast, our study strongly supports the idea that a few bacterial groups, specifically Sphingobacteriia and Acidobacteria_Gp7, rather than overall diversity, play a key role in sustaining stream multifunctionality.These results highlight the importance of drying specialists, selection effects and functional compensation as the primary mechanisms for maintaining stream multifunctionality in response to drying.Consequently, we did not find any evidence supporting niche complementarity and facilitation processes as drivers of multifunctionality.Sphingobacteriia and Acidobacteria_Gp7 are able to thrive in challenging and fluctuating environments, such as those with limited water and nutrients (Roger et al. 2016;Fan et al. 2021).Sphingobacteriia, which belong to the phylum Bacteroidota, are widely distributed Gram-negative bacteria that can thrive in both anaerobic and aerobic environments.These drying-resistant bacteria have been linked to abrupt wet events occurring in arid soil communities and can maintain their activity during dry periods (Barberan et al. 2014;Aslam et al. 2016).Acidobacteria_Gp7 belongs to the phylum Acidobacteria, which is known for its metabolic capabilities related to carbon degradation and its sensitivity to aridity (Maestre et al. 2015), potentially explaining the observed positive relationships with functional dimensions under wet conditions.Furthermore, we found a strong negative association between stream multifunctionality and most bacterial groups and bacterial phylogenetic diversity, which reinforces contrasting contributions of bacteria to stream multifunctionality.The fact that phylogenetic diversity showed a more negative association than bacterial ASV richness and diversity could be reflecting that the functional performance of stream bacteria is linked to their phylogenetic relatedness.Hence, taxa occurring in phylogenetically diverse communities appear to exhibit lower functional performance than Sphingobacteriia and Acidobacteria_Gp7 taxa, which showed a positive association with multifunctionality.However, taxonomic richness and diversity were unable to capture this pattern because they do not consider the phylogenetic similarity among taxa.In addition, other organisms (e.g., fungi and diatoms) may have influenced the functional measurements, and the focus on bacteria might have limited the capacity to detect complementarity effects on multifunctionality.The consideration of a wider set of stream taxa and functions in future studies can help to better assess the ecological mechanisms driving stream multifunctionality.
In conclusion, our findings contribute to a deeper understanding of the mechanisms governing stream multifunctionality under drying stress.We show, for the first time, that certain drying-resistant bacteria can maintain multiple ecosystem functions in stream sediments under both wet and drying conditions.Our study underscores the crucial role of bacterial biodiversity in providing functional resistance to human-induced drying in stream ecosystems and will help to predict biogeochemical responses to global change.

Fig. 1 .
Fig. 1.Responses of stream multifunctionality (A) and bacterial biodiversity (B) to drying duration (Spearman correlations).The regression fit (solid line), Spearman correlation coefficient (r S ) and p-value are shown for the relationships displayed in the scatterplots.Dashed lines in barplots represent the significance threshold for correlation values.n = 32 sites.

Table 1 .
Indicators used to characterize the three functional dimensions of stream multifunctionality.Their units and their contribution to stream multifunctionality are shown in the table.
DW and RAZ indicate the grams of sediment dry weight and the micromoles of resazurin, respectively.