TY - JOUR
AU1 - T, Le, Huong
AU2 - Emma, Rochelle-Newall,
AU3 - Yves, Auda,
AU4 - Olivier, Ribolzi,
AU5 - Oloth, Sengtaheuanghoung,
AU6 - Elisa, Thébault,
AU7 - Bounsamay, Soulileuth,
AU8 - Thomas, Pommier,
AB - ABSTRACT Impact of land use (LU) change on stream environmental conditions and the inhabiting bacterial community remains rarely investigated, especially in tropical montane catchments. We examined the effects of LU change and its legacy along a tropical stream by comparing seasonal patterns of dissolved organic carbon (DOC) / colored dissolved organic matter (CDOM) in relation to variations in structure, diversity and metabolic capacities of particle-attached (PA) and free-living (FL) bacterial communities. We hypothesized that despite seasonal differences, hydrological flows that accumulate allochthonous carbon along the catchment are a major controlling factor of the bacterial community. Surprisingly, local environmental conditions that were largely related to nearby LU and the legacy of LU change were more important for stream bacterial diversity than hydrological connectivity. DOC was strongly correlated with PA richness and diversity. The legacy of LU change between teak plantation and annual crops induced high DOC and high diversity and richness of PA in the adjacent waters, while banana plantations were associated with high diversity of FL. The community structures of both PA and FL differed significantly between seasons. Our results highlight the importance of vicinal LU change and its legacy on aquatic bacterial communities in mixed used tropical watersheds. LU legacy, DOC, CDOM, bacterial diversity, land management, hyporheic zone INTRODUCTION Although rarely the main focus of studies on aquatic microbial communities, land use (LU) change represents one of the most significant factors controlling stream diversity (Zeglin 2015). LU change is an anthropogenic activity referring to any transition from one LU category e.g. annual crops/fallow into another category e.g. agricultural lands, planted forests or urban areas. LU change can therefore have local and more wide ranging impacts on a variety of biological, chemical and physical processes. In catchments, LU change can affect downstream ecosystems due to the interactive links between soils and streams played by hydrological processes such as stream-groundwater interactions through the hyporheic zone (HZ) and overland flow processes (Febria et al.2012; Ward 2016; Lacombe et al.2017). In streams at base flow, water follows a complex path alternating between free flow along the surface of the stream bed and sub-surface transient storage throughout the streamed sediments of the HZ (Bencala and Walters 1983). LU change can induce changes in soil organic carbon, i.e. loss of soil organic carbon caused by conversion of primary forest into cropland (Axel, Jens and Annette 2011) or release of the recalcitrant soil organic carbon fraction due to conversion of peatlands to paddies (Wang et al.2017). These changes drive the composition of the soil microbial community (Tian et al.2017) and strongly impact catabolic responses of microbial communities (Mazzetto et al.2016). In aquatic environments, LU change is also acknowledged to strongly modify hydrological processes—i.e. evapotranspiration rates (Zhang et al.2013), runoff and flood peak discharge (Bahram et al.2007; Zare, Samani and Mohammady 2016), detachment rate (Lacombe et al.2017), overland flow and sediment yields (Ribolzi et al.2017). Recent work has shown that complex along-stream water paths through transient storage zones may influence microbial community composition at base flow (Kim et al.2017). Therefore, aquatic community structure might be driven by the local streambed sediment composition and characteristics (carbon content, texture, permeability, etc.) that vary spatially and temporally along the stream as a function of surrounding soil sources (Evrard et al.2016). Consequently, LU change plays an important role in the transport along streams of dissolved organic matter (DOM) (Wilson and Xenopoulos 2008; Williams et al.2010), as well as impacting the organisms in the aquatic ecosystem, such as planktonic microbes (Kamjunke, Herzsprung and Neu 2015) and pathogenic bacteria (Ribolzi et al.2015; Rochelle-Newall et al.2015). Bacteria play a key role in the functioning of freshwater ecosystems (Allison and Martiny 2008). Aquatic microbial communities may be affected by spatial and temporal variations in hydrological processes or in the type and degree of environmental change. In particular, seasonal shifts in water temperature and river flow can influence the diversity of temperate river bacterial communities (Crump et al.2003). Moreover, hydrological conditions have a differential impact on particle-attached (PA) and free-living (FL) bacteria (Luef et al.2007). Whereas the abundance of PA bacteria is influenced by the quality of particle matter imported from different flow paths, the abundance of FL bacteria is more strongly impacted by flow connectivity (Luef et al.2007). Changes in bacterial community composition are also linked to both autochthonous processes (e.g. phytoplankton blooms) and allochthonous factors (inputs of terrestrially derived organic matter), both of which can induce shifts in bacterial community structure (Crump et al.2003; Allgaier and Grossart 2006). Some studies have indicated that the sources and characteristics of dissolved organic carbon (DOC) could drive aquatic bacterial community functioning. For example in temperate streams, allochthonous DOM from agricultural activities has lower molecular weight compounds that were more labile to microbial community than DOM from forest and wetlands (Williams et al.2010). Similarly, fresh humic DOC originating from forest headwaters supported anabolism at a higher level of bacterial growth efficiency in adjacent rivers than peat harboring pre-processed humic DOC (Berggren and del Giorgio 2015). As part of the bulk DOM pool, chromophoric (or colored) dissolved organic matter (CDOM) can be used a proxy for changes in the chemical characteristics of the DOM pool (Fellman, Hood and Spencer 2010; Yamashita et al.2011; Ishii and Boyer 2012). Several indices, based on the absorption or fluorescence at specific wavelengths have been proposed. For example, the absorption at 254 nm divided by DOC concentration, also known as the SUVA254, provides a proxy for aromaticity (Weishaar et al.2003). The spectral slopes (S), determined over selected wavelength ranges such as between 275 and 295 nm or between 355 and 400 nm are surrogates for average molecular size and bioavailability and hence can provide information on the sources and diagenetic state of the DOM (Helms et al.2008). The spectral slope ratio, Sr, determined as S275–295 / S355–400, provides information on the apparent molecular weight of the DOM (Helms et al.2008; Fasching et al.2016). These indices can also be viewed as proxies for bioavailability. SUVA254 is negatively correlated with biodegradable DOC (Fellman et al.2008) and positively correlated with the hydrophobic organic acid fraction of DOM (Spencer, Butler and Aiken 2012), while higher S values indicate low molecular weight material and decreasing aromaticity and, hence, potentially increased bioavailability for bacteria (Fichot and Benner 2012). Fluorescence three-dimensional excitation-emission matrices (EEMs) of CDOM can also provide information on DOM source and on its bioavailability (Fellman, Hood and Spencer 2010). For example, EEMs have been used to study the effects of LU on stream biogeochemistry (Wilson and Xenopoulos 2009) or to follow DOM bioavailability in streams (Wickland, Neff and Aiken 2007; Fellman et al.2009). Yet, to our knowledge, there is no integrated understanding of how current and past LU change (or the legacy of LU) via hydrological processes, in relation to DOC, impacts PA and FL stream bacteria communities, especially in the tropical regions. In Southeast Asia, LU change such as the replacement of annual crops with tree plantations in upland catchment areas is increasing rapidly (Ribolzi et al.2017). Moreover, these ecosystems have specific geographical, pedological and meteorological characteristics (high volume, intense rainfall events, high slope and low soil organic carbon contents). The combination of LU change with these specific environmental factors results in high rates of soil erosion, a process known to be predominantly driven by LU change rather than environmental characteristics at the catchment scale (Valentin et al.2008). Here we investigated how LU change and its legacy impacted the concentration and characteristics of DOC and CDOM in the stream, and how these two factors subsequently drove the dynamics of aquatic bacterial metabolic capacity and diversity along the Houay Pano stream, Laos. This archetypal mixed LU catchment harbors typical seasonal variations such as found in the tropics. In the past two decades, it has experienced rapid LU change from upland crops (e.g. upland rice) with long fallow periods to annual crops (e.g. job's tears and maize) or forests (e.g. teak and bananas plantations) with reduced fallows and traditional slash and burn management techniques that removed the presence of a viable understory (Patin et al.2012). The loss of the understory led to patches of bare soil with increased soil crusting rates and decreased soil permeability resulting in increased overland flows and soil losses (Ribolzi et al.2017). The recurrence of annual crops without any fallow or rotation had a strong negative impact on soil fertility at the catchment scale (Valentin et al.2008). It can therefore be hypothesized that the legacy of such LU change strongly modified stream microbial metabolic capacity and diversity, as a consequence of disrupted recycling of dissolved organic matter. As the streams in such steep catchments drain relatively large spatial areas under diverse LU practices, we hypothesized that despite seasonal differences, hydrological flow that accumulates allochthonous carbon along the catchment was a major controlling factor of the stream bacterial community structure. MATERIALS AND METHODS Study site The study region is located in the Houay Pano catchment area (19°51′N–102°10′E), 10 km South-West of Luang Prabang city, Laos (Fig. 1). The regional climate is classified as monsoon tropical with two seasons: a wet season from April to September and a dry season from October to March. The average annual precipitation is 1585 mm year−1, more than 80% of which occurs during the wet season (Ribolzi et al.2008). Figure 1. View largeDownload slide Study site: (A) location of Laos in Southeast Asia, (B) location of the Houay Pano catchment and land LU in 2015 Figure 1. View largeDownload slide Study site: (A) location of Laos in Southeast Asia, (B) location of the Houay Pano catchment and land LU in 2015 Five sites along the river were sampled (Table 1). Around each sampling site, concentric drainage areas with radius 50, 100, 150, 200 and 250 m were constructed from the TanDEM-X Digital Elevation Model (Fig. S1, Supporting Information). Each drainage area is cumulative i.e. it includes the smaller areas. This spatial structure is used to analyze soil type, LU and its changes. Table 1. Description of UTM coordinates and LU characteristics around sampling sites. UTM coordinates (WGS84) Sampling sites Easting Northing LU characteristics RIB72 204 158.163 2 198 513.314 Upmost site of the catchment, surrounded mostly by fallow and teak plantations S1 203 896.417 2 198 512.136 First gauging stations on the stream, surrounded by bananas and teak plantations as well as rotating lands RIB48 203 670.637 2 198 364.471 Typical upland bog that is surrounded with teak plantation S4 203 501.245 2 197 660.376 Second gauging station of the stream that is surrounded with teak plantations S9 203 597 2 197 416 Outlet of the village Lak Sip UTM coordinates (WGS84) Sampling sites Easting Northing LU characteristics RIB72 204 158.163 2 198 513.314 Upmost site of the catchment, surrounded mostly by fallow and teak plantations S1 203 896.417 2 198 512.136 First gauging stations on the stream, surrounded by bananas and teak plantations as well as rotating lands RIB48 203 670.637 2 198 364.471 Typical upland bog that is surrounded with teak plantation S4 203 501.245 2 197 660.376 Second gauging station of the stream that is surrounded with teak plantations S9 203 597 2 197 416 Outlet of the village Lak Sip View Large Table 1. Description of UTM coordinates and LU characteristics around sampling sites. UTM coordinates (WGS84) Sampling sites Easting Northing LU characteristics RIB72 204 158.163 2 198 513.314 Upmost site of the catchment, surrounded mostly by fallow and teak plantations S1 203 896.417 2 198 512.136 First gauging stations on the stream, surrounded by bananas and teak plantations as well as rotating lands RIB48 203 670.637 2 198 364.471 Typical upland bog that is surrounded with teak plantation S4 203 501.245 2 197 660.376 Second gauging station of the stream that is surrounded with teak plantations S9 203 597 2 197 416 Outlet of the village Lak Sip UTM coordinates (WGS84) Sampling sites Easting Northing LU characteristics RIB72 204 158.163 2 198 513.314 Upmost site of the catchment, surrounded mostly by fallow and teak plantations S1 203 896.417 2 198 512.136 First gauging stations on the stream, surrounded by bananas and teak plantations as well as rotating lands RIB48 203 670.637 2 198 364.471 Typical upland bog that is surrounded with teak plantation S4 203 501.245 2 197 660.376 Second gauging station of the stream that is surrounded with teak plantations S9 203 597 2 197 416 Outlet of the village Lak Sip View Large LU change and its legacy The Houay Pano catchment is characteristic in that it is experiencing the rapid changes in LU occurring over Southeast Asia. This catchment has been monitored since 1998 and during this time LU has changed from shifting cultivation with long fallows to planted forests, notably teak plantations. The area occupied by teak plantations has increased from 20% in 2008 to nearly 40% area in 2012 (Huon et al.2013). Between June 2014 and March 2015, LU did not change significantly in Houay Pano and fourteen LU types were recorded (Table 2). Table 2. LU in the wet season and dry season (unit: ha). LU Wet season Dry season Fish 0.12 0.12 Sweet potato 0.1 - Vegetable gardent and fruit tree 0.245 0.245 Broom grass field 0.345 0.345 Maize 0.092 0.092 Job's stear 0.589 0.111 Banana 2.782 2.782 Forest (decidious) 0.832 0.832 Forest (Dipterocarpe) 4.693 4.693 Teak plantation 23.98 23.87 Upland rice 1.788 1.18 Village 1.428 1.428 Rotating land 29.55 31.16 LU Wet season Dry season Fish 0.12 0.12 Sweet potato 0.1 - Vegetable gardent and fruit tree 0.245 0.245 Broom grass field 0.345 0.345 Maize 0.092 0.092 Job's stear 0.589 0.111 Banana 2.782 2.782 Forest (decidious) 0.832 0.832 Forest (Dipterocarpe) 4.693 4.693 Teak plantation 23.98 23.87 Upland rice 1.788 1.18 Village 1.428 1.428 Rotating land 29.55 31.16 View Large Table 2. LU in the wet season and dry season (unit: ha). LU Wet season Dry season Fish 0.12 0.12 Sweet potato 0.1 - Vegetable gardent and fruit tree 0.245 0.245 Broom grass field 0.345 0.345 Maize 0.092 0.092 Job's stear 0.589 0.111 Banana 2.782 2.782 Forest (decidious) 0.832 0.832 Forest (Dipterocarpe) 4.693 4.693 Teak plantation 23.98 23.87 Upland rice 1.788 1.18 Village 1.428 1.428 Rotating land 29.55 31.16 LU Wet season Dry season Fish 0.12 0.12 Sweet potato 0.1 - Vegetable gardent and fruit tree 0.245 0.245 Broom grass field 0.345 0.345 Maize 0.092 0.092 Job's stear 0.589 0.111 Banana 2.782 2.782 Forest (decidious) 0.832 0.832 Forest (Dipterocarpe) 4.693 4.693 Teak plantation 23.98 23.87 Upland rice 1.788 1.18 Village 1.428 1.428 Rotating land 29.55 31.16 View Large The most important LU change occurred between 2012 and 2014 when the transition from crop to teak occurred in some areas of the catchment. To track the LU change, LU classes were simplified (Table S1, Supporting Information) e.g. cultivated parcel followed with fallow the next year was not regarded as a LU change. According to this new LU typology, spatial zones were delimited as a function of the changes occurring between 2012 and 2015. Taking into account ground surveys and geo-referencing accuracies, a treatment based on a majority filter (window 5 × 5, pixel size 10 m) suppressed isolated pixels. Plots smaller than 350 m2 were removed from the calculation (Fig. S2, Supporting Information). Sample collection Stream water samples were collected by hand in 1.5 L new, clean plastic bottles that were rinsed 3 times with sample before collection. Five sites were sampled along the stream during base flow during the wet season (12th June 2014) and the dry season (22th March 2015). Within a maximum of 2 h after collection, the samples were returned to the laboratory for filtration. Samples were maintained at ∼20°C–22°C (in situ water temperature) and in the dark pending filtration (<2 h). Immediately prior to sub-sampling, samples were vigorously shaken to limit sedimentation biases; i.e. to insure sampling of suspended particles. Measurement of DOC and CDOM A subsample of 200 mL of each sample was filtered through Whatman GF/F glass fiber filters to remove large particles. For the determination of DOC concentration, duplicate 30 mL of this filtrate was kept in pre-combusted (450°C, overnight) glass tubes, preserved with 36 µL 85% phosphoric acid (H3PO4) and sealed with a Teflon lined cap in order to remove inorganic carbon. Samples were stored at ambient temperature and in the dark until measurement. DOC concentration was measured on a Shimadzu total organic carbon VCPH analyzer following the method described in Rochelle-Newall et al. (2011). DOC concentration is expressed as mg of organic C per liter of stream water. For CDOM, 100-mL filtered samples were stored in pre-clean 125-mL amber glass bottles seal with Teflon lines cap. After collection, the samples were stored frozen (−20°C) until measurement. Before the optical measurements, the samples were thawed slowly to room temperature and re-filtered at 0.2 μm (Sartorius Minisart NML Syringe filters) to remove particles. CDOM absorption was measured with a spectrophotometer (Analytica.Jena Specord 205 UV-VIS) from 200 to 750 nm using a 10 cm quartz cell and Milli-Q water as the blank. The absorbance values were converted to absorption coefficient, α(λ), m−1, calculated as \begin{eqnarray*} 2.303*{\rm{A}}/{\rm{l}} \end{eqnarray*} (1) Where A is the absorbance (log Io/L), l is the path length (in m), and 2.303 converts between log10 and natural log. SUVA254, a surrogate for DOM aromaticity, was also calculated from DOC-normalized absorption coefficient at wavelength 254 nm (a(254):DOC), expressed in units of mg−1 L m−1. The dependence of α(λ) on λ is described using equ (2) \begin{eqnarray*} \alpha \left( \lambda \right) = \alpha \left( {{\lambda _0}} \right){e^{ - {\rm{S}}\left( {\lambda - \lambda 0} \right)}} \end{eqnarray*} (2) Where α(λ) is the absorption coefficient at wavelength λ, α(λ0) is the absorption coefficient at a reference wavelength, and S is the spectral slope coefficients in the λ0 – λ-nm spectral range. The spectral slopes of absorption coefficient for 275–295 nm (S275–295), 350–400 nm (S350–400) was estimated using a linear fit of the log-linearized a (λ) spectrum (equ2) over spectral ranges and the ratio of the two slopes (Sr) was calculated. Excitation—Emission—Matrices (EEM) measurements were made on a Gilden Fluorosens fluorometer using a 1 cm quart cuvette with 5 nm bandwidths for excitation and emission at an integration time of 100 ms. Excitation scans were made over a range of 200 to 450 nm at 5 nm increments and emission from 220 to 600 nm at 2.5 nm increments. EEMs were corrected for inner filter effects and the manufacturers’ machine correction was applied. EEM fluorescence of Milli-Q water blank was subtracted from that of sample EEM. PARAFAC analysis was carried out in MATLAB (version R2016a 9.0.0) with the DOMFluor toolbox for MATLAB (Murphy et al.2013) in order to decompose the fluorescence signal into a series of tri-linear structures. Processing data were done to minimize the impact of scatter lines that caused some mathematical difficulties, including removal of Rayleigh and Raman scatter. EEM wavelength ranges were reduced to excitation 290–450 nm and emission 300–600 nm. PARAFAC model generated three components using EEMs from our samples (n = 10). The three-component model (C1–C3) was validated using split-half and random initialization methods. Metabolic capacities of bacterial communities Biolog® Ecoplate was used to determine catabolic capacity of bacterial communities. These 96-well micro-plates include triplicates of 31 different carbon-based substrates and one water control. Each well also contains a tetrazolium violet dye that turns purple upon substrate oxidation (Stefanowicz 2006). One mL water sample was diluted 10 times with 0.2 μm filtered stream water to avoid osmotic shock and allow detection of color changes. One hundred and fifty (150) μL of this diluted sample was added to each well, and the plates were then incubated in the dark at 25°C–27°C without shaking. Following a short shaking (2 s at 100 rpm) well color development was measured at 590 nm using a portable spectrophotometer (Bio-Rad laboratories, iMark Microplate Reader) after 12 h incubation and then every 24 h for 72 h. The water blank was subtracted from the carbon substrates, which were then grouped into six substrate families (amine, polymers, phenolic acids, carboxylic acids, carbohydrates and amino acids) (Pommier et al.2014). Metabolic potential of each community was normalized to each substrate family relative to the highest performance in the substrate family. Assessment of bacterial diversity DNA extraction and 16S rRNA gene sequencing Fifty (50) mm of each sample was first filtered through a 3 μm pore size filter (Polycarbonate Whatman) and a 0.2 μm (Polycarbonate Supor) filters to separate Particles Attached (PA) fraction (>3.0 µm) from FL fraction (<3.0 µm and >0.2 µm) (Crump, Armbrust and Baross 1999). All filters were stored at −20°C until DNA extraction that followed a modified protocol adapted from Fuhrman et al. (1988). Briefly, the filters were cut in two halves using ethanol cleaned scissors and placed in separate Eppendorf® tubes. One tube was placed on ice and 525 μL of lysis buffer was added to start breaking Gram negative bacterial cells. The other tube was stored at −20°C as a back-up filter. Three cycles of freeze-thaw switches (65°C for 2 min—ice for 5 min) were then performed to facilitate cell membrane breakage. Then 0.5 g of glass beads (0.5 mm) was added into the tubes. The tubes were shaken for 45 s at 6 m s−1 using a Fastprep® (Millipore, Fastprep®-24, USA) and were incubated at 4°C for 5 min; these steps were performed twice. 11 μL of Lyzozyme (1 mg mL−1, final concentration) was added to the tube and left for 30 min at 37°C to break down bacterial cell wall (cell lysis). Sodium dodecyl sulfate (SDS 10%) and proteinase K (final concentration, 100 μg mL−1) were added to the tubes and were incubated at 55°C for 2 h under constant shaking conditions (180 rpm) to remove lipid membrane. A sodium chloride (NaCl 5M) and a cetyltrimethylammonium bromide (CTAB) solution (final concentration, 1% in a 0.7 M NaCl solution) was added to the tubes, mixed and incubated at 65°C for 10 min to separate DNA from protein. Nucleic acids were then extracted twice from digestion products with phenol-chloroform-isoamyl alcohol (25:24:1); the aqueous phase containing nucleic acids was kept and purified by adding phenol-chloroform-isoamyl alcohol (25:24:1). After isopropanol (0.6 volume of tube) addition, the nucleic acids were precipitated at −20°C for 12 h. After centrifugation, the DNA pellet was rinsed with 99% pure ethanol to remove the salt previously added. The samples were spun 20 min at maximum speed (15 000 rpm) at 4°C and the supernatant was removed. The DNA pellets were dried in a Speed Vac® (Labconco, USA) for 10 min and re-suspended in 50 μL of molecular cleaned distilled water. Nucleic acid extracts were stored at −20°C and sent overnight to Molecular Research Laboratories (Texas, USA) for further PCR amplification, product cleaning, library construction and high throughput sequencing. The PCR primers 515F AGRGTTTGATCMTGGCTCAG and 806R GTNTTACNGCGGCKGCTG (producing a 291 bp amplicons) (Capone et al.2011) with sample-specific barcodes on the forward primer were used to amplify the V4 variable region of the 16S rRNA gene. Thirty (30) cycles of PCR were performed using the HotStarTaq Plus Master Mix Kit (Qiagen, USA) under the following conditions: 94°C for 3 min, followed by 28 cycles of 94°C for 30 s, 53°C for 40 s and 72°C for 1 min, after which a final elongation step at 72°C for 5 min was performed. After amplification, PCR products were checked in 2% agarose gel to determine the success of amplification, the relative intensity of bands and to check the expected amplicon size (291 bp). The 10 samples included in this study were pooled together in equal proportions based on their molecular weight and DNA concentrations. Pooled PCR products were purified using calibrated Agencourt AMPure XP magnetic beads according to the manufacturer's guidelines. Briefly, the pooled PCR products were incubated with magnetic beads (1.8:1 ratio of beads to sample) that pull down the PCR product to the bottom of a tube for subsequent washing steps. The clean PCR product is then release from the magnetic beads for elution. The pooled and purified PCR product was then used to prepare a DNA library by following Illumina TruSeq DNA library preparation protocol. Sequencing was performed at MR DNA (www.mrdnalab.com, Shallowater, TX, USA) on an Illumina MiSeq machine following the manufacturer's guidelines. Sequences processing and data analysis The MOTHUR software (v. 1.33) (Schloss et al.2009) was used to process 16S rRNA gene sequence reads following the SOP. From the original 632,196 and 1313 266 reads for PA and FL respectively, short reads (<250 bp) and reads with ambiguous primer or barcode sequences were discarded. Corresponding reads were paired in single sequences with an average length of 486 bp. Sequencing errors were reduced by aligning remaining reads to the SILVA database (Pruesse et al.2007), screening the alignment to the overlapping region, and pre-clustering sequences distant by <2 bp. Chimeric sequences were identified using the integrated version of UChime (Edgar et al.2011) and removed accordingly. To avoid misinterpretation, sequences that were classified as ‘Chloroplast’, ‘Mitochondria’ or ‘unknown’ lineages were removed before clustering into Operational Taxonomic Units (OTUs). The remaining 494 724 and 1023 897 sequences for PA and FL respectively were clustered in OTUs with a pairwise distance <0.03 substitutions per nucleotide with average neighbor method and considered for further analyses. Taxonomic assignments were performed on the alignment of consensus sequences with the RDP database (Cole et al.2005). After this streaming process, 90 858 and 115 573 OTUs were found for PA and FL respectively from which 18 359 sequences were randomly subsampled for each site to perform community comparisons. Diversity indexes were calculated to compare richness and evenness of bacterial community between different sites. Bray–Curtis distances were also calculated to estimate the dissimilarity in structure between all samples, and represented as a dendrogram with tree.shared command in MOTHUR. Accession numbers: The raw sequence files were deposited in the NCBI sequence Read Archive under BioProject no. PRJNA416158. This study project has been deposited at DDBJ/ENA/Genbank under the accession KBTU00000000. The version described in this paper is the version KBTU01000000. Statistics and partial least square analysis DOC concentrations and optical indices between different sites and seasons were compared using ANOVA in R studio after having checked the normal distribution of the data and the homoscedasticity of the residuals. When necessary, the data were log-transformed to assure normality. Turkey HSD was used to test pairwise comparison between sites. Correlations between DOM concentrations and optical index, alpha diversity (i.e. Chao, Invsimpson, Shannon index) was estimated using the Pearson correlation coefficient with the Hmisc package in R (Harrell 2006). AMOVA test was performed using MOTHUR software to determine the significance of the difference in community structure between dry and wet season for PA and FL communities. All differences were considered significant when the P-value of the tests was lower than to 0.05. To explain the influence of environment on the bacterial fauna, the data were organized in two tables. The first table described the environmental variables obtained by physicochemical measurements or built from soil type, soil occupation and its changes within concentric drainage areas. The second table characterized the bacterial communities. To disentangle the seasonal from the land-use effects, we considered the samples taken during the dry and the wet season separately. The analysis focuses on factors explaining the typology of the stations. The typology of seasons was considered after as it mainly relies on physicochemical variables with low weight. Two reasons justify the use of Partial Least Square (PLS) analysis to explain the relationship between bacterial diversity and environmental variables: (i) the number of variables in both of the tables was much greater than the number of individuals; and (ii) the low sensitivity of PLS to multicollinearity (Abdi 2010) was adapted to our dataset which presented some redundancy, especially for data obtained from the concentric drainage areas. All factor computing between the bacterial diversity and the environment tables was conducted using the R packages Plsdepot (Sanchez 2012) and the graphical identification of the relevant variables was performed using ADE-4 (Thioulouse et al.1997). As the computing procedures produced numerous and redundant data that were hard to interpret, the PLS was applied using a two-step procedure. A first PLS was performed and the relevant variables retained based on the VIP (Variable Importance for the Projection) criterion where values of VIP must be greater than 0.8. If two variables describe the same environmental characteristic and differ only by the size of the concentric drainage areas, only the variable with the smallest size was kept. For example, as Banana builds LU variables with VIP equal to 2.24 for three concentric areas of radius 50, 100 and 150, only ‘Banana 50’ was retained. Q2 was also used to estimate the number of factors to be retained. A factor was retained if its Q2 value was greater than an arbitrary value generally set to 0.0975 (Abdi 2010). All selected variables are presented in the Supplementary information (Tables S1, S2 and S3, Supporting Information). A second PLS was applied with the variables retained from the first PLS to select the most important factors among the environmental variables that could potentially drive bacterial metabolic capacities and the diversity of the most abundant OTUs. Co-occurrence network To identify the specific response of the OTUs to environmental variables (i.e. DOC/CDOM characteristics), co-occurrence network analysis was performed on 4 datasets that separated the OTUs occurring at each of the sample sites into wet and dry seasons and PA from FL fractions separately. To limit the number of correlations appearing in the network, only OTUs with relative abundances >0.1% in at least one sample from the wet or dry season for the PA or FL fraction was retained. Randomness in co-occurrence of OTUs in the dry and wet seasons datasets was tested against a null model using the quasiswap algorithm (Miklós and Podani 2004). Checkerboard score (C-score) metric that measures species segregation was performed on 50 000 simulations (Stone and Roberts 1990) using the oecosimu command in the vegan package in R (Oksanen et al.2007). The C-score represents the mean number of checkerboard units (CUs) per pairs of OTUs in each dataset and does not require perfect checkerboard distributions (Claire et al.2007). The number of CUs for any species pair can be calculated as (Stone and Roberts 1990): \begin{equation*} {\rm{CU}} = \left( {{{\rm{R}}_{\rm{i}}} - {\rm{S}}} \right)\left( {{{\rm{R}}_{\rm{j}}} - {\rm{S}}} \right) \end{equation*} where Ri and Rj are the total abundances of OTU i and j, and S is the number of sites containing both species. The co-occurrence network for each season was built based on the Spearman correlation matrix with the Hmisc package in R (Harrell 2006). All results with high Spearman correlation coefficients (i.e. R >0.8) and significance levels (P <0.05) were used to build the network. The nodes in each network represent OTUs and the edges that connect these OTUs show correlation coefficient between OTUs. The connectivity of the network is equal to the number of edges in the network divided by the potential number of edges if the network was fully connected ( = number of nodes × (number of nodes − 1)/2). Network images were generated using Gephi v. 0.9.1 using the Fruchterman–Reingold layout algorithm (Bastian, Heymann and Jacomy 2009). Methodological approach Our method of 16S rRNA sequencing allows identification of the community structure and taxonomy of stream bacteria. Though having limited functional resolution and lower sensitivity, the 16S rRNA gene gives broad shifts in community diversity over time with a wide, comprehensive 16S rRNA gene database for deep taxonomy as compared to metagenomic data (Poretsky et al.2014). In our study, 16S rRNA allowed the assignation of the bacteria community that was correlated with DOC/CDOM to genus level and the construction of a co-occurrence network. RESULTS Legacy of LU change In 2014 and 2015 when our sampling occurred, teak plantations covered a large proportion of the catchment surface area (approximately 23.3 ha (35%), distributed from middle to downstream sections of the catchment. Shifting cultivation with rotating LU including periods of fallow remained consequent with approximately 30 ha (46%), mostly distributed in the upstream sections (Table 2). A part of the catchment was also covered by upland rice and crops such as banana, maize and Job's tears that are scattered along the stream length (Fig. 1). Overall, LU was stable between 2014 and 2015. The only exception was a slight decrease in the surface covered by upland rice and increase of that covered by fallow areas/rotating LU in 2015 (Table 2). Between 2012 and 2015, only 7% of the area was affected by LU change: annual crops to teak plantations (3 ha), teak plantations to annual crops (1.5 ha) and the transition from forests into annual crops (0.5 ha) (Table S2, Supporting Information). Nevertheless, even though these changes apply to small areas, the following analyzes show their significant influence, probably because they were located near our sampling sites. Variations in DOC/CDOM characteristics DOC concentrations differed significantly between seasons (ANOVA, P <0.001) in the gauging stations (S1 and S4), in the upland bog (RIB48) and notably, at the outlet of the village (S9), where DOC concentration was two times higher in the dry season. DOC concentration was highest in the upland bog (RIB48) and at the outlet of the village (S9) in the wet season (3.2 ± 0.06 mg L−1 and 3.05 ± 0.03 mg L −1, respectively; Table 3). Similarly in the dry season, the sample from the outlet of the village (S9) showed the highest DOC concentration (6.25 ± 0.16 mg L−1), which was six times higher than the lowest concentrations at the two uppermost sites (RIB72 and S1). Table 3. DOC concentration (mg L−1) and its optical indices (SUVA254 (mg−1 L m−1) and Sr) in different sites in the wet and dry season. Season Site DOC (mg L−1) SUVA254 (mg−1 L m−1) Sr Wet season RIB72 1.28±0.04 2.61±0.05 1.07±0.07 S1 1.47±0.02 7.67±0.06 0.57±0.007 RIB48 3.21±0.06 4.60±0.08 0.87±0.01 S4 2.25±0.02 7.27±0.07 0.83±0.001 S9 3.05±0.04 5.25±0.08 0.93±0.01 Dry season RIB72 1.17±0.01 2.59±0.01 1.1±0.04 S1 1.11±0.01 5.53±0.07 0.80±0.01 RIB48 2.13±0.01 5.09±0.01 0.83±0.001 S4 2.00±0.01 6.62±0.02 0.75±0.007 S9 6.25±0.16 3.77±0.02 0.99±0.02 Season Site DOC (mg L−1) SUVA254 (mg−1 L m−1) Sr Wet season RIB72 1.28±0.04 2.61±0.05 1.07±0.07 S1 1.47±0.02 7.67±0.06 0.57±0.007 RIB48 3.21±0.06 4.60±0.08 0.87±0.01 S4 2.25±0.02 7.27±0.07 0.83±0.001 S9 3.05±0.04 5.25±0.08 0.93±0.01 Dry season RIB72 1.17±0.01 2.59±0.01 1.1±0.04 S1 1.11±0.01 5.53±0.07 0.80±0.01 RIB48 2.13±0.01 5.09±0.01 0.83±0.001 S4 2.00±0.01 6.62±0.02 0.75±0.007 S9 6.25±0.16 3.77±0.02 0.99±0.02 View Large Table 3. DOC concentration (mg L−1) and its optical indices (SUVA254 (mg−1 L m−1) and Sr) in different sites in the wet and dry season. Season Site DOC (mg L−1) SUVA254 (mg−1 L m−1) Sr Wet season RIB72 1.28±0.04 2.61±0.05 1.07±0.07 S1 1.47±0.02 7.67±0.06 0.57±0.007 RIB48 3.21±0.06 4.60±0.08 0.87±0.01 S4 2.25±0.02 7.27±0.07 0.83±0.001 S9 3.05±0.04 5.25±0.08 0.93±0.01 Dry season RIB72 1.17±0.01 2.59±0.01 1.1±0.04 S1 1.11±0.01 5.53±0.07 0.80±0.01 RIB48 2.13±0.01 5.09±0.01 0.83±0.001 S4 2.00±0.01 6.62±0.02 0.75±0.007 S9 6.25±0.16 3.77±0.02 0.99±0.02 Season Site DOC (mg L−1) SUVA254 (mg−1 L m−1) Sr Wet season RIB72 1.28±0.04 2.61±0.05 1.07±0.07 S1 1.47±0.02 7.67±0.06 0.57±0.007 RIB48 3.21±0.06 4.60±0.08 0.87±0.01 S4 2.25±0.02 7.27±0.07 0.83±0.001 S9 3.05±0.04 5.25±0.08 0.93±0.01 Dry season RIB72 1.17±0.01 2.59±0.01 1.1±0.04 S1 1.11±0.01 5.53±0.07 0.80±0.01 RIB48 2.13±0.01 5.09±0.01 0.83±0.001 S4 2.00±0.01 6.62±0.02 0.75±0.007 S9 6.25±0.16 3.77±0.02 0.99±0.02 View Large SUVA254, a proxy for DOC aromaticity, ranged from 2.61 ± 0.05 to 7.67 ± 0.06 in the wet season, and 2.59 ± 0.01 to 6.62 ± 0.02 in the dry season. The highest SUVA254 index was found at the two gauging stations S1 and S4 (7.67 ± 0.06 and 7.27 ± 0.07, respectively) in the samples from the wet season, and at S4 (6.62 ± 0.02) in the samples from the dry season (Table 3). Sr, which is negatively correlated with molecular weight, ranged in the wet season from 0.57 ± 0.007 at the first gauging station (S1) to 1.07 ± 0.07 at the uppermost site (RIB72). The samples from the dry season ranged from 0.75 ± 0.007 at S4 to 1.1 ± 0.04 at the uppermost site (RIB72) (Table 3). SUVA254 was negatively correlated with Sr (Correlation coefficient, R = −0.75, P <0.01). The PARAFAC model on 10 EEMs identified and validated three components. Component 2 (C2) presented a fluorescent peak at an excitation/emission (Ex/Em) wavelength of 300/355 nm, which is similar to Tryptophan-like fluorescence (Yamashita and Tanoue 2003; Zhang et al.2010). Component 1 (C1) has two peaks at an Ex/Em wavelength of 330(340)/452(467) that respectively resemble fulvic acid (Coble 1996, 2007; Yamashita et al.2011) and humic fluorophore group exported from agricultural sub-catchments (Stedmon and Markager 2005). Component 3 (C3) showed a peak at an Ex/Em of 290(390)/470 that was also identical to the previously identified humic-like component (Stedmon and Markager 2005). While the Tryptophan-like component (C2) was positively correlated with SUVA254 (Correlation coefficient, R = 0.6, P<0.01) and negatively correlated with Sr (Correlation coefficient, R = -0.6, P <0.01), fulvic acid-like (C1) and humic-like (C3) components were highly and positively correlated with each other (Correlation coefficient, R = 0.99, P <0.01), and they also showed strong correlation with DOC concentration (Correlation coefficient, R = 0.74–0.87, P <0.01). Patterns of carbon metabolic capacities of bacterial communities Except for the uppermost site (RIB72), the pattern of metabolic activity (after 72 h incubation) was similar for the different samples at both seasons but did not increase from upstream to downstream (Fig. 2). For all sites but except for RIB72, potential metabolic capacities were higher at dry than at wet season (ANOVA, P = 0.0013). Substrate utilization was the highest in the upland bog (RIB48), followed by the outlet of the village (S9), then the reference gauging station S4, and was the lowest at the most elevated gauging station (S1). This order was similar to the one observed for total DOC concentration in the wet season. Figure 2. View largeDownload slide Potential metabolic capacity in different sites along the stream in the wet and dry seasons (as measured by Biolog Ecoplate). The carbon substrates were grouped into six substrate families. The position on each of the six axes indicates the highest potential degradation rate of that group of substrates by the community after a 72-h incubation. Figure 2. View largeDownload slide Potential metabolic capacity in different sites along the stream in the wet and dry seasons (as measured by Biolog Ecoplate). The carbon substrates were grouped into six substrate families. The position on each of the six axes indicates the highest potential degradation rate of that group of substrates by the community after a 72-h incubation. Bacterial community structure and diversity Similar to metabolic capacities, the patterns of community structure were more related to local site specificities than to hydrological processes. Moreover, bacterial communities in both FL and PA fractions showed clear seasonal patterns in structure (Fig. 3A,B; AMOVA, P<0.01). In the wet season, both FL and PA bacterial communities sampled at the sites harboring the highest DOC concentrations (i.e. RIB48 and S9) were highly similar, and those sampled at the two gauging stations (S1 and S4) showed alike structures. As observed above for the metabolic capacities, the communities sampled at the uppermost site (RIB72) did not follow this pattern. The FL fraction and PA fraction in the dry season had low similarity between sites along stream path from upstream to downstream in the study. Figure 3. View largeDownload slide Bray–Curtis dissimilarity of community structure based on relative species abundances was used to cluster all samples in the dry and wet seasons for (A) PA and (B) FL fraction. Figure 3. View largeDownload slide Bray–Curtis dissimilarity of community structure based on relative species abundances was used to cluster all samples in the dry and wet seasons for (A) PA and (B) FL fraction. The 90 858 and 115 573 OTUs from the FL and PA fraction belonged to 35 and 39 different phyla, respectively (Fig. S3, Supporting Information). The dominant bacterial phyla were affiliated to Proteobacteria, Actinobacteria, Bacteroidetes and Firmicutes. Overall, the relative abundance of OTUs belonging to Firmicutes was higher in the dry season than in the wet season for both FL and PA fraction. In the PA fraction (Fig. S3A, Supporting Information), OTUs belonging to Proteobacteria were relatively more abundant in the wet season than in the dry season. In the wet season, Proteobacteria dominated in all sites, in particularly at the first gauging station S1 (85.4% of all OTUs were classified as Proteobacteria). OTUs classified as Actinobacteria and Bacteroidetes were also highly abundant phyla in the communities sampled in the wet season. Similar to the pattern found in the wet season, Proteobacteria found in the dry season dominated the communities at the outlet of the village (S9). Notably, the community structure from the dry season sampled at the second gauging station (S4) differed from the wet season, with a dominance of OTUs belonging to Bacteroidetes and Firmicutes, although they both were under the influence of teak plantation. In the FL fraction (Fig. S3B, Supporting Information), the relative abundance of OTUs belonging to Proteobacteria increased in the wet season, but decreased in the dry season from the uppermost station (RIB72) to the outlet of the village (S9). For this fraction, a large proportion (27.3%–56% of total) of the OTUs found in the samples from the wet season could not be classified, except for the outlet of the village (S9) where Proteobacteria dominated. Similar to the metabolic capacity, the community structure sampled at the uppermost site (RIB72) in the wet season differed from the other communities due to a large extent to the relative dominance of OTUs belonging to Bacteroidetes. In contrast at the dry season, Proteobacteria dominated the communities sampled in all sites, followed with Firmicutes and Bacteroidetes. Remarkably, high relative abundance of OTUs belonging to Firmicutes (40%) was found at the outlet of the village (S9). The richness and diversity of bacteria also did not increase from upstream to downstream and showed differences between sites (Table S4, Supporting Information). The richness of PA bacteria was highest at RIB48 and S9 as compared to the other sites in the wet season, whereas in the dry season it was highest at S1. For FL bacteria, the highest richness occurred at S9 in the wet season and at S1 in the dry season. The diversity of PA bacteria was high at RIB48 and S4 for both the wet and dry seasons. Similarly, the diversity of FL bacteria was high at S4 for both seasons (Table S4, Supporting Information). DOC concentration was significantly, positively correlated with the richness and the diversity of PA community in the wet season (Correlation coefficient R = 0.9 and 0.81, P<0.01, respectively), whereas Sr was negatively correlated with Shannon diversity (Correlation coefficient R = -0.8, P<0.01). Impact of LU, the legacy of LU change and environmental variables on aquatic bacterial communities The first PLS regression analysis (data not shown) indicated that only the first axis was interpretable (axis 1, Q2 = 0.03, axis 2, Q2 = -0.07). The second PLS regression analysis indicated that the sample sites were grouped on the first axis according to site characteristics rather than according to seasonal variations (axis 1, Q2 = 0.19, axis 2, Q2 = -0.002). Fig. 4 splits the graph of the explanatory variables into three groups (physicochemical measurements, LU change and its legacy) and that of the target variables (substrate activity, diversity indexes for the PA and the FL fraction). The positive (right) side of the correlation circle is associated with teak plantation (Teak 50 for LU change), LU change legacy of teak to crops (CropToTeak 50), from crops to teak (TeakToCrop 100), forest to crops (ForestToCrop 150), DOC, Sand Leptosol 150 and Sr (molecular weight). The negative (left) side of the circle is associated with banana plantation (Banana 50), fallow (Fallow 50 or CropOrFallow 50), low-drained luvisol (Luvisol 150) and swamp (Swamp 250 for LU and LU change) and SUVA254 (DOM aromaticity). Figure 4. View largeDownload slide Correlation circles of partial least squares (PLS) regression, dividing explanatory variables into three groups in different graphs: physicochemical measurements (DOC, SUVA254, Sr); soil type, LU change & its legacy; and the target variables including groups of carbon substrates; diversity indexes (Chao, Invsimpson, Shannon, Shannoneven and sobs: number of observed OTU) for PA (A) and FL (F) fraction. Only the first axis was interpretable (axis 1, Q2 = 0.19, axis 2, Q2 = -0.002). Sample sites from different seasons were grouped according to the first axis. Figure 4. View largeDownload slide Correlation circles of partial least squares (PLS) regression, dividing explanatory variables into three groups in different graphs: physicochemical measurements (DOC, SUVA254, Sr); soil type, LU change & its legacy; and the target variables including groups of carbon substrates; diversity indexes (Chao, Invsimpson, Shannon, Shannoneven and sobs: number of observed OTU) for PA (A) and FL (F) fraction. Only the first axis was interpretable (axis 1, Q2 = 0.19, axis 2, Q2 = -0.002). Sample sites from different seasons were grouped according to the first axis. The PLS analysis confirmed that metabolic capacities were high in the samples with high DOC inputs (RIB48, S9), and that PA bacterial diversity was strongly associated with the soil occupation (mostly teak plantation) (Teak 50) and recent LU change from crop to teak plantations (CropToTeak 50) and from teak to crop (TeakToCrop 100) during 4 years (2012–2015). Though being sampled at the upland bog, the community structure in RIB48 was much more affected by surrounding LU (teak plantation) rather than swamp factors. The swamp area (Swamp 250) had high value of SUVA254. On the other hand, the FL bacterial diversity was associated with the contemporary LU with banana plantation and fallow being important around the first gauging station (S1). Co-occuring OTUs and stream environmental analyses The OTU co-occurrence patterns were assessed in relation with DOC/CDOM characteristics using co-occurrence network analysis (Fig. 5A, B). C-scores and the properties of co-occurrence network are shown in Table 4. C-scores observed for both PA and FL fraction and both seasons, except for FL in the wet season, were higher than the C-score produced by the null model (Table 4). This indicated that pairs of OTUs tend to co-occur less often than expected by chance (i.e. there was segregation of OTUs) in the PA community in both seasons and in FL community in the dry season. Figure 5. View largeDownload slide Co-occurrence networking of (A) PA and (B) FL fraction in dry (right side) and wet season (left side). A connection means that the correlation is strong (Spearman's R> 0.8) and significant (P-value <0.05). Node color represents OTUs linked to a site where the relative abundance of OTUs was more than 0.1% than was found for the site alone. OTUs that appeared in at least two sites with relative abundance more than 0.1% were considered to build the network. The color of edges represents correlation between two nodes including negative correlation (red), positive correlation (grey). The positive correlation between OTUs and DOC/CDOM properties, LU change legacy was shown in black edges. Each node's size is proportional to the number of connections (degree). Figure 5. View largeDownload slide Co-occurrence networking of (A) PA and (B) FL fraction in dry (right side) and wet season (left side). A connection means that the correlation is strong (Spearman's R> 0.8) and significant (P-value <0.05). Node color represents OTUs linked to a site where the relative abundance of OTUs was more than 0.1% than was found for the site alone. OTUs that appeared in at least two sites with relative abundance more than 0.1% were considered to build the network. The color of edges represents correlation between two nodes including negative correlation (red), positive correlation (grey). The positive correlation between OTUs and DOC/CDOM properties, LU change legacy was shown in black edges. Each node's size is proportional to the number of connections (degree). Table 4. C-score and typology properties of co-occurrence network for PA and FL fraction in the wet and dry season. Dry—PA Wet—PA Dry—FL Wet—FL Observed C-score 1.29 0.62 1.35 1.5 Simulated C-score 1.28 0.59 1.31 1.49 P value 0.01 0.001 0.001 0.14 Number of nodes 170 281 182 114 Number of edges 1490 3853 1888 509 Connectivity 0.103 0.098 0.114 0.079 Dry—PA Wet—PA Dry—FL Wet—FL Observed C-score 1.29 0.62 1.35 1.5 Simulated C-score 1.28 0.59 1.31 1.49 P value 0.01 0.001 0.001 0.14 Number of nodes 170 281 182 114 Number of edges 1490 3853 1888 509 Connectivity 0.103 0.098 0.114 0.079 View Large Table 4. C-score and typology properties of co-occurrence network for PA and FL fraction in the wet and dry season. Dry—PA Wet—PA Dry—FL Wet—FL Observed C-score 1.29 0.62 1.35 1.5 Simulated C-score 1.28 0.59 1.31 1.49 P value 0.01 0.001 0.001 0.14 Number of nodes 170 281 182 114 Number of edges 1490 3853 1888 509 Connectivity 0.103 0.098 0.114 0.079 Dry—PA Wet—PA Dry—FL Wet—FL Observed C-score 1.29 0.62 1.35 1.5 Simulated C-score 1.28 0.59 1.31 1.49 P value 0.01 0.001 0.001 0.14 Number of nodes 170 281 182 114 Number of edges 1490 3853 1888 509 Connectivity 0.103 0.098 0.114 0.079 View Large The number of nodes and edges of the co-occurrence network of PA was higher in the community sampled in the wet season, with 281 nodes (i.e. OTUs) and 3853 edges (i.e. significant positive or negative correlations), than in the dry season (170 nodes and 1490 edges; Table 4). The co-occurrence network of the FL community showed an opposite trend, with the community sampled in the dry season having a higher number of nodes and edges (182 and 1888, respectively) than that sampled in the wet season (114 nodes and 509 edges; Table 4). The connectivity between OTUs of co-occurrence network was higher during the dry season than the wet season for both PA and FL community (Table 4). In general, the network of PA in the dry season was more separated and structured than the network of the wet season (Fig. 5). OTUs (i.e. >0.1% relative abundance) that were found in at least two sites accounted for a large proportion (37.01%) in the network of PA in the wet season and only 15.88% in the PA network in the dry season (Fig. 5). Moreover, there was more negative correlation between OTUs in the wet season PA network (37.3%) compared to the dry season (24.03%). OTUs that appeared in at least two sites were also more common in the network built from the FL community sampled during the wet season (34.21%) than in that the dry season (33.52%, Fig. 5). DOC/CDOM properties, LU occupation and the legacy of LU change were also included in the networks to decipher that abundant OTUs (i.e. >0.1% relative abundance) were related to these driving factors. We found that DOC concentration was highly correlated with 59 OTUs including 16 OTUs in the dry and 43 OTUs in the wet season for PA network. These OTUs were mainly classified as Alphaproteobacteria, Actinobacteria and Sphingobacteria, and could be considered as well adapted to the two sites harboring the highest DOC concentration (i.e. S9 and RIB48). The fulvic acid and humic-like fluorescence components (C1 and C3) showed positive correlations with 24 PA OTUs that appeared in the second gauging station S4 and at the outlet of the village S9 in the wet season. Meanwhile, these components were positively correlated with 22 OTUs from the PA network in the dry season that occurred preferentially at the outlet of the village (S9) and that were in most cases classified as Beta—Proteobacteria. Moreover, DOC concentration and the tryptophan-like fluorescence component (C2) were negatively correlated with 9 OTUs that occurred at the first gauging station (S1) in the dry season (Fig 5A). Compared to PA OTUs, fewer FL OTUs correlated with DOC: only 5 FL OTUs and 6 FL OTUs for the wet and dry season, respectively. These OTUs occurred preferentially in the upland bog (RIB48), the second gauging station (S4) and at the outlet of the village (S9) in wet season but only at S9 in the dry season. They were classified as Alpha or Gamma -Proteobacteria for wet season and Bacilli, Clostridia and Alpha and Gamma—Proteobacteria in the dry season. The fulvic acid and humic-like fluorescence components (C1 and C3) correlated with 16 OTUs that appeared mainly at the gauging station S4 and at the S9 outlet of the village in the FL community network for the wet season. These OTUs were classified as Alpha, Gamma—Proteobacteria and Actinobacteria. Similarly, in the dry season, fulvic acid and humic-like fluorescence components (C1 and C3) were associated with 10 OTUs at S9 that are classified as Bacilli, Alpha, Gamma—Proteobacteria (Fig. 5B). LU change legacy was correlated with various OTUs in the networks. Transition from teak plantation to crops in the past 4 years prior to sampling (TeakToCrop 100) appeared to be the most important as it was positively correlated with 17 and 31 PA OTUs for the dry and wet season at the outlet of the village (S9) and negatively correlated with 19 PA OTUs for both seasons. These OTUs were mainly classified as Alpha, Beta, Gamma—Proteobacteria, Actinobacteria, Sphingobacteria and Flavobacteria. The teak plantation (Teak 50) was positively correlated with 7 PA OTUs in the wet and 8 PA OTUs in the dry season (Gamma, Beta—Proteobacteria, Sphingobacteria and Bacilli) that were highly abundant in the upland bog RIB48. Banana plantation (Banana 50) was positively associated with 3 PA OTUs (2 Gamma and 2 Beta—Proteobacteria) occurring in the first gauging station (S1) at the wet season. Fallow (Fallow 50) showed also a positive relationship with abundant OTUs occurring in S1 at the dry season. They were mainly classified as Deinococcales, Epsiloproteobacteria, Betaproteobacteria, Bacilli, Actinobacteria and Sphingobacteria (Fig. 5A). DISCUSSION LU change as a driving factor of bacterial community structure To assess which factors preferentially drove bacterial community diversity and structure, the legacy of LU, DOC/CDOM characteristics and seasonal variation were considered. Indeed, although only five sampling locations were sampled along the stream, they were chosen because they represent the main LU in this catchment, and because they were close to instrumented outlets. Although not fully satisfying from a statistical point-of-view, this sampling strategy was sufficient to cover the main representative LU in the catchment while at the same time avoiding redundancy. The results obtained from a small number of sites clearly indicate an influence of LU and its changes on the diversity of bacterial community. Moreover, the sampling strategy also allowed the comparison of the characteristics of DOC and stream bacterial community between the wet and dry seasons during an inter-storm period. We hypothesized that the hydrological flow (and direction) more strongly influenced the structure of the bacterial community than vicinal LU change and its legacy. However, in contrast to our hypothesis, the results showed that although stream bacterial composition, structure and functioning varied with the seasonal variations of hydrological conditions, LU change and its legacy was the main determinant driving DOC/CDOM and the bacterial community along the stream. Many previous studies have indicated that seasonal hydrology is an important factor controlling freshwater bacterial communities (Crump et al.2007; Nino-Garcia, Ruiz-Gonzalez and Del Giorgio 2016). Most studies have tended to focus on assessing the microbial assemblage in interconnected habitats or along hydrological fragmented habitats, and have provided conclusions as to the effect of local environmental factors (Fierer et al.2007) and regional processes (e.g. geographic distance and microbial dispersion) (Lindstrom et al.2006; Fazi et al.2013). The processes that affect bacterial structure in streams include regional hydrological connectivity, local environmental conditions and non-directed local processes (Freimann et al.2015). In other words, the regional hydrological connectivity related to the dispersal of bacterial community (mass effect) or to changes of solute characteristics along the flow path shifts the bacterial community composition in streams. Local environmental conditions are important for forming the structure of the bacterial community when the mechanism of species sorting dominates over mass effect during the low flow. Therefore, we found here that the structure of the bacterial community was more related to local site specificities (local surrounding conditions) than to hydrological processes (flow path) because of low discharge, especially in dry season. However, few studies have considered these processes under effects of vicinal landscape factors on the local scale. The nutrient inputs from the surrounding agricultural watershed have been shown to increase DOC, influencing bacterial community structure (Song and Li 2016). Bacterial community composition in the water column of a temperate river was driven by regional factors related to watershed LU (Hosen et al.2017). Yet, the influence of LU is further complicated when cycles of change occur, such as when agricultural land converted to forest or when forest land is first converted to agriculture (Allan 2004). In the present study, the local morphological and physico-chemical characteristics were heterogeneous along the stream. For example, RIB48 was characterized by a gentle slope and muddy sediments in the upland bog whereas S4 was characterized by coarse sediments and a steep slope combined with a sinuous surface-subsurface water flow. The streambed sediments at a given spatial point along the stream-path principally originated from the surrounding hillslope soils that had their own vegetation cover, LU and LU legacies. Our results found that the upland bog station, RIB48, was associated with teak plots with high DOC and high richness and diversity of PA bacterial community. The environment of this station has undergone a crop / fallow transition to teak plantations during the 4 years prior to sampling (CropToTeak). Station S9 also had high DOC and was located in the village, but it was affected by a teak growing area subject to a transition towards crops during the 4 years prior to sampling (TeakToCrop). We also found high diversity of PA bacterial community at S4 that was surrounded with teak plantation although it was not significantly affected by teak plantation according to the PLS analysis. Interestingly, LU change from crop/fallow to teak and from teak to crop/fallow led to increases in DOC and in the richness and diversity of the PA bacterial community. The crop/fallow transition to teak intensified drastically in this catchment in recent time. During the period 2008 to 2014, teak plantations covered up to 36% of catchment area, most of which were more than 3 years old and were characterized by limited understory vegetation cover and increasingly degraded soils (Ribolzi et al.2017). According to this study, overland flow detached more soil particles in areas of teak plantation as compared to that generated under fallow. Moreover, the poor forest management may be the cause of the high rates of erosion rather than the nature of the teak itself (Fernández-Moya et al.2014). In the Houay Pano catchment, most farmers frequently burn the vegetation understory in the teak plantations. Such land management results in increasing surface crusting, lower rain infiltration and higher overland flow and higher rates of soil erosion (Lacombe et al.2017). During storm events, the overland flow from surrounding hillslope soils entered and accumulated in local streamed sediment in the HZ where exchange of surface and ground water stream occurs (Febria et al.2012). This accumulation represents a potential source of DOC and soil PA bacteria from surrounding LU, which probably explains the observed high richness and diversity of PA bacteria at sites associated with the transition of crops to teak plantations, or teak plantation to crops. In other words, the presence of teak plantations, either currently or in the past, results in high DOC concentrations and richness and diversity of PA fraction in the stream. In addition, the high DOC concentrations and richness and diversity found at RIB48 and S9 are probably linked with the longer water residence times and riparian zones in the upland bog (RIB48) and from effluent from the village (S9). In contrast, the station S1 was adjacent to a banana plantation on low draining luvisol soils, which may be related to the low diversity of PA bacteria and low metabolic capacity at this site, especially in the wet season. Interestingly, the diversity of FL bacteria was high in this sample, as well as at the second gauging station (S4) that is surrounded by teak plantations. These results suggested that PA bacteria community was more associated with LU change via local streamed sediments rather than FL bacterial community. Seasonal difference in community structure and functioning of PA and FL bacteria Seasonal changes inducing variability in water temperature, hydrological connectivity and organic matter quality and quantity are known to affect microbial diversity and composition in aquatic systems (Crump et al.2003; Febria et al.2012, 2015; Zeglin 2015). However, these effects become more complicated in stream ecosystems due to their variable hydrology and associated potential for cell dispersal (Crump, Amaral-Zettler and Kling 2012). For example, the seasonal shift in a temporary stream subject to hydrological fragmentation reduced microbial dispersion and affected the quality of DOM thereby shaping the bacterioplankton community composition (Fazi et al.2013). Moreover, as transient storage does not vary with discharge (Ward 2016), the proportion of bacteria that originate (or transit) the streambed will decrease at higher flow thus potentially leading to lower connectivity between stations along the stream. We also observed differences in community structure between the dry and wet seasons for both PA and FL. There was a larger degree of separation in PA community structure between upstream and downstream in the dry season. As suggested above this indicates that the community structure of PA bacteria was more affected than FL bacteria by local environmental conditions. This has also been observed in a river-floodplain system where the PA bacterial community was more heterogeneous and more dependent upon changes in environmental conditions and terrestrial organic matter flow than was the FL bacterial community (Besemer et al.2005). Our co-occurrence networks also showed that sampling sites along stream were more separated in the dry season than in the wet season. The OTUs of the PA network in the dry season showed site specificity, especially in RIB72 (the uppermost site) and S9 (outlet of the village). This can be explained by the longer residence time of local inputs of allochthonous carbon in the dry season leading to a greater spatial heterogeneity. A previous study has also suggested that freshwater inputs from the upper-river catchment and adjacent terrestrial runoffs decreased the influence of localized factors on biological composition, although a clear spatial heterogeneity could be found during low inflow periods (Carney et al.2015). In the wet season, more negative correlations appeared in the network (36.4%) indicating either competitive interactions or non-overlapping niches between microbes (Faust and Raes 2012). Relationship between DOC/CDOM and stream bacterial community under LU change Our study emphasized that DOC concentration was an important factor that directly impacted stream bacterial diversity, richness and co-occurrence networks. Recent studies have shown that DOC is more correlated with the bacterial community in streams in the wet as opposed to the dry season (Febria et al.2012) and that DOC plays a key role in the bacterial co-occurrence network of ponds and lakes (Comte et al.2016). Variations of DOM composition in soils and surface waters also influence bacterial community dynamics (Judd, Crump and Kling 2006). In our study, the role of DOC was stronger for the PA bacterial community than for the FL community. DOC was significantly correlated with diversity and richness of PA fraction (Correlation coefficient, R >0.8, P<0.01) while no correlation with the richness and diversity of FL fraction was found. Networking structure also suggested that DOC played an important role in connecting OTUs in the PA fraction network, especially in the wet season. In addition, DOC was correlated with fulvic acid and humic-like fluorescence components (C1 and C3). This suggests that DOC is of allochthonous origin, probably from upland soils. During precipitation events, overland flow exports the PA fraction and soil derived DOC from the soils surface into the stream thereby influencing stream DOC and bacterial community composition. A previous study also indicated that the soil runoff carried unique attached bacteria taxa associated with low molecular weight DOC into the HZ (Sabater, Meyer and Edwards 1993). The site at the outlet of the village (S9) that contained the most OTUs that correlated with high DOC and the components of fulvic acid and humic-like fluorescence components (C1 and C3) was impacted by the transition from teak to crops (TeakToCrop) in both the wet and dry networks. Potential bacterial metabolic capacity was also higher in RIB48 and S9. Although the humic-like component (C1) is considered to be recalcitrant DOC (Fellman, Hood and Spencer 2010; Khan et al.2010), it may be more preferred as a carbon source by some bacteria than other more labile compounds (Rosenstock, Zwisler and Simon 2005). Recalcitrant compounds that cannot be easily taken up by others are more reliable sources of carbon for these bacteria (Biasi et al.2005). Meanwhile, the tryptophan-like component, considered as labile DOC for bacteria, was positively correlated with DOC aromaticity (SUVA254) and high molecular weight (lower Sr). The high tryptophan-like fluorescence probably originates from dissolved amino acids in the non-living, high molecular mass, DOM pool released by the flushing of soil humic substances from the adjacent landscape in the stream (Fellman et al.2009; Yamashita and Tanoue 2003). Our results also showed that the tryptophan-like fluorescence component (C2) and SUVA254 were negatively correlated with some PA—OTUs and positively correlated with FL—OTUs. This suggests that specific FL bacteria may have preferentially utilized this carbon substrate. CONCLUSION AND PERSPECTIVES This is the first study that gives a holistic view on the impact of LU change and its legacy on seasonal changes of stream DOC and PA and FL bacterial diversity and functioning in a tropical catchment under base flow. Although seasonal variation of DOC, bacterial community structure for both PA and FL fractions as well as bacterial metabolic capacity was found, LU change and its legacy in Houay Pano catchment is considered as driving factor for stream aquatic bacterial community. The PA fraction was much more affected by LU change and its legacy than the FL fraction in our study. LU therefore can exert both a direct and an indirect control on bacterial communities via streambed sediments and overland flow as a mechanism of erosion as well as a vector of solid particles from the surrounding slopes. Moreover, it is interesting to note that changes in LU, particularly the transition between teak and crops that occurred between 2012 and 2015, and the presence of teak plantations led to an increase in the biodiversity of the PA fraction. This study emphasized the importance of LU change and its legacy for a stream bacterial community in a tropical catchment. Indeed, proximal factors need to be taken account if we are to have a complete understanding of the drivers influencing stream bacterial community assemblages. This work also underlines the important role of LU management in upland tropical catchments, especially in developing countries where LU has drastically changed in recent years. Future work addressing the specific soil bacterial groups that characterize different LU and the local stream sediment community is essential if we are to fully understand the interactions between stream and soil communities in the catchment via local streamed sediments in the hyporheic zone. It will also be interesting to conduct similar studies during stormflow periods as these hydrological events are known to have a high impact on sediment dynamics and on sediment bound bacterial transfer. SUPPLEMENTAL DATA Supplementary data are available at FEMSEC online. ACKNOWLEDGEMENTS This work was financed by the French National Research Agency (TecItEasy project; ANR-13-AGRO-0007). The authors would like to thank Mr Keooudone Latsachack and his team at Laksip and the MSEC (http://msec.obs-mip.fr/) project (Multi-Scale Environment Changes) for their support. The TerraSAR-X / Tandem-X data used to build the drainage area, were provided by DLR under the proposal DEM_HYDR1512. This work forms part of the Ph.D. thesis requirements of HL who is financed by a Ph.D. grant from the USTH, Vietnam and the IRD (ARTS). Conflicts of interest. None declared. REFERENCES Abdi H . Partial least squares regression and projection on latent structure regression (PLS Regression) . Focus Article . 2010 , DOI: https://doi.org/10.1002/wics.051 . Allan JD . Landscapes and riverscapes: the influence of land use on stream ecosystems . Annu Rev Ecol Evol Syst . 2004 ; 35 : 257 – 84 . Google Scholar Crossref Search ADS Allgaier M , Grossart HP . Seasonal dynamics and phylogenetic diversity of free-living and particle-associated bacterial communities in four lakes in northeastern Germany . Aquat Microb Ecol . 2006 ; 45 : 115 – 28 . Google Scholar Crossref Search ADS Allison SD , Martiny JB . Colloquium paper: resistance, resilience, and redundancy in microbial communities . Proc Natl Acad Sci USA . 2008 ; 105 : 11512 – 9 . Google Scholar Crossref Search ADS PubMed Axel D , Jens S , Annette F . Impact of tropical land-use change on soil organic carbon stocks – a meta-analysis . Global Change Biol . 2011 ; 17 : 1658 – 70 . Google Scholar Crossref Search ADS Bahram S , Hassan F , Babak B et al. Flood intensification due to changes in land use . Water Resour Manag . 2007 ; 22 : 1051 – 67 . Bastian M , Heymann S , Jacomy M . Gephi: an open source software for exploring and manipulating networks . ICWSM . 2009 ; 8 : 361 – 2 . Bencala KE , Walters RA . Simulation of solute transport in a mountain pool-and-riffle stream – a transient storage model . Water Resour Res . 1983 ; 19 : 718 – 24 . Google Scholar Crossref Search ADS Berggren M , del Giorgio PA . Distinct patterns of microbial metabolism associated to riverine dissolved organic carbon of different source and quality . J Geophys Res Biogeosci . 2015 ; 120 : 989 – 99 . Google Scholar Crossref Search ADS Besemer K , Moeseneder MM , Arrieta JM et al. Complexity of bacterial communities in a river-floodplain system (Danube, Austria) . Appl Environ Microbiol . 2005 ; 71 : 609 – 20 . Google Scholar Crossref Search ADS PubMed Biasi C , Rusalimova O , Meyer H et al. Temperature-dependent shift from labile to recalcitrant carbon sources of arctic heterotrophs . Rapid Commun Mass Spectrom . 2005 ; 19 : 1401 – 8 . Google Scholar Crossref Search ADS PubMed Capone KA , Dowd SE , Stamatas GN et al. Diversity of the human skin microbiome early in life . J Invest Dermatol . 2011 ; 131 : 2026 – 32 . Google Scholar Crossref Search ADS PubMed Carney RL , Mitrovic SM , Jeffries T et al. River bacterioplankton community responses to a high inflow event . Aquat Microb Ecol . 2015 ; 75 : 187 – 205 . Google Scholar Crossref Search ADS Claire H-DM , Silver JM , Leibold MA et al. A comparison of taxon co-occurrence patterns for macro- and microorganisms . Ecology . 2007 ; 88 : 1345 – 53 . Google Scholar Crossref Search ADS PubMed Coble PG . Characterization of marine and terrestrial DOM in seawater using excitation emission matrix spectroscopy . Mar Chem . 1996 ; 51 : 325 – 46 . Google Scholar Crossref Search ADS Coble PG . Marine optical biogeochemistry: the chemistry of ocean color . Chem Rev . 2007 ; 107 : 402 – 18 . Google Scholar Crossref Search ADS PubMed Cole JR , Chai B , Farris RJ et al. The ribosomal database project (RDP-II): sequences and tools for high-throughput rRNA analysis . Nucleic Acids Res . 2005 ; 33 : D294 – 6 . Google Scholar Crossref Search ADS PubMed Comte J , Lovejoy C , Crevecoeur S et al. Co-occurrence patterns in aquatic bacterial communities across changing permafrost landscapes . Biogeosciences . 2016 ; 13 : 175 – 90 . Google Scholar Crossref Search ADS Crump BC , Armbrust EV , Baross JA . Phylogenetic analysis of particle-attached and free-living bacterial communities in the Columbia River, its estuary, and the adjacent coastal ocean . Appl Environ Microbiol . 1999 ; 65 : 3192 – 204 . Google Scholar PubMed Crump BC , Amaral-Zettler LA , Kling GW . Microbial diversity in arctic freshwaters is structured by inoculation of microbes from soils . Isme J . 2012 ; 6 : 1629 – 39 . Google Scholar Crossref Search ADS PubMed Crump BC , Kling GW , Bahr M et al. Bacterioplankton community shifts in an arctic lake correlate with seasonal changes in organic matter source . Appl Environ Microbiol . 2003 ; 69 : 2253 – 68 . Google Scholar Crossref Search ADS PubMed Crump BC , Adams HE , Hobbie JE et al. Biogeography of bacterioplankton in lakes and streams of an arctic tundra catchment . Ecology . 2007 ; 88 : 1365 – 78 . Google Scholar Crossref Search ADS PubMed Edgar RC , Haas BJ , Clemente JC et al. UCHIME improves sensitivity and speed of chimera detection . Bioinformatics . 2011 ; 27 : 2194 – 200 . Google Scholar Crossref Search ADS PubMed Evrard O , Laceby JP , Huon S et al. Combining multiple fallout radionuclides (137Cs, 7Be, 210Pbxs) to investigate temporal sediment source dynamics in tropical, ephemeral riverine systems . J Soils Sediments . 2016 ; 16 : 1130 – 44 . Google Scholar Crossref Search ADS Fasching C , Ulseth AJ , Schelker J et al. Hydrology controls dissolved organic matter export and composition in an Alpine stream and its hyporheic zone . Limnol Oceanogr . 2016 ; 61 : 558 – 71 . Google Scholar Crossref Search ADS PubMed Faust K , Raes J . Microbial interactions: from networks to models . Nat Rev Microbiol . 2012 ; 10 : 538 – 50 . Google Scholar Crossref Search ADS PubMed Fazi S , Vazquez E , Casamayor EO et al. Stream hydrological fragmentation drives bacterioplankton community composition . PLoS One . 2013 ; 8 : e64109 . Google Scholar Crossref Search ADS PubMed Febria CM , Beddoes P , Fulthorpe RR et al. Bacterial community dynamics in the hyporheic zone of an intermittent stream . Isme J . 2012 ; 6 : 1078 – 88 . Google Scholar Crossref Search ADS PubMed Febria CM , Hosen JD , Crump BC et al. Microbial responses to changes in flow status in temporary headwater streams: a cross-system comparison . Front Microbiol . 2015 ; 6 : 522 . Google Scholar Crossref Search ADS PubMed Fellman JB , Hood E , Spencer RGM . Fluorescence spectroscopy opens new windows into dissolved organic matter dynamics in freshwater ecosystems: a review . Limnol Oceanogr . 2010 ; 55 : 2452 – 62 . Google Scholar Crossref Search ADS Fellman JB , D'Amore DV , Hood E et al. Fluorescence characteristics and biodegradability of dissolved organic matter in forest and wetland soils from coastal temperate watersheds in southeast Alaska . Biogeochemistry . 2008 ; 88 : 169 – 84 . Google Scholar Crossref Search ADS Fellman JB , Hood E , Edwards RT et al. Changes in the concentration, biodegradability, and fluorescent properties of dissolved organic matter during stormflows in coastal temperate watersheds . J Geophys Res . 2009 ; 114 : G01021:1 – 14 . Google Scholar Crossref Search ADS Fernández-Moya J , Alvarado A , Forsythe W et al. Soil erosion under teak (Tectona grandis L.f.) plantations: general patterns, assumptions and controversies . Catena . 2014 ; 123 : 236 – 42 . Google Scholar Crossref Search ADS Fichot CG , Benner R . The spectral slope coefficient of chromophoric dissolved organic matter (S275-295) as a tracer of terrigenous dissolved organic carbon in river-influenced ocean margins . Limnol Oceanogr . 2012 ; 57 : 1453 – 66 . Google Scholar Crossref Search ADS Fierer N , Morse J , Berthrong S et al. Environmental controls on the landscape-scale biogeography of stream bacterial communities . Ecology . 2007 ; 88 : 2162 – 73 . Google Scholar Crossref Search ADS PubMed Freimann R , Burgmann H , Findlay SE et al. Hydrologic linkages drive spatial structuring of bacterial assemblages and functioning in alpine floodplains . Front Microbiol . 2015 ; 6 : 1221 . Google Scholar Crossref Search ADS PubMed Fuhrman JA , Comeau DE , Hagstrom A et al. Extraction from natural planktonic microorganisms of DNA suitable for molecular biological studies . Appl Environ Microbiol . 1988 ; 54 : 1426 – 9 . Google Scholar PubMed Harrell FEJ . Hmisc: Harrel miscelaneous , R package version . 2018 ; 4 : 1 . Helms JR , Stubbins A , Ritchie RD et al. Absorption spectral slopes and slope ratios as indicators of molecular weight, source, and photobleaching of chromophoric dissolved organic matter . Limnol Oceanogr . 2008 ; 53 : 955 – 69 . Google Scholar Crossref Search ADS Hosen DJ , Febria C , Crump BC et al. Watershed urbanization linked to differences in stream bacterial community composition . Front Microbiol . 2017 ; 8 : 1452 . Google Scholar Crossref Search ADS PubMed Huon S , De Rouw A , Bonte P et al. Long-term soil carbon loss and accumulation in a catchment following the conversion of forest to arable land in northern Laos . Agric Ecosyst Environ . 2013 ; 169 : 43 – 57 . Google Scholar Crossref Search ADS Ishii SKL , Boyer TH . Behavior of reoccurring PARAFAC components in fluorescent dissolved organic matter in natural and engineered systems: a critical review . Environ Sci Technol . 2012 ; 46 : 2006 – 17 . Google Scholar Crossref Search ADS PubMed Judd KE , Crump BC , Kling GW . Variation in dissolved organic matter controls bacterial production and community composition . Ecology . 2006 ; 87 : 2068 – 79 . Google Scholar Crossref Search ADS PubMed Kamjunke N , Herzsprung P , Neu TR . Quality of dissolved organic matter affects planktonic but not biofilm bacterial production in streams . Sci Total Environ . 2015 ; 506–507 : 353 – 60 . Google Scholar Crossref Search ADS PubMed Khan MGM , Wu F , Liu C-Q et al. Characterization of Nanming River (southwestern China) sewerage-impacted pollution using an excitation-emission matrix and PARAFAC . Limnology . 2010 ; 11 : 217 – 31 . Google Scholar Crossref Search ADS Kim H , Dietrich WE , Thurnhoffer BM et al. Controls on solute concentration-discharge relationships revealed by simultaneous hydrochemistry observations of hillslope runoff and stream flow: the importance of critical zone structure . Water Resour Res . 2017 ; 53 : 1424 – 43 . Google Scholar Crossref Search ADS Lacombe G , Valentin C , Sounyafong P et al. Linking crop structure, throughfall, soil surface conditions, runoff and soil detachment: 10 land uses analyzed in Northern Laos . Sci Total Environ . 2017 ; 616–617 : 1330 – 8 . Google Scholar PubMed Lindstrom E , Forslund M , Algesten G et al. External control of bacterial community structure in lakes . Limnol Oceanogr . 2006 ; 51 : 399 – 42 . Google Scholar Crossref Search ADS Luef B , Aspetsberger F , Hein T et al. Impact of hydrology on free-living and particle-associated microorganisms in a river floodplain system (Danube, Austria) . Freshwater Biol . 2007 ; 52 : 1043 – 57 . Google Scholar Crossref Search ADS Mazzetto AM , Feigl BJ , Cerri CE et al. Comparing how land use change impacts soil microbial catabolic respiration in Southwestern Amazon . Braz J Microbiol . 2016 ; 47 : 63 – 72 . Google Scholar Crossref Search ADS PubMed Miklós I , Podani J . Randomization of presence-absence matrices: comments and new algorithms . Ecology . 2004 ; 85 : 86 – 92 . Google Scholar Crossref Search ADS Murphy KR , Stedmon CA , Graeber D et al. Fluorescence spectroscopy and multi-way techniques. PARAFAC . Anal Methods . 2013 ; 5 : 6557 . Google Scholar Crossref Search ADS Nino-Garcia JP , Ruiz-Gonzalez C , Del Giorgio PA . Interactions between hydrology and water chemistry shape bacterioplankton biogeography across boreal freshwater networks . Isme J . 2016 ; 10 : 1755 – 66 . Google Scholar Crossref Search ADS PubMed Oksanen J , Blanchet G , Friendly M et al. Vegan: community ecology package . R package version 1 . 2017 ; 2 : 4 – 5 . Patin J , Mouche E , Ribolzi O et al. Analysis of runoff production at the plot scale during a long-term survey of a small agricultural catchment in Lao PDR . J Hydrol . 2012 ; 426 : 79 – 92 . Google Scholar Crossref Search ADS Pommier T , Merroune A , Bettarel Y et al. Off-site impacts of agricultural composting: role of terrestrially derived organic matter in structuring aquatic microbial communities and their metabolic potential . FEMS Microbiol Ecol . 2014 ; 90 : 622 – 32 . Google Scholar Crossref Search ADS PubMed Poretsky R , Rodriguez RL , Luo C et al. Strengths and limitations of 16S rRNA gene amplicon sequencing in revealing temporal microbial community dynamics . PLoS One . 2014 ; 9 : e93827 . Google Scholar Crossref Search ADS PubMed Pruesse E , Quast C , Knittel F et al. SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB . Nucleic Acids Res . 2007 ; 35 : 7188 – 96 . Google Scholar Crossref Search ADS PubMed Ribolzi O , Evrard O , Huon S et al. Use of fallout radionuclides ((7)Be, (210)Pb) to estimate resuspension of Escherichia coli from streambed sediments during floods in a tropical montane catchment . Environ Sci Pollut Res Int . 2015 ; 23 : 3427 – 35 . Google Scholar Crossref Search ADS PubMed Ribolzi O , Cuny J , Sengsoulichanh P et al. Assessment of water quality along a tributary of the Mekong River in a mountainous, mixed land-use environment of the Lao P.D.R . Lao J Agricult Forest . 2008 ; 17 : 91 – 111 . Ribolzi O , Evrard O , Huon S et al. From shifting cultivation to teak plantation: effect on overland flow and sediment yield in a montane tropical catchment . Sci Rep . 2017 ; 7 : 3987 . Google Scholar Crossref Search ADS PubMed Rochelle-Newall EJ , Nguyen TMH , Le TPQ et al. A short review of faecal indicator bacteria in tropical aquatic ecosystems: knowledge gaps and future directions . Front Microbiol . 2015 ; 6 : 308 . Google Scholar Crossref Search ADS PubMed Rochelle-Newall EJ , Chu VT , Pringault O et al. Phytoplankton diversity and productivity in a highly turbid, tropical coastal system (Bach Dang Estuary, Vietnam) . Mar Poll Bull . 2011 ; 62 : 2317 – 29 . Google Scholar Crossref Search ADS Rosenstock B , Zwisler W , Simon M . Bacterial consumption of humic and non-humic low and high molecular weight DOM and the effect of solar irradiation on the turnover of labile DOM in the Southern Ocean . Microb Ecol . 2005 ; 50 : 90 – 101 . Google Scholar Crossref Search ADS PubMed Sabater F , Meyer J , Edwards R . Longitudinal patterns of dissolved organic-carbon concentration and suspended bacterial density along a blackwater river . Biogeochemistry . 1993 ; 21 : 73 – 93 . Google Scholar Crossref Search ADS Sanchez G . Plsdepot: Partial Least Squares (PLS) Data Analysis Methods , R package version , version 0.1.17 , 2012 . Schloss PD , Westcott SL , Ryabin T et al. Introducing mothur: open source, platform-independent, community-supported software for describing and comparing microbial communities . Appl Environ Microbiol . 2009 ; 75 : 7537 – 41 . Google Scholar Crossref Search ADS PubMed Song L , Li L . Variations in bacterial community in a temperate lake associated with an agricultural watershed . Microb Ecol . 2016 ; 72 : 277 – 86 . Google Scholar Crossref Search ADS PubMed Spencer RGM , Butler KD , Aiken GR . Dissolved organic carbon and chromophoric dissolved organic matter properties of rivers in the USA . J Geophys Res-Biogeo . 2012 ; 117 : G03001:1 – 14 . Google Scholar Crossref Search ADS Stedmon CA , Markager S . Resolving the variability in dissolved organic matter fluorescence in a temperate estuary and its catchment using PARAFAC analysis . Limnol Oceanogr . 2005 ; 50 : 686 – 97 . Google Scholar Crossref Search ADS Stefanowicz A . Review: the biolog plates technique as a tool in ecological studies of microbial communities . Polish J Environ Stud . 2006 ; 15 : 669 – 76 . Stone L , Roberts A . The checkerboard score and species distributions . Oecologia . 1990 ; 85 : 74 – 9 . Google Scholar Crossref Search ADS PubMed Thioulouse J , Chessel D , Dol ́edec S et al. ADE-4: a multivariate analysis and graphical display software . Stat Comput . 1997 ; 7 : 75 – 83 . Google Scholar Crossref Search ADS Tian Q , Taniguchi T , Shi WY et al. Land-use types and soil chemical properties influence soil microbial communities in the semiarid Loess Plateau region in China . SciRep . 2017 ; 7 : 45289 . Valentin C , Agus F , Alamban R et al. Runoff and sediment losses from 27 upland catchments in Southeast Asia: impact of rapid land use changes and conservation practices . Agric Ecosyst Environ . 2008 ; 128 : 225 – 38 . Google Scholar Crossref Search ADS Wang Z , Liu S , Huang C et al. Impact of land use change on profile distributions of organic carbon fractions in peat and mineral soils in Northeast China . Catena . 2017 ; 152 : 1 – 8 . Google Scholar Crossref Search ADS Ward AS . The evolution and state of interdisciplinary hyporheic research . Wires Water . 2016 ; 3 : 83 – 103 . Google Scholar Crossref Search ADS Weishaar JL , Aiken GR , Bergamaschi BA et al. Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition and reactivity of dissolved organic carbon . Environ Sci Technol . 2003 ; 37 : 4702 – 8 . Google Scholar Crossref Search ADS PubMed Wickland KP , Neff JC , Aiken GR . Dissolved organic carbon in Alaskan boreal forest: sources, chemical characteristics, and biodegradability . Ecosystems . 2007 ; 10 : 1323 – 40 . Google Scholar Crossref Search ADS Williams CJ , Yamashita Y , Wilson HF et al. Unraveling the role of land use and microbial activity in shaping dissolved organic matter characteristics in stream ecosystems . Limnol Oceanogr . 2010 ; 55 : 1159 – 71 . Google Scholar Crossref Search ADS Wilson HF , Xenopoulos MA . Landscape influences on stream fish assemblages across spatial scales in a northern Great Plains ecoregion . Can J Fish Aquat Sci . 2008 ; 65 : 245 – 57 . Google Scholar Crossref Search ADS Wilson HF , Xenopoulos MA . Effects of agricultural land use on the composition of fluvial dissolved organic matter . Nat Geosci . 2009 ; 2 : 37 – 41 . Google Scholar Crossref Search ADS Yamashita Y , Tanoue E . Chemical characterization of protein-like fluorophores in DOM in relation to aromatic amino acids . Mar Chem . 2003 ; 82 : 255 – 71 . Google Scholar Crossref Search ADS Yamashita Y , Kloeppel BD , Knoepp J et al. Effects of watershed history on dissolved organic matter characteristics in headwater streams . Ecosystems . 2011 ; 14 : 1110 – 22 . Google Scholar Crossref Search ADS Zare M , Samani AAN , Mohammady M . The impact of land use change on runoff generation in an urbanizing watershed in the north of Iran . Environ Earth Sci . 2016 ; 75 : 1279 . Google Scholar Crossref Search ADS Zeglin LH . Stream microbial diversity in response to environmental changes: review and synthesis of existing research . Front Microbiol . 2015 ; 6 : 454 . Google Scholar Crossref Search ADS PubMed Zhang C , Takase K , Oue H et al. Effects of land use change on hydrological cycle from forest to upland field in a catchment, Japan . Front Struct Civ Eng . 2013 ; 7 : 456 – 65 . Google Scholar Crossref Search ADS Zhang G-L , Zhang J , Liu S-M et al. Nitrous oxide in the Changjiang (Yangtze River) Estuary and its adjacent marine area: rerine input, sediment release and atmospheric fluxes . Biogeosciences . 2010 ; 7 : 3505 – 16 . Google Scholar Crossref Search ADS © FEMS 2018. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Vicinal land use change strongly drives stream bacterial community in a tropical montane catchment
JF - FEMS Microbiology Ecology
DO - 10.1093/femsec/fiy155
DA - 2018-11-01
UR - https://www.deepdyve.com/lp/oxford-university-press/vicinal-land-use-change-strongly-drives-stream-bacterial-community-in-t6v8klnrvV
VL - 94
IS - 11
DP - DeepDyve
ER -