The meta-gut: Hippo inputs lead to community coalescence of animal and environmental microbiomes

All animals carry specialized microbiomes, and their gut microbiotas in particular are continuously released into the environment through excretion of waste. Here we propose the meta-gut as a novel conceptual framework that addresses the ability of the gut microbiome released from an animal to function outside the host and potentially alter ecosystem processes mediated by microbes. An example considered here is the hippopotamus (hippo) and the pools they inhabit. Hippo pool biogeochemistry and fecal and pool water microbial communities were examined through field sampling and an experiment. Sequencing using 16S RNA methods revealed that the active microbial communities in hippo pools that received high inputs of hippo feces are more similar to the hippo gut microbiome than other nearby aquatic environments. The overlap between the microbiomes of the hippo gut and the waters into which they excrete therefore constitutes a meta-gut system with potentially strong influence on the biogeochemistry of pools and downstream waters. We propose that the meta-gut may be present where other species congregate in high densities, particularly in aquatic environments. Significance Animals can have considerable impacts on biogeochemical cycles and ecosystem attributes through the consumption of resources and physical modifications of the environment. Likewise, microbial communities are well known to regulate biogeochemical cycles. This study links those two observations by showing that the gut microbiome in waste excreted by hippos can persist ex-situ in the environment and potentially alter biogeochemical cycles. This “meta-gut” system may be present in other ecosystems where animals congregate, and may have been more widespread in the past before many large animal populations were reduced in range and abundance.


60
Animals alter the functioning of ecosystems by consuming plant and animal matter, through the 61 transport and excretion of nutrients, and in myriad other ways [1][2][3][4][5][6]. In many cases, the activities 62 of animals directly or indirectly influence microorganisms, which in turn regulate the major 63 biogeochemical cycles, although such linkages remain to be fully investigated [7][8][9]. Abundant 64 evidence exists for altered decomposition and nutrient cycling in response to organic matter and 65 nutrient excretion and egestion by animals, but few studies have explicitly examined the role of 66 the externally released animal gut microbiome in mediating those changes [5,6,[10][11][12][13], but see 67 [14]. 68 Community coalescence theory is an ecological framework for investigating the mixing of entire 69 microbial communities and their surrounding environments [15][16][17], but much of the focus has 70 been on metacommunity dynamics among the microbiota, with less attention on the ecosystem 71 implications of resource flows that accompany this mixing [18]. Animal excretion and egestion 72 present a unique case of community coalescence by effectively mixing the animal gut 73 microbiome with preexisting microbial communities, together with organic matter, nutrients, and 74 metabolic byproducts that may also be excreted or egested. Heavy rates of such loading can 75 shape the external environment in ways that support the persistence of gut microbiota outside the 76 host gut, particularly in aquatic ecosystems, increasing the likelihood that ex situ gut microbes 77 could influence ecosystem processes and be re-ingested by other consumers. We propose that the 78 resulting meta-gut system is a dynamic interplay of abiotic resources and microbial communities 79 between the host gut, the external environment, and possibly the guts of other individual hosts 80 that inhabit the same environment.

81
Hippos (Hippopotamus amphibius) have profound effects on aquatic ecosystems in which they 82 wallow during the day by adding large amounts of organic matter and nutrients from their 83 nighttime terrestrial grazing via defecation and urination [19][20][21][22][23], and their fecal inputs are 84 accompanied by abundant enteric microbes (Fig. 1A). These resource subsidies interact with 85 environmental characteristics of the recipient ecosystem to alter ecosystem function [6,23,24]. 86 The organic matter excreted by hippos accumulates at the bottom of hippo pools under low to 87 moderate discharge, and in pools with high hippo densities the decomposition of this organic 88 matter often depletes dissolved oxygen in the water column [23]. As hippo pools become anoxic, 89 the pool environment becomes more similar to the hippo gut, increasing the likelihood that some 90 of the enteric microbes will survive and even function outside the host gut. Anoxia is a persistent 91 state in the bottom waters of many hippo pools until high flow events flush organic matter 92 downstream and reaerate the water column [23,25], briefly exposing the microbial community 93 to oxic conditions before organic matter loading by hippos once again drives pools towards 94 anoxia. The conditions produced by hippos provide a unique opportunity to investigate the 95 importance of the meta-gut system.

97
The Mara River flows through the Serengeti Mara Ecosystem in Kenya and Tanzania (Fig. S1). 98 There are over 4,000 hippos distributed among approximately 170 hippo pools in the Kenyan 99 portion of the Mara River and its seasonal tributaries [26]. Water is present in hippo pools year-100 round even in the seasonal tributaries, some of which may exhibit very low or no flow during dry 101 periods. Hippo number and water residence time of hippo pools (pool volume / discharge) 102 interact to shape in-pool and downstream biogeochemistry in the Mara River system [23]. We 103 classified pools by the magnitude of subsidy inputs: high-, medium-, and low-subsidy pools [38].

104
Low-subsidy pools remain oxic, while high-subsidy pools are typically anoxic, except during 105 periodic flushing flows.

106
Here we examine this meta-gut phenomenon in hippo pools by sequencing the microbial 107 communities of hippo guts and hippo pools in the field across a range of environmental 108 conditions, and we consider the biogeochemical implications of the meta-gut system. We also 109 conducted a microcosm experiment to investigate the role of both microbiomes and viruses 110 (specifically bacteriophages) from the hippo gut in driving biogeochemical processes and 111 microbial community changes within hippo pools.

113
Community changes over space 114 We characterized the microbial communities in the hippo gut by collecting fresh hippo feces 115 from multiple individuals across multiple pools on the landscape. There was little variability in 116 the structure of the hippo gut microbiome among the samples from 10 individuals, and the 117 microbiomes of all samples were dominated by Firmicutes, Bacteroidetes, Proteobacteria and 118 Tenericutes (Fig. 1A). The gut microbiomes of hippos sharing a high-subsidy pool were not 119 distinct relative to those of hippos from other high-subsidy pools and had similar dispersions 120 (PERMANOVA: F = 068524, P = 0.886, PERMDISP: F = 7.3918, P = 0.013, Fig. S3). 121 We characterized the microbial communities in the water column of hippo pools across a 122 gradient of hippo subsidy. Aquatic microbial communities in the bottom waters of high-, 123 medium-, and low-subsidy hippo pools were distinctly different from one another 124 (PERMANOVA: F = 3.3146, P = 0.002). They also had different dispersions (PERMDISP: F = 125 0.3532, P = 0.72); however, their differences in diversity were supported through clustering 126 within a NMDS ordination (NMDS Stress = 0.1, Fig. 2A). The microbial communities in high-127 subsidy pools were more similar to those of the hippo gut microbiome than to the aquatic 128 microbial community sampled from an area outside the influence of hippos (tributary, Fig. 2A). 129 We observed large differences between the three types of hippo pools when constrained by 130 microbial communities and biogeochemistry (PERMANOVA: F = 1.3641, P = 0.011). High-, 131 medium-, and low-subsidy hippo pools were strongly separated by the CCA ordination, which 132 accounted for approximately 90% of the variability in the active microbial community structure 133 within the first two axes of the constrained ordination (Fig. 2B). We observed strong relationships between the microbial community and biogeochemical constituents affected by 135 microbial metabolism. Concentrations of dissolved methane, sulfate, and BOD were correlated 136 and had effects (influence) that were opposite from nitrate, all of which was strongly related to 137 axis 1 (Fig. 2B). Soluble reactive phosphorus (SRP) was strongly related to axis 2. Low-subsidy 138 hippo pools were strongly associated with higher concentrations of nitrate. Medium-subsidy 139 hippo pools were strongly associated with higher concentrations of sulfate, methane and BOD, 140 but low levels of SRP. High-subsidy hippo pools were strongly associated with higher 141 concentrations of methane, BOD, sulfate and SRP. Firmicutes were more closely associated with 142 high-subsidy pools and higher concentrations of methane, BOD, and SRP. 143 We characterized the microbial communities in the water column upstream of and within hippo 144 pools, and along a gradient of hippo density in the Mara River and its tributary, the Talek River.   Table S1). Notably, 161 Clostridia, obligate anaerobes, were responsible for approximately 19% of the overlap between 162 the active microbial communities in the bottom of the hippo pools and the hippo gut microbiome.

163
Macellibacteroides, an obligate anaerobe, was responsible for approximately 16% of the overlap, 164 and was previously described from anaerobic wastes from an abattoir in Tunisia [27]. 165 Prevotellaceae, commonly found in the intestines of animals and which helps break down 166 proteins and carbohydrates, was responsible for 11% of the overlap [28].

167
Community changes over time 168 We characterized the microbial communities in the water column of hippo pools during  Influence of hippo gut microbiome on biogeochemistry in a mesocosm experiment 184 We conducted a mesocosm experiment to understand the changes in microbial communities that 185 occur as a hippo pool goes anoxic, and to test the role of microbial taxa associated with the hippo 186 gut in driving biogeochemical changes within the hippo pools. An additional goal was to test the 187 impact of fecal bacteriophages from the hippo gut on the microbial communities (which are 188 composed almost entirely of bacteria) and the biogeochemical processes they mediate in hippo 189 pools. We used a destructive sampling design with control (sterilized), bacteria, and 190 bacteria+virus treatments (Fig. S2). 191 We found distinct differences in the composition of the active aquatic microbial communities and declined thereafter. The total taxa (including active and inactive microorganisms) included 195 more taxa derived from upstream than we detected in just the active taxa (Figs. 5A-C and S5). 196 We found a statistically significant effect of treatment on biogeochemical variables including 197 Fe(II), pH, H2S, BOD, CH4, and SO4 2- (Table 1). The bacteria treatment had higher

208
Hippos excrete microbes into the water of hippo pools, and a portion of the microbial taxa 209 continue to function and may appreciably alter environmental biogeochemical processes ( Fig. 1)

210
[23]. The collective hippo-environmental microbiome, the meta-gut, is reinforced through the 211 constant loading of hippo feces into hippo pools, which depletes dissolved oxygen and increases 212 the similarity between the host gut and the aquatic environment.

213
Biogeochemistry within pools was driven by interactions between hippo loading and 214 environmental characteristics of the pool. Microbial communities in high-subsidy hippo pools 215 were strongly associated with higher levels of soluble reactive phosphorus (Fig. 2B), which is 216 likely due to inputs of P in hippo urine and to the release of SRP from sediments under anoxic 217 conditions. The oxic waters of low-subsidy hippo pools had higher concentrations of nitrate than 218 found in the main river and there was higher sulfate found in the medium-subsidy pools in the 219 tributaries [23]. The tributaries are likely draining catchments that have more geological sources 220 of sulfur, while the higher nitrate concentrations in the low-subsidy hippo pools in the river may 221 be due in part to oxic conditions that result in less denitrification. These biogeochemical 222 differences likely interact with the rate of loading by hippos to alter microbial communities 223 within the pools, which in turn may further alter the biogeochemistry. 224 We hypothesized that hippos within a pool would have more similar gut microbiomes than 225 hippos across pools. Data from this study did not support that hypothesis-hippos from the same 226 high-subsidy pool did not have more similar microbiomes to one another than to hippos from 227 other high-subsidy pools (Fig. S2). However, our sample size was small (N=10, 3 feces samples  previous meta-analysis of fish microbiomes found an unexpected similarity in fermentative 289 bacteria between fish gut communities and vertebrate gut communities, highlighting the potential 290 for this effect [39]. Aquatic insects also have been known either to ingest gut microbiota from 291 the environment that can then aid directly in digestion or to rely on external microbial 292 decomposition to process recalcitrant organic matter prior to ingestion [40].

293
Hippos provide an ideal case study for examining the meta-gut system in wild populations, as 294 they congregate in high densities in pools that quickly become anoxic due to a subsidy overload

307
Altogether our research demonstrates that community coalescence between the hippo gut 308 microbiome and the river microbiome can occur, forming a meta-gut system in which the hippo 309 gut microbiota can continue to function ex situ in certain environmental contexts. Thus, the role 310 of hippos in subsidizing tropical rivers extends beyond the loading of nutrients and organic 311 matter to also include the successful transference of microbial communities that may influence 312 biogeochemical cycling. In hippo pools, two factors are required for the meta-gut system. First, 313 there is a continual loading of waste products (organic matter and nutrients) and microbial 314 communities. Without continual loading, the resulting microbial community may coalesce into 315 an anaerobic generalist community that does not represent the functioning gut microbial 316 communities from the donor host. Second, loading occurs in an environmental patch that has 317 similar environmental characteristics to the gut, which in this case develops in response to  below). We also collected water samples in four of the high-subsidy hippo pools every 2-3 days 338 starting immediately after a flushing event until the next flushing event (Fig. S1) [23]. 339 We sampled the aquatic microbial community and biogeochemical variables along a longitudinal 340 transect down both the Mara and Talek rivers (Fig. S1, Table S1). For the Mara River, we  We prepared 15 bottles for each of three treatments-control, bacteria, and bacteria+virus-as 366 follows: Control -Unfiltered river water, 5 g wet weight sterilized hippo feces, and two blank 367 Supor filters; Bacteria -Unfiltered river water, 5 g wet weight sterilized hippo feces, two Supor 368 filters containing bacteria, and 4 ml sterilized filtrate; Virus -Unfiltered river water, 5 g wet 369 weight sterilized hippo feces, two Supor filters containing bacteria, and 4 ml unsterilized filtrate 370 containing viruses.

371
We ran the experiment for 27 days from September to October 2017. Initial microbial samples of 372 the river water, hippo feces bacteria and hippo fecal liquid filtrate were taken on day 0, and three 373 replicate samples per treatment were destructively sampled on day 3, 9, 15, 21, and 27. During 374 each time step, the bacterial communities were sampled using the methods detailed above, and 375 chemical analyses were done on the water samples as described below. We also measured 376 chlorophyll a, dissolved oxygen, temperature, conductivity, total dissolved solids, turbidity, and 377 pH with a Manta2 water quality sonde (Eureka Water Probes, Austin, TX, USA).

379
All water samples collected in the field and in the experiment were analyzed for dissolved 380 ferrous iron (Fe(II)), hydrogen sulfide (H2S), dissolved organic carbon (DOC), inorganic 381 nutrients, major ions, dissolved gases, and biochemical oxygen demand following the standard 382 methods provided in detail in Dutton et al. (2020) and briefly summarized in Supplementary 383 Information 1.  We used the Bray-Curtis dissimilarity matrix followed by ordination with NMDS to examine 392 differences between individual hippo gut microbiomes; between low-, medium-, and high-393 subsidy hippo pools; and between a gradient of hippo pools and the environment. We used a 394 CCA to test for the influence of biogeochemical drivers. We used PERMANOVA and 395 PERMDISP to test for significant differences between groups [58]. See Supplementary   396 Information 1 for more details on analyses. 397 We compared aquatic microbial communities from the bottom of high-subsidy hippo pools, from 398 hippo feces, and upstream of high-subsidy hippo pools (free of hippo gut microbiome influence) 399 using the Bray-Curtis dissimilarity matrix followed by ordination with NMDS, and we quantified 400 the proportion of taxa shared between the hippo gut microbiome and the bottom of the high-401 subsidy hippo pool but not present in the upstream samples. 402 We used SourceTracker to quantify the contribution of the hippo gut, upstream waters, or