Draft

Bacterial diversity in the waterholes of the Kruger National Park: an eDNA metabarcoding approach

Journal: Genome

Manuscript ID gen-2018-0064.R1

Manuscript Type: Article

Date Submitted by the Author: 08-Jul-2018

Complete List of Authors: Farrell, Maxwell; McGill University, Department of BiologyGovender, Danny; Scientific Services, Kruger National Park, SANParks; University of Pretoria, Department of Paraclinical Sciences, Faculty of Veterinary ScienceHajibabaei, Mehrdad; University of GuelphVAN DER BANK, Michelle; University of Johannesburg, African Centre for DNA BarcodingDavies, T; University of British Columbia, Botany, Forestry & Conservation Sciences; University of Johannesburg, African Centre for DNA Barcoding

Keyword: microbiome, conservation, biomonitoring, 16S, watering holes

Is the invited manuscript for consideration in a Special

Issue? :7th International Barcode of Life

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

Bacterial diversity in the waterholes1

of the Kruger National Park:2

an eDNA metabarcoding approach3

Maxwell J. Farrell1∗, D. Govender2,3,M. Hajibabaei4, M. van der Bank5, T. J. Davies5,6

1Department of Biology, McGill University2Scientific Services, Kruger National Park, SANParks

3Department of Paraclinical Sciences, Faculty of Veterinary Science, University of Pretoria4Integrative Biology & Biodiversity Institute of Ontario, University of Guelph

5African Centre for DNA Barcoding, University of Johannesburg6Botany, Forest & Conservation Sciences, University of British Columbia

∗To whom correspondence should be addressed; e-mail: maxwell.farrell@mail.mcgill.ca

4

September 21, 20185

1

Page 1 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

Abstract6

Bacteria are essential components of natural environments. They contribute to ecosystem functioning7

through roles as mutualists and pathogens for larger species, and as key components of food webs8

and nutrient cycles. Bacterial communities respond to environmental disturbances, and the tracking9

of these communities across space and time may serve as indicators of ecosystem health in areas10

of conservation concern. Recent advances in DNA sequencing of environmental samples allow for11

rapid and culture-free characterization of bacterial communities. Here we conduct the first metabar-12

coding survey of bacterial diversity in the waterholes of the Kruger National Park, South Africa.13

We show that eDNA can be amplified from waterholes and find strongly structured microbial com-14

munities, likely reflecting local abiotic conditions, animal ecology, and anthropogenic disturbance.15

Over timescales from days to weeks we find increased turnover in community composition, indicating16

bacteria may represent host-associated taxa of large vertebrates visiting the waterholes. Through taxo-17

nomic annotation we also identify pathogenic taxa, demonstrating the utility of eDNA metabarcoding18

for surveillance of infectious diseases. These samples serve as a baseline survey of bacterial diversity19

in the Kruger, and in the future, spatially distinct microbial communities may be used as markers of20

ecosystem disturbance, or biotic homogenization across the park.21

2

Page 2 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

Introduction22

Traditional programs that monitor for early signs of ecosystem degradation require baseline data on the23

distributions and ecology of species in an ecosystem. DNA barcoding uses differences in conserved24

regions of genomes to classify sequences as belonging to particular taxonomic units, regardless of25

whether or not they have been described formally by taxonomists (Hebert et al. 2003; Blaxter et al.26

2005; Ratnasingham and Hebert 2013). DNA barcoding is thus a particularly powerful tool for explor-27

ing microbial diversity, where there are many undescribed taxa that cannot be cultured using traditional28

methods (Rappe and Giovannoni 2003). Molecular barcoding coupled with recent advances in genetic29

sequencing have allowed for unprecedented exploration of microbial communities and the ability to30

characterize organisms of interest from environmental samples with great sensitivity (Shokralla et al.31

2012). In particular, sequencing of cellular and extracellular DNA that can be extracted from envi-32

ronmental samples, collectively known as environmental DNA (eDNA) (Taberlet et al. 2012), is an33

emerging approach for exploring diversity in aquatic ecosystems (Rees et al. 2014; Lodge et al. 2012).34

Microbial diversity in freshwater systems responds to environmental conditions (Lozupone and35

Knight 2007), and perturbations (Zeglin 2015) including multiple anthropogenic impacts such as ur-36

banization (Fisher et al. 2015) and pollution (Bouskill et al. 2010). In addition to acting as indicators of37

ecosystem health, changes in microbial diversity may be important in themselves. Bacteria are essen-38

tial components of ecosystems and play important roles in food webs, nutrient recycling, disease, and39

as important mutualists for larger multicellular species. Bacteria are thus integral to maintaining the40

natural balance of ecosystems and shifts in taxonomic composition due to environmental change may41

severely impact connectivity, functioning (Laforest-Lapointe et al. 2017; Delgado-Baquerizo et al.42

2016) and increase exposure to pathogens (Cabral 2010). However, in most ecosystems of conserva-43

tion priority, microbial diversity is poorly described.44

Surface waters are a vital resource for savannah ecosystems (Redfern et al. 2005; Owen-Smith45

1996), but frequent use by a large variety of species means they can also be a source of cross-species46

infection and spread of harmful pathogens (Bengis and Erasmus 1988). These ecosystems provide47

an ideal context for refining eDNA metabarcoding approaches as they act as sources and sinks of48

microbial species for larger animals, however baseline information about microbial diversity in these49

3

Page 3 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

systems is lacking. Here we conduct a survey of bacterial diversity among watering holes of Kruger50

National Park, South Africa (KNP) through spatio-temporal sampling and sequencing of the V3-V451

region of 16S genes present in water. Water can be scarce in the KNP throughout the dry season and52

periods of drought (Redfern et al. 2005), and the park has a long history of water provisioning that53

included the construction of a series of more than 300 artificial waterholes beginning in the 1930s54

(Smit et al. 2007). These waterholes were intended to increase game numbers by stabilizing water55

availability year-round and are frequently visited by a diversity of birds and mammals (Smit and56

Grant 2009). However, they have proven to alter the distributions of wildlife, which in turn have57

negative impacts on vegetation dynamics and the park-wide ecosystem (Smit et al. 2007; Smit and58

Grant 2009). As a result, a number of artificial waterpoints have been closed since 1994 as the park59

began reverting to a more natural cycle of water availability (Smit et al. 2007; Van Wyk 2011). A60

subset of the waterholes still open are small concrete troughs which are well mixed, largely mud and61

silt-free, and experience limited inflow from nearby surface waters. This means that eDNA samples62

will largely represent microbes in sourcewater and those dispersed by air and animals, allowing us to63

capture snapshots of local bacterial diversity across the park.64

This study provides the first survey of bacterial diversity in the waterholes of the KNP, and is65

among the first studies using next-generation sequencing to describe aquatic microbial diversity in66

Africa (see Jordaan and Bezuidenhout (2016; 2013); Tekere, M., Prinsloo, A., Olivier, J., Jonker, N.,67

Venter, S. (2012); Mwirichia, R., Cousin, S., Muigai, A. W., Boga, H. I., Stackebrandt, E. (2011);68

Tekere, M., Lotter, A., Olivier, J., Jonker, N., Venter, S. (2011)). Here we explore bacterial diversity69

across the southern half of the park and describe variation across, space, time, sample volume, and70

abiotic influences.71

Methods72

Study Site73

Waterholes were sampled in June and July of 2015 in the Kruger National Park, South Africa (KNP), a74

large protected savannah ecosystem and a global diversity hotspot (Lahaye et al. 2008). Sampling was75

4

Page 4 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

conducted during the dry season when natural sources of surface water are largely dry and watering76

hole visitation rates by medium and large vertebrates are highest. The park is divided into twenty-two77

ranger sections, which range in size from roughly 520 to 1,170 square kilometers. Across the southern78

half of the reserve below the Olifants river, ten concrete bottom artificial waterholes were selected79

from five of these sections (Table 1, Fig. 1). The waterholes varied in shape with some mimicking the80

contours of natural pans, making volume estimations difficult. However, the generic design included81

longer and shorter axes, with comparable dimensions across waterholes. Each waterhole is equipped82

with a ball-valve, which regulates water levels and re-fills the trough from nearby reservoirs when83

water levels drop. Water is sourced predominantly from groundwater via boreholes, but three sites use84

pipeline troughs filled with diverted river water.85

Water Sampling and Processing86

At each site, samples were taken once per week for three weeks. For one site (NWA), samples were87

also taken every day for five consecutive days. Sampling consisted of two 1L water samples collected88

in autoclaved, UV sterilized glass jars from opposite ends of the waterhole, approximately one foot89

from the nearest edge. These two within-waterhole samples (A/B) were taken along the waterhole’s90

longest axis that maximized the distance and upwind-downwind gradient between them, if a strong91

wind was present. Water samples were placed on icepacks in a cooler and kept between 4-8◦C until92

returning to the laboratory, where they were placed in the fridge.93

Water quality parameters were taken during each sampling period using a YSI 650QS multi-94

parameter sonde. Temperature (◦C), conductivity (mS/cm), dissolved oxygen (in mg/L and % sat-95

uration), and pH were recorded. Three measurements were taken along the same axis that the A/B96

water samples were drawn, and then averaged to measure quality per sample-time.97

In the lab, the outside of water sample collection bottles were washed with ELIMINase (Decon98

Labs) and rinsed with deionized (DI) water to limit contamination. For each A/B sample, 150 mL of99

water was sub-sampled and filtered through gamma-irradiated 0.2 µm Supor hydrophilic polyether-100

sulfone membranes (Pall no. 66234). The filtration apparatus consisted of three 300 mL Advantec101

polysulfone 47mm filter funnels fitted to a Pall vacuum manifold with vacuum pressure maintained by102

5

Page 5 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

a Pall filtration vacuum/pressure pump (model no. 13158). After filtration, filters were stored in sterile103

15 mL Falcon tubes and placed in a freezer at -60◦C. On one sampling date for six sites, additional104

volumes of 50 ml and 15 ml were filtered from each 1L sample to asses the impact of sample volume105

filtered. Twice throughout sampling, BLANK samples were generated by filtering 1L of deionized106

water used in the laboratory.107

Prior to and between filtrations, all funnel components and tweezers used to manipulate the filters108

were sterilized by soaking with 10% bleach for 10 minutes, rinsing with DI water, washing with109

ELIMINase, rinsing with DI water, and subsequent exposure to UV radiation for a minimum of 30110

minutes. Gloves were worn at all times and changed between samples to minimize cross-sample111

contamination. To avoid sample freezing and bacterial growth in collection jars, all samples were112

processed within 12 hours of collection. Frozen 50 mL unfiltered voucher samples were kept and113

placed at -80◦C at the University of Johannesburg’s African Centre for DNA Barcoding for long term114

storage.115

DNA Extraction, Amplification, and Sequencing116

DNA was isolated from filter papers using MO BIO PowerWater DNA Isolation Kits. Universal bac-117

terial primer sets designed by Sundquist et al. (2007) (V3-F: 5’ACTCCTACGGGAGGCAGCAG 3’;118

V4-R: 5’GGACTACARGGTATCTAAT 3’) tagged with an Illumina adapter sequence were used to119

amplify the V3-V4 hypervariable region of the 16S ribosomal RNA gene through polymerase chain120

reaction (PCR). The PCR used a standard mix of 17.8µL molecular grade water, 2.5µL 10 reaction121

buffer (200mM Tris HCl, 500mM KCl, pH 8.4), 1µL MgCl2 (50mM), 0.5µL dNTP (10mM), 0.5µL122

forward primer (10mM), 0.5µL reverse primer (10mM), 0.2µL Platinum Taq DNA polymerase (Invit-123

rogen), and 2µL DNA as template for a total volume of 25µL. PCRs underwent the following cycler124

conditions: initial 94◦C for 5 minutes, then 30 cycles of 94◦C for 40 seconds, 46◦C for 1 minute, 72◦C125

for 30 seconds, and a final temperature of 72◦C for 2 minutes. Amplification success was confirmed126

through gel electrophoresis, using a 1.5% agarose gel. PCR products were purified using MinElute127

PCR purification kit (Qiagen), and quantified through flurometry using a Quant-iT PicoGreen dsDNA128

assay kit (Invitrogen). Samples were normalized, then multiplexed with the Nextera XT Index kit (96129

6

Page 6 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

indexes) (Illumina) and sequenced on an Illumina MiSeq flowcell using a V2 sequencing chemistry130

kit (2 x 250) making up approximately 1/8th of the run.131

Sequence Processing, Taxonomy Assignment, and Phylogeny Construction132

Across all samples, we generated a total of 2,164,262 Illumina reads. Primer sequences were removed133

using the trim.seqs function in mothur (Schloss et al. 2009). Reads were then processed in R (version134

3.4.3) (R Development Core Team 2008) using the package dada2 version 1.6.0 (Callahan et al. 2016)135

following a modified version of the DADA2 Bioconductor workflow (Callahan et al. 2017) and online136

tutorials v1.6 and workflow for big data v1.4 (benjjneb.github.io/dada2/tutorial.html). Reads were137

filtered by quality, removing sequences with maximum expected error (maxEE) greater than 6 for both138

forward and reverse reads, and reads with any base pair having Q of 6 or lower. Reads were truncated139

to a length of 230bp and 220bp for forward and reverse reads respectively, consistent with dropoffs in140

quality profiles, and reads shorter than this were removed. Since the samples were sequenced across141

four different runs, subsequent steps of learning error rates, dereplication, denoising and Amplicon142

Sequence Variant (ASV) calling (Callahan et al. 2017) using pooled samples, and merging of paired143

reads were performed separately for each run. Tables of ASV sequences per sample within each run144

were then combined and chimera detection using all pooled samples was performed (see SM Table S2145

for the number of reads retained across each step). In total 1,184,831 reads were retained, representing146

3533 ASVs.147

Taxonomy assignment from Kingdom to Genus was performed using the RDP classifier and148

SILVA nr v128 reference database (Quast et al. 2013) formatted for DADA2 (benjjneb.github.io/dada2/training.html),149

using the assignTaxonomy function (Fig. 5). ASVs assigned as Archaea, Eukarya, Chloroplast, or Mi-150

tochondria were removed. Species level assignments were added by exact sequence matching using151

the addSpecies function. ASV sequences were aligned with the pynast algorithm via align seqs.py in152

QIIME (Caporaso et al. 2010) and sequences with poor alignment automatically removed. A phylo-153

genetic tree was constructed using the GTRCAT model in FastTree version 2.1.3 (Price et al. 2010)154

after filtering nucleotides with greater than 90% gap fraction and removing the 5% highest entropy155

positions with filter alignment.py in QIIME (Caporaso et al. 2010). This resulted in a phylogenetic156

7

Page 7 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

tree of 3393 ASVs which were used in subsequent community analyses.157

Community Analyses158

The ASV sequence table was merged with the phylogeny and sample metadata using the R pacakge159

phyloseq version 1.22.3 (McMurdie and Holmes 2013). Negative controls (BLANK samples) used to160

investigate contamination during sample filtration contained 43 ASVs collectively (with 9 ASVs found161

in both samples). The sequence reads in each filtration blank were both dominated by the same ASV162

(53% and 86% respectively), however none of the 43 ASVs identified in the blanks were identified in163

any of the other samples. These control samples were removed prior to community analyses.164

A subset of core samples was created by removing the first four daily NWA samples and samples165

of differential volume (S & XS samples), resulting in 54 samples of 150 mL each (Table 2). An ASV166

accumulation curve for core samples was generated using the specaccum function in the R package167

vegan version 2.4.6 (Oksanen et al. 2018) using the “exact” method, and extrapolated to total ASV168

richness using the Chao and Bootstrap methods in vegan’s specpool function. Alpha diversity was169

calculated for the core samples as observed ASV richness and Shannon diversity using the phyloseq170

package. Additive partitioning of Shannon diversity across core samples was investigated using the171

adipart function in vegan (Table 3).172

Taxonomic composition was assessed by merging core samples at each site, and plotting rela-173

tive abundances of reads for the most common taxa at levels of phylum, class, and order (Fig. 6).174

Temporal variation in taxonomic composition across core samples was assessed by merging A and B175

samples and plotting relative abundances of reads for sites with two or more weekly samples, for the176

levels of phylum (Fig. 7), and class (Fig. S6). To further investigate fine-scale temporal variation in177

taxonomic composition (phylum, class, and order), relative abundance of reads were plotted across178

the daily samples at site NWA (Fig. 8). We also explored temporal turnover among samples with179

Sorensen’s dissimilarity calculated using the beta.pair function from the betapart package (Baselga180

et al. 2018) (Fig. S9) and significant differences among daily and weekly samples was assessed using181

permutationsl multivariate analysis of variance (adonis in vegan) with 999 permutations each.182

Community composition across sites in the core samples was described with nonmetric multidi-183

8

Page 8 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

mentional scaling (NMDS) ordinations on relative ASV abundances per sample using the Bray-Curtis184

dissimilarity, and the abundance weighted Unifrac dissimilarity (Figs. 9, S10 & S11). Statistically,185

associations between dissimilarities and both water quality properties and common taxonomic groups186

were assessed using the envfit function in vegan for bacterial classes (Fig. 10) and orders (Fig. S12).187

Phylogenetic community structure across core samples was calculated using standardized effect188

sizes of mean pairwise phylogenetic distances (MPD) and mean nearest taxon distances (MNTD) in189

the R package picante version 1.6.2 (Kembel et al. 2010) using the abundance weighted “richness”190

null model and 999 randomizations in the ses.mpd and ses.mntd functions (Fig. S13). For a given sam-191

ple, MPD calculates the mean phylogenetic distance among each pair of taxa present, while MNTD192

calculates the mean phylogenetic distance from each taxa to its closest relative. These raw metrics193

give an estimate of how closely related community members are to each other, and are then compared194

to randomized communities to determine whether the observed metrics are different than what would195

expected if communities were assembled at random from taxa pooled across all samples.196

To assess the effect of differential sample volumes, S (50 mL) and XS (15 mL) samples were subset197

along with their corresponding full volume samples (150 mL). Alpha diversity, calculated as observed198

ASV richness and Shannon diversity were calculated as described above (Fig. S14). Variation in199

taxonomic composition was investigated by comparing relative read abundances of bacterial phyla in200

A/B samples across sites and differential volumes (Fig. S15).201

Raw reads with primers removed are available via the NCBI Sequence Read Archive BioProject202

PRJNA490450 (accession numbers SRR7822814 to SRR7822901). The ASV table, taxonomic as-203

signments, phylogenetic tree, sample metadata, and scripts necessary to reproduce the results can be204

found at 10.6084/m9.figshare.7119668.205

Results206

Our sampling design aimed to sequence a core set of samples with all ten sites being sampled once207

per week for three weeks. Due to logistic constraints of sample storage and extremely low water208

levels from drawdown by animals, we were only able to process 27 of the planned 30 weekly samples209

(Table 2). In addition to this core sampling, we sequenced differential volumes for six samples, and210

9

Page 9 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

an additional four daily samples from Nwaswitshaka (site NWA). A/B samples were taken at each211

site-time, resulting in a total of 88 sequenced samples, including the two filtration blanks. Across all212

88 samples, we identified a total of 3393 ASVs. Roughly 15% of ASVs (n=524) were represented213

by a single read, together comprising fewer than 0.05% of all reads. The DADA2 approach infers the214

biological sequences in the sample prior to the introduction of amplification and sequencing errors,215

and can distinguish sequence variants differing by as little as one nucleotide. As such, we included all216

ASVs, including those represented by single reads, in subsequent analyses of biodiversity.217

The ASV accumulation curve generated for the core sample set (2603 ASVs) does not appear218

to saturate (Fig. 2). Estimates of total richness using the Chao estimator predicts 6164 ASVs (+/-219

262 SE) among the core samples, indicating we may be capturing less than half of the total bacterial220

diversity present among our sites. However, estimates of total diversity using the Bootstrap method221

were more conservative, with 3260 (+/- 146 SE) estimated ASVs. ASV diversity varied across sites222

(Figs. 3, Fig. 4), but the largest turnover (β diversity) was observed among park sections (Table223

3). Variation among A/B samples contributed very small amounts to β diversity, indicating that at a224

particular time, microbial diversity within each waterhole was fairly well mixed.225

In terms of taxonomic composition, 99.2% of ASVs were assigned to a known phylum, with the226

proportion of assignments decreasing at lower taxonomic levels (Fig. 5). The majority of bacterial227

ASVs were classified as Proteobacteria (∼ 59%), followed by Bacteroidetes (∼ 14%), Firmicutes (∼228

9%), Actinobacteria (∼ 6%), and Verrumicrobia (∼ 2%). For bacterial classes, ASVs were largely229

classified as Betaproteobacteria (∼ 34%), Alphaproteobacteria (∼ 11%), Gammaproteobacteria (∼230

10%), and Sphingobacteriia (∼ 6%). Among core samples, relative abundances of phyla, classes,231

and orders varied across sites (Figs. 6, S4, S5). Across weeks, relative abundances of phyla varied232

within each site (Fig. 7), with some sites displaying more stability (IMB & HOY) compared to others233

(NYA & NGO). Patterns among bacterial classes (Fig. S6) largely reflected variation among phyla,234

though one site (HOY) displayed much more variation in relative abundances among classes, reflecting235

substantial turnover within Proteobacteria. Comparing weekly turnover with the five daily samples236

taken at Nwaswitshaka (NWA) (Fig. 8), taxonomic composition appeared more stable across days than237

weeks. Using hierarchical clustering of Sorensen’s dissimilarity, we find that samples taken within238

10

Page 10 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

a single week cluster together (Fig. S9). Permutational multivariate ANOVAs on these distances239

revealed a significant difference in beta diversity among weekly samples (NWA 2,7,8; p = 0.02), with240

49% of the variance explained by sample date, but no significant difference among the additional daily241

samples (NWA 3,4,5,6; p = 0.54), with 14% of the variance explained by sample date.242

Community composition visualized through NMDS ordinations reflected results from the additive243

partitioning of diversity, with core samples clustering by site (Fig. 9) and section (Fig. S10) for both244

Bray-Curtis and abundance weighted UniFrac dissimilarities. Interestingly, waterholes filled by water245

from pipeline troughs (NGO, NYA, WIT) grouped together (Fig. S10), although these three sites are246

situated on a different geological type than sites fed by boreholes, making us unable to differentiate247

the effects of each factor (Fig. S11). Bacterial community composition was significantly structured248

by conductivity and pH for both Bray-Curtis and UniFrac dissimilarities, and dissolved oxygen also249

had an influence on UniFrac dissimilarity (Figs. 10 & S12).The dissimilarity of high conductivity250

sites (particularly HOY & IMB in the Kingfisherspruit section) was associated with high abundances251

of Clostridia, Gammaproteobacteria, and Bacteroidia, while sites with high pH and dissolved oxygen252

were positively related to the abundances of Actinobacteria and Alphaproteobacteria (Fig. 10).253

Reflecting the NMDS structure of the abundance weighted UniFrac dissimilarities, MNTD, which254

is most sensitive to phylogenetic structure towards the tips of the tree (Mazel et al. 2015), indicated255

strong phylogenetic clustering within the majority of samples (Fig. S13). The strength of clustering256

was weaker for MPD, which is more sensitive to phylogenetic structure deeper in the tree (Mazel et al.257

2015).258

We did not find any clear decrease in alpha diversity with smaller sample volumes (Fig. S14),259

and one of the 15 mL samples returned the largest richness of ASVs, though the median value for260

15 mL samples was lower and had a larger interquartile range than the 50 mL and 150 mL samples.261

The major phyla detected within samples was also relatively consistent, with most groups represented262

across different sample volumes, though not always in the same proportions (Fig. S15).263

11

Page 11 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

Discussion264

Biological monitoring is an essential aspect of conservation for tracking contemporary changes in265

ecosystems as well as providing a historical baseline for making management decisions. The Kruger266

National Park, established in 1898, has a long history of management practices revolving around267

maintenance of large mammals (Venter et al. 2008). While bacterial diversity has been explored for268

important infectious agents in the system (Michel et al. 2007; Bengis and Erasmus 1988; Smith et al.269

2000), recent advances in next generation sequencing methods now allow for the rapid and culture-free270

description of bacterial diversity throughout the park.271

Here we present the first description of bacterial diversity in the waterholes of the Kruger Na-272

tional Park. In total we identified over 3000 unique taxa (referred to as amplicon sequence variants, or273

ASVs), only about half of which could be assigned to a previously described genus. The relative domi-274

nance of bacterial phyla was consistent with bacterial surveys of the Vaal River in central South Africa275

(Jordaan and Bezuidenhout 2016). However, bacterial diversity was strongly structured across space,276

with the largest turnover in diversity occurring among park sections. This is not surprising consider-277

ing the distances from site to site range from 3km to 115km and represent a gradient in large animal278

density, rainfall, vegetation, and major subsurface geology (Chirima et al. 2012; Van Wilgen et al.279

2000; Smit and Grant 2009; Smit et al. 2013). Samples also clustered by site, displaying substantial280

variation in taxonomic composition across sites. This variation was associated with physico-chemical281

properties of the water, with conductivity and pH being important explanatory variables. In addition282

to water quality, variability in taxonomic composition is likely influenced by the origin of the water283

used to fill each waterhole, differences in the surrounding soil and vegetation types, and the particular284

species and populations of animals using the waterholes.285

We assessed daily turnover in composition at Nwaswitshaka, which appeared to be more stable286

over this shorter timescale when compared to turnover across weeks. However, Nwaswitshaka was287

less variable across weeks than other sites, indicating that daily variation in bacterial communities288

could be greater in other locations. Important water quality variables (conductivity and pH) were289

largely consistent across weeks (Table S1), suggesting that the observed temporal heterogeneity may290

be driven by differences in external factors influencing bacterial input and removal from the system,291

12

Page 12 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

such as variation in animal visitation throughout the sampling period. Between sampling events,292

water levels would sometimes drop substantially, indicating major drawdown by animals and likely293

removing bacteria deposited by animals visiting earlier in the week. Large mammal communities294

vary across the sampled regions of the park (Chirima et al. 2012), which may contribute to observed295

spatial variation in bacterial communities. However, different species also differ in their dependence296

on water, which is reflected in their rates of visitation to water points (Redfern et al. 2005). Variation297

in samples taken across subsequent weeks may therefore reflect different components of local animal298

communities, each with their unique host-associated bacterial taxa (Ley et al. 2008). By pairing299

bacterial composition with animal visitation prior to sampling (either through direct observation, or300

presence of genetic material), it may be feasible to build statistical probabilities of associations using301

co-occurrences of microbial and animal signatures.302

Across samples, patterns of phylogenetic clustering were consistent with observed taxonomic vari-303

ation. Multiple phyla were present in all samples, consistent with an even representation of deep304

bacterial lineages. However, turnover at lower taxonomic levels shown by significantly low mean305

nearest taxon distances indicate that there are distinct subsets of closely related taxa present at each306

site. This structuring may reflect filtering of bacterial communities by local environmental conditions,307

or the deposition of microbes by particular animal populations or individuals. Many vertebrate species308

have expansive home ranges, but during the dry season drought-intolerant animals will restrict their309

movement so as to stay close to permanent water bodies (Redfern et al. 2005). Thus the maintenance310

of major bacterial taxa may reflect both free-living environmental bacteria, and the core microbiome311

of water-dependent species. By taking repeat temporal samples, it may be possible to build associ-312

ation networks between bacterial taxa and host species, or even their local populations, solely from313

environmental DNA.314

Through examination of taxonomic assignments we identified taxa belonging to genera that in-315

clude important pathogens (Arcobacter, Bacillus, Burkholderia, Coxiella, Legionella, Neisseria, Pas-316

teurella, Rickettsia, and Yersinia). While many of these genera include both pathogenic as well as be-317

nign species found in environmental samples, some of these genera are comprised solely of pathogenic318

species. For example, the genus Coxiella is represented by one species, Coxiella burnetii, the causative319

13

Page 13 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

agent of Q fever, which has previously been documented as causing disease in the park (Van Heerden320

et al. 1995). Additionally, taxa in the order Chlamydiales are all obligate intracellular pathogens of321

eukaryotes (Ball et al. 2015), and taxa in the genus Neisseria colonize mucosal surfaces of animals,322

some of which are pathogenic in humans (Liu et al. 2015). Interestingly, we also identified sequences323

classified as Streptococcus urinalis, a recently described species linked to urinary tract infections in324

humans (Peltroche-Llacsahuanga et al. 2012). While hypervariable regions of the 16S gene may not325

be the optimal genomic regions for detecting the presence of particular pathogenic species or strains,326

our findings indicate that broad scale surveys of microbial diversity may be useful in determining the327

presence of potential pathogens across vastly divergent groups of bacteria. This can in turn guide more328

targeted sampling of both pathogenic and commensal bacteria across the park.329

In addition to the presence of pathogens, surveys of bacterial diversity may be used to detect330

anthropogenic influences in the park. For example, we found that bacterial diversity was quite different331

for the two sites sampled in the Kingfisherspruit section (Hoyo Hoyo and Imbali). These sites were332

dominated by an ASV assigned to the genus Arcobacter, but which did not exactly match any sequence333

in the SILVA reference database. Three of the five described members of this genus are known to334

be pathogenic (Fera et al. 2004) and include A. butzleri, which can cause severe diarrhea (Lerner335

et al. 1994) and was detected in two of the weekly samples at Hoyo Hoyo. The two waterholes in336

Kingfisherspruit are fed in part with greywater that is passed through reed beds. Greywater is untreated337

household wastewater that typically has not been contaminated by toilet waste and is often used as338

year-round sources of water, especially in water scarce areas (Nganga et al. 2012). Compared to source339

water, kitchen and laundry greywater can have elevated conductivity (Nganga et al. 2012), which may340

explain the high conductivity of water at these sites, and present a strong selective environment driving341

their unique bacterial communities.342

We did not have sufficient sampling of sites to explore all possible drivers of differences in bacte-343

rial communities. Nonetheless, some features differed obviously among sites. For example, Witpens344

is a heavily vulture-dominated site. Bathing by vultures likely results in large influxes of nutrients345

such as blood, and vulture feces has been found to alter soil bacterial communities through elevated346

nitrogen and decreased pH (Ganz et al. 2012). Visually, water from Witpens was bright green, indi-347

14

Page 14 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

cating high abundance of photosynthetic species and consistent with the large variations observed in348

dissolved oxygen (Table S1). However, we found no evidence of elevated abundances of Microcystis349

or other Cyanobacteria, though samples from Witpens strongly clustered together and had relatively350

high abundances of Rhizobiales and Rhodocylales, both of which include species known to fix nitro-351

gen (Carvalho et al. 2010; Loy et al. 2005).352

Witpens, along with Nyamarhi and Ngotso North are filled by pipeline troughs that divert river353

water to waterholes many kilometers away. While pipeline troughs are likely to reflect a subsample of354

the diversity found in river water, the acts of pipeline transport and storage themselves may have strong355

filtering effects on bacterial communities. Our results indicate that waterholes represent locally unique356

bacterial communities, thus the practice of diverting river water to waterholes kilometers away may357

homogenize microbial diversity across the landscape, ultimately disrupting local communities. The358

consequences of such shifts in community structure are difficult to assess without comparing pipeline359

troughs with their source waters, but the diversion of river water may have unintended impacts on360

microbial diversity. For example, genes conferring antimicrobial resistance have been shown to spread361

from river water to impala in the Kruger National Park (Mariano et al. 2009).362

Here we show that eDNA can be amplified from waterholes in the Kruger National Park, and find363

strongly structured microbial communities, likely reflecting local abiotic conditions, animal ecology,364

and anthropogenic disturbance. We suggest that disruption of spatially distinct microbial communi-365

ties may be used as a marker of ecosystem disturbance, or biotic homogenization across the park.366

We find that for artificial waterholes, bacterial diversity is surprisingly insensitive to sample volume,367

with even small volumes useful for capturing major components of bacterial communities, though368

larger volumes are necessary to detect rare taxa. Replicating this study across different seasons, and369

expanding sampling to include natural waterpoints will provide improved understanding of the roles370

micro-organisms play in ecosystem stability and resilience, and offer an effective method for mon-371

itoring of shifting species interactions in the face of environmental change. Just as studies of the372

microbiome have revolutionized our understanding of human health, metagenomic analysis of envi-373

ronmental DNA have the potential to revolutionize our understanding of ecosystem health. Tracking374

of bacterial communities can provide a template for monitoring ecosystem disturbance through their375

15

Page 15 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

response to biological contaminants, documenting the spread of invasive species or infectious organ-376

isms, and better understanding the impacts ecological disturbances have on the composition of native377

communities.378

Acknowledgements379

This work would not have been possible without the logistical support of the African Centre for DNA380

Barcoding and Purvance Shikwambana, molecular work by Shannon Eagle, encouragement and sup-381

port from Lea Berrang-Ford, and the incredible knowledge and protection provided by Kruger game382

guards Thomas, Velly, and Martin. We also thank Steven Kembel, his lab, Alexis Carteron, and383

Benjamin Callahan for constructive feedback on bioinformatic analyses. Special thanks are due to384

B.S. Bezeng for volunteering his friendship, dedication, and long days in the field. This work was385

performed with the support and permission of SANParks (project DAVIJ1256). MJF was supported386

by a Vanier NSERC CGS and the CIHR Systems Biology Training Program, with project funding387

supported by the Quebec Centre for Biodiversity Science, the McGill Biology Department, and an388

NSERC Discovery Grant awarded to TJD.389

References390

Ball, S. G., C. Colleoni, D. Kadouche, M. Ducatez, M. C. Arias, and C. Tirtiaux. 2015. Toward an391

understanding of the function of Chlamydiales in plastid endosymbiosis. Biochimica et Biophysica392

Acta - Bioenergetics 1847(6-7), 495–504.393

Baselga, A., D. Orme, S. Villeger, J. De Bortoli, and F. Leprieur. 2018. betapart: Partitioning Beta394

Diversity into Turnover and Nestedness Components. R package version 1.5.0.395

Bengis, R. G. and J. M. Erasmus. 1988. Wildlife diseases in South Africa: a review. Revue Scientifique396

et Technique, (OIE) 7(4), 807–821.397

Blaxter, M., J. Mann, T. Chapman, F. Thomas, C. Whitton, R. Floyd, et al.. 2005. Defining operational398

16

Page 16 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

taxonomic units using DNA barcode data. Philosophical Transactions of the Royal Society B:399

Biological Sciences 360(1462), 1935–1943.400

Bouskill, N. J., J. Barker-Finkel, T. S. Galloway, R. D. Handy, and T. E. Ford. 2010. Temporal bacterial401

diversity associated with metal-contaminated river sediments. Ecotoxicology 19(2), 317–328.402

Cabral, J. P.. 2010. Water microbiology. Bacterial pathogens and water. International Journal of403

Environmental Research and Public Health 7(10), 3657–3703.404

Callahan, B. J., P. J. McMurdie, and S. P. Holmes. 2017. Exact sequence variants should replace405

operational taxonomic units in marker-gene data analysis. ISME Journal 11(12), 2639–2643.406

Callahan, B. J., P. J. McMurdie, M. J. Rosen, A. W. Han, A. J. A. Johnson, and S. P. Holmes. 2016.407

DADA2: High-resolution sample inference from Illumina amplicon data. Nature Methods 13(7),408

581–583.409

Callahan, B. J., K. Sankaran, J. A. Fukuyama, P. J. McMurdie, and S. P. Holmes. 2017. Bio-410

conductor Workflow for Microbiome Data Analysis: from raw reads to community analyses.411

F1000Research 5(3), 1492.412

Caporaso, J. G., J. Kuczynski, J. Stombaugh, K. Bittinger, F. D. Bushman, E. K. Costello, et al.. 2010.413

QIIME allows analysis of high- throughput community sequencing data Intensity normalization414

improves color calling in SOLiD sequencing. Nature Publishing Group 7(5), 335–336.415

Carvalho, F. M., R. C. Souza, F. G. Barcellos, M. Hungria, and A. T. R. Vasconcelos. 2010. Genomic416

and evolutionary comparisons of diazotrophic and pathogenic bacteria of the order Rhizobiales.417

BMC Microbiology 10.418

Chirima, G. J., N. Owen-Smith, and B. F. Erasmus. 2012. Changing distributions of larger ungulates419

in the Kruger National Park from ecological aerial survey data. Koedoe 54(1), 1–11.420

Delgado-Baquerizo, M., F. T. Maestre, P. B. Reich, T. C. Jeffries, J. J. Gaitan, D. Encinar, et al.. 2016.421

Microbial diversity drives multifunctionality in terrestrial ecosystems. Nature Communications 7,422

1–8.423

17

Page 17 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

Fera, M. T., T. L. Maugeri, C. Gugliandolo, C. Beninati, E. L. Camera, M. Carbone, et al.. 2004.424

Detection of Arcobacter spp . in the Coastal Environment of the Mediterranean Sea Detection of425

Arcobacter spp . in the Coastal Environment of the Mediterranean Sea. Applied and Environmental426

Microbiology 70(3), 1271–1276.427

Fisher, J. C., R. J. Newton, D. K. Dila, and S. L. McLellan. 2015. Urban microbial ecology of a428

freshwater estuary of Lake Michigan. Elementa: Science of the Anthropocene 3, 000064.429

Ganz, H. H., U. Karaoz, W. M. Getz, W. Versfeld, and E. L. Brodie. 2012. Diversity and structure of430

soil bacterial communities associated with vultures in an African savanna. Ecosphere 3(6), art47.431

Hebert, P. D. N., A. Cywinska, S. L. Ball, and J. R. DeWaard. 2003. Biological identifications through432

DNA barcodes. Proceedings. Biological sciences / The Royal Society 270(1512), 313–21.433

Jordaan, K. and C. C. Bezuidenhout. 2013. The impact of physico-chemical water quality parameters434

on bacterial diversity in the Vaal River, South Africa. Water SA 39(3), 385–396.435

Jordaan, K. and C. C. Bezuidenhout. 2016. Bacterial community composition of an urban river in the436

North West Province, South Africa, in relation to physico-chemical water quality. Environmental437

Science and Pollution Research 23(6), 5868–5880.438

Kembel, S., P. Cowan, M. Helmus, W. Cornwell, H. Morlon, D. Ackerly, et al.. 2010. Picante: R tools439

for integrating phylogenies and ecology. Bioinformatics 26, 1463–1464.440

Laforest-Lapointe, I., A. Paquette, C. Messier, and S. W. Kembel. 2017. Leaf bacterial diversity441

mediates plant diversity and ecosystem function relationships. Nature 546(7656), 145–147.442

Lahaye, R., M. van der Bank, D. Bogarin, J. Warner, F. Pupulin, G. Gigot, et al.. 2008. DNA barcoding443

the floras of biodiversity hotspots. Proceedings of the National Academy of Sciences 105(8), 2923–444

2928.445

Lerner, J., V. Brumberger, and V. Preac-Mursic. 1994. Severe diarrhea associated with Arcobacter446

butzleri. European Journal of Clinical Microbiology & Infectious Diseases 13(8), 660–662.447

18

Page 18 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

Ley, R. E., M. Hamady, C. Lozupone, P. J. Turnbaugh, R. R. Ramey, J. S. Bircher, et al.. 2008.448

Evolution of mammals and their gut microbes. Science 320(5883), 1647–1651.449

Liu, G., C. M. Tang, and R. M. Exley. 2015. Non-pathogenic Neisseria: members of an abundant,450

multi-habitat, diverse genus. Microbiology (Reading, England) 161(7), 1297–1312.451

Lodge, D., C. Turner, and C. Jerde. 2012. Conservation in a cup of water: estimating biodiversity and452

population abundance from environmental DNA. Molecular Ecology 21(11), 2555–8.453

Loy, A., C. Schulz, S. Lucker, A. Schopfer-Wendels, K. Stoecker, C. Baranyi, et al.. 2005. 16S rRNA454

gene-based oligonucleotide microarray for environmental monitoring of the betaproteobacterial or-455

der ”Rhodocyclales”. Applied and Environmental Microbiology 71(3), 1373–1386.456

Lozupone, C. A. and R. Knight. 2007. Global patterns in bacterial diversity. Proceedings of the457

National Academy of Sciences of the United States of America 104(27), 11436–11440.458

Mariano, V., C. M. E. McCrindle, B. Cenci-Goga, and J. A. Picard. 2009. Case-control study to deter-459

mine whether river water can spread tetracycline resistance to unexposed impala (aepyceros melam-460

pus) in Kruger National Park (South Africa). Applied and Environmental Microbiology 75(1),461

113–118.462

Mazel, F., T. J. Davies, L. Gallien, J. Renaud, M. Groussin, T. Munkemuller, et al.. 2015. Influence of463

tree shape and evolutionary time-scale on phylogenetic diversity metrics. Ecography.464

McMurdie, P. J. and S. Holmes. 2013. Phyloseq: An R Package for Reproducible Interactive Analysis465

and Graphics of Microbiome Census Data. PLoS ONE 8(4).466

Michel, A. L., L.-M. de Klerk, N. van Pittius, R. M. Warren, and P. D. van Helden. 2007. Bovine467

tuberculosis in African buffaloes: observations regarding Mycobacterium bovis shedding into water468

and exposure to environmental mycobacteria. BMC Veterinary Research 3(1), 23.469

Mwirichia, R., Cousin, S., Muigai, A. W., Boga, H. I., Stackebrandt, E.. 2011. Bacterial diversity in470

the haloalkaline lake elmenteita, Kenya. Current Microbiology 62(1), 209–221.471

19

Page 19 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

Nganga, V. G., F. W. Kariuki, and K. Kotut. 2012. A comparison of the physico-chemical and bac-472

teriological quality of greywater from water deficient households in Homabay town and Githurai473

estates in Kenya. Open Environmental Engineering Journal 5, 110–118.474

Oksanen, J., F. G. Blanchet, M. Friendly, R. Kindt, P. Legendre, D. McGlinn, et al.. 2018. vegan:475

Community Ecology Package. R package version 2.4-6.476

Owen-Smith, N.. 1996. Ecological guidelines for waterpoints. South African Journal of Wildlife477

Research 26(4), 107–112.478

Peltroche-Llacsahuanga, H., B. Frye, and G. Haase. 2012. Isolation of Streptococcus urinalis from a479

human blood culture. Journal of Medical Microbiology 61(5), 740–742.480

Price, M. N., P. S. Dehal, and A. P. Arkin. 2010. FastTree 2 - Approximately maximum-likelihood481

trees for large alignments. PLoS ONE 5(3).482

Quast, C., E. Pruesse, P. Yilmaz, J. Gerken, T. Schweer, P. Yarza, et al.. 2013. The SILVA riboso-483

mal RNA gene database project: Improved data processing and web-based tools. Nucleic Acids484

Research 41(D1), 590–596.485

R Development Core Team. 2008. R: A Language and Environment for Statistical Computing. Vienna,486

Austria: R Foundation for Statistical Computing. ISBN 3-900051-07-0.487

Rappe, M. S. and S. J. Giovannoni. 2003. The Uncultured Microbial Majority. Annual Review of488

Microbiology 57(1), 369–394.489

Ratnasingham, S. and P. D. N. Hebert. 2013. A DNA-Based Registry for All Animal Species: The490

Barcode Index Number (BIN) System. PLoS ONE 8(7).491

Redfern, J., C. Grant, A. Gaylard, and W. Getz. 2005. Surface water availability and the management492

of herbivore distributions in an African savanna ecosystem. Journal of Arid Environments 63(2),493

406–424.494

Rees, H. C., B. C. Maddison, D. J. Middleditch, J. R. M. Patmore, and K. C. Gough. 2014. The495

20

Page 20 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

detection of aquatic animal species using environmental DNA - a review of eDNA as a survey tool496

in ecology. Journal of Applied Ecology.497

Schloss, P. D., S. L. Westcott, T. Ryabin, J. R. Hall, M. Hartmann, E. B. Hollister, et al.. 2009. Intro-498

ducing mothur: Open-source, platform-independent, community-supported software for describing499

and comparing microbial communities. Applied and Environmental Microbiology 75(23), 7537–500

7541.501

Shokralla, S., J. L. Spall, J. F. Gibson, and M. Hajibabaei. 2012. Next-generation sequencing tech-502

nologies for environmental DNA research. Molecular ecology 21(8), 1794–805.503

Smit, I. and C. Grant. 2009. Managing surface-water in a large semi-arid savanna park: Effects on504

grazer distribution patterns. Journal for Nature Conservation 17(2), 61–71.505

Smit, I. P., C. C. Grant, and B. J. Devereux. 2007. Do artificial waterholes influence the way herbivores506

use the landscape? Herbivore distribution patterns around rivers and artificial surface water sources507

in a large African savanna park. Biological Conservation 136(1), 85–99.508

Smit, I. P., E. S. Riddell, C. Cullum, and R. Petersen. 2013. Kruger National Park research supersites:509

Establishing long-term research sites for cross-disciplinary, multiscaled learning. Koedoe 55(1),510

1–7.511

Smith, K. L., V. DeVos, H. Bryden, L. B. Price, M. E. Hugh-Jones, and P. Keim. 2000. Bacillus512

anthracis diversity in Kruger National Park. Journal of Clinical Microbiology 38(10), 3780–3784.513

Sundquist, A., S. Bigdeli, R. Jalili, M. L. Druzin, S. Waller, K. M. Pullen, et al.. 2007. Bacterial514

flora-typing with targeted, chip-based Pyrosequencing. BMC Microbiology 7(1), 108.515

Taberlet, P., E. Coissac, M. Hajibabaei, and L. H. Rieseberg. 2012. Environmental DNA. Molecular516

ecology 21(8), 1789–93.517

Tekere, M., Lotter, A., Olivier, J., Jonker, N., Venter, S.. 2011. Metagenomic analysis of bacte-518

rial diversity of Siloam hot water spring, Limpopo, South Africa. African Journal of Biotechnol-519

ogy 10(78), 18005–18012.520

21

Page 21 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

Tekere, M., Prinsloo, A., Olivier, J., Jonker, N., Venter, S.. 2012. An evaluation of the bacterial521

diversity at Tshipise, Mphephu and Sagole hot water springs, Limpopo Province, South Africa.522

African Journal of Microbiology Research 6(23), 4993–5004.523

Van Heerden, J., M. G. Mills, M. J. Van Vuuren, P. J. Kelly, and M. J. Dreyer. 1995. An investigation524

into the health status and diseases of wild dogs (Lycaon pictus) in the Kruger National Park. Journal525

of the South African Veterinary Association 66(1), 18–27.526

Van Wilgen, B. W., H. C. Biggs, S. P. O’Regan, and N. Mare. 2000. A fire history of the savanna527

ecosystems in the Kruger National Park, South Africa, between 1941 and 1996.528

Van Wyk, L.. 2011. Review of the ecological implications of artificial waterhole closures in the Kruger529

National Park and the effect thereof on tourism. pp. 1–80.530

Venter, F. J., R. J. Naiman, H. C. Biggs, and D. J. Pienaar. 2008. The evolution of conservation531

management philosophy: Science, environmental change and social adjustments in Kruger National532

Park. Ecosystems 11(2), 173–192.533

Zeglin, L. H.. 2015. Stream microbial diversity in response to environmental changes: Review and534

synthesis of existing research. Frontiers in Microbiology 6(MAY), 1–15.535

22

Page 22 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

Figures & Tables536

Section Site Type GeologyTshokwane (TSH) Nhlanguleni (NHL) Borehole GraniteSkukuza (SKZ) Nwaswitshaka (NWA) Borehole GraniteSkukuza (SKZ) De LaPorte (DLP) Borehole GraniteSkukuza (SKZ) Kwaggas Pan (KWA) Borehole GraniteSatara (SAT) Girivana (GIR) Borehole GraniteSaTara (SAT) Witpens (WIT) Pipeline trough BasaltKingfisherpruit (KFI) Imbali (IMB) Borehole GraniteKingfisherpruit (KFI) Hoyo Hoyo (HOY) Borehole GraniteHoutboschrand (HOU) Nyamarhi (NYA) Pipeline trough BasaltHoutboschrand (HOU) Ngosto North (NGO) Pipeline trough Basalt

Table 1: Sample locations

●●●●

-24.5

-25.0

31.25 31.5 31.75 32.0Longitude

Latit

ude

KWA

NWA DLP

NHL

HOY IMB

GIRWIT

NYA

NGO

Figure 1: Map of site locations with park boundary indicated by dashed line. Circles represent sitesfilled by boreholes while triangles represent sites filled by river water via pipeline troughs.

23

Page 23 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

Site Weeks S XS Daily A/B TotalNhlanguleni (NHL) 3 0 0 0 Yes 6Nwaswitshaka (NWA) 3 1 1 4 Yes 18De LaPorte (DLP) 1 1 1 0 Yes 6Kwaggas Pan (KWA) 2 1 1 0 Yes 8Girivana (GIR) 3 0 0 0 Yes 6Witpens (WIT) 3 0 0 0 Yes 6Imbali (IMB) 3 0 0 0 Yes 6Hoyo Hoyo (HOY) 3 1 1 0 Yes 10Nyamarhi (NYA) 3 1 1 0 Yes 10Ngosto North (NGO) 3 1 1 0 Yes 10BLANK 2 0 0 0 No 2

29 6 6 4 88

Table 2: Samples sequences, broken down by number of weekly samples, number of site-times forwhich S (50 mL) and XS (15 mL) samples were filtered, additional daily samples taken, whether A/Bsamples were taken, and the resulting total number of samples sequenced per site.

0 10 20 30 40 50

050

015

0025

00

Number of 150mL samples

Num

ber

of B

acte

rial A

SV

s

Figure 2: ASV accumulation curve of bacterial ASV richness using the “exact” method. Bars representtwo standard deviations around mean estimates.

24

Page 24 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft0.8300.4490.0000.0000.988 0.6460.9060.7370.9040.1290.9230.7930.7550.8770.0000.0000.8510.0000.855

0.850

1.0000.9640.8320.9380.0000.0000.0000.000

0.9250.8990.0000.0000.0000.7890.7680.7900.8470.921

0.9470.9480.8540.0000.8610.7270.888

0.8980.4350.9130.455

0.0000.369

0.851

0.951

0.9030.9700.3840.9510.7750.878 0.9770.4000.8460.946 0.9980.0000.9390.7740.7730.0000.0000.9050.7750.840

0.6370.5300.3120.8420.7820.0000.8650.7440.796

0.0000.0000.9230.0000.7830.8160.8450.000

0.843

0.883

0.8640.8290.9740.7950.6600.9510.7620.7960.9310.8560.0000.9570.0000.8130.9320.8440.9350.0000.9450.7970.7650.8770.8040.7080.4610.9210.964

0.800

0.838

0.897

0.812

0.641

0.8920.8300.0000.8430.9810.8710.000 1.000 0.963 0.8900.861 0.9580.859

0.776

0.757

0.876

0.894

0.967

0.830

0.832

0.579 0.7830.738 0.921 0.930 0.765 0.9960.9760.848 1.000 1.000 0.994 0.9940.983 0.991 0.557

1.000 1.000

0.985

0.928

0.917 0.951 0.942 0.777

0.846

0.916

0.977

0.616

0.901

0.042 0.774 0.9750.944

0.9380.850

0.711 0.8550.562 1.0000.912 1.000 0.8120.908 1.0000.8190.622 1.000 0.9350.777 0.9890.3670.9060.022 0.776 0.9600.876 0.9370.8680.8790.649 0.9980.2080.891

0.961 0.921 0.883 0.946 0.9951.0000.951

0.952 0.767 0.979 0.9860.8710.8070.929 0.8610.7810.815

0.993 0.749 0.9730.863 0.991

0.079 0.4880.926 0.890

0.8730.7040.739 0.885 0.9850.7420.971 0.9991.000

0.758 0.8460.828 1.0000.572 0.897 0.6590.9840.558 0.923 0.997 0.8870.8110.779 0.9980.4840.0000.0000.9970.236

0.783 0.883 0.638 0.9980.892 0.9450.853 1.0000.9360.861

0.977 0.952 0.314 0.962 0.987 0.9930.9461.0000.982 0.9700.5680.7580.8750.893

0.585

0.980

0.859

0.922

0.925

0.612

0.000

0.970 0.999 0.999 0.982 0.963 0.911 0.960 1.0000.768 0.868 0.712 1.0000.517 0.783 0.951 1.0000.934 0.823 0.9990.535

0.912

0.944

0.988

0.957 0.9980.978 0.906

0.936

0.837

0.708 0.930 0.978 0.0000.808 0.9940.992 0.9840.644 0.9090.7290.000

0.759

0.731 0.7350.000 0.881 0.981 0.9940.8420.8120.994 0.436

0.9690.670 0.980 1.000 0.9630.8550.943 0.9150.9160.7740.210 1.0000.982 0.9050.4220.8150.779 0.9950.8390.793 0.9480.7510.000

0.817

0.915

0.9350.636

0.987 0.776

0.7580.853 0.959 0.7020.4170.738 0.9960.996

0.771 0.705 0.943 0.9450.874 0.9840.9540.000 0.9980.3300.7730.8760.7590.000

0.951 0.7990.8610.5330.153 0.908 0.893 1.0000.547 0.917 0.9900.7150.999 0.916 0.9970.4610.8591.0000.898 0.934 0.989

1.000 0.977 0.923 1.0000.858 0.5570.701 0.8970.7790.7930.0000.0000.0000.767

0.897

0.817

0.975

0.9900.6940.3390.8900.2240.944 0.9820.8910.840 0.9850.9800.9940.889 0.9980.7510.7540.903

0.789

0.878

0.6770.4260.852 0.6910.6120.687 0.874 1.0000.938

0.933 0.924 0.9950.0000.931 0.935 0.9990.8610.940

0.004

0.8120.886

0.703 0.512 0.919 0.9930.7540.716 0.9590.793 0.9190.964 0.950 0.914 0.9890.9570.7640.7410.958 0.918

0.573 0.941 0.8880.654 0.969 0.904 0.9690.0000.9980.4380.4770.878 0.970 0.8890.8710.8430.9840.992 0.5900.9990.9980.0000.917

0.782

0.9480.633 0.981

0.4720.776 0.798 0.8920.6620.809 0.943 0.9970.978 0.9990.754

0.710 0.9990.998

0.750 0.968 0.9680.400 0.913 0.2340.9230.9040.9040.905 0.7650.8470.9040.583 0.9740.0000.8520.722

0.994 0.961

0.969 0.2500.779 0.986 1.0000.038 0.9800.8981.000

0.9250.2590.6370.408 0.879 0.967 0.9830.833 0.686 0.997 0.980 0.9700.245 0.9280.865 0.9540.885 0.969 0.9900.865 0.9500.949 0.996 0.954 0.969

0.8410.393 0.940 0.998 1.0000.6400.813 0.962 0.913 0.9810.981 1.0000.8200.8850.7620.028 0.9100.8960.8730.7670.8320.098

0.9770.7630.761 0.9950.0000.9810.954 1.000

0.230 0.998

0.387 0.9660.671 0.951 0.941 0.9951.000 0.7860.575 0.996 0.8540.574 1.0000.869 0.9860.000 0.847

0.9880.482 0.898 0.9950.7620.890 0.338 0.935 0.388

0.563 0.9980.8150.390

0.0000.9960.1860.7660.0000.7800.7690.0000.0000.000

0.5450.7930.868 0.940 0.944 0.8600.8750.207 1.0000.849 0.845 0.9480.1500.983 0.960

0.881 1.000 0.9770.245 0.9930.8610.8840.7580.8450.9040.756 0.997 1.0000.2450.976 0.9420.370 0.9920.985

0.8340.136 1.000 0.999

0.945 1.000

0.999 0.774 0.9380.453 0.4890.167 0.9961.0000.722

0.7090.905

0.974 0.9920.9540.916 0.996 0.7070.9280.8360.9830.9100.852 0.998

0.945 0.958 1.0000.7320.0000.9520.0000.1720.846 0.949 0.699 0.881 0.927 0.9800.8050.9090.8480.4200.667 0.9970.8360.8020.9020.852

0.871 0.9960.426 1.000

0.968 1.0000.5470.9880.9490.283

0.423 0.794 0.999 0.825 0.672 0.9570.880 0.982 0.246 0.9780.811 1.000 0.6710.728 0.9040.880

0.976 0.848

0.935

0.6840.817 0.993 0.914 0.995 0.6850.994

0.769 0.546 0.9960.945 0.998 0.840 0.952 0.9510.7780.7500.984 0.9920.491 0.9960.974

0.8770.968

0.2740.572 0.919 0.999

0.830 0.998 0.916 0.9660.301 1.0000.914 1.0000.5790.826 0.885 0.9990.753 0.955

0.913 0.819 0.706 0.824 1.0000.902 0.9270.213 0.994

1.0000.967 0.858 0.802 0.9990.648 0.454 0.9990.000 0.9320.957 0.969

0.943 0.9890.911 0.982 0.791 0.921 1.0000.9741.0000.688 0.921 0.635

0.9970.602 1.0001.000 0.9290.880

0.978 0.468 0.958 0.963 0.9690.422

0.954 0.998 0.983 0.9970.854 0.579 0.704 0.7040.806 0.479 0.9480.9580.992 1.0000.7750.915

0.565 1.0000.7750.000

1.0000.6830.8540.9220.646 0.956 0.769 0.9710.8480.9580.6190.874 0.9900.941

0.000 1.0000.978 0.879 0.927 0.9820.846 0.971 0.970 0.365 0.988 0.9990.745 1.000 1.0000.827 0.992

0.856 1.000 0.998 0.948 0.9930.845 0.9990.881 0.8860.888 0.850 0.818

0.933

0.924

0.9150.796

0.539 0.929 1.0000.8810.985 0.8500.738 0.878 0.997 0.9070.9750.721 0.9440.8590.744 0.613 0.947

0.9290.970 0.944 0.9950.847 0.998 0.961 0.9420.6490.9750.9440.912

0.395 0.972 0.733 0.956 0.9850.991 0.3780.897 0.9980.821 0.9970.517 0.9440.8820.770 0.6770.467 0.905 0.9440.6920.6860.877 0.9830.9660.954 0.951

0.935

0.968

0.907

0.747

0.369 0.9870.816 0.9070.670 0.613 0.9900.852 0.9740.8430.7540.8900.8240.755 0.561 0.9960.9980.909 0.9910.854 0.9610.956

0.9911.0000.842

0.0000.820

0.6610.765 0.965 0.9020.8620.996 0.959 0.990 0.9990.985 0.911 0.999 0.860 0.9410.862 0.9820.6930.7880.746

0.9930.801 0.8870.656 0.9030.7380.458 0.8830.675 0.9200.936

0.9560.856 0.818 0.9890.8450.8750.811 0.976 0.9240.995

0.513 0.996 0.933 0.9070.8540.663 0.983 1.0000.4990.863 0.9700.743

0.684 0.969 0.9041.000 0.9230.4330.637

0.903 0.950

0.880

0.826

0.8040.995

0.999 0.867 1.000

0.7340.552

0.851 0.957

0.8060.695 1.0000.915 0.9490.4270.8720.781 0.7800.8220.892 0.9920.8650.7490.894 0.9780.0000.7420.864

0.955

0.9720.402 0.8860.900 0.9390.759 0.9790.8700.7480.859

0.3890.956 0.9780.680 0.9990.834

0.8930.9360.7950.000

0.9560.0000.915 0.9720.3320.7480.968 0.9880.9330.0000.7340.8780.8330.9990.8640.9010.7920.8240.8730.8420.3760.9150.9200.945 0.9760.0000.7350.8940.7620.9610.8900.9020.8630.000

1.0000.656 0.9630.8050.990 0.992

0.906 0.8240.8920.8550.9300.000 0.9960.957 0.9090.740 0.9970.7910.399

0.672 0.963 0.999 0.901 0.997 0.808 0.9930.8810.988 0.875 0.509 0.8550.8420.8140.908 0.9960.968

0.943 0.895 0.961 0.084 0.9990.835 0.9990.9100.722 0.000

0.854 0.8660.0390.990 0.8710.853 0.999 0.9980.945

0.9630.898 0.940 0.921 0.7310.000 0.969 0.9550.850 0.990

0.987 0.9970.9400.9820.715 0.9990.950

0.924 0.961 0.9940.7610.7210.834 1.0000.2100.986

0.962 0.955 0.9900.9890.8900.7410.8140.9400.9180.8230.1080.8590.8740.9490.7960.777

0.7400.7540.8540.9800.9600.775

0.8740.897 0.907

0.9710.361

0.7670.432 0.927 0.990 0.9130.8460.9280.9530.0000.8450.973 0.9870.816 0.998

0.988

0.920 0.950 0.994 0.977 0.995 0.822

0.912

0.8190.841 0.759 0.8820.6150.750 0.994 0.8960.821 0.9370.9820.271 0.902 0.8860.2640.882 0.9820.8820.9380.9210.0000.7480.7830.7980.935

0.8330.1560.8840.8770.857

0.9330.993 0.909 0.902 0.9590.7350.9160.9660.8780.643 0.919 0.978

0.9570.919

0.7650.3630.128 0.919 0.9730.9700.735 0.9790.3250.960

0.731 0.9420.273 0.949 0.819 0.976 0.9440.9120.9560.0000.0000.0000.8211.000

0.9930.7640.7690.0000.0000.8300.664 0.9990.9410.9660.0000.7960.5330.512

0.9450.7770.9210.9730.0000.9550.0000.928 0.9740.222 1.0000.9670.931 0.474 0.683 0.953 1.0000.9830.8140.8890.7860.9340.7770.0000.8480.8880.8540.0000.0000.4550.905 0.9830.790 0.8560.687 0.995 0.811 0.893 0.973 0.974 0.9790.745

0.9970.314 0.945 0.9970.928 0.962 0.905 0.947

0.870 0.9660.7940.0000.9090.1330.857

0.786 0.963 0.977 0.9590.804 0.9120.2930.0000.9291.0000.9921.0000.9300.762

0.621 0.993

0.9960.7720.7450.857 0.976 0.991

0.473 1.000 0.9380.425 0.994 0.981 0.9550.9270.938 0.997 0.7680.872 0.483 0.999

0.9750.231 0.893 0.997 0.276 0.9930.579 0.995 0.8730.993 0.483 0.9940.850 0.882 0.9820.960 0.922

0.788

0.8050.929 0.282 0.877

0.953 0.987 0.990 0.9810.770 0.879 0.9860.436 0.9380.8440.815

0.313 0.0890.924 0.9050.9350.9071.0000.9110.936 1.000

0.9820.733

0.278 0.993 0.263 1.0000.669 0.9870.8780.8870.756 0.7370.8050.166 0.949 0.9920.8310.0000.9950.9980.306

0.877 0.963 0.929 0.9970.8060.8790.8430.8700.3520.6890.8570.829

0.9790.552 0.8580.9930.9730.9330.882

0.714 0.963 0.970 0.8720.829 0.993 0.8980.960 0.9700.977 0.9710.7760.9800.9290.8910.9130.7670.856

0.994

0.918

0.822

0.919

0.791

0.928

0.151

0.939

0.762

0.9890.908

0.7100.7850.8800.8230.790 0.9880.9960.9140.955 0.961 0.977 0.9950.7530.883

0.9370.5970.0000.748 0.974 0.9260.000 0.9440.0000.0000.9131.0000.926 0.9930.0530.494

0.7370.839 0.9830.531

0.9940.801 0.984 0.9540.3320.416 0.9660.9800.8350.8190.305 0.9170.0000.8680.929 0.9130.885 0.532

0.8300.7940.9200.9130.0000.8340.7820.8240.534

0.888

0.7670.9960.599 0.956 0.8810.8820.0000.958 0.9410.972 0.9840.0000.7750.8480.924

0.8780.9680.9400.9370.9740.9170.893 0.9820.3490.8340.0990.7760.7900.8230.8550.8450.794

0.9760.7490.7840.797 0.9770.7760.9040.8020.0000.9140.0000.227

0.976 0.703 0.9870.016 0.8670.933 0.997 0.897 0.991 0.7800.926

0.9630.8800.7290.730 0.9690.0000.9430.861 0.8650.8550.5060.000 0.9940.246

0.9490.7290.8840.9210.0000.7160.6630.8730.9570.4560.3260.8890.793

0.9750.7220.7310.9280.7770.8900.881 0.958 0.3070.919 0.9690.887 0.9850.850 0.984

0.8030.7190.8290.7410.922 0.9860.201 0.983 0.9820.9500.8040.7610.9320.869

0.9400.9390.842 0.981 0.961 1.000 0.2760.990

0.8500.7440.7470.9690.9320.664 0.9890.890 0.9170.7170.7710.9460.7570.767

0.9000.842 0.828 0.668 0.855 0.5990.9340.9280.862

0.653 0.745 0.785 0.848 0.9920.442 0.987 0.939 0.9870.0000.8690.8130.8730.825 0.973 0.9640.6820.784 0.852 0.794 0.9840.727 0.9980.725

0.936 0.9400.4100.9030.510 0.9790.7690.9420.8370.808

0.877 0.9510.7120.903 0.954 0.9840.7560.995

0.9110.837 0.890 0.938 0.833 0.9090.8820.716 0.9860.986 0.7850.790 0.996

0.871

0.8680.593 0.855 0.946 0.8950.745 0.9980.9650.444 0.985 0.9980.911 0.9110.9400.6110.8650.7670.796 0.9590.365

0.761 0.9380.771 0.9730.374 0.470 0.816 0.904 0.9990.960 0.7670.1600.656

0.817 1.0000.572 0.610

0.756 0.6500.719 0.9980.973 0.7580.620 0.910 0.8870.717 0.9730.9510.7960.961 1.000

0.8540.864 0.916 1.0000.179 0.4360.412 0.935 0.9600.650 0.991

0.838 0.885 0.918 0.8750.9040.3950.650 0.9970.944 0.969 0.000 0.999 0.9190.7670.161

0.098

0.983 1.000 0.9800.153 0.844 0.8800.715 0.947 0.8550.814 0.9570.998 0.9900.9220.520 0.9020.5000.841 0.602 0.982 0.988 0.9850.0000.8500.9240.9420.802 0.8560.6740.8760.0000.0000.8470.0000.855

1.0000.9840.8460.481

0.671 1.0000.989 0.9260.846 0.984 0.9710.2850.921 0.9090.9930.9190.8330.3780.641

0.5870.730

0.999 0.997 0.9750.838 0.9990.7660.5370.712 0.7680.9750.8470.840 0.935 0.5610.056 0.890 1.0000.834 0.986 0.836 0.9750.926 1.000

0.9320.828 0.9200.061 1.000 0.992 0.9120.567 0.9900.8700.2320.610 0.9750.8290.8280.5440.790 0.998

0.679 0.999 1.000 0.428

0.8820.972 0.897 0.986 0.9710.0000.921 0.9210.879 0.885 0.9340.940 0.922 1.0000.3560.967

0.3710.9640.8470.8600.034

0.5160.975 0.862

0.725

0.930 0.422 0.991 1.000

0.972

0.881 0.9570.281 0.9960.878 0.9650.0000.8690.7310.8450.0000.937

0.391

0.6970.8930.2730.159 0.770 0.995 0.8170.6790.9390.906 0.987 0.870 0.9750.358 0.9980.637

0.002 0.923

0.7330.748

0.9640.730 0.892 0.8350.8190.755

0.9530.9310.7650.9670.9230.8320.3400.8010.9640.9240.789

0.9310.8130.802 0.977 0.9370.0000.8210.4490.994 0.733

0.9870.236 0.9900.883 0.947 0.969

0.795 1.0000.734 0.997 0.8071.0000.874 0.926 0.935 0.814 0.5750.877 0.879 0.990 0.9110.473 0.160

0.8800.578 0.522 1.0000.531 0.993 0.997 0.9370.4570.795

0.833

0.928 1.0000.0000.8600.038

0.832

0.632

0.970 0.9870.8260.899 0.938 0.9950.7730.763 0.827 0.999 0.7440.9140.7040.967 0.9480.7510.972 0.9990.7370.0000.7820.9480.6180.0000.6430.0000.9030.9200.9990.7970.969 0.9690.8170.000

0.894 0.9880.8840.797

0.972

0.955 0.9700.6200.000 0.9850.7800.8200.000 1.0000.9160.865 0.9370.8630.9330.0000.8060.9450.7360.9280.7710.9770.9120.0000.3000.9740.000

0.906

0.9850.9010.2510.9640.9700.955 0.172 0.9250.8840.8060.461 1.0000.0000.534 0.958 0.9750.0000.8560.9350.6290.9480.9450.9440.796 0.9560.7670.9390.6840.9410.8770.8150.9380.8670.511

0.8520.7830.8560.8510.7810.7660.8260.0000.9120.912

0.923 0.958 0.988 1.0000.8170.9630.4460.7400.789

0.741 0.914 1.000 0.9550.8790.8410.7250.6900.5421.0000.9370.7780.9050.9780.0000.9310.8710.0000.8950.0000.3940.594 0.998 1.0001.000

0.8810.8320.827 0.9760.7750.967 0.949 0.9970.7780.7740.0000.8490.8010.000

0.947 0.640

0.987 0.878 0.992

0.846 0.9280.7500.368 0.991 0.9050.9750.5430.8990.8170.8010.7670.737

0.7950.9270.956 0.0970.985 0.9860.788 0.994 0.9720.9080.8460.9570.860 1.0000.792 0.9850.940 0.9270.969 1.0000.281

0.691

1.000

0.0000.947 0.7660.8950.8610.797 0.981 0.907 1.000 0.9980.2950.8290.9670.9080.6450.9140.8330.8650.9330.7650.9310.758

0.7450.7810.7530.8690.8620.9650.8730.0000.9410.8500.9760.6960.6840.8360.876 0.9150.912

0.9740.8700.922

0.8610.8680.657 0.861 0.9970.8880.9050.916 0.992 0.9930.4550.065 0.9980.7410.9820.0000.925

0.735 0.9960.9260.8570.775

0.9040.000

0.9500.9450.6860.624 0.9840.4730.9280.650 0.9300.8460.9150.9560.990 0.942 1.000

0.9540.901 0.941

0.1910.832 0.987 0.9460.5550.311 0.9840.9440.9540.7980.9830.9470.1530.874

0.8930.7640.9600.9090.856

0.7470.9480.9250.9530.8110.552 0.9730.9460.8440.0000.464

0.7610.814 0.9690.9120.874 0.9730.6280.8890.8540.873

0.2740.436

0.916 0.9980.302

0.996 0.996

0.737 0.994 0.9770.8580.0000.9360.8510.880 0.918 0.9400.4940.7970.7970.3650.9940.0450.8130.6230.9990.8510.8460.916 0.971

0.465

0.8200.900

0.753

0.944

0.8570.9700.000 0.9910.9790.5500.8870.783

0.9820.734

0.9960.853

1.000

0.9630.450

0.860

0.859

0.8450.0000.8230.0000.8690.7750.0000.8030.0000.7960.9740.7140.8980.8580.930

0.7980.9010.0000.8430.0001.0000.0000.9310.0000.8950.9520.6770.6930.715

0.7970.9190.8390.0000.7670.8700.909 0.9910.8700.8740.0000.7500.0000.0000.8870.6490.8800.9560.0000.8540.0000.858

0.0000.8710.9560.0000.0000.5150.000

0.8840.0000.9310.9430.0000.6680.8500.0000.000

0.9130.0000.866

0.854 0.9820.0000.9100.5330.4560.7930.9370.939

0.0000.8460.8820.8580.0000.8500.935

0.0000.000

0.000

0.9520.0000.8390.877

0.959

0.775

0.0000.729 0.9170.7700.343 0.909 0.9660.885 0.9880.8020.5970.894 0.9950.7640.977 0.9830.4220.9000.0000.531 0.939 0.937 0.9930.721

0.7700.432 0.9920.921 0.552 0.9530.7450.903 0.873 1.0000.9400.9970.655 0.9620.862 0.987 0.992

0.997 0.9790.909 0.988

0.7600.538

0.978 0.983

0.2660.883

0.879 1.0000.326

0.8610.8980.9210.918 0.9830.9110.792

0.7780.7810.9600.952

0.744

0.983

0.741

0.9010.8200.9330.9100.9770.9130.0000.0000.7360.897

0.886

0.8790.860 0.911 0.999

0.9580.235 0.9940.5090.8370.8900.000 0.9940.740 0.9570.9310.7640.9180.5180.8560.9790.915

0.8940.725 0.9990.0000.7990.883

0.803 0.9540.8760.832 0.975

0.7540.8060.0000.7260.7930.9420.8290.0000.9360.9970.7750.9510.0000.5110.8700.6530.9030.7960.9580.9240.0000.9280.0000.9670.8460.000

0.778

0.750

0.9100.3710.880

0.9730.769

0.8240.8560.9240.0000.8430.1590.0000.0000.8730.9420.0000.7570.7680.7970.8240.0000.7550.9221.0000.8300.0000.9060.924

0.7820.729 0.9430.7110.0000.0000.9770.9640.9130.3180.9540.0000.0000.903

0.7740.9740.8470.8210.0200.9140.7830.9130.934

0.8900.944 0.997

0.8080.8750.6430.733 0.9350.983 0.985

0.7520.8860.848 0.9490.925 0.925 0.9110.840

0.9730.871

0.9380.8500.8980.7620.9620.9140.914 0.9830.7430.0000.9750.9670.9870.000

0.3150.575 0.9640.7610.8600.7730.9320.9410.9800.9360.946 0.9930.1860.8470.0000.8640.7430.893

0.8990.7210.8730.9640.826 0.9880.8960.7460.000 0.9810.9970.9040.851

0.9840.7430.9410.8530.6870.8840.478 0.943 0.931 0.9990.7410.9800.7680.2680.9580.8900.9280.9340.9180.8700.7740.7480.988

0.980 0.973

0.9490.8790.9370.8280.9060.8370.7620.7360.763

0.7390.904 0.9840.4520.9320.7150.9880.8020.8750.0000.4020.8310.214

0.9440.8390.6720.9920.9070.000

0.8900.9500.3540.8670.8370.0000.7830.8920.8480.8390.8630.9140.943

0.9630.0000.8480.0000.9220.8650.0000.8580.9570.0000.8990.7570.9740.3940.9010.9770.7890.789

0.9980.0000.769

0.787

0.0000.8840.7830.8870.9100.0000.7480.723

0.9920.7950.3000.0000.8800.509 0.9940.7450.7280.000

0.9350.9220.7550.873 0.9680.740 0.958 0.9470.8610.8700.8380.9610.7990.0000.0000.9650.8850.0000.7510.9600.7410.8790.3060.7580.5890.995

0.9980.743

0.8930.8800.809 0.9650.755 0.9940.765 0.989 0.8860.267 0.930 0.7640.396 0.999 0.9840.8270.3950.521 0.9590.8730.9670.864 0.928 0.9170.8680.9140.000

0.7610.5110.9780.7490.8840.8870.7600.0000.810

0.9700.7750.8210.2040.9340.000 1.0000.7880.7950.7450.9350.000

0.8250.916

0.976 0.9970.729 0.359 0.9390.9760.9420.7510.8050.8800.7660.7920.7790.339

0.901 0.998

0.9440.855

0.5040.602

0.9660.750 0.965 0.928 0.9440.9410.803 0.960 0.995 0.219 0.9870.924

0.1080.7230.8850.774

0.917 0.9670.0800.7880.9450.7850.9820.9650.8750.8940.7560.7140.857 0.956 0.944 1.0000.8120.871 0.9600.8150.1140.772 0.9870.874 0.991

0.8140.0000.789

0.999

0.9720.1300.9710.788 1.0000.9800.8310.8850.8670.7930.929

0.9790.0000.7560.880 0.998 0.984 0.8750.000 0.9790.8680.7780.765 0.9770.9140.3770.980 0.8550.622 0.9450.8360.7500.9550.0000.7770.7510.9010.738 0.824 0.832

0.9680.9060.9040.7700.9720.9840.8150.000

0.8670.836 0.952 0.9360.7270.9420.2770.5180.8670.747 1.0000.000

0.9420.7500.9580.0000.910

0.0000.8560.9270.7840.8000.8680.802 0.9990.8540.7820.0000.8580.0000.9250.926

0.9830.0000.7160.9120.8320.8540.7620.8460.8860.0000.000

0.8540.0000.8570.0000.000

● ●● ●

● ● ● ● ●

● ●

●

●

●

● ●

●

●

● ● ● ● ●

●

● ● ●

● ●

●

●

●

●

● ● ● ● ● ●

●

●

●

●

●

●

●

● ● ● ●

● ● ●

● ● ●

●

●

●

●

● ●

● ●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

● ●

● ●

●

●

●

● ●

●

●

●

● ● ● ●● ●

●

●

●

●

●

●

●

● ●

● ●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

● ● ● ●

●

●

●

●

●

● ●

●

●

●

●● ●

●

●

●

●

●

●

● ● ● ●

●

● ●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

● ●

● ●

●

●

● ●

● ●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●

● ● ● ●

●

●

●

●

●

●

●

● ●

●

●

●

●●

●

●

●

●

●

●

●

●

●

● ●

●

●

● ●

● ●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●●

●

●

●

● ● ● ●

● ● ●

●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

●●

●

● ●

●

●

●

●

● ● ●

●

● ● ●

● ●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

● ●

●

●

●

● ●

●

●

●

●

● ●

● ●

●

●●

●

●

●

●

●

● ●

● ● ● ● ●

● ●

●

●

●

●

● ● ●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

● ● ● ● ●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

●

● ● ● ● ● ● ●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

● ● ● ●

●

●

●

●●

●

●

●

●

●

● ●

●

●

●

●●

●

● ● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

● ● ●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

● ●

●

● ●

●

●●

●

●

● ●

●

●●

● ●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ● ●● ● ●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

● ● ● ●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●●

●

● ● ● ● ● ●

● ● ● ●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ● ● ●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ● ● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

● ● ●●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ●

●

●

●

●

●

●

●

● ● ● ●

●

●

●

●

●

●

●

●

●

●

● ● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

● ●

●

●

●

●

●

●●

●

●

● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ● ● ● ●

● ● ●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

● ● ● ● ●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●●

●

● ● ● ● ●

● ●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

● ●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ● ● ● ●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ● ●

●

● ● ● ● ● ● ●

● ● ● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●●

● ●

●

●

●

●

●

●

●

●

●●

● ●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ●

●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

● ●

● ● ●

●

●

●

●

●

●

●

●

●

● ● ● ● ●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

● ●

● ● ● ●

●

●

●

●

●

●

●●

●

●

●

●

●

●

● ●

●

●

●

● ● ●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

●●

●

●

●●

● ● ●

●

●

● ●

● ●

●

●

● ●

●

●

●

● ● ●

● ● ●

●

● ● ●

●

●

●

● ● ●

● ●

● ● ●●

● ● ●

●

● ●

●

●

●

● ● ●

●

● ● ●

●

●

●

●

● ●

● ●

●

●

● ●

● ●

● ●

●

●

● ●

●

●

● ● ● ●

● ● ●

● ●

● ●

●

● ●

●

●

● ●

● ●

●

●

●

● ● ● ●

● ●

●

● ●

● ●

● ●

●

●

●

●

● ●

●

● ●

● ●

● ●

●

● ● ●

● ●

● ●

●

●

● ●

●

● ●

● ●

●

●

● ●

●

●

●

●

●

● ●

● ● ● ● ●

● ●

●

●

●

● ●

●

●

● ●

●

●

●

● ●

● ●

●

●

●

●

● ●

●

●

● ● ●

●

● ●

●

● ●

● ●

●

●

●

●

● ●

●

●

●●

●

●

●

● ●

● ● ● ● ● ●

●

●

● ●

●

●

●

● ●

●

● ●

● ● ●

●

●

●

●

●

●●

●

●

●

● ● ● ●

●

●

● ●

●

● ●

●

●

●

●

● ●

● ● ●

● ●

●

●

● ●

●

●

● ●

● ● ● ●

● ●

●

●

● ●

●

●

●

● ●

● ● ●

● ●

●

●

●

●

● ●

● ●

●

●

●

●

●

●

● ●

●

● ●

● ●

●

●

● ●

●

● ●

●

● ●

●

●

●

● ●

●

● ●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

● ●

●

●

●

● ●

●

●

● ●

● ●

●

● ● ● ●

● ●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

● ●

● ●

● ●

● ●

●

● ● ●

● ●

●

●

●

●

● ● ●

●

●

●

●

●

●

●

● ● ● ● ●

● ●

● ●

●

●●

●

●

● ● ● ● ● ●

●

● ●

●

●

● ● ●

● ●

●

●

● ●

●

●

●

● ●

● ● ● ● ● ●

● ●

●

●

●

● ●

●

●

● ● ●

● ● ● ●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

● ●

●

● ●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ●

●

●

●

●

●

● ●

●

●

● ●

●

●

●

●

●

● ● ● ●

●

●

●

●

●

●

● ●

● ●

● ●

●

●

● ●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

● ●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ● ●●

● ● ● ● ●

●

● ●

●

●

●

● ● ● ●

●

●

●

●

●

●

●

● ● ● ● ●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ● ●

●

●

●

●

● ●

● ●

●

● ●

●

●

●

● ●

● ● ● ● ● ● ●

● ● ●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ● ● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

● ● ● ● ● ●

●

●

●

●

●

●

●

● ●

● ●

●

●

●

●

●

●

●

●

● ●

● ●

●

●

●

●

●

●

● ●

●

Abundance●

●●

0.2

0.4

0.6

Site●

●

●

●

●

●

●

●

●

●

NHL

NWA

DLP

KWA

GIR

WIT

IMB

HOY

NYA

NGO

SectionTSH SKZ SAT KFI HOU

Figure 3: Phylogenetic tree of 16S ASV sequences in the core samples, paired with their relativeabundances at each site. Sites are ordered by section.

Diversity Level Shannon SES 2.5% 97.5% Pr(sim.)α A/B samples 2.93 -9335.9 5.10 5.10 0.01α Temporal samples 2.99 -12535.9 5.13 5.13 0.01α Sites 3.39 -16873.6 5.15 5.15 0.01α Sections 3.94 -14919.9 5.16 5.16 0.01γ (Total) 5.17 0.0 5.17 5.17 1.00β A/B samples 0.05 153.3 0.027 0.027 0.01β Temporal samples 0.40 2688.2 0.022 0.022 0.01β Sites 0.55 6528.7 0.008 0.008 0.01β Sections 1.23 14919.9 0.008 0.008 0.01

Table 3: Additive partitioning of Shannon diversity into α, β, and γ diversities across sections, sites,temporal samples, and within site-time samples (A/B) as components of the total diversity observedacross all core samples. Observed diversity is compared to 99 simulations and the standardized effectsize (SES) using “r2dtable” null model with the adipart function in the R package vegan.

25

Page 25 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft

●

●

●

●

●

●

●

● ●

●●

●

●●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

● ●

●●●

●●

●

●

●

●

●●

●

●

●

●

●

●●

●

●

●

●

●●

●

●

●

●

●

●

●

●●

● ●

●

●

●

●●●

●

●●●

●●

●

●

●●

●

●●

●

●●

●●●

●

●● ●

●

●

●

●●

● ●

●

●

●●●●

●

●●●

●●

●●

●●●

●●

●

●●

●● ●

●

●●

Observed Shannon

NH

L

NW

A

DLP

KW

A

GIR

WIT

IMB

HO

Y

NYA

NG

O

NH

L

NW

A

DLP

KW

A

GIR

WIT

IMB

HO

Y

NYA

NG

O

1

2

3

4

50

100

150

200

250

site

Alp

ha D

iver

sity

Mea

sure

type●● Borehole

Pipeline trough

section●●

●●

●●

●●

●●

HOU

KFI

SAT

SKZ

TSH

Figure 4: Plots of observed ASV richness and Shannon diversity across samples. Samples are groupedand coloured by park section and with shape indicating waterhole type.

Phylum Class Order Family Genus

Cla

ssifi

ed s

eque

nces

up

to ta

xono

mic

leve

l (%

)

020

4060

8010

0

Figure 5: The proportion of ASVs assigned a given taxonomic level using the SILVA database v128and a pre-trained RDP classifier with minimum 50% bootstrap support.

26

Page 26 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft0%

25%

50%

75%

100%

HO

Y

IMB

GIR

WIT

DLP

NW

A

NH

L

KW

A

NYA

NG

O

Site

Rel

ativ

e ab

unda

nce

(%)

Phylum

Actinobacteria

Bacteroidetes

Chlamydiae

Cyanobacteria

Firmicutes

Other

Proteobacteria

Verrucomicrobia

Figure 6: Relative abundances of bacterial phyla across sites. Sites are ordered by relative abundanceof phylum Proteobacteria.

NHL NWA KWA GIR WIT IMB HOY NYA NGO

0%

25%

50%

75%

100%

Rel

ative

abu

ndan

ce (%

) PhylumActinobacteria

Bacteroidetes

Chlamydiae

Firmicutes

Other

Proteobacteria

Verrucomicrobia

Week1 2 3 1 2 3 1 2 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3

Figure 7: Relative abundances of bacterial phyla across weekly samples.

27

Page 27 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft0%

25%

50%

75%

100%

3A −

Jun

e 29

3B −

Jun

e 29

4A −

Jun

e 30

4B −

Jun

e 30

5A −

Jul

y 1

5B −

Jul

y 1

6A −

Jul

y 2

6B −

Jul

y 2

7A −

Jul

y 3

7B −

Jul

y 3

Rel

ativ

e ab

unda

nce

(%) Phylum

Actinobacteria

Bacteroidetes

Chlamydiae

Firmicutes

Other

Proteobacteria

Verrucomicrobia

Figure 8: Relative abundances of bacterial phyla across five days at a single site (NWA).

●

●

●●

●

●

● ●

●●

●

●

●

●

● ●

●

● ●

●●

●●

●●

●

●

●

● ●●●●

●●

●●●

●●

●

●●

●●

●●

●●

●●

●

●●

●

●● ●

●

●

● ●●●

●

●

●●

● ●●

● ●

●●

●●

●●

●

●●● ●●●●

●●

●●●●●

●●●●●

●●

●●

●●●●●

−2

−1

0

1

2

−2 0 2NMDS1

NM

DS2

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●●

●

●

●

● ●

●●

●●●

●●●●● ●●

●

●

●●●

●●

●

●

●

●●

●●●●

●●●

●●●●

●

●

●●

●

●

●

●●●

●

●

●●

●

●

●

●●●●●●●

●●●●● ●●

●●

●●●

●●

●●

●

●●

●●●●●●●●●

−0.02

−0.01

0.00

0.01

0.02

−0.02 0.00 0.02NMDS1

NM

DS2

site●●●●●●●●●●●●●●●●●●●●

DLP

GIR

HOY

IMB

KWA

NGO

NHL

NWA

NYA

WIT

a) b)

Figure 9: NMDS plots of a) Bray-Curtis and b) abundance-weighted UniFrac distances. Coloursrepresent site.

28

Page 28 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome

Draft−3 −2 −1 0 1 2

−2−1

01

2

NMDS1

NM

DS2

Actinobacteria

AlphaproteobacteriaBacteroidia

Betaproteobacteria

Clostridia

Flavobacteriia

Gammaproteobacteria

Sphingobacteriia

Conductivity

pH

(a) Bray-Curtis NMDS

−0.03 −0.01 0.01 0.02 0.03

−0.0

20.

000.

010.

020.

03

NMDS1

NM

DS2

ActinobacteriaAlphaproteobacteria

Bacteroidia

Clostridia

Gammaproteobacteria

Sphingobacteriia

Conductivity

Dissolved Oxygen

pH

(b) Weighted UniFrac NMDS

Figure 10: Nonmetric multidimensional scaling (NMDS) ordination of variation in bacterial com-munity structure across 54 samples based on a) Bray-Curtis and b) abundance-weighted UniFracdistances. Arrows indicate the direction of significant (p < 0.05) correlations among variables andthe NMDS axes, with arrow length indicating the strength of the correlation. Blue arrows indicateenvironmental variables, while black arrows indicate relative abundances of sequences from differ-ent microbial classes. The ordination axes explain 96.8% (a) and 98.1% (b) of the variance in thedissimilarities (Fig. S17).

29

Page 29 of 29

https://mc06.manuscriptcentral.com/genome-pubs

Genome