Microb Ecol (2014) 68:47–59 DOI 10.1007/s00248-014-0401-x
MICROBIOLOGY OF AQUATIC SYSTEMS
Diversity of Benthic Biofilms Along a Land Use Gradient in Tropical Headwater Streams, Puerto Rico Sofía Burgos-Caraballo & Sharon A. Cantrell & Alonso Ramírez
Received: 6 May 2013 / Accepted: 21 February 2014 / Published online: 20 March 2014 # Springer Science+Business Media New York 2014
Abstract The properties of freshwater ecosystems can be altered, directly or indirectly, by different land uses (e.g., urbanization and agriculture). Streams heavily influenced by high nutrient concentrations associated with agriculture or urbanization may present conditions that can be intolerable for many aquatic species such as macroinvertebrates and fishes. However, information with respect to how benthic microbial communities may respond to changes in stream ecosystem properties in relation to agricultural or urban land uses is limited, in particular for tropical ecosystems. In this study, diversity of benthic biofilms was evaluated in 16 streams along a gradient of land use at the Turabo watershed in Puerto Rico using terminal restriction fragment length polymorphism. Diversity indices and community structure descriptors (species richness, Shannon diversity, dominance and evenness) were calculated for both bacteria and eukaryotes for each stream. Diversity of both groups, bacteria and eukaryotes, did not show a consistent pattern with land use, since it could be high or low at streams dominated by different land uses. This suggests that diversity of biofilms may be more related to
S. Burgos-Caraballo (*) Department of Biology, University of Puerto Rico, P.O. Box 70377, San Juan, PR 00936-8377, USA e-mail: [email protected] S. A. Cantrell Department of Biology, Universidad del Turabo, P.O. Box 3030, Gurabo, PR 00778, USA A. Ramírez Department of Environmental Sciences, University of Puerto Rico, P.O. Box 190341, San Juan, PR 00919, USA
site-specific conditions rather than watershed scale factors. To assess this contention, the relationship between biofilm diversity and reach-scale parameters (i.e., nutrient concentrations, canopy cover, conductivity, and dissolved oxygen) was determined using the Akaike Information Criterion (AICc) for small sample size. Results indicated that nitrate was the variable that best explained variations in biofilm diversity. Since nitrate concentrations tend to increase with urban land use, our results suggest that urbanization may indeed increase microbial diversity indirectly by increasing nutrients in stream water.
Introduction Natural landscapes are transformed by land use activities that are necessarily to satisfy human needs (e.g., agriculture and urbanization) [15]. These landscape modifications may affect stream ecosystems due to the close relationship that streams have with their surrounding landscapes [1]. In streams heavily influenced by agricultural or urban land uses, elevated nutrient concentrations and conductivity could be expected, while oxygen concentrations decrease [1]. Light availability may increase as canopy cover of riparian vegetation is eliminated, but can decrease with depth due to increased turbidity in altered streams. Other factors associated to land use, such as leaky pipes and combined sewer overflow entering streams, can have similar results [1, 37, 41, 48]. These degraded conditions may be intolerable for many aquatic organisms, such as macroinvertebrates and fishes, as well as other organisms with life cycles linked to freshwaters ecosystems (e.g., amphibians and reptiles) [7]. Therefore, sensitive species could be
48
eliminated, while more tolerant species may persist and become more abundant causing a shift in community composition [26]. Likewise, previous studies from temperate regions demonstrated that microbial communities can change in composition as well as in diversity in streams influenced by urban land use [8, 39, 49]. For example, Dopheide et al. found that streams more impacted by urbanization showed higher diversity of ciliates than undisturbed forested areas [11]. A similar pattern was reported by Carrino-Kyker et al. for different functional groups of eukaryotes (algae, fungi, and protist) in urban vernal pools [8]. However, it is not clear if the same patterns observed in microbial communities from temperate regions will be similar in benthic biofilms from tropical streams. This is of major concern, as changes in species composition can result in alterations in ecosystem function [19]. The stream benthic environment is inhabited by complex assemblages of microbial communities, including bacteria and eukaryotes such as algae, fungi, and other protists. This conglomerate of microbial organisms, known as biofilms, is responsible for many essential ecosystem processes such as organic matter decomposition and nutrient cycling [3, 29, 37]. In streams on Caribbean islands, food webs are primarily based on biofilms and are the main food source for different species of fish and shrimp, which in turn represent an important protein source for humans [24]. However, the quality of these food sources can be affected by land uses [5]. Therefore, changes in the community composition of benthic biofilm, as a result of human impact to watershed and reach conditions, may result in the alteration in streams ecosystem function [2]. The purpose of this study was to analyze benthic biofilm community composition in tropical streams along a land use gradient. At the same time, we explored the relationship between reach and watershed factors in order to determine how these variables influenced community composition and diversity of benthic biofilms. We expected to find differences in community composition by considering forest and urbanization as extremes of the land use gradient. However, we did not predict a direction of change in diversity since previous studies indicate that microbial community richness can be positively or negatively influenced by urban land use [8, 11, 50].
Materials and Methods Watershed and Reach-Scale Parameters The Turabo watershed is located in the municipality of Caguas, in the central northeastern region of Puerto Rico (Fig. 1). The geology of the zone is dominated by volcanic and volcano clastic rocks from the late Cretaceous together with some
S. Burgos-Caraballo et al.
fragments of plutonic and alluvium deposits from the Quaternary [47]. In general, the watershed is dominated by forest land use, with agriculture and urbanization as secondary land uses. Forest cover consists of montane, submontane, and lower montane evergreen forest combined with abandoned coffee plantations [18]. Agricultural activities consist of pastures with small fragments used for small-scale farming. The lower portions of the watershed are close to downtown Caguas where urbanization is concentrated. The areas of high elevation in the watershed consist mainly of suburban development. A closer examination of data revealed that no relationship exists between elevation and urban land use. Land Use Analyses For our analyses, the term “land use” refers to land use/land cover data. Land use classifications were made for each study stream using geographic information systems (GIS, ESRI ver. 9.1). The subcatchment of each study stream was delimited by identifying the topographic divide following contour lines on topographic maps (scale 1:10,000) to estimate the area. GIS tools were used to classify, delineate, and digitize land uses. Photo interpretation mapping scale was 1:10,000 and was performed by creating polygons of areas no smaller than 4,000 m2. Dominant land uses were identified by using 2007 aerial images provided by the Department of Environmental and Natural Resources of the Puerto Rico Commonwealth. Each polygon was categorized into one of four general land use types: urban, forest, agriculture, and other (e.g., bare soil; Table 1). Physicochemical Parameters At each stream, we measured water temperature, dissolved oxygen (O2) (using a luminescent dissolved oxygen probe, HQ10–HQ20 meters, Hach Company, Colorado), pH (using a Hanna Instrument, HI 98150 model), conductivity, total dissolved solids, and sodium chloride (all using a Hanna Instrument, HI 9835). Incident light was measured with a LI-Cor quantum sensor (LI-193 Spherical Quantum Sensor), and canopy cover was estimated taking four measurements with a concave spherical densiometer. Nutrients and major ions were measured by collecting a filtered water sample (Whatman® glass microfiber filters, GF/ F 47 mm) at each stream. Samples were frozen until analysis at the University of New Hampshire Water Quality Analysis Laboratory. Samples were analyzed for chloride (Cl−), nitrate (NO3−), sulfate (SO4−3), sodium (Na+), potassium (K+), magnesium (Mg+2), and calcium (Ca+2) using a Dionex ICS1000 ion chromatograph. Nutrients analysis included ammonium (NH4+) and phosphate (PO4−3) (Westco Scientific SmartChem 200 discrete automated analyzer), total dissolved nitrogen (TDN) and dissolved organic carbon (DOC) (Shimadzu
Benthic Biofilm Diversity
49
TOCV with total nitrogen module), and dissolved organic nitrogen (DON) (calculated as the differences between TDN and NO3−-N + NH4+-N). Benthic Biofilm Sampling Samples were collected from September to November in 2008 during the rainy season. The northeastern region of Puerto Rico, where these studies were conducted, lacks strong climatic seasonality; therefore, it is probable that the time of year played a minor role. In each stream, biofilms were collected from rocks at four different riffles separated by at least one sequence of pools and riffle to reduce influence between them. Collected rocks were placed in individual sterile Whirl Packs bags and refrigerated during transport. Once in the laboratory, all benthic material was removed from the rocks by scraping the surface using a sterile metal brush and rinsed with autoclaved distillated water. From all removed material, approximately 1.5 ml was collected and centrifuged at 1,000 rpm velocity to create pellets which were frozen at −20° until DNA extraction. DNA Extraction, PCR, and TRFLP’s Analysis
Fig. 1 Study site location. The shaded regions represent the subwatershed associated to each stream. Each number corresponds to the each study site as is indicated in the text and figures
In order to determine benthic community structure, terminal restriction fragment length polymorphism (TRFLP) was analyzed for bacteria 16S recombinant DNA (rDNA) genes and Eukarya 18S rDNA; a fingerprint approach that is useful to differentiate communities, compare phylotype richness, and community structure among sites [14, 28, 36].
Table 1 Land use percent at each study site Study stream
Forest
Urban
Agriculture
Other
Catchment area (km2)
Elevation (m)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
100 58 59 57 74 76 86 64 55 66 45 35 23 23 24 14
0 0 0 0 2 2 7 13 25 27 29 34 49 67 70 86
0 42 40 43 24 22 7 18 19 7 22 27 7 3 0 0
0 0 1 0 0 1 0 5 0 0 4 4 0 7 6 0
0.23 1.01 0.37 1.92 1.97 2.13 0.20 1.02 1.36 2.51 0.69 1.30 0.17 0.30 0.28 0.21
100 210 290 330 140 90 120 120 110 90 110 330 330 410 90 420
Each number corresponds to the each study site as is indicated in the text and figures
50
Biofilm DNA was extracted using MoBio UltraClean soil kit (MoBio Laboratories, Solana Beach, CA) following manufacturer protocols. DNA concentrations were quantified by fluorometry. Amplifications were conducted using 12.5 μg REDTaq Polymerase (Sigma-Aldrich, St. Louis, MO), 2–4 μg of DNA template, 0.625 μg of primers, and 0.5 μg of bovine serum albumin (BSA; Fermentas, Vilnius, Lithuania) to reduce interference with contaminants. Bacteria 16S rDNA genes were amplified using the following set of primers 16S-27F (5′AGAGTTTGATCMTGGCTCAG) and 16S1525R (5′AAGGAGGTGWTCCARCC). Eukarya 18S rDNA was amplified using universal set of primers 18S-82F (5′GAAACTGCGAATGGCTC) and 18S-1520R (5′CYGC AGGTTCACCTAC). For each set of primers, only the forward primer was fluorescently labeled at the 5′ end with FAM dye (Sigma-Aldrich, Atlanta, GA). Polymerase chain reactions (PCR) were performed with a model 2720 thermal cycler (ABI, USA) using the following program: 1-min hot start at 80 °C, 94 °C for 5 min followed by 30 cycles of denaturation at 94 °C for 30 s, followed by annealing at 52 °C for 30 s, at 72 °C for 1 min 30 s, with a final extension step at 72 °C for 10 min. Amplified DNA was verified by electrophoresis of PCR mixtures in 1.0 % agarose in 1X TAE buffer. Fluorescently labeled amplicons were enzymatically digested with BsuRI (HaeIII), Rsa, and MnlI (Fermentas Life Science, Maryland, USA) to find the enzymes with the higher resolution for both, bacteria and eukarya. MnlI was chosen as it provided the greatest numbers of terminal restriction fragments (T-RFs) for bacteria, as well as for eukarya. Digestions were performed at 37 °C for 120 min with 2.0 μl of buffer, 0.2 μl of enzyme, and 5.0 μl of amplified product and water to a final volume of 20 μl. Samples were precipitated in ethanol to eliminate impurities, dried, and resuspended in Hi-Di formamide with GeneScan Liz 500 (ABI, USA) using a size standard that range between 50 and 500 bp. Those fragments that differed by 1 bp were considered different. Samples were run on ABI 3130 genetic analyzer (ABI, USA). Data were normalized by using fragments with a fluorescence >1 % of the total fluorescence. The area of each peak was standardized with respect to the sum of all peak areas of the sample. We considered the fragments that were present at three of the four replicate analyzed at each study site. Restriction fragments were analyzed using the web page MICA (Microbial community analysis [45]). The program APLAUS was selected to infer the community structure based on the TRFLP data of communities present in the biofilms [45], but since one TRFLP could represent different organisms, bacteria were identified at the Phyla level, unless they could be identified to a lower level. Databases of Eukarya were more limited and therefore no eukaryote could be identified.
S. Burgos-Caraballo et al.
Statistical Analysis To reduce the influence of most common Operational Taxonomic Units (OTUs), a square root transformation was used for data obtained at each study site [17]. All OTUs present in a sample were sorted by abundance to identified shared OTUs (by considering those that were present in 10 of the 16 study sites) unique OTUs (by considering the phylotypes that were present in 1 of the 16 study sites). OTUs richness (S) was determined by counting the total number of OTUs present at each study site. Shannon diversity index (H′), Simpson’s diversity index (D), and Pielou’s evenness index (E) were determined using the average abundance of OTUs per study site in PC-Ord version 4.25 [4]. In addition, Smith and Wilson evenness (Evar), previously recommended as a measure of diversity for TRFLP’s data, was calculated [4]. A nonmetric multidimensional scaling (NMS) ordination was used to analyze the profiles of OTUs among study streams, using a square root transformation of the relative height for each OTUs. We choose this ordination for various reasons. It is an efficient ordination method where data does not have to be linearly related [32]. NMS also alleviates problems associated to data sets containing zeros [32]. In addition, the user can choose the distance measure appropriate for the data [32]. We run the NMS considering the following parameters: 6 axes as maximum, 500 iterations, 0.20 steps in length, and 50 runs with real data and randomized data. The dissimilarity distance measure selected for this ordination was Bray-Curtis because it considers the presence of a species as more informative than its absence, as the absence of species from the data is not necessarily associated with environmental factors [25]. It also detects differences in community composition between study sites and has been recommended for TRFLP’s data [43]. The stability criterion used was 0.0005. The final model selected had a stress value less than 20 and was run in PC-Ord. Three NMS analyses were performed, one for bacteria, other for eukarya, and a third one including bacteria and eukarya together. We considered a set of candidate models consisting of watershed (e.g., land uses) and reach-scale variables to determine how these variables influenced bacteria, eukarya, and the complete biofilm community. Three main competing hypotheses were considered as follows: (a) microbial communities are influenced by land use, (b) microbial communities are influenced by reach-scale parameters, and (c) microbial communities are influenced by the interactions that exist between land uses and reach-scale parameters. We calculated the variance inflation factor to eliminate redundant variables from the analysis. Interactions among variables were also considered. Model selection was used to identify the hypothesis that best supported benthic biofilm diversity data. We used the Akaike Information Criterion corrected for small sampling size (AICc). Variables were checked for normality and
Benthic Biofilm Diversity
transformed when needed. All possible models, included the intercept and error, were obtained from R (version 15.2.0) using the AICcmodavq package to compare and select the best models. Delta AIC (ΔAIC) was determined as the difference between one possible model and the model with the lowest AIC. Akaike weights (wI) were determined as the ratio that one model is best among other possible models using the likelihood of one model in comparison to the likelihood of all possible models. Evidence ratio was determined by dividing the Akaike weights of one model divided by the Akaike weights of the best possible model. Models were selected based on AICc and Akaike weights (wI) [21]. Simple linear regressions were performed to test for relationships suggested by the model.
Results Subcatchment Land Use In general, the Turabo catchment was dominated by forest cover. The area of 16 subcatchments considered in this study ranged from 0.21 to 2.50 km2 and formed a continuous gradient from forested to urban streams. Secondary forest was the dominant land cover in 10 of the 16 study streams with >50 % of forest cover (Table 1). Forest cover ranged from 14 to 100 % cover among all study sites (Table 1). Urban land use was variable and ranged from 0 to 86 %. Urbanization was characterized by being dense in the lower parts of the catchment, in particular near the city of Caguas, and more disperse in the upper areas, consisting of small clusters of houses and residential developments. Subcatchment area ranges from 0.17 to 2.51 km2, and no relationship was found with urban land use (R 2 = 0.17, p = 0.07). Elevation ranges from 90 to 420 m. Agricultural activities in the Turabo catchment were variable and consisted of small crop patches and low-intensity cattle pastures dispersed throughout the catchment, reaching a maximum of 43 %. Exposed soils did not appear to be associated with agricultural or urban land use and comprised
MICROBIOLOGY OF AQUATIC SYSTEMS
Diversity of Benthic Biofilms Along a Land Use Gradient in Tropical Headwater Streams, Puerto Rico Sofía Burgos-Caraballo & Sharon A. Cantrell & Alonso Ramírez
Received: 6 May 2013 / Accepted: 21 February 2014 / Published online: 20 March 2014 # Springer Science+Business Media New York 2014
Abstract The properties of freshwater ecosystems can be altered, directly or indirectly, by different land uses (e.g., urbanization and agriculture). Streams heavily influenced by high nutrient concentrations associated with agriculture or urbanization may present conditions that can be intolerable for many aquatic species such as macroinvertebrates and fishes. However, information with respect to how benthic microbial communities may respond to changes in stream ecosystem properties in relation to agricultural or urban land uses is limited, in particular for tropical ecosystems. In this study, diversity of benthic biofilms was evaluated in 16 streams along a gradient of land use at the Turabo watershed in Puerto Rico using terminal restriction fragment length polymorphism. Diversity indices and community structure descriptors (species richness, Shannon diversity, dominance and evenness) were calculated for both bacteria and eukaryotes for each stream. Diversity of both groups, bacteria and eukaryotes, did not show a consistent pattern with land use, since it could be high or low at streams dominated by different land uses. This suggests that diversity of biofilms may be more related to
S. Burgos-Caraballo (*) Department of Biology, University of Puerto Rico, P.O. Box 70377, San Juan, PR 00936-8377, USA e-mail: [email protected] S. A. Cantrell Department of Biology, Universidad del Turabo, P.O. Box 3030, Gurabo, PR 00778, USA A. Ramírez Department of Environmental Sciences, University of Puerto Rico, P.O. Box 190341, San Juan, PR 00919, USA
site-specific conditions rather than watershed scale factors. To assess this contention, the relationship between biofilm diversity and reach-scale parameters (i.e., nutrient concentrations, canopy cover, conductivity, and dissolved oxygen) was determined using the Akaike Information Criterion (AICc) for small sample size. Results indicated that nitrate was the variable that best explained variations in biofilm diversity. Since nitrate concentrations tend to increase with urban land use, our results suggest that urbanization may indeed increase microbial diversity indirectly by increasing nutrients in stream water.
Introduction Natural landscapes are transformed by land use activities that are necessarily to satisfy human needs (e.g., agriculture and urbanization) [15]. These landscape modifications may affect stream ecosystems due to the close relationship that streams have with their surrounding landscapes [1]. In streams heavily influenced by agricultural or urban land uses, elevated nutrient concentrations and conductivity could be expected, while oxygen concentrations decrease [1]. Light availability may increase as canopy cover of riparian vegetation is eliminated, but can decrease with depth due to increased turbidity in altered streams. Other factors associated to land use, such as leaky pipes and combined sewer overflow entering streams, can have similar results [1, 37, 41, 48]. These degraded conditions may be intolerable for many aquatic organisms, such as macroinvertebrates and fishes, as well as other organisms with life cycles linked to freshwaters ecosystems (e.g., amphibians and reptiles) [7]. Therefore, sensitive species could be
48
eliminated, while more tolerant species may persist and become more abundant causing a shift in community composition [26]. Likewise, previous studies from temperate regions demonstrated that microbial communities can change in composition as well as in diversity in streams influenced by urban land use [8, 39, 49]. For example, Dopheide et al. found that streams more impacted by urbanization showed higher diversity of ciliates than undisturbed forested areas [11]. A similar pattern was reported by Carrino-Kyker et al. for different functional groups of eukaryotes (algae, fungi, and protist) in urban vernal pools [8]. However, it is not clear if the same patterns observed in microbial communities from temperate regions will be similar in benthic biofilms from tropical streams. This is of major concern, as changes in species composition can result in alterations in ecosystem function [19]. The stream benthic environment is inhabited by complex assemblages of microbial communities, including bacteria and eukaryotes such as algae, fungi, and other protists. This conglomerate of microbial organisms, known as biofilms, is responsible for many essential ecosystem processes such as organic matter decomposition and nutrient cycling [3, 29, 37]. In streams on Caribbean islands, food webs are primarily based on biofilms and are the main food source for different species of fish and shrimp, which in turn represent an important protein source for humans [24]. However, the quality of these food sources can be affected by land uses [5]. Therefore, changes in the community composition of benthic biofilm, as a result of human impact to watershed and reach conditions, may result in the alteration in streams ecosystem function [2]. The purpose of this study was to analyze benthic biofilm community composition in tropical streams along a land use gradient. At the same time, we explored the relationship between reach and watershed factors in order to determine how these variables influenced community composition and diversity of benthic biofilms. We expected to find differences in community composition by considering forest and urbanization as extremes of the land use gradient. However, we did not predict a direction of change in diversity since previous studies indicate that microbial community richness can be positively or negatively influenced by urban land use [8, 11, 50].
Materials and Methods Watershed and Reach-Scale Parameters The Turabo watershed is located in the municipality of Caguas, in the central northeastern region of Puerto Rico (Fig. 1). The geology of the zone is dominated by volcanic and volcano clastic rocks from the late Cretaceous together with some
S. Burgos-Caraballo et al.
fragments of plutonic and alluvium deposits from the Quaternary [47]. In general, the watershed is dominated by forest land use, with agriculture and urbanization as secondary land uses. Forest cover consists of montane, submontane, and lower montane evergreen forest combined with abandoned coffee plantations [18]. Agricultural activities consist of pastures with small fragments used for small-scale farming. The lower portions of the watershed are close to downtown Caguas where urbanization is concentrated. The areas of high elevation in the watershed consist mainly of suburban development. A closer examination of data revealed that no relationship exists between elevation and urban land use. Land Use Analyses For our analyses, the term “land use” refers to land use/land cover data. Land use classifications were made for each study stream using geographic information systems (GIS, ESRI ver. 9.1). The subcatchment of each study stream was delimited by identifying the topographic divide following contour lines on topographic maps (scale 1:10,000) to estimate the area. GIS tools were used to classify, delineate, and digitize land uses. Photo interpretation mapping scale was 1:10,000 and was performed by creating polygons of areas no smaller than 4,000 m2. Dominant land uses were identified by using 2007 aerial images provided by the Department of Environmental and Natural Resources of the Puerto Rico Commonwealth. Each polygon was categorized into one of four general land use types: urban, forest, agriculture, and other (e.g., bare soil; Table 1). Physicochemical Parameters At each stream, we measured water temperature, dissolved oxygen (O2) (using a luminescent dissolved oxygen probe, HQ10–HQ20 meters, Hach Company, Colorado), pH (using a Hanna Instrument, HI 98150 model), conductivity, total dissolved solids, and sodium chloride (all using a Hanna Instrument, HI 9835). Incident light was measured with a LI-Cor quantum sensor (LI-193 Spherical Quantum Sensor), and canopy cover was estimated taking four measurements with a concave spherical densiometer. Nutrients and major ions were measured by collecting a filtered water sample (Whatman® glass microfiber filters, GF/ F 47 mm) at each stream. Samples were frozen until analysis at the University of New Hampshire Water Quality Analysis Laboratory. Samples were analyzed for chloride (Cl−), nitrate (NO3−), sulfate (SO4−3), sodium (Na+), potassium (K+), magnesium (Mg+2), and calcium (Ca+2) using a Dionex ICS1000 ion chromatograph. Nutrients analysis included ammonium (NH4+) and phosphate (PO4−3) (Westco Scientific SmartChem 200 discrete automated analyzer), total dissolved nitrogen (TDN) and dissolved organic carbon (DOC) (Shimadzu
Benthic Biofilm Diversity
49
TOCV with total nitrogen module), and dissolved organic nitrogen (DON) (calculated as the differences between TDN and NO3−-N + NH4+-N). Benthic Biofilm Sampling Samples were collected from September to November in 2008 during the rainy season. The northeastern region of Puerto Rico, where these studies were conducted, lacks strong climatic seasonality; therefore, it is probable that the time of year played a minor role. In each stream, biofilms were collected from rocks at four different riffles separated by at least one sequence of pools and riffle to reduce influence between them. Collected rocks were placed in individual sterile Whirl Packs bags and refrigerated during transport. Once in the laboratory, all benthic material was removed from the rocks by scraping the surface using a sterile metal brush and rinsed with autoclaved distillated water. From all removed material, approximately 1.5 ml was collected and centrifuged at 1,000 rpm velocity to create pellets which were frozen at −20° until DNA extraction. DNA Extraction, PCR, and TRFLP’s Analysis
Fig. 1 Study site location. The shaded regions represent the subwatershed associated to each stream. Each number corresponds to the each study site as is indicated in the text and figures
In order to determine benthic community structure, terminal restriction fragment length polymorphism (TRFLP) was analyzed for bacteria 16S recombinant DNA (rDNA) genes and Eukarya 18S rDNA; a fingerprint approach that is useful to differentiate communities, compare phylotype richness, and community structure among sites [14, 28, 36].
Table 1 Land use percent at each study site Study stream
Forest
Urban
Agriculture
Other
Catchment area (km2)
Elevation (m)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
100 58 59 57 74 76 86 64 55 66 45 35 23 23 24 14
0 0 0 0 2 2 7 13 25 27 29 34 49 67 70 86
0 42 40 43 24 22 7 18 19 7 22 27 7 3 0 0
0 0 1 0 0 1 0 5 0 0 4 4 0 7 6 0
0.23 1.01 0.37 1.92 1.97 2.13 0.20 1.02 1.36 2.51 0.69 1.30 0.17 0.30 0.28 0.21
100 210 290 330 140 90 120 120 110 90 110 330 330 410 90 420
Each number corresponds to the each study site as is indicated in the text and figures
50
Biofilm DNA was extracted using MoBio UltraClean soil kit (MoBio Laboratories, Solana Beach, CA) following manufacturer protocols. DNA concentrations were quantified by fluorometry. Amplifications were conducted using 12.5 μg REDTaq Polymerase (Sigma-Aldrich, St. Louis, MO), 2–4 μg of DNA template, 0.625 μg of primers, and 0.5 μg of bovine serum albumin (BSA; Fermentas, Vilnius, Lithuania) to reduce interference with contaminants. Bacteria 16S rDNA genes were amplified using the following set of primers 16S-27F (5′AGAGTTTGATCMTGGCTCAG) and 16S1525R (5′AAGGAGGTGWTCCARCC). Eukarya 18S rDNA was amplified using universal set of primers 18S-82F (5′GAAACTGCGAATGGCTC) and 18S-1520R (5′CYGC AGGTTCACCTAC). For each set of primers, only the forward primer was fluorescently labeled at the 5′ end with FAM dye (Sigma-Aldrich, Atlanta, GA). Polymerase chain reactions (PCR) were performed with a model 2720 thermal cycler (ABI, USA) using the following program: 1-min hot start at 80 °C, 94 °C for 5 min followed by 30 cycles of denaturation at 94 °C for 30 s, followed by annealing at 52 °C for 30 s, at 72 °C for 1 min 30 s, with a final extension step at 72 °C for 10 min. Amplified DNA was verified by electrophoresis of PCR mixtures in 1.0 % agarose in 1X TAE buffer. Fluorescently labeled amplicons were enzymatically digested with BsuRI (HaeIII), Rsa, and MnlI (Fermentas Life Science, Maryland, USA) to find the enzymes with the higher resolution for both, bacteria and eukarya. MnlI was chosen as it provided the greatest numbers of terminal restriction fragments (T-RFs) for bacteria, as well as for eukarya. Digestions were performed at 37 °C for 120 min with 2.0 μl of buffer, 0.2 μl of enzyme, and 5.0 μl of amplified product and water to a final volume of 20 μl. Samples were precipitated in ethanol to eliminate impurities, dried, and resuspended in Hi-Di formamide with GeneScan Liz 500 (ABI, USA) using a size standard that range between 50 and 500 bp. Those fragments that differed by 1 bp were considered different. Samples were run on ABI 3130 genetic analyzer (ABI, USA). Data were normalized by using fragments with a fluorescence >1 % of the total fluorescence. The area of each peak was standardized with respect to the sum of all peak areas of the sample. We considered the fragments that were present at three of the four replicate analyzed at each study site. Restriction fragments were analyzed using the web page MICA (Microbial community analysis [45]). The program APLAUS was selected to infer the community structure based on the TRFLP data of communities present in the biofilms [45], but since one TRFLP could represent different organisms, bacteria were identified at the Phyla level, unless they could be identified to a lower level. Databases of Eukarya were more limited and therefore no eukaryote could be identified.
S. Burgos-Caraballo et al.
Statistical Analysis To reduce the influence of most common Operational Taxonomic Units (OTUs), a square root transformation was used for data obtained at each study site [17]. All OTUs present in a sample were sorted by abundance to identified shared OTUs (by considering those that were present in 10 of the 16 study sites) unique OTUs (by considering the phylotypes that were present in 1 of the 16 study sites). OTUs richness (S) was determined by counting the total number of OTUs present at each study site. Shannon diversity index (H′), Simpson’s diversity index (D), and Pielou’s evenness index (E) were determined using the average abundance of OTUs per study site in PC-Ord version 4.25 [4]. In addition, Smith and Wilson evenness (Evar), previously recommended as a measure of diversity for TRFLP’s data, was calculated [4]. A nonmetric multidimensional scaling (NMS) ordination was used to analyze the profiles of OTUs among study streams, using a square root transformation of the relative height for each OTUs. We choose this ordination for various reasons. It is an efficient ordination method where data does not have to be linearly related [32]. NMS also alleviates problems associated to data sets containing zeros [32]. In addition, the user can choose the distance measure appropriate for the data [32]. We run the NMS considering the following parameters: 6 axes as maximum, 500 iterations, 0.20 steps in length, and 50 runs with real data and randomized data. The dissimilarity distance measure selected for this ordination was Bray-Curtis because it considers the presence of a species as more informative than its absence, as the absence of species from the data is not necessarily associated with environmental factors [25]. It also detects differences in community composition between study sites and has been recommended for TRFLP’s data [43]. The stability criterion used was 0.0005. The final model selected had a stress value less than 20 and was run in PC-Ord. Three NMS analyses were performed, one for bacteria, other for eukarya, and a third one including bacteria and eukarya together. We considered a set of candidate models consisting of watershed (e.g., land uses) and reach-scale variables to determine how these variables influenced bacteria, eukarya, and the complete biofilm community. Three main competing hypotheses were considered as follows: (a) microbial communities are influenced by land use, (b) microbial communities are influenced by reach-scale parameters, and (c) microbial communities are influenced by the interactions that exist between land uses and reach-scale parameters. We calculated the variance inflation factor to eliminate redundant variables from the analysis. Interactions among variables were also considered. Model selection was used to identify the hypothesis that best supported benthic biofilm diversity data. We used the Akaike Information Criterion corrected for small sampling size (AICc). Variables were checked for normality and
Benthic Biofilm Diversity
transformed when needed. All possible models, included the intercept and error, were obtained from R (version 15.2.0) using the AICcmodavq package to compare and select the best models. Delta AIC (ΔAIC) was determined as the difference between one possible model and the model with the lowest AIC. Akaike weights (wI) were determined as the ratio that one model is best among other possible models using the likelihood of one model in comparison to the likelihood of all possible models. Evidence ratio was determined by dividing the Akaike weights of one model divided by the Akaike weights of the best possible model. Models were selected based on AICc and Akaike weights (wI) [21]. Simple linear regressions were performed to test for relationships suggested by the model.
Results Subcatchment Land Use In general, the Turabo catchment was dominated by forest cover. The area of 16 subcatchments considered in this study ranged from 0.21 to 2.50 km2 and formed a continuous gradient from forested to urban streams. Secondary forest was the dominant land cover in 10 of the 16 study streams with >50 % of forest cover (Table 1). Forest cover ranged from 14 to 100 % cover among all study sites (Table 1). Urban land use was variable and ranged from 0 to 86 %. Urbanization was characterized by being dense in the lower parts of the catchment, in particular near the city of Caguas, and more disperse in the upper areas, consisting of small clusters of houses and residential developments. Subcatchment area ranges from 0.17 to 2.51 km2, and no relationship was found with urban land use (R 2 = 0.17, p = 0.07). Elevation ranges from 90 to 420 m. Agricultural activities in the Turabo catchment were variable and consisted of small crop patches and low-intensity cattle pastures dispersed throughout the catchment, reaching a maximum of 43 %. Exposed soils did not appear to be associated with agricultural or urban land use and comprised
