Patterns of Microbial Diversity and Community Composition in Slot Canyons, Rock Pools, and Other Ephemeral and Perennial Aquatic Habitats

Utah State University DigitalCommons@USU

All Graduate Theses and Dissertations Graduate Studies

5-2020

Marley Madsen Utah State University

Follow this and additional works at: https://digitalcommons.usu.edu/etd

Part of the Ecology and Evolutionary Biology Commons

Recommended Citation Madsen, Marley, "Patterns of Microbial Diversity and Community Composition in Slot Canyons, Rock Pools, and Other Ephemeral and Perennial Aquatic Habitats" (2020). All Graduate Theses and Dissertations. 7750. https://digitalcommons.usu.edu/etd/7750

This Thesis is brought to you for free and open access by the Graduate Studies at DigitalCommons@USU. It has been accepted for inclusion in All Graduate Theses and Dissertations by an authorized administrator of DigitalCommons@USU. For more information, please contact [email protected]. PATTERNS OF MICROBIAL DIVERSITY AND COMMUNITY COMPOSITION IN SLOT CANYONS, ROCK POOLS, AND OTHER EPHEMERAL AND PERENNIAL AQUATIC HABITATS

Marley Madsen

A thesis submitted in partial fulﬁllment of the requirements for the degree

MASTER OF SCIENCE

Ecology

Approved:

Bonnie Waring, Ph.D. Paul Wolf, Ph.D. Major Professor Committee Member

Will Pearse, Ph.D. Richard S. Inouye, Ph.D. Committee Member Vice Provost for Graduate Studies

UTAH STATE UNIVERSITY Logan, Utah

2020 ii

Abstract

Patterns of Microbial Diversity and Community Composition in Slot Canyons,

Rock Pools, and Other Ephemeral and Perennial Aquatic Habitats

Marley Madsen, Master of Science

Utah State University, 2020

Major Professor: Dr. Bonnie Waring Department: Biology

In general, species sorting is the dominant community assembly mechanism for microbes, and this pattern holds true over a diverse range of habitats and spatial scales. However, the relative importance of species sorting and other community assembly mechanisms, such as dispersal limitation, varies with the habitat being studied. To attempt to understand the conditions that determine dominant community assembly processes, I investigated the diversity and community composition of bacteria in slot canyons and rock pools of the Colorado Plateau, Utah. Though similar to one another, each habitat is actually subject to vastly different environmental conditions and dispersal regimes. The results of my study suggest that species sorting plays the predominant role in structuring bacterial communities in both habitats. However, dispersal limitation appears to be more important to slot canyons than to rock pools. I also found that perennial pools within slot canyons have significantly higher alpha diversity than ephemeral open rock pools. In my second study, I used meta-analysis to determine the effect of drying disturbance on microbial diversity in other perennial and ephemeral lentic environments. The results of the meta-analysis support my previous results and indicate that, globally, perennial systems tend to have higher microbial alpha diversity than ephemeral systems.

(45 pages) iv

Public Abstract

Patterns of Microbial Diversity and Community Composition in Slot Canyons, Rock Pools, and Other Ephemeral and Perennial Aquatic Habitats

Marley Madsen

Microbes are the most diverse life forms on the planet and perform many impor-tant ecological functions. However, despite the abundance, diversity, and ecological importance of microbes they are often overlooked and understudied in many natural systems, including freshwater habitats. This thesis details the ﬁrst ever investigation of the microbial diversity and community composition within freshwater rock pools and slot canyons of the Colorado Plateau, Utah. The purpose of the study was to determine the relative importance of various microbial community assembly processes. This thesis also includes a meta-analysis of the microbial alpha diversity in other perennial and ephemeral aquatic systems around the globe. The purpose of the meta-analysis was to identify the relationship between microbial alpha diversity and disturbance from drying. Together, these studies complement one another by describing the microbial ecology of a very speciﬁc habitat type, rock pools, as well as a diverse group of globally distributed aquatic habitats. (45 pages) v

Acknowledgments

First, I want to express gratitude to God. It’s amazing how much more productive

I am after a short prayer. Thank you for guiding me to the right program and helping me ﬁnish.

Jarom, thank you for being my ﬁeld crew, my coding tutor, my therapist, my proofreader, my “carrier of things”, the best ham sandwich maker, and my absolute favorite human. Thank you also for all your help with Theo. I know there were a lot of sleepless nights, early mornings, and days you were late to work so I could have time to

ﬁnish my degree. I’m one lucky lady to have you for my husband!

Bonnie, Paul, and Will, thank you for being such fantastic advisors. Graduate school was a bit of a shock to the system at ﬁrst. Thanks for sticking with me.

Well...mostly. Being on my committee has the side eﬀect of people getting new jobs thousands of kilometers away from where I live.

Bonnie and Will, thanks for having a baby at the same time as me. First of all, she’s super cute. Second, it has been tremendously comforting to know that you really

“get it” when it comes to all the hurdles of getting things done as a new parent.

And Dad, thank you for teaching me to love being outside. It was all the adven- turing we have done together that helped me ﬁgure out what I wanted to study. I had a blast turning a regular recreational activity for us into a full blown research project.

Marley Madsen vi

Contents

Page

Abstract...... iii

Public Abstract...... iv

Acknowledgments...... v

List of Figures...... vii

List of Tables...... x

1 Introduction...... 1

2 Species Sorting Determines Bacterial Community Composition of Rock Pools and Slot Canyons of the Colorado Plateau, Utah...... 5 2.1 Introduction...... 5 2.2 Methods...... 7 2.2.1 Experimental Design and Sampling...... 7 2.2.2 DNA Extraction and 16s rRNA Gene Sequencing...... 8 2.2.3 Analysis of 16s rRNA Gene Sequences...... 8 2.2.4 Statistical Analysis...... 10 2.3 Results...... 11 2.4 Discussion...... 11 2.4.1 Hypothesis 1...... 11 2.4.2 Hypotheses 2 and 3...... 16 2.4.3 Hypothesis 4...... 18 2.5 Conclusion...... 18

3 A Meta-Analysis of Microbial Alpha Diversity in Perennial and Ephemeral Pool Habitats...... 19 3.1 Introduction...... 19 3.2 Methods...... 20 3.2.1 Overall Experimental Design...... 20 3.2.2 Data Selection...... 20 3.2.3 Statistical Analysis...... 21 3.2.4 Results...... 21 3.3 Discussion...... 24 3.4 Conclusion...... 26

Bibliography...... 27 vii

List of Figures

Figure Page

1.1 Example of a slot canyon rock pool...... 2 1.2 Examples of open rock pools...... 2 1.3 Looking into a slot canyon from above. The narrow slit in the center of the photo is the opening to the canyon which is approx. 20m deep. The narrowness of the opening greatly limits the amount of direct sunlight that reaches the rock pools in the canyon bottom. In this particular canyon the average water temperature was 15.9◦C...... 3 1.4 A large log jam brought into a slot canyon by a ﬂash ﬂood. Image courtesy of Canyon Collective (Collective 2018)...... 3

2.1 Sample Areas - The numbers correspond to (1) Eardley Canyon, (2) Entrajo Canyon, and (3) Fry Canyon. The hatched area is the boundary of the Colorado Plateau. The two photos were taken at the Fry Canyon Area. The top photo is an example of a SCRP. The bottom photo is an example of an ORP. Lake Powell, Utah is visible in the background....9 2.2 NMDS plot of beta diversity - PerMANOVA indicated that bacterial community composition differed significantly by habitat type (F1,90=24.34, 2 2 p<0.001, r =0.179) and sample area (F1,90=6.54, p<0.001, r =0.096), with habitat type explaining more of the variation in community composition than sample area. This can be seen in the NMDS plot (stress=0.076) which shows samples clustering first, by habitat type, and second, by sample area within each habitat cluster. Additionally, ORPs appear to have greater spatial separation than SCRPs. The perMANOVA analyses of environmental parameters showed that bacterial community composition 2 also varied significantly by pH (F1,90=6.67, p<0.001, r =0.049) and max- 2 imum pool volume (F1,90=3.78, p<0.001, r =0.028)...... 12 viii

2.3 Differential Abundance of Phyla - This bar chart shows the top 9 most abundant phyla with all other phyla (60 total) grouped together under “Other”. ANCOM identified 19 bacterial phyla and 2 archaeal phyla that were differentially abundant by habitat type. Of note is that the phyla Cyanobacteria and Chloroflexi, both composed of photosynthesizers, were approximately 4.5X more abundant in ORPs than SCRPs. Both phyla had a combined relative frequency of 36.6% and 8.0% for ORP and SCRP bacterial communities, respectively. Two bacterial phyla, Chlamy- diae and BRC1, were identified as differentially abundant within the Eard- ley Canyon area. Chlamydiae was approximately 4X more abundant and had a relative frequency of 0.08% within the Eardley Canyon area. BRC1 was approximately 3X more abundant and had a relative frequency of 0.15% within the Eardley Canyon area. Chlamydiae was also differentially abundant within the sample month of September. BRC1 had a W value of 40, which was slightly below the threshold to be considered differentially abundant. Only Eardley Canyon was sampled in September and it is therefore unclear whether the the differential abundance in Chlamy- diae and BRC1 is occurring by sample area or sample month since both ANCOM tests produced essentially the same results...... 13 2.4 Results of RDA and variance partitioning - The plot was generated from the results of the RDA analysis from the full data set. The numbers in the biplot represent the following explanatory variables: 1) latitude, 2) longitude, 3) pool replicate number, 4) habitat, 5) maximum pool volume, 6) pH, 7) Entrajo sample area, and 8) Fry sample area. The length of each vector corresponds to the strength of its predictive power. Across the whole data set, all environmental and spatial variables together explained 37% of the compositional variation, with 63% left unexplained. Environ- mental factors alone explained 26% and spatial factors alone explained 11%. These results were mirrored by the variable partitioning analysis which showed environmental factors accounted for more of the variation (r2=0.26) than spatial factors (r2=0.11). When the data set was split, the RDA results showed that spatial factors explained 34% and 32% of the compositional variation in SCRPs and OPRs, respectively, while environmental factors explained 15% of the variation for both habitat types. The model was able to explain more of the compositional variation for SCRPs (53%) than ORPs (51%). The variance partitioning analysis produced similar results. In SCRPs, space alone accounts for 27% (r2=0.27), environment alone accounts for 34% (r2=0.34), and space + environment accounts for 49% (r2=0.49) of the variation in the data set. In ORPs, space alone accounts for 20%, environment alone accounts for 32%, and space + environment accounts for 47% of the variation in the data set. These results indicate that spatial factors are able to explain 7% more of the variation in SCRPs than in ORPs...... 14 ix

2.5 Linear regression results - The numeric labels on the box plots are means for the respective habitat types. a) SCRPs have greater ASV 2 richness than ORPs (F1,90=12.21, p<0.001, r =0.12). Alpha diversity 2 was also positively correlated with pH (F1,90=19.95, p<0.001, r =0.18). No statistically significant effect on alpha diversity was detected from month of sampling (p=0.43), sample area (p=0.28), or maximum pool volume (p=0.29). b) Inundated ORPs had significantly higher water ◦ ◦ temperatures (mean = 20 C) than SCRPs (mean = 17 C) (F1,59=34.26, p<0.001, r2=0.37). c) SCRPs had a higher mean pH (7.72) than ORPs 2 (7.13) (F1,90=83.66, p<0.001, r =0.48)...... 15

3.1 Eﬀect of Pool Persistence on Alpha Diversity - The median value of alpha diversity is lower for the perennial group (364) than the ephemeral group (1126). However, the mean is much higher at 1417 and 1003, respectively...... 23 3.2 Eﬀect of Pool Persistence by Microbe Type on Alpha Diversity - The mean and median alpha diversity values are: 955 & 802 for ephemeral eukaryotes; 171 & 102 for perennial eukaryotes; 1140 & 1436 for ephemeral prokaryotes; and 1798 & 1336 for perennial prokaryotes, respectively.... 24 x

List of Tables

Figure Page

3.1 Effect of Pool Persistence on Alpha Diversity - This model included the following fixed effects: an interaction between persistence and microbe type, absolute latitude, and sample type. Study ID was included in the model as a random effect...... 22 3.2 Effect of Salinity on Alpha Diversity - This model included the following fixed effects: persistence, absolute latitude, and salinity. Study ID was included in the model as a random effect...... 22 3.3 Effect of Precipitation on Alpha Diversity - This model included the following fixed effects: persistence, absolute latitude, and mean annual precipitation. Study ID was included in the model as a random effect... 23 Chapter 1

Introduction

Slot canyons are extremely narrow and deep geological features that form in layers of continuous bedrock where stream incision occurs faster than channel widening (Holland 1977). Slot canyons can be found worldwide in various types of bedrock including sandstone and limestone formations (Selkirk et al. 2001; Harvey et al. 2011; Sanders et al. 2014). Some slot canyons contain perennial streams that continuously erode the bedrock. Other slot canyons contain intermittent streams that erode bedrock via pulses of ﬂooding activity following a precipitation event (Limaye and Lamb 2014). Canyons formed from intermittent streams often contain pools of water within the bedrock formation that persist between precipitation events (see Figure 1.1).

Most research on slot canyons pertains to the geological sciences (Carter and An- derson 2006; Pratt-Sitaula et al. 2007; Sanders et al. 2014). There are very few published biological studies that refer to slot canyons and those that do typically do so only in passing (Lindley et al. 2017; Mullet et al. 2008). Despite extensive literature searches, I have been unable to ﬁnd any research pertaining to the ecology of slot canyon rock pools. However, while slot canyons have not received much attention other freshwater rock pool environments have. The Colorado Plateau region of Utah is one of the best studied open rock pool areas in the world (Jocque et al. 2010), and unlike slot canyons, peer reviewed information about the chemical, physical, and biological properties of open rock pools has been published (Baron et al. 1998; Chan et al. 2005). Open rock pools are globally distributed and are typically characterized as depressions that have been eroded into the exposed surface of rocky outcrops. These depressions periodically ﬁll with water via precipitation and may persist only a few hours to several months depending on the size of the pool (see Figure 1.2). 2

Figure 1.1: Example of a slot canyon Figure 1.2: Examples of open rock rock pool. pools.

Both open rock pools and slot canyon rock pools could be classified as “rock pool ecosystems”, but slot canyons are characteristically different from open rock pools in many respects. Slot canyons are deep and extremely narrow. This morphology greatly limits the amount of sunlight that reaches the canyon bottom. As a result, many slot canyon rock pools hold water perennially and water temperatures in slot canyons are often less than 15◦C even in midsummer when ambient temperatures exceed 30◦C [personal observation] (see Figure 1.3). Tall, narrow canyon walls also protect slot canyons from disturbance from wind and instead promote powerful flash floods. Floods are likely the primary dispersal mechanism within slot canyons capable of dispersing sediment and propagules linearly between pools of a single canyon. Flash floods in slot canyons have been known to drastically change canyon morphology by flushing out enormous amounts of sediment and large boulders and by bringing in logs and other organic material (see Figure 1.4).

Conversely, the morphology of open rock pools leaves them extremely exposed. Recorded midsummer water temperatures in open rock pools of the Colorado Plateau are known to exceed 30◦C(Baron et al. 1998; Carter and Anderson 2006; Jocque et al. 2010). This intense heat causes most open rock pools to be ephemeral. Even large open rock pools typically only hold water for only a few days or weeks following a precipitation event (Baron et al. 1998; Jocque et al. 2010). Disturbance from wind is also common (Netoﬀ and Shroba 1993; Kurtz Jr and Netoﬀ 2001). Wind is speculated to be the dominant dispersal mechanism for open rock pools, capable of transporting propagules vast distances across the landscape (Graham and Wirth 2008).

There is very little information available about the microbial ecology of open rock 3

Figure 1.3: Looking into a slot canyon from above. The narrow slit in the center of the photo is the opening to the canyon which is approx. 20m deep. The narrowness of the opening greatly limits the amount of direct Figure 1.4: A large log jam sunlight that reaches the rock pools in brought into a slot canyon by the canyon bottom. In this particular a flash flood. Image courtesy canyon the average water temperature of Canyon Collective (Collective was 15.9◦C. 2018). pools on the Colorado Plateau, Utah. Chan et al.(2005) found that open rock pools of the Colorado Plateau often contain black rings of desiccated bacteria which either line the bottom of the pool or appear as a high-water mark. However, that study does not specify the specific bacterial taxa composing the biofilms and instead uses very general terms like “aerobic heterotrophs”. To my knowledge, the microbial ecology of slot canyons has never been studied anywhere in the world. In Chapter2 of this thesis I address this knowledge gap by investigating the bacterial community composition of slot canyons and open rock pools of the Colorado Plateau. Chapter2 describes the microbial alpha diversity of each habitat and discusses differences in the relative importance of various community assembly processes between habitat types.

In Chapter3 of this thesis I conducted a meta-analysis to investigate patterns of microbial alpha diversity across globally distributed ephemeral and perennial pool environments. Previous meta-analyses of perennial and ephemeral habitats have focused almost exclusively on large organisms such as macro-invertebrates (Rosset et al. 2017; Soria et al. 2017). There is much debate in the ﬁeld of ecology about whether microbes follow the same biodiversity patterns as macro-organisms (Finlay 2002; Fierer et al. 2011; Shen et al. 2014; Wang et al. 2011). By conducting a meta-analysis that focused on the microbial diversity this thesis contributes meaningfully to the aforementioned debate. 4

The meta-analysis also adds to the research I conducted in Chapter2 by establishing that slot canyons and open rock pools follow the same patterns of microbial diversity generally seen in other ephemeral and perennial pool environments. Together, these chapters complement one another by describing the microbial ecology of very speciﬁc ephemeral and perennial habitats, slot canyons and rock pools, as well as a diverse group of globally distributed ephemeral and perennial aquatic habitats. 5

Chapter 2

Species Sorting Determines Bacterial Community Composition of Rock

Pools and Slot Canyons of the Colorado Plateau, Utah

2.1 Introduction Species sorting and dispersal limitation have both been shown to play pivotal roles in determining aquatic microbial community composition (Van der Gucht et al. 2007; Lindströmand Ostman¨ 2011). In general, species sorting seems to play the predominant role in structuring microbial communities (Langenheder and Ragnarsson 2007; Van der Gucht et al. 2007; Lindströmand Ostman¨ 2011; Wang et al. 2013a). However, the relative importance of species sorting and dispersal limitation varies with the characteristics of the habitat being studied. For example, species sorting has been shown to have increased importance in eutrophic waters because high nutrient levels can facilitate rapid reproduction and allow bacteria to quickly track environmental changes (Van der Gucht et al. 2007). Environmental heterogeneity also increases the importance of species sorting by selecting for microbes based on specific niche requirements (Maloufi et al. 2016).

Less is known about the conditions that cause dispersal limitation to be dominant. Studies that report a stronger signal of spatial factors than environmental factors tend to sample densely from within the same habitat type (Lindstr¨omand Ostman¨ 2011). Other studies suggest that dispersal limitation may become increasingly important in homogeneous and oligotrophic habitats (Lear et al. 2014). There is also evidence that a habitat’s rarity and degree of isolation may increase the importance of dispersal limitation (Reche et al. 2005). Ultimately, additional research is needed to better understand when and where species sorting or dispersal limitation should be expected to play the greater role in determining aquatic microbial community composition. 6

Freshwater rock pools of the Colorado Plateau are ideal habitats for studying the relative importance of species sorting and dispersal limitation. Open rock pools (ORPs) are widely distributed and less complex than other aquatic habitats (Jocque et al. 2010). These characteristics have allowed previous studies on ORPs to contribute substantially to to the ﬁeld of microbial community ecology (Therriault and Kolasa 2001; Langen- heder and Ragnarsson 2007; Sz´ekely and Langenheder 2014). However, some rock pool habitats, such as slot canyon rock pools (SCRPs), have never been studied by microbial ecologists. This presents a unique opportunity to study a previously unstudied habitat while simultaneously teasing apart the mechanisms governing microbial community composition.

SCRPs are subject to vastly different environmental conditions and dispersal regimes than typical ORPs. Slot canyons are extremely narrow and deep due to incision occurring faster than channel widening (Holland 1977) (see Figure 2.1). Slot canyons are so narrow that very little light and heat can reach the canyon bottom. As a result many SCRPs hold water perennially and midsummer water temperatures often stay below 15◦C [personal observation]. Tall canyon walls also protect SCRPs from wind, but in turn facilitate violent flash floods. These floods disperse propagules linearly from pool to pool within a canyon. In contrast, ORPs are depressions in the exposed surface of rocky outcrops (Jocque et al. 2010) (see Figure 2.1). Their extreme exposure means ORPs are typically ephemeral with water temperatures frequently exceeding 30◦C midsummer (Baron et al. 1998; Chan et al. 2005). Disturbance from wind is common (Netoff and Shroba 1993; Kurtz Jr and Netoff 2001) and is speculated to be the dominant dispersal mechanism for ORPs by transporting propagules to new pools across the landscape (Graham and Wirth 2008).

In this study, I compare the bacterial community composition of naturally occurring ORP and SCRP habitats on the Colorado Plateau to seek to understand the conditions that affect the relative importance of species sorting and dispersal limitation. I hypothesize: (1) that SCRPs and ORPs are indeed distinct habitat types with significantly different bacterial community compositions; (2) that species sorting is the dominant mechanism determining community composition in both SCRPs and ORPs, as seen in other aquatic microbial communities; but (3) that SCRPs are more dispersal limited than ORPs due to their differential disturbance regimes and increased rarity 7 and isolation. Because ORPs experience prolonged periods of extreme heat and desiccation while SCRPs are inundated year-round, I also hypothesize (4) that bacterial alpha diversity will be higher in SCRPs than ORPs. Disturbance from drying has been shown to significantly decrease alpha diversity in microbes (Gionchetta et al. 2019) and macroorganisms (Del Rosario and Resh 2000; Therriault and Kolasa 2001; Leigh and Datry 2017; Rosset et al. 2017; Soria et al. 2017) and I anticipate that the same pattern will hold true for rock pool bacteria.

2.2 Methods

2.2.1 Experimental Design and Sampling My overall experimental design sought to differentiate between the effects of space (dispersal limitation) and environment (species sorting) on bacterial community composition by sampling SCRPs and nearby ORPs in different geographic regions on the Colorado Plateau. Environmental parameters were measured at each sample location to help identify potential drivers of species sorting.

Sampling took place in three areas of the Colorado Plateau identified here as: Entrajo Canyon (lat = 38.472551 long = -109.39623), Fry Canyon (lat = 37.635438 long = -110.149417), and Eardley Canyon (lat = 38.791396, long = -110.523958) (see Figure 2.1). The Entrajo Canyon and Fry Canyon areas were sampled June 15-17, 2018. Due to flash floods creating unsafe field conditions, the Eardley Canyon area was sampled approximately three months later on September 22-23, 2018.

Within each area, soil samples were collected from the sediment at the bottom of both SCRPs and ORPs. Four pools of each habitat type were sampled within each area, and four replicates were taken from each pool for a total of 96 soil samples. Replicates were collected starting from the edge of the pool and ending towards the middle of the pool at 0m, 0.5m, 1m, and 1.5m, respectively. Between each sample, collection equipment was sterilized by removing all visible solid mass with a cloth wipe and then thoroughly rinsing the equipment using 97% isopropyl alcohol. Samples were stored in sterile 118ml Nasco R Whirl-Pak bags on ice while in the ﬁeld, then placed in a -20◦C freezer in the lab.

To help determine potential drivers of species sorting the following environmental parameters were measured at each pool: maximum length, maximum width, maximum 8 depth, water temperature, ambient air temperature, and pH. Maximum pool volume was estimated from the maximum depth, width, and length of each pool, assuming the pools were cuboidal. Temperatures were measured using a YSI Model 85 water quality meter. Ambient air temperature was collected at every pool and water temperature was collected at all inundated pools. The exact geographic location of each sample pool was recorded using a hand held GPS. The pH of all 96 samples was recorded in the lab upon return from ﬁeld work. For each sample, 3g of soil were suspended in 15ml of deionized water. Suspended samples were allowed to settle for 24hrs before measuring the pH using a bench top pH meter.

2.2.2 DNA Extraction and 16s rRNA Gene Sequencing To determine the bacteria present in SCRPs and ORPs I used 16s rRNA gene sequencing on environmental DNA (eDNA) occurring in my soil samples. eDNA was extracted from the samples using the DNeasy R PowerSoil R Kit (QIAGEN, German- town, MD). Extractions took place according to the manufacturer’s protocol, excluding the optional five minute incubation steps. Additionally, extractions on soil samples collected from inundated pools used 400mg of the sample to account for water weight. Extractions on dry samples used the standard 250mg specified in the manufacturer’s protocol. DNA samples were sent to the Center for Integrated Biosystems, Utah State University, Logan, UT for amplification of the V4 region of the 16s rRNA gene. Am- plification took place using the primer pair 515F/806R as described by Parada et al. (2016) and Apprill et al.(2015), respectively. Amplicons were sequenced on Illumina MiSeq using paired-end 250 + 250bp reads.

2.2.3 Analysis of 16s rRNA Gene Sequences Sequences were processed using the bioinformatics data pipeline QIIME2 (version 2018.11) (Caporaso et al. 2010). All QIIME2 commands are included in the supplemental materials, but briefly, sequence quality filtering was performed using DADA2 (Callahan et al. 2016) on default settings. Filtering was performed after trimming the forward and reverse primer sequences. Amplicon sequence variants (ASVs) were used in favor of clustering operational taxonomic units (OTUs) due to their superior reproducibility (Callahan et al. 2017) and specificity (Thompson et al. 2017). Taxonomic classifications were assigned to ASVs using the naive Bayesian classifier classify-sklearn (scikit- learn version 0.19.1) (Bokulich et al. 2018) found in QIIME2’s feature-classifier 9 is an example of an ORP. Lake Powell, Utah is visible in the background. The numbers correspond to (1) Eardley Canyon, (2) Entrajo Canyon, and (3) Fry Canyon. The hatched area is the Sample Areas - boundary of the Colorado Plateau. The two photos were taken at the Fry Canyon Area. The top photo is an example of a SCRP. The bottom photo Figure 2.1: 10 plugin. The classifier was trained using the GreenGenes 13 8 99 OTU reference database containing only 515F/806R sequences (McDonald et al. 2012). All sequences classified as “Unassigned” were removed from the data set using QIIME2’s filter-table method in the taxa plugin. Differential sequencing depth between samples was accounted for by rarefying to 94,150 sequences after confirming saturation of the rarefaction curve. Three of the samples did not achieve adequate sequencing depth and were removed from the data set prior to statistical analysis.

2.2.4 Statistical Analysis To determine the bacterial community composition of my samples I calculated beta diversity using Bray-Curtis dissimilarity. I then used perMANOVA to test my hypothesis that bacterial community composition is significantly different between SCRPs and ORPs. To better visualize results I plotted beta diversity using non-metric multidimen- sional scaling (NMDS). PerMANOVA was also used to identify which of the measured environmental parameters might be contributing to species sorting. I used ANCOM (Mandal et al. 2015) to identify any bacterial phyla that were differentially abundant by habitat type. Because there was a three month delay in sample collection at the Eard- ley Canyon area, ANCOM was also used to determine if any phyla were differentially abundant by month of sampling or sample location.

I used redundancy analysis (RDA) and variance partitioning to test my second and third hypotheses and determine the relative importance of species sorting and dispersal limitation by habitat type. In order to determine if the amount of variance explained by spatial and environmental factors differed between the habitats, I used RDA and variance partitioning on the whole data set then split the data set into two and conducted each test separately by habitat type. To test my fourth hypothesis that SCRPs have greater alpha diversity than ORPs, I used linear models. I also used linear models to determine which environmental parameters significantly affected alpha diversity and/or were significantly different between habitat types.

Except for ANCOM, all statistical analyses were performed using R version 3.4.0. ANCOM was performed using the ANCOM method within QIIME2’s composition plugin. 11

2.3 Results The results of the perMANOVA analyses supported my hypothesis that SCRPs and ORPs are distinct habitats with significantly different bacterial community compositions (see Figure 2.2). The perMANOVA results also showed that bacterial community composition differed significantly by sample area, pH, and maximum pool volume. The ANCOM results identified 19 bacterial phyla and 2 archaeal phyla that were differentially abundant by habitat type. ANCOM also identified 2 phyla that were differentially abundant by sample area and 1 phylum that was differentially abundant by month of sampling (see Figure 2.3).

The results of my RDA and variance partitioning analyses supported my hypothesis that species sorting is the dominant mechanism determining bacterial community composition in both SCRPs and ORPs (see Figure 2.4). The results also suggest that dispersal limitation may be more important to SCRPs than ORPs, but additional study is needed to be certain since other factors could have produced the same pattern in the data.

My hypothesis that bacterial alpha diversity is higher in SCRPs than ORPs was supported by the results of my linear models. Linear models also indicated that alpha diversity is positively correlated with pH and that mean pH and water temperature were signiﬁcantly diﬀerent between SCRPs and ORPs (see Figure 2.5).

2.4 Discussion The results of my study support my first, second, and fourth hypotheses that: (1) SCRPs and ORPs are distinct habitats with significantly different bacterial community compositions; (2) species sorting is the dominant community assembly mechanism for both habitat types; and, (4) SCRPs have greater alpha diversity than ORPs. My results also suggest my third hypothesis, that SCRPs are more dispersal limited than ORPs, may be true.

2.4.1 Hypothesis 1 Bacterial community composition differed significantly by habitat type indicating that ORPs and SCRPs are indeed distinct rock pool habitats. The exact environmental conditions responsible for the observed differences in bacterial community composition are unknown, but the literature points to several likely factors. For example, pH is known 12

Open Slot Canyon

Fry Eardley Entrajo NMDS2

NMDS1

Figure 2.2: NMDS plot of beta diversity - PerMANOVA indicated that bacterial community composition differed significantly by habitat type (F1,90=24.34, p<0.001, 2 2 r =0.179) and sample area (F1,90=6.54, p<0.001, r =0.096), with habitat type explaining more of the variation in community composition than sample area. This can be seen in the NMDS plot (stress=0.076) which shows samples clustering first, by habitat type, and second, by sample area within each habitat cluster. Additionally, ORPs appear to have greater spatial separation than SCRPs. The perMANOVA analyses of environmental parameters showed that bacterial community composition also var- 2 ied significantly by pH (F1,90=6.67, p<0.001, r =0.049) and maximum pool volume 2 (F1,90=3.78, p<0.001, r =0.028). to be an important driver of variation in aquatic bacterial communities (Yannarell and Triplett 2005). The results of my perMANOVA analyses indicate that SCRPs have a higher mean pH than ORPs, and even very slight changes in pH can cause substantial shifts in aquatic bacterial community composition (Krause et al. 2012). It is possible this is due to the narrow pH ranges ideal for the growth of individual bacterial species. Changes of ±1.7 pH units consistently reduce growth by 50% (Fernández-Calviñoand B˚a˚ath 2010) and reductions as small as 25% are enough to give other bacterial species a competitive advantage and become more abundant community members (Rousk et al. 2010).

Light intensity is another factor that can cause substantial variation in the community composition. Wagner et al.(2015) demonstrated that photosynthetic species were more abundant in aquatic bioﬁlm located in high intensity light. In my study, 13

Proteobacteria Chloroflexi Bacteroidetes Cyanobacteria Acidobacteria Firmicutes Verrucomicrobia Acidobacteria Planctomycetes Other

Ent Slot Ent

Ear Slot Ear

Fry Slot Fry Ent Open Ent This bar chart shows the top 9 most abundant phyla with all other phyla (60 total) grouped

tests produced essentially the same results.

Ear Open Ear

Fry Open Fry

1.0 0.8 0.6 0.4 0.2 0.0

Differential Abundance of Phyla - Proportion of Total Sequences Total of Proportion together under “Other”.that ANCOM the identified phyla 19 Cyanobacteria bacterial andBoth phyla Chloroflexi, phyla and both had composed 2 aChlamydiae of archaeal and combined photosynthesizers, phyla BRC1, were were relative that approximately identified frequency were asand 4.5X differentially differentially had of more abundant abundant a abundant within 36.6% relative by in the and frequency Eardley0.15% ORPs habitat Canyon of 8.0% within than type. area. 0.08% SCRPs. the for within Chlamydiae Of Eardley was ORP theof note approximately Canyon Eardley and 4X 40, area. is Canyon more SCRP which area. Chlamydiae abundant bacterial was wastherefore BRC1 also slightly unclear was communities, approximately differentially whether below respectively. 3X the abundant the more the within Two threshold abundant differential the bacterial and to abundance sample had phyla, in be month a Chlamydiae of considered and relative September. frequency differentially BRC1 is of abundant. BRC1 occurring had Only by a sample Eardley W area Canyon value or was sample sampled month since in both September ANCOM and it is Figure 2.3: 14

100 Eardley Entrajo

80 Fry Open

60 Slot 40 RDA2

20 6 4 7

0 8 2 5

-20 3 1 -40

-50 0 50 100

RDA1

Figure 2.4: Results of RDA and variance partitioning - The plot was generated from the results of the RDA analysis from the full data set. The numbers in the biplot represent the following explanatory variables: 1) latitude, 2) longitude, 3) pool replicate number, 4) habitat, 5) maximum pool volume, 6) pH, 7) Entrajo sample area, and 8) Fry sample area. The length of each vector corresponds to the strength of its predictive power. Across the whole data set, all environmental and spatial variables together explained 37% of the compositional variation, with 63% left unexplained. Environmen- tal factors alone explained 26% and spatial factors alone explained 11%. These results were mirrored by the variable partitioning analysis which showed environmental factors accounted for more of the variation (r2=0.26) than spatial factors (r2=0.11). When the data set was split, the RDA results showed that spatial factors explained 34% and 32% of the compositional variation in SCRPs and OPRs, respectively, while environmental factors explained 15% of the variation for both habitat types. The model was able to explain more of the compositional variation for SCRPs (53%) than ORPs (51%). The variance partitioning analysis produced similar results. In SCRPs, space alone accounts for 27% (r2=0.27), environment alone accounts for 34% (r2=0.34), and space + environment accounts for 49% (r2=0.49) of the variation in the data set. In ORPs, space alone accounts for 20%, environment alone accounts for 32%, and space + environment accounts for 47% of the variation in the data set. These results indicate that spatial factors are able to explain 7% more of the variation in SCRPs than in ORPs. 15

Figure 2.5: Linear regression results - The numeric labels on the box plots are means for the respective habitat types. a) SCRPs have greater ASV richness than 2 ORPs (F1,90=12.21, p<0.001, r =0.12). Alpha diversity was also positively correlated 2 with pH (F1,90=19.95, p<0.001, r =0.18). No statistically significant effect on alpha diversity was detected from month of sampling (p=0.43), sample area (p=0.28), or maximum pool volume (p=0.29). b) Inundated ORPs had significantly higher water ◦ ◦ temperatures (mean = 20 C) than SCRPs (mean = 17 C) (F1,59=34.26, p<0.001, 2 r =0.37). c) SCRPs had a higher mean pH (7.72) than ORPs (7.13) (F1,90=83.66, p<0.001, r2=0.48). 16 photosynthetic Cyanobacteria and Chloroflexi were 4.5X more abundant in high light (i.e., ORPs) than in low light (i.e., SCRPs).

Water permanence is also highly influential on bacterial community composition. In one study of Australian rock pools, pool drying served as an environmental filter that excluded drought intolerant taxa in ephemeral pools and caused ephemeral pool communities to be a nested subset of perennial pool communities (Brendonck et al. 2015). Bacterial phylogentic beta diversity is known to have strong environmental associations due to habitat specific adaptations (Wang et al. 2013a). It is possible that the bacterial community composition of ORPs reflects those taxa with life history strategies most adapted to desiccation. As previously stated, Cyanobacteria were more abundant in ORPs than SCRPs. They are also known to form hygroscopic extracellular polysaccharides which are resilient to desiccation (Chan et al. 2005; Mager and Thomas 2011) making them better adapted to conditions in ephemeral habitats.

2.4.2 Hypotheses 2 and 3 The results of my study support most of the published literature which identifies species sorting as the primary assembly mechanism in many aquatic bacterial communities (Langenheder and Ragnarsson 2007; Lindströmand Ostman¨ 2011; Van der Gucht et al. 2007; Wang et al. 2013a). A recent review of microbial biogeography found that over 90% of published studies reported a significant effect of habitat and that across studies more variation was explained by environmental factors than by geographic distance (Hanson et al. 2012). This pattern is easily visualized in the NMDS plot in Figure 2.2 which shows bacterial community composition clustering first by habitat type and then by sample area with each habitat cluster. SCRPs and ORPs are subject to contrasting environmental conditions and evidently, those conditions are varied enough to support significantly different bacterial communities.

Studies that do report a stronger signal of spatial factors than environmental factors tend to sample densely from within the same habitat type (Lindströmand Ostman¨ 2011; Lear et al. 2014). In these cases dispersal limitation and/or mass effects dominate community composition because the selective strength of local environmental conditions is beneath a habitat specific threshold (Lindströmand Langenheder 2012; Wang et al. 2013a). The same pattern holds true for my study. When each habitat type was analyzed individually the effect of space became more important (see Figure 2.4). In other words, 17 spatial factors played a greater role in bacterial community assembly within a single habitat than they did across the habitat types.

Interestingly, the NMDS plot shows habitat type separating along the ﬁrst axis, and space separating along the second. The spread of the data points along the second axis is a possible indication that dispersal limitation is more important to SCRPs than ORPs, as stated in my third hypothesis. Additionally, space explained more of the compositional variation in SCRPs for both the RDA and variance partitioning analyses (see Figure 2.4), adding further support to my hypothesis. If dispersal limitation is more important to the assembly of SCRP bacterial communities it could be because the primary dispersal mechanism of slot canyons is ﬂooding rather than wind. Floods are local whereas wind is regional and can transport propagules thousands of kilometers (Choudoir et al. 2018). SCRPs are also much more isolated and rare than ORPs which has been hypothesized to increase the importance of dispersal limitation (Reche et al. 2005).

However, while dispersal limitation could indeed be more important in SCRPs than in ORPs, there are other possible explanations for the patterns I observed in my data. In general, bacteria are not greatly limited by their ability to disperse (Choudoir et al. 2018). Sample areas for my study were located a maximum of 150km apart (see Figure 2.1) and prokaryotes are known to have mean range sizes spanning thousands of km2 (Choudoir et al. 2018). This places my sample areas well within the geographic range of most bacteria and archaea. Even still, 11% of the variation of my entire data set could be explained by spatial factors alone. Therefore, dispersal limitation is almost certainly involved in rock pool community assembly because any effect of geographic distance is evidence of dispersal limitation (Hanson et al. 2012). Mass effects may also periodically impact bacterial community composition. Heavy precipitation and flooding may introduce new microbes to rock pools and increase the likelihood of colonization by flushing out other community members. However, Lindströmand Ostman¨ (2011) demonstrated that dispersal is only capable of consistently changing bacterial community composition if it is maintained at very high levels. Otherwise, locally adapted taxa quickly exclude introduced taxa through their superior growth rates. If mass effects from flooding do change rock pool communities it is likely only for a short time since disturbance from flooding is relatively brief and very intermittent. 18

2.4.3 Hypothesis 4 My study found that bacterial alpha diversity was significantly higher in SCRPs than in ORPs. Several factors are likely responsible for this pattern. For example, temperature and pH are both known to affect microbial alpha diversity (Griffiths et al. 2011; Wang et al. 2013b; Ren et al. 2015; Thompson et al. 2017) and both were significantly different between habitat types. Water permanence is likely another major factor. While some studies have reported that drying and rewetting has little effect on the diversity and richness of bacteria (Fierer et al. 2003; McHugh and Schwartz 2016) these studies were conducted in temperate environments. Bacteria are known to be resilient to non-extreme changes in moisture (Kaisermann et al. 2015), but moisture fluctuations within ORPs regularly alternate between complete inundation and desiccation. Such extreme drying limits diffusion causing restricted nutrient supply to bacteria (Stark and Firestone 1995). It also inhibits cellular functions by dehydrating enzymes and changing their conformations (Csonka 1989) and can even cause plasmolysis and cell death (Beney and Gervais 2001). One study conducted by Gionchetta et al.(2019) found that extreme changes in flow intermittency and drying decreased microbial alpha diversity in intermittent streams (Gionchetta et al. 2019). Disturbance from drying has also been shown to significantly decrease alpha diversity in macroinvertebrates (Therriault and Kolasa 2001; Del Rosario and Resh 2000; Leigh and Datry 2017; Soria et al. 2017). My results provide additional evidence that microbes follow the trends observed in mac- robes and that perennial systems (i.e. SCRPs) have higher microbial alpha diversity than ephemeral systems (i.e., ORPs).

2.5 Conclusion In conclusion, my research provides additional evidence of the importance of species sorting for aquatic bacterial communities. It also supports the claim that perennial systems have higher microbial alpha diversity than ephemeral systems, as is observed in larger organisms. Additionally, my study identifies slot canyons as distinct habitats that vary significantly from other rock pool environments. It is one of the few studies to explore slot canyon ecology and the first to investigate slot canyon microbial ecology. The effects of climate change are likely to impact rock pool ecosystems as arid regions of North America become increasingly dry (Seager et al. 2007). Currently, very little is known about how slot canyons function ecologically. It is my hope that this study will spark more research interest in these unique habitats to aid their conservation. 19

Chapter 3

A Meta-Analysis of Microbial Alpha Diversity in Perennial and Ephemeral

Pool Habitats

3.1 Introduction Drylands make up a large portion of the Earth’s land surface (Reynolds et al. 2007), and total dryland area is expected to expand due to climate change (Seager et al. 2007; Fu and Feng 2014; Huang et al. 2016). Drying has been shown to significantly decrease alpha diversity in macro-organisms (Therriault and Kolasa 2001; Del Rosario and Resh 2000; Leigh and Datry 2017; Soria et al. 2017). However, microbes do not allows follow the same ecological trends as larger organisms, and the effect of drying on microbial diversity is less clear. For example, Gionchetta et al.(2019) showed significant decreases in bacterial alpha diversity in intermittent stream soils due to drying. In contrast Fierer et al.(2003) and McHugh and Schwartz(2016) found no effect of drying on microbial alpha diversity in soils.

Reviewing the current literature is obviously not enough to determine a clear relationship of drying on microbes One possible solution is to use meta-analysis to attempt to identify patterns. Meta-analysis is a powerful statistical approach that allows the results of independent studies to be compared quantitatively. Meta-analysis is funda- mentally different from vote counting (i.e., tallying which effect appears most frequently in the published literature) because meta-analysis takes into account both within study and between study variances. By accounting for both sources of variance, meta-analysis has the ability to summarize the general effect of an explanatory variable on a response variable using information from studies that may at first appear to have confounding results.

In this study, I used meta-analysis of perennial and ephemeral pool habitats to test the hypothesis that microbial diversity is greater in perennial pools than in ephemeral 20 pools. Perennial and ephemeral pools are ideal study systems because ephemeral pools experience regular disturbance from drying while perennial pools can act as a kind of control due to their consistent inundation. I also tested the hypothesis that microbes follow the typical latitudinal diversity gradient (i.e., diversity is highest at the equator and decreases towards the poles).

3.2 Methods

3.2.1 Overall Experimental Design This meta-analysis was performed in four basic steps. First, specific search terms were used to find relevant literature in a peer reviewed database. Second, each of the search results was thoroughly reviewed to determine if it should be included in the analysis. Third, methodological and biological data was gathered and recorded from each included study. And fourth, the data was run through mixed effects models to account for both within and between study variances and to test the hypothesis that perennial pools exhibit greater microbial alpha diversity than ephemeral pools. Each of these steps is described in greater detail in the paragraphs below.

3.2.2 Data Selection Two separate Google Scholar searches were made to locate published literature for inclusion in the meta-analysis. The ﬁrst search included the the following search terms: alpha diversity AND microb* AND aquatic AND pool* AND ephemeral. The second search used the terms: alpha diversity AND microb* AND aquatic AND pool* AND perennial. Both searches were performed on October 1, 2019. All search results were then recorded, and any research published after the search date was not included in the analysis. Each of the search results was thoroughly reviewed to determine if it is appropriate to include in the meta-analysis. Appropriate studies need to (a) be primary literature, (b)use sequence based approaches to quantify microbial diversity, (c) include alpha diversity measurements, and (d) have samples that were collected from an ephemeral or perennial pool environment. If a study did not meet each of these criteria, it was excluded from the analysis and the speciﬁc reasons for exclusion were recorded. The data I obtained from my research on slot canyons and open rock pools, detailed in Chapter2 of this thesis document, was also included in the analysis.

Alpha diversity, measured as the number of operational taxonomic units (OTUs) or 21 amplicon sequence variants (ASVs), was recorded for every study. For all studies that did not explicitly state mean annual temperature or mean annual precipitation, values were extracted from global climate data compiled by Willmot and Matsuura (Willmot and Matsuura 2001) using ArcMAPTM (Esri ArcGIS R 10.7.1). Climate data from PRISM Climate Group (PRISM Climate Group 2004) was then used for any studies conducted in the United States that were still missing temperature or precipitation data.

3.2.3 Statistical Analysis Linear regression was used to determine whether diﬀerences in methodology between studies were correlated with alpha diversity. I used linear models to test each of the following methodological variables: sequences per sample, sequencing platform, OTU percent, and taxonomic assignment database.

Alpha diversity was used as the measure of effect size for this study. I assumed that true effect sizes were different between studies and, therefore, used random effects models. A weighted mean effect size was calculated so studies with the greatest sample sizes contributed most meaningfully to the analysis. I used variance inflation factors (VIF) to account for multicollinearity between variables. Only variables that were found to be orthogonal were included in the model. Variables for microbe type (i.e., eukaryotic vs prokaryotic) and sample type (i.e., water vs sediment) were also included in the model to account for well-known differences in alpha diversity between these factors (Whitman et al. 1998; Torsvik and Øvre˚as 2002). All statistical analyses were conducted using R version 3.6.1.

3.2.4 Results The results of the linear regression analysis indicated that there was no effect to alpha diversity from methodological differences between studies for sequences per sample (p=0.35); sequencing platform (p=0.44); and OTU percent (p=0.09). The effect of taxonomic assignment database was determined to be significant (F4,83=4.94, p=0.0001, r2=0.16). However, the VIF analysis showed this variable to be collinear with the sample type and microbe type variables, which have a well documented effect on alpha diversity (Whitman et al. 1998; Torsvik and Øvre˚as 2002). To further support this, the results of the linear regression analysis showed that eukaryotic diversity was lower than prokaryotic diversity in both perennial and ephemeral habitats. 22

Table 3.1: Effect of Pool Persistence on Alpha Diversity - This model included the following fixed effects: an interaction between persistence and microbe type, absolute latitude, and sample type. Study ID was included in the model as a random effect.

Parameter Name Estimate Standard Error Lower-95 Upper-95 Intercept 2891.618 619.335 1677.743 4105.493 Latitude -12.471 0.008 -12.485 -12.456 Persistence: Perennial -254.265 1.491 -257.186 -251.343 Microbe: Prokaryotes 150.001 1.414 147.229 152.773 Sample: Water -653.292 0.479 -654.230 -652.354 Perennial and Prokaryotes 465.996 1.434 463.186 468.806

Table 3.2: Effect of Salinity on Alpha Diversity - This model included the following fixed effects: persistence, absolute latitude, and salinity. Study ID was included in the model as a random effect.

Parameter Name Estimate Standard Error Lower-95 Upper-95 Intercept -1605.356 8990.189 -19225.803 16015.090 Latitude 84.479 1.238 82.054 86.905 Persistence: Perennial -707.246 10381.051 -21053.732 19639.240 Salinity 0.066 0.005 0.056 0.076

As expected, prokaryotic diversity was greater than eukaryotic diversity for both perennial and ephemeral habitats. The results of the random effects models show an overall positive effect of perennial persistence on microbial alpha diversity (see Table 3.1 and Figure 3.1). However, there is an inverse interaction between pool persistence and microbe type. Prokaryotes have higher alpha diversity in perennial pools, but eukaryotes have lower diversity in perennial pools (see Figure 3.2). Only one study included in the meta-analysis sampled for eukaryotes in perennial pools, and that study took place in a permanently frozen Antarctic lake. It is possible that if my data set was larger and included more studies for eukaryotes in perennial pools that this inverse relationship would no longer be observed. The models also showed an overall negative effect of latitude on alpha diversity. In other words, microbial alpha diversity tends to decrease as distance from the equator increases (see Table 3.1). Overall, salinity had a negligible affect on alpha diversity, with the model showing estimating a positive effect of <0.1 (see Table 3.2). Mean annual precipitation appeared to have a slight positive effect on alpha diversity with alpha diversity increasing with increased precipitation (see Table 3.3). 23

Table 3.3: Effect of Precipitation on Alpha Diversity - This model included the following fixed effects: persistence, absolute latitude, and mean annual precipitation. Study ID was included in the model as a random effect.

Parameter Name Estimate Standard Error Lower-95 Upper-95 Intercept 469.538 410.435 -334.900 1273.977 Latitude -3.122 0.054 -3.228 -3.017 Persistence: Perennial 117.641 0.453 116.754 118.530 Precipitation 28.138 0.059 28.023 28.252

12500

10000

7500

5000 Alpha Diversity Alpha

2500

0 Ephemeral Perennial

Figure 3.1: Eﬀect of Pool Persistence on Alpha Diversity - The median value of alpha diversity is lower for the perennial group (364) than the ephemeral group (1126). However, the mean is much higher at 1417 and 1003, respectively. 24

12500

Ephemeral 10000 Perennial

7500

5000 Alpha Diversity Alpha

2500

0 Euk Euk Prok Prok

Figure 3.2: Eﬀect of Pool Persistence by Microbe Type on Alpha Diversity - The mean and median alpha diversity values are: 955 & 802 for ephemeral eukaryotes; 171 & 102 for perennial eukaryotes; 1140 & 1436 for ephemeral prokaryotes; and 1798 & 1336 for perennial prokaryotes, respectively.

3.3 Discussion The results of this meta-analysis support my first hypothesis that perennial pools have greater alpha diversity than ephemeral pools. This may be because desiccation often leads to cell death in microbes. Cell death can occur by inhibiting molecular diffusion and restricting microbial access to energy and nutrients (Stark and Firestone 1995). It can also occur because enzymes dehydrate, change shape, and become less effective (Csonka 1989). In extreme cases, drying can even cause plasmolysis (Beney and Gervais 2001).

Our results indicate that perennial and ephemeral pools may be useful model systems to help scientists draw additional parallels between micro- and macro-organisms. Microbial organisms do not always follow the same ecological patterns observed in larger species (Fierer et al. 2011; Shen et al. 2014; Wang et al. 2011). However, drying is also 25 known to reduce the diversity of macro-organisms (Del Rosario and Resh 2000; Ther- riault and Kolasa 2001; Leigh and Datry 2017; Rosset et al. 2017; Soria et al. 2017). Increased mean annual precipitation was another factor associated with increased microbial diversity that is also true for macro-organisms (Tilman and El Haddi 1992; McCain and Colwell 2011; Yan et al. 2015). Higher levels of precipitation are likely correlated with lower magnitudes and frequencies of desiccation. In this way, the positive eﬀect of precipitation may be “the other side of the coin” in relation to the negative eﬀects of drying.

My second hypothesis was also supported. I was able to identify a pattern of increased microbial diversity at lower latitudes which is another well documented pattern in macro-organisms (Hillebrand 2004). Previous microbial studies investigating the latitudinal diversity gradient have been mixed (Fuhrman et al. 2008; Moss et al. 2019). It is possible that latitudinal diversity differences in perennial and ephemeral pools are more dramatic than in other habitat types. However, a massive meta-analysis conducted by Thompson et al.(2017) found a weak but significant trend of increasing microbial diversity from poles to tropics across diverse habitat types. Given that microbes are already incredibly diverse (Locey and Lennon 2016), it may be difficult to detect increases in diversity within the sampling scale of a singe study. The field of microbial ecology may benefit from additional large-scale meta-analyses that can specifically address this issue.

Changes in salinity were found to have little to no eﬀect on microbial alpha diversity whereas macro-organism diversity tends to decrease with increases in salinity (Briggs and Taws 2003; Carrete Vega and Wiens 2012). Our ﬁndings are consistent with the results of other microbial studies. For example, a study of Tibetan lakes with broad salinity ranges found that bacterioplankton richness actually increased with increased salinity up to 1 PSU and then remained relatively unchanged up to 280 PSU (Wang et al. 2011). Telesh et al.(2013) proposed that this inverse diversity pattern could be related to community shifts towards smaller, more quickly-evolving microbes. They argued that accelerated evolution allows for high physiological adaptability to extreme salinity (Stock et al. 2002) and facilitates unusually high taxonomic diversity (Telesh et al. 2011).

In addition to the parallels between micro- and macro-organisms that can be drawn from the results of this meta-analysis, my study has important implications for the future 26 microbial diversity in the face of climate change. Many areas of the globe are predicted to become drier due to global change (Seager et al. 2007; Fu and Feng 2014; Huang et al. 2016), and my results mirror those of other studies that have shown decreases in microbial diversity from drying (Maestre et al. 2015; Gionchetta et al. 2019). Not only do increasingly dry conditions pose a threat to microbial diversity, but by extension they have the potential to negatively impact microbial functional productivity (Fenner et al. 2005; Maestre et al. 2015). Microbes are responsible for many ecological processes essential to the environments in which they are found. For example, microorganisms are the major link between primary and secondary production for detritus based food webs (Fenchel and Jørgensen 1977), and bacteria are especially important in the cycling of carbon and energy from lignocellulose in aquatic ecosystems (Benner et al. 1986). In the freshwater ecosystem of Lake Cadagno, Switzerland, green and purple sulfur bacteria have even been labeled keystone species for their role in biogeochemical cycling (Musat et al. 2008). Potential impacts to ecosystem functions from microbial diversity loss due to drying could have serious consequences for overall environmental health and should not be ignored.

3.4 Conclusion In conclusion, this meta-analysis shows that perennial pools have greater microbial diversity than ephemeral pools, and implies that drying is generally detrimental to diversity. Additionally, the meta-analysis provides evidence that microbes adhere to the latitudinal diversity gradient and have higher diversity in areas with greater mean annual precipitation. A relationship between salinity levels and microbial diversity was not identiﬁed. This study also demonstrates the value of perennial and ephemeral pool habitats as model systems for studying the eﬀects of drying in the face of global change. 27

Bibliography

A. Apprill, S. Mcnally, R. Parsons, and L. Weber. Minor revision to V4 region SSU rRNA 806R gene primer greatly increases detection of SAR11 bacterioplankton. Aquatic Microbial Ecology, 75:129–137, 2015. doi: 10.3354/ame01753.

J.S. Baron, T. Lafrancois, and B.C. Kondratieﬀ. Chemical and biological characteristics of desert rock pools in intermittent streams of Capitol Reef National Park, Utah. Great Basin Naturalist, 58(3):250–264, 1998.

L. Beney and P. Gervais. Inﬂuence of the ﬂuidity of the membrane on the response of microorganisms to environmental stresses. Applied Microbiology and Biotechnology, 57(1-2):34–42, 2001.

R. Benner, M.A. Moran, and R.E. Hodson. Biogeochemical cycling of lignocellulosic carbon in marine and freshwater ecosystems: Relative contributions of procaryotes and eucaryotes. Limnology and Oceanography, 31(1):89–100, 1986.

N.A. Bokulich, B.D. Kaehler, J.R. Rideout, M. Dillon, E. Bolyen, R. Knight, G.A. Hutt- ley, and G.J. Caporaso. Optimizing taxonomic classiﬁcation of marker-gene amplicon sequences with QIIME2’s q2-feature-classiﬁer plugin. Microbiome, 6(1):1–17, 2018. doi: 10.1186/s40168-018-0470-z.

L. Brendonck, M. Jocqu´e,K. Tuytens, B.V. Timms, and B. Vanschoenwinkel. Hydro- logical stability drives both local and regional diversity patterns in rock pool meta- communities. Oikos, 124(6):741–749, 2015.

S.V. Briggs and N. Taws. Impacts of salinity on biodiversity—clear understanding or muddy confusion? Australian Journal of Botany, 51(6):609–617, 2003.

B.J. Callahan, P.J. McMurdie, M.J. Rosen, A.W. Han, A.A. Johnson, and S.P. Holmes. DADA2: High-resolution sample inference from Illumina amplicon data. Nature Meth- ods, 13(7):581–583, 2016. 28

B.J. Callahan, P.J. McMurdie, and S.P. Holmes. Exact sequence variants should replace operational taxonomic units in marker-gene data analysis. ISME Journal, 11(12):1–5, 2017.

G.J. Caporaso et al. QIIME allows analysis of high-throughput community sequencing data. Nature Methods, 7(5):335–336, 2010.

G. Carrete Vega and J.J. Wiens. Why are there so few ﬁsh in the sea? Proceedings of the Royal Society B: Biological Sciences, 279(1737):2323–2329, 2012.

C.L. Carter and R.S. Anderson. Fluvial erosion of physically modeled abrasion- dominated slot canyons. Geomorphology, 81(1-2):89–113, 2006.

M.A. Chan, K. Moser, J.M. Davis, G. Southam, K. Hughes, and T. Graham. Desert potholes: ephemeral aquatic microsystems. Aquatic Geochemistry, 11(3):279–302, 2005.

M.J. Choudoir, A. Barber´an,H.L. Menninger, R.R. Dunn, and N. Fierer. Variation in range size and dispersal capabilities of microbial taxa. Ecology, 99(2):322–334, 2018. doi: 10.1002/ecy.2094.

Canyon Collective. Black Hole of White Canyon. http://canyoncollective.com/ threads/black-hole-of-white-canyon.19537/, 2018. Accessed: June 28, 2018.

L.N. Csonka. Physiological and genetic responses of bacteria to osmotic stress. Micro- biology and Molecular Biology Reviews, 53(1):121–147, 1989.

R.B. Del Rosario and V.H. Resh. Invertebrates in intermittent and perennial streams: is the hyporheic zone a refuge from drying? Journal of the North American Benthological Society, 19(4):680–696, 2000.

T.M. Fenchel and B.B. Jørgensen. Detritus food chains of aquatic ecosystems: the role of bacteria. In Advances in microbial ecology, pages 1–58. Springer, 1977.

N. Fenner, C. Freeman, and B. Reynolds. Hydrological eﬀects on the diversity of phenolic degrading bacteria in a peatland: implications for carbon cycling. Soil Biology and Biochemistry, 37(7):1277–1287, 2005.

D. Fernández-Calviñoand E. B˚a˚ath.Growth response of the bacterial community to ph in soils differing in ph. FEMS Microbiology Ecology, 73(1):149–156, 2010. 29

N. Fierer, J.P. Schimel, and P.A. Holden. Inﬂuence of drying-rewetting frequency on soil bacterial community structure. Microbial Ecology, 45(1):63–71, 2003.

N. Fierer, C.M. McCain, P. Meir, M. Zimmermann, J.M. Rapp, M.R. Silman, and R. Knight. Microbes do not follow the elevational diversity patterns of plants and animals. Ecology, 92(4):797–804, 2011.

B.J. Finlay. Global dispersal of free-living microbial eukaryote species. Science, 296 (5570):1061–1063, 2002.

Q. Fu and S. Feng. Responses of terrestrial aridity to global warming. Journal of Geophysical Research: Atmospheres, 119(13):7863–7875, 2014.

J.A. Fuhrman, J.A. Steele, I. Hewson, M.S. Schwalbach, M.V. Brown, J.L. Green, and J.H. Brown. A latitudinal diversity gradient in planktonic marine bacteria. Proceedings of the National Academy of Sciences, 105(22):7774–7778, 2008.

G. Gionchetta, A.M. Roman´ı,F. Oliva, and J. Artigas. Distinct responses from bacterial, archaeal and fungal streambed communities to severe hydrological disturbances. Scientiﬁc reports, 9(1):1–13, 2019.

T.B. Graham and D. Wirth. Dispersal of large branchiopod cysts: potential movement by wind from potholes on the colorado plateau. Hydrobiologia, 600(1):17–27, 2008.

R.I. Griﬃths, B.C. Thomson, P. James, T. Bell, M. Bailey, and A.S. Whiteley. The bacterial biogeography of British soils. Environmental Microbiology, 13(6):1642–1654, 2011.

C.A. Hanson, J.A. Fuhrman, M.C. Horner-Devine, and J.B.H. Martiny. Beyond biogeo- graphic patterns: Processes shaping the microbial landscape. Nature Reviews Micro- biology, 10(7):497–506, 2012. doi: 10.1038/nrmicro2795.

Jonathan E. Harvey, Joel L. Pederson, and Tammy M. Rittenour. Exploring relations between arroyo cycles and canyon paleoﬂood records in Buckskin Wash, Utah: Rec- onciling scientiﬁc paradigms. Bulletin, 123(11-12):2266–2276, 2011.

H. Hillebrand. On the generality of the latitudinal diversity gradient. The American Naturalist, 163(2):192–211, 2004.

W.N. Holland. Slot Valley. Australian Geographer, 13(5):338–339, 1977. 30

J. Huang, H. Yu, X. Guan, G. Wang, and R. Guo. Accelerated dryland expansion under climate change. Nature Climate Change, 6(2):166–171, 2016.

M. Jocque, B. Vanschoenwinkel, and L. Brendonck. Freshwater rock pools : a review of habitat characteristics, faunal diversity and conservation value. Freshwater Biology, 55(8):1587–1602, 2010.

A Kaisermann, P.A. Maron, L. Beaumelle, and J.C. Lata. Fungal communities are more sensitive indicators to non-extreme soil moisture variations than bacterial communities. Applied Soil Ecology, 86:158–164, 2015.

E. Krause, A. Wichels, L. Gim´enez,M. Lunau, M.B. Schilhabel, and G. Gerdts. Small changes in ph have direct eﬀects on marine bacterial community composition: a mi- crocosm approach. PloS One, 7(10):1–12, 2012.

H.D. Kurtz Jr and D.I. Netoﬀ. Stabilization of friable sandstone surfaces in a desiccat- ing, wind-abraded environment of south-central utah by rock surface microorganisms. Journal of Arid Environments, 48(1):89–100, 2001.

S. Langenheder and H. Ragnarsson. The role of environmental and spatial factors for the composition of aquatic bacterial communities. Ecology, 88(9):2154–2161, 2007. doi: 10.1890/06-2098.1.

G. Lear, J. Bellamy, B.S. Case, J.E. Lee, and H.L. Buckley. Fine-scale spatial patterns in bacterial community composition and function within freshwater ponds. ISME Journal, 8(8):1715–1726, 2014. doi: 10.1038/ismej.2014.21.

C. Leigh and T. Datry. Drying as a primary hydrological determinant of biodiversity in river systems: A broad-scale analysis. Ecography, 40(4):487–499, 2017.

A.B.S. Limaye and M.P. Lamb. Journal of Geophysical Research : Earth Surface. Journal of Geophysical Research: Earth Surface, 119:927–950, 2014.

T.T. Lindley, J. Molinari, R.M. Shelley, and B.N. Steger. A fourth account of centipede (Chilopoda) predation on bats. Insecta Mundi, 1064:1–4, 2017.

E.S. Lindstr¨omand S. Langenheder. Local and regional factors inﬂuencing bacterial community assembly. Environmental Microbiology Reports, 4(1):1–9, 2012. 31

E.S. Lindstr¨omand O.¨ Ostman.¨ The importance of dispersal for bacterial community composition and functioning. PLoS ONE, 6(10):1–7, 2011. doi: 10.1371/journal.pone. 0025883.

K.J. Locey and J.T. Lennon. Scaling laws predict global microbial diversity. Proceedings of the National Academy of Sciences, 113(21):5970–5975, 2016.

F.T. Maestre, M. Delgado-Baquerizo, T.C. Jeﬀries, D.J. Eldridge, V. Ochoa, B. Gozalo, J.L. Quero, M. Garcia-Gomez, A. Gallardo, W. Ulrich, et al. Increasing aridity reduces soil microbial diversity and abundance in global drylands. Proceedings of the National Academy of Sciences, 112(51):15684–15689, 2015.

D.M. Mager and A.D. Thomas. Extracellular polysaccharides from cyanobacterial soil crusts: a review of their role in dryland soil processes. Journal of Arid Environments, 75(2):91–97, 2011.

S. Malouﬁ, A. Catherine, D. Mouillot, C. Louvard, A. Cout´e,C. Bernard, and M. Trous- sellier. Environmental heterogeneity among lakes promotes hyper β-diversity across phytoplankton communities. Freshwater Biology, 61(5):633–645, 2016.

S. Mandal, W. Van Treuren, R.A. White, M. Eggesbø, R. Knight, and S.D. Peddada. Analysis of composition of microbiomes: a novel method for studying microbial composition. Microbial Ecology in Health and Disease, 26(1):27663, 2015.

C.M. McCain and R.K. Colwell. Assessing the threat to montane biodiversity from discordant shifts in temperature and precipitation in a changing climate. Ecology letters, 14(12):1236–1245, 2011.

D. McDonald, M.N. Price, J. Goodrich, E.P. Nawrocki, T.Z. Desantis, A. Probst, G.L. Andersen, R. Knight, and P. Hugenholtz. An improved Greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. ISME Journal, 6(3):610–618, 2012. doi: 10.1038/ismej.2011.139.

T.A. McHugh and E. Schwartz. A watering manipulation in a semiarid grassland induced changes in fungal but not bacterial community composition. Pedobiologia, 59(3):121– 127, 2016. 32

J.A. Moss, N.L. Henriksson, J.D. Pakulski, R.A. Snyder, and W.H. Jeﬀrey. Oceanic microplankton do not adhere to the latitudinal diversity gradient. Microbial ecology, pages 1–5, 2019.

T.C. Mullet, B. Zank, F. Armstong, and C.M. Ritzi. Predicting Viola guadalupensis (Violaceae) habitat in the Guadalupe Mountains using GIS evidence of a new isolated population. Journal of the Botanical Research Institute of Texas, 2(1):677–684, 2008.

N. Musat, H. Halm, B. Winterholler, et al. A single-cell view on the ecophysiology of anaerobic phototrophic bacteria. Proceedings of the National Academy of Sciences, 105(46):17861–17866, 2008.

D.I. Netoﬀ and R.R. Shroba. Morphology and possible origin of giant weathering pits in the Entrada Sandstone, southeastern Utah; preliminary ﬁndings. Technical report, US Geological Survey, 1993.

A.E Parada, D.M. Needham, and J.A. Fuhrman. Every base matters : assessing small subunit rRNA primers for marine microbiomes with mock communities , time series and global ﬁeld samples. Environmental Microbiology, 18(5):1403–1414, 2016. doi: 10.1111/1462-2920.13023.

B. Pratt-Sitaula, M. Garde, D.W. Burbank, M. Oskin, A. Heimsath, and E. Gabet. Bedload-to-suspended load ratio and rapid bedrock incision from Himalayan landslide- dam lake record. Quaternary Research, 68(1):111–120, 2007.

PRISM Climate Group. http://prism.oregonstate.edu, 2004.

I. Reche, E. Pulido-Villena, R. Morales-Baquero, and E.O. Casamayor. Does ecosystem size determine aquatic bacterial richness? Ecology, 86(7):1715–1722, 2005.

L. Ren, E. Jeppesen, D. He, J. Wang, L. Liboriussen, P. Xing, and Q.L. Wu. ph inﬂuences the importance of niche-related and neutral processes in lacustrine bacterioplankton assembly. Applied Environmental Microbiology, 81(9):3104–3114, 2015.

J.F. Reynolds, D.M.S. Smith, E.F. Lambin, B.L. Turner, M. Mortimore, S.P.J. Bat- terbury, T.E. Downing, H. Dowlatabadi, R.J. Fern´andez,J.E. Herrick, et al. Global desertiﬁcation: building a science for dryland development. Science, 316(5826):847– 851, 2007. 33

V. Rosset, A. Ruhi, M.T. Bogan, and T. Datry. Do lentic and lotic communities respond similarly to drying? Ecosphere, 8(7):1–14, 2017.

J. Rousk, E. B˚a˚ath,P.C. Brookes, C.L. Lauber, C. Lozupone, J.G. Caporaso, R. Knight, and N. Fierer. Soil bacterial and fungal communities across a ph gradient in an arable soil. The ISME journal, 4(10):1340, 2010.

Diethard Sanders, Lukas Wischounig, Alfred Gruber, and Marc Ostermann. Inner gorge- slot canyon system produced by repeated stream incision (eastern Alps): Signiﬁcance for development of bedrock canyons. Geomorphology, 214:465–484, 2014.

R. Seager, M. Ting, I. Held, Y. Kushnir, J. Lu, G. Vecchi, H. Huang, N. Harnik, A. Leet- maa, N. Lau, et al. Model projections of an imminent transition to a more arid climate in southwestern north america. Science, 316(5828):1181–1184, 2007.

P.M. Selkirk, D.A. Adamson, and A.J. Downing. Landform and Vegetation Change in the Greaves Creek Basin : an asymmetric hanging valley in the Blue Mountains , New South Wales. Australian Geographer, 32(1):45–75, 2001.

C. Shen, W. Liang, Y. Shi, X. Lin, H. Zhang, X. Wu, G. Xie, P. Chain, P. Grogan, and H. Chu. Contrasting elevational diversity patterns between eukaryotic soil microbes and plants. Ecology, 95(11):3190–3202, 2014.

M. Soria, C. Leigh, T. Datry, L.M. Bini, and N. Bonada. Biodiversity in perennial and intermittent rivers: A meta-analysis. Oikos, 126(8):1078–1089, 2017.

J.M. Stark and M.K. Firestone. Mechanisms for soil moisture eﬀects on activity of nitrifying bacteria. Applied Environmental Microbiology, 61(1):218–221, 1995.

C. Stock, H.K. Grønlien, R.D. Allen, and Y. Naitoh. Osmoregulation in Paramecium: in situ ion gradients permit water to cascade through the cytosol to the contractile vacuole. Journal of Cell Science, 115(11):2339–2348, 2002.

A.J. Sz´ekely and S. Langenheder. The importance of species sorting diﬀers between habitat generalists and specialists in bacterial communities. FEMS microbiology ecology, 87(1):102–112, 2014.

I. Telesh, H. Schubert, and S. Skarlato. Revisiting remane’s concept: evidence for high plankton diversity and a protistan species maximum in the horohalinicum of the baltic sea. Marine Ecology Progress Series, 421:1–11, 2011. 34

I. Telesh, H. Schubert, and S. Skarlato. Life in the salinity gradient: discovering mechanisms behind a new biodiversity pattern. Estuarine, Coastal and Shelf Science, 135: 317–327, 2013.

T.W. Therriault and J. Kolasa. Desiccation frequency reduces species diversity and predictability of community structure in coastal rock pools. Israel Journal of Zoology, 47(4):477–489, 2001.

L.R. Thompson et al. A communal catalogue reveals Earth’s multiscale microbial diversity. Nature, 551(7681):457–463, 2017. doi: 10.1038/nature24621.

D. Tilman and A. El Haddi. Drought and biodiversity in grasslands. Oecologia, 89(2): 257–264, 1992.

V. Torsvik and L. Øvre˚as. Microbial diversity and function in soil: from genes to ecosystems. Current Opinion in Microbiology, 5(3):240–245, 2002.

K. Van der Gucht, K. Cottenie, K. Muylaert, N. Vloemans, S. Cousin, S. Declerck, E. Jeppesen, J.M. Conde-Porcuna, K. Schwenk, G. Zwart, H. Degans, W. Vyverman, and L. De Meester. The power of species sorting: Local factors drive bacterial community composition over a wide range of spatial scales. Proceedings of the National Academy of Sciences, 104(51):20404–20409, 2007. doi: 10.1073/pnas.0707200104.

K. Wagner, K. Besemer, N.R. Burns, T.J. Battin, and M.M. Bengtsson. Light avail- ability aﬀects stream bioﬁlm bacterial community composition and function, but not diversity. Environmental Microbiology, 17(12):5036–5047, 2015.

J. Wang, D. Yang, Y. Zhang, J. Shen, C. Van Der Gast, M.W. Hahn, and Q. Wu. Do patterns of bacterial diversity along salinity gradients diﬀer from those observed for macroorganisms? PloS one, 6(11), 2011.

J. Wang, J. Shen, Y. Wu, C. Tu, J. Soininen, J.C. Stegen, J. He, X. Liu, L. Zhang, and E. Zhang. Phylogenetic beta diversity in bacterial assemblages across ecosystems: Deterministic versus stochastic processes. ISME Journal, 7(7):1310–1321, 2013a. doi: 10.1038/ismej.2013.30. URL http://dx.doi.org/10.1038/ismej.2013.30.

S. Wang, W. Hou, H. Dong, H. Jiang, L. Huang, G. Wu, C. Zhang, Z. Song, Y. Zhang, H. Ren, et al. Control of temperature on microbial community structure in hot springs of the Tibetan Plateau. PLoS One, 8(5):1–14, 2013b. 35

W.B. Whitman, D.C. Coleman, and W.J. Wiebe. Prokaryotes: the unseen majority. Proceedings of the National Academy of Sciences, 95(12):6578–6583, 1998.

C.J. Willmot and K. Matsuura. Terrestrial Air Temperature and Precipitation: Monthly Annual Time Series (1950-1999). 2001. URL http://climate.geog.udel.edu/

~climate/html_pages/README.ghcn_ts2.html.

H. Yan, C. Liang, Z. Li, Z. Liu, B. Miao, C. He, and L. Sheng. Impact of precipitation patterns on biomass and species richness of annuals in a dry steppe. PLoS One, 10 (4), 2015.

A.C. Yannarell and E.W. Triplett. Geographic and environmental sources of variation in lake bacterial community composition. Applied Environmental Microbiology, 71(1): 227–239, 2005.