Consequences of Environmental Variation for Fish and Their Skin Associated Microbial Communities

(1)

Linnaeus University Dissertations No 417/2021

Hanna Berggren

Consequences of Environmental

Variation for Fish and Their Skin

Associated Microbial Communities

linnaeus university press

Lnu.se

isbn: 978-91-89283-84-8 (print), 978-91-89283-85-5 (pdf)

List of papers:

I. Berggren, H., Nordahl, O., Tibblin, P., Larsson, P., Forsman, A. (2016) “Testing for local adaptation to spawning habitat in sympatric subpopulations of pike by reciprocal translocation of embryos” PLoS ONE 11:5.

II. Nordahl, O., Tibblin, P., Koch-Schmidt, P., Berggren, H., Larsson, P., Forsman, A. (2018) “Sun basking fish benefit from body temperatures in excess of ambient water” Proceedings of the Royal Society B 285:1879.

III. Berggren, H., Tibblin, P., Yıldırım, Y., Broman, E., Larsson, P., Lundin, D., Forsman, A. “Fish skin microbiomes are highly variable among individuals and populations but not within individuals” Manuscript.

IV. Berggren, H., Yıldırım, Y., Nordahl, O., Larsson, P., Dopson, M., Tibblin, P., Lundin, D., Pinhassi, J., Forsman, A., “Ecological filtering drives rapid spatiotemporal dynamics in fish skin microbiomes” Submitted manuscript. V. Berggren, H., Nordahl, O., Yıldırım, Y., Larsson, P., Tibblin, P., Forsman,

A., “Effects of environmental translocation and host characteristics on skin associated microbiomes of sun-basking fish” Submitted manuscript.

Co nseq uen ce s o f En vir onment al Va ria tio n f or F ish and

Their Skin Asso

cia ted Mi cr ob ia l Co mm unit ie s Ha nn a Ber gg ren

(2)

Consequences of Environmental Variation

for Fish and Their Skin Associated Microbial

(3)

(4)

Linnaeus University Dissertations

No 417/2021

_C

_{ONSEQUENCES OF}

_E

_NVIRONMENTAL

V

ARIATION FOR

F

ISH AND

T

HEIR

S

KIN

A

SSOCIATED

M

ICROBIAL

C

OMMUNITIES

H

ANNA

B

ERGGREN

(5)

Consequences of Environmental Variation for Fish and Their Skin Associated Microbial Communities

Doctoral Dissertation, Department of Biology and Environmental Science, Linnaeus University, Kalmar, 2021

ISBN: 978-91-89283-84-8 (print), 978-91-89283-85-5 (pdf) Published by: Linnaeus University Press, 351 95 Växjö

(6)

Abstract

Environmental conditions that vary in space and time influence the distribution, abundance, diversity and evolution of individuals, populations, species and communities. This thesis explores how environmental variation affects diversity at different levels of biological organization, and across a wide range of spatiotemporal scales, by studying fish and their associated microbiomes. The specific aims were to investigate i) effects of coarse- and fine-scale environmental variation for the performance of fish populations and individuals, and ii) ecological drivers impacting the structure and dynamics of microbial communities associated with fish hosts.

For the first aim, I studied effects of environmental variation both within and between local habitats, by comparing populations of spawning migrating pike and monitor sun-basking behaviour of carp individuals. Results revealed that natal spawning site fidelity can promote evolution of local adaptations and population differentiation on relatively fine spatial scales in relation to the species dispersal capacity. I also demonstrated that fish can actively thermo-regulate and attain body temperatures in excess of the surrounding water by sun-basking, and that this translates into faster growth. Homing and sun-basking behaviour thus are important drivers of phenotypic diversity among and within populations and can also - as it turned out - influence the microbial communities associated with fish skin.

For the second aim, I used a mixture of observational and experimental approaches to characterize and identify sources of variation in microbial communities associated with fish skin of perch, roach and carp. An important finding was that fish skin microbiomes are highly dynamic biodiversity hotspots. Results further suggested that variation in the assembly, composition, spatial structure, and temporal shifts of these microbiomes are influenced by stochastic events in combination with ecological filtering imposed by environment and host phenotype, most notably behaviour.

A key conclusion that emerges from this thesis is that diversity at one level of biological organisation seems to support and increase diversity at a higher hierarchical level of organisation. My thesis thus adds to the knowledge, and contribute new understanding and insight into, how environmental heterogeneity and the complex interplay between different species and hierarchical levels generate and maintain biodiversity.

Keywords: biodiversity, community ecology, Cyprinus carpio, environmental

heterogeneity, Esox lucius, Perca fluviatilis, phenotypic flexibility, Rutilus

(7)

(8)

Konsekvenser av miljömässig variation för fisk och deras mikrobiom Sammanfattning

Organismers livsmiljö innefattar samverkande icke-biologiska faktorer som klimat men även biologiska faktorer som samspel mellan individer och arter. Miljömässig variation kan vara tydlig när man jämför olika geografiska platser, men den finns även i mindre skala inom habitat. Dessutom förändras miljöbetingelserna ständigt över tid. Sammantaget får det här konsekvenser för utbredning, abundans, diversitet och evolution av individer, populationer, arter och samhällen. Detta innebär att miljön bidrar till, och är en del av, biodiversiteten på vår jord. Denna avhandling syftar till att utforska hur rumslig och tidsmässig miljövariation påverkar diversitet på olika nivåer av biologisk organisation genom att studera fisk, och de mikrobiella samhällen (mikrobiom) som florerar i slemmet på fiskarna. Det övergripande syftet var att studera hur miljömässig variation påverkar i) framgången för populationer och individer av fisk, samt ii) artrikedom, artsammansättning och dynamik i fiskars yttre mikrobiom.

För att undersöka hur nyttjandet av olika lekhabitat påverkade genetisk och fenotypisk variation inom och mellan två fiskpopulationer, genomförde jag ett kläckningsförsök där jag lät befruktade ägg utvecklas både i sin naturliga miljö och i ett främmande habitat. Jag jämförde därefter kläckningsgrad och ynglens överlevnad men även honornas investering i reproduktionen, d.v.s. mängden och storleken på romkornen, från de två populationerna. Det visade sig att dessa populationer hade olika reproduktionsstrategier, och resultaten från fältexperimentet indikerade att det faktiskt verkar röra sig om lokala anpassningar till respektive lekområde dit de vuxna individerna årligen återvänder för att leka (hemvändarbeteende). Sådan kunskap har betydelse för bevarandet av biologisk mångfald och restaurering av livsmiljöer. Vi vet nämligen att genetisk och fenotypisk variation bidrar till att arter kan klara störningar och förändringar i miljön bättre, samt att de lättare kan etablera sig i nya livsmiljöer.

Vidare studerade jag miljömässig variation inom habitat med fokus på temperatur - en avgörande miljövariabel som påverkar fiskars beteende och kroppsliga funktioner. Temperatur kan påverka utbredningen av individer inom ett habitat genom att de beteendemässigt uppsöker varmare eller kallare områden utifrån behov. Tidigare har man trott att fiskars kroppstemperatur är helt beroende av temperaturen i vattnet, och att vattnets kylande effekt innebar att det var omöjligt för fiskar att höja sin kroppstemperatur genom att solbada. Det visade sig dock att solning kunde bidra till att öka fiskens kroppstemperatur i förhållande till vattnet, och att de individer som höjde sin kroppstemperatur mest under solning också växte snabbare. Detta innebär att solningsbeteende,

(9)

förutom att det kan påverka utbredning av individer, populationer och arter, kan ha stor betydelse genom att bidra till variation i fiskars framgång.

Så långt visar denna avhandling att hemvändar- och solningsbeteende kan bidra till, och upprätthålla, variation inom arter genom att populationer och individer utsätts för olika miljöförhållanden. Mikrober som lever i fiskens slemskikt påverkas också av dessa miljöfaktorer, och därtill av miljöbetingelser som har sitt ursprung i fiskens inneboende fenotypiska egenskaper, t.ex., dess immunförsvar. Hur påverkar då fiskens livsmiljö, fenotyp och habitatval dess mikrobiom?

För att ta reda på vad som reglerar fiskars mikrobiom tog jag prover på fiskar från olika populationer, och jämförde mikrobiomens artsammansättning mellan populationer, individer och kroppsdelar. Det visade sig att variationen i både artrikedom och artsammansättning var stor, särskilt mellan fiskindivider. För att ytterligare utröna om detta berodde främst på den yttre miljön (d.v.s. fiskens livshabitat) eller fiskens inneboende egenskaper, utförde jag ett experiment där fiskar förflyttades mellan olika miljöer. Experimentet avslöjade att den yttre miljön var en mycket stark drivkraft och även om det fanns skillnader mellan olika individers mikrobiom var det inte lika tydligt under kontrollerade förhållanden som för vildfångade fiskar. Jag undersökte därför om olikheter i mikrobiomens artsammansättning samvarierade med fiskarnas fenotypiska egenskaper och beteenden, som färg, tillväxt, kön, solning och aktivitetsmönster. Resultatet tydde på att fiskarnas tillväxt och beteenden kunde förklara en del av variationen i mikrobiom mellan individer. Sammantaget visar mina resultat att mikrobiella samhällen i fiskars slemskikt hyser en enorm diversitet och kännetecknas av en snabb omsättning av arter. Denna mångfald påverkas av en kombination av slumpmässiga och selektiva processer orsakade av den omgivande miljön samt fiskens egenskaper – framför allt dess beteende och val av habitat. Framtida studier bör undersöka om mikrobiomens funktion skiljer sig åt lika mycket som dess artsammansättning, men också om och hur mikrobiomens artrikedom och artsammansättning påverkar värdens välbefinnande och fitness.

Sammanfattningsvis illustrerar denna avhandling hur variation på en lägre nivå av biologisk organisation, såsom inomartsvariation mellan individer och populationer av fisk, kan bidra till variationen högre upp i hierarkin genom att påverka artrikedom och struktur i mikrobiella samhällen. Således bidrar min avhandling till kunskapen om hur det komplexa samspelet mellan arter och hierarkiska nivåer medverkar till och upprätthåller naturens biodiversitet.

(10)

List of papers

This thesis is based on the following papers, referred to in the text by their Roman numbers. The published papers were reprinted according to the Creative Commons Attribution License (CC BY 4.0).

I. Berggren, H., Nordahl, O., Tibblin, P., Larsson, P., Forsman, A.

(2016) “Testing for local adaptation to spawning habitat in sympatric subpopulations of pike by reciprocal translocation of embryos” PLoS ONE 11:5.

II. Nordahl, O., Tibblin, P., Koch-Schmidt, P., Berggren, H., Larsson, P., Forsman, A. (2018) “Sun basking fish benefit from body temperatures in excess of ambient water” Proceedings of

the Royal Society B 285:1879.

III. Berggren, H., Tibblin, P., Yıldırım, Y., Broman, E., Larsson, P.,

Lundin, D., Forsman, A. “Fish skin microbiomes are highly variable among individuals and populations but not within individuals” Manuscript.

IV. Berggren, H., Yıldırım, Y., Nordahl, O., Larsson, P., Dopson,

M., Tibblin, P., Lundin, D., Pinhassi, J., Forsman, A., “Ecological filtering drives rapid spatiotemporal dynamics in fish skin

microbiomes” Submitted manuscript.

V. Berggren, H., Nordahl, O., Yıldırım, Y., Larsson, P., Tibblin, P.,

Forsman, A., “Effects of environmental translocation and host characteristics on skin associated microbiomes of sun-basking fish” Submitted manuscript.

(11)

(12)

“What I love about science is that as you learn, you

don’t really get answers. You just get better

questions.”

(13)

(14)

(15)

(16)

Introduction

To understand how and why biodiversity change through space and time are central aims in ecology, and environmental variation is a fundamental driver of genetic and phenotypic variation among individuals, populations, species and communities (Levins 1968, Bell 2010, Kraft et al. 2011). Environmental conditions typically vary among, but also over time within, habitats (Bell 2010), resulting in spatiotemporally varying selective pressures. Such environmental heterogeneity has the potential to induce evolutionary modifications of populations (i.e., change the frequency of different alleles) and affect the composition of communities (i.e., change the frequency of different species).

Besides selection, movement and dispersal is of interest because of the different effects they can have on the structure of populations and communities. In principle, random dispersal can increase diversity within and reduce differences between geographically adjacent populations (Wright 1943), whereas genotype or phenotype dependent dispersal can instead contribute to population differentiation (Edelaar et al. 2008, Shine et al. 2011a, Berggren et al. 2012, Edelaar and Bolnick 2012). Similarly, high dispersal can potentially have a homogenizing effect on the species composition of communities (Leibold et al. 2004, Miller et al. 2018), whereas species-specific requirements and ecological filtering (i.e., species sorting) may result in differential colonization and establishment (Sloan et al. 2006). Moreover, most animals have the ability to adjust their behaviour and move between habitat patches to reside in environments that meet their requirements (i.e. phenotypic flexibility). This can lead to differential environmental exposure and ecology of individuals even within populations at a fine spatial scale (Shine et al. 2003, Svanbäck and Eklöv 2003, Bolnick et al. 2007).

This thesis explores patterns and ecological drivers of diversity across different spatiotemporal scales, and among levels of biological organization. First, I demonstrate that populations can differ in fitness related traits (connected to reproduction) on a relatively fine spatial and temporal scale (spawning period) due to homing behaviour (Paper I). I also show that variation in micro-habitat utilization among individuals mediated by sun-basking

(19)

4

behaviour affects fitness in terms of growth (Paper II). Next, I explore patterns of diversity in species richness and community composition of microbial communities associated with fish hosts (Paper III), and investigate whether and how these patterns are affected by external environmental variation, inter-host dispersal, or phenotypic dimensions of the host (Papers IV & V). The results show that these microbial communities are highly diverse and dynamic within and among individual fish hosts (Papers III-V). It is also evident that they quickly respond to changes in the external environment (Papers IV & V). Collectively, the findings indicate that these microbial communities are governed by interacting effects of the external environment, host-specific factors (i.e., ecological filtering) and stochastic events.

Taken together, this thesis has investigated the consequences of environmental variation at several levels of biological organization, and across a wide range of spatiotemporal scales. It spans from how weekly to yearly (among spawning events) variation in fish behaviour affect the performance of individuals and the evolution of populations, and finally how this variation, from coarse to fine spatiotemporal scales can translate to higher hierarchical levels (i.e., communities) and contribute to the diversity and dynamics of fish associated microbial communities (Figure 1).

Figure 1. Hierarchical levels of biological organization and the fundamental impact of extrinsic and intrinsic environmental variation - a schematic summary of the studies in

this thesis. Environmental variation can impact the performance, distribution, dynamics and interactions among individuals, populations and communities (Paper I-V). Host associated microbial communities varied between populations and individuals (Paper III-V). Environmental variation and host behaviour jointly affect the composition and dynamics of the microbial communities associated with fish (Paper IV-V).

(20)

Environmental variation – a fundamental driver of

biodiversity

The environmental conditions an organism experience comprises the interaction between abiotic and biotic factors, and it varies in both space and time. For instance, the environment typically varies along geographic and altitudinal gradients, and it can fluctuate over time within habitats leading to more or less unpredictable conditions (e.g., daily and seasonal patterns versus extreme weather events). The ever changing environment is the fundamental phenomenon that all organisms need to cope with in order to persist and spread their genes (Darwin 1859). This means the environment imposes natural selection on individuals – but also selection on species within a community by shaping their distribution and abundance (i.e., ecological filtering). Accordingly, environmental conditions influence distributional patterns of species, populations and individuals (Levins 1968, Hooper et al. 2005, Bell 2010). This also entails that environmental heterogeneity has the potential to influence the processes that shape biodiversity (genotypic and phenotypic variation). Studies of consequences of environmental variation are warranted more than ever due to the anthropogenic pressure on the world’s ecosystems, for example climate change, habitat fragmentations and exploitation of natural resources.

Organisms that face variation or changes in environmental conditions may in principle adapt, disperse, enter dormancy, or go extinct, and the various ways of coping depend partly on their biology. Environmental heterogeneity is perceived differently depending on what organism that is under study because the consequences of temporal variation are likely to differ depending on longevity, and the responses to spatial variation are likely to differ depending on dispersal capability (Levins 1968). Larger organisms (e.g., fish) can show signs of population structure on relatively fine spatial scales (i.e., without physical dispersal boundaries), due to, for example, variation among individuals in habitat preferences or sorting according to dispersal capacity (Wilson and Dugatkin 1997, Bolnick et al. 2003, Austin et al. 2004, Helfman et al. 2009, Shine et al. 2011b, a). In contrast, for microbial organisms, the distribution, abundance, and interactions among species and communities, can change within a centimetre (Petro et al. 2019). The appropriate spatial and temporal scale for studies of consequences of environmental variation and heterogeneity will therefore differ among organisms. The scale considered for diversity patterns is important in ecological studies also because the processes involved can switch among hierarchical levels (e.g., local/regional patterns, individual/population) (Levin 1992, Rosenzweig 1995, Leibold et al. 2004). Using fish and their associated microbial communities as a model system allows for studying consequences of environmental variation from different perspectives: i)

(21)

6

different hierarchical levels of biological organization (from fish individuals to microbial communities) and ii) spatiotemporal environmental variation.

Phenotypic variation within and among

populations and individuals

Understanding causes and consequences of intraspecific variation is important because such variation can help populations and species to withstand environmental disturbances and promote establishment success – and these traits are of vital importance for the conservation of biodiversity (Forsman 2014, Des Roches et al. 2018). Fish display remarkable variation in ecology, morphology and behaviour (Helfman et al. 2009). To this date, more than 15,000 species of bony fish (class Actinopterygii), that inhabit marine, brackish and freshwater environments, have been described (https://obis.org/). Fish therefore offer good opportunities to study phenotypic variation within and among populations.

Phenotypic variation may reflect genetically based differences, result from environmentally induced plasticity, or be a combination of the two (Stearns 1989, West-Eberhard 2003, Forsman 2015). Genetically based phenotypic variation makes up the raw material for selection to act upon, for example enabling the evolution of adaptations to local environments among populations. In theory, divergence among populations may ultimately lead to speciation (Schluter 2001). However, there are many factors involved in population divergence that jointly influence whether ongoing structuring will proceed or cease, for example gene flow, genetic drift, and whether the environmental conditions are stable or fluctuating (Schluter 2001, Kawecki and Ebert 2004). Population structures typically arise when individuals are isolated due to dispersal barriers, but structuring can also emerge when gene flow is possible but not realized for some reason. For instance, genotype- and phenotype-dependent habitat preferences or dispersal capacity can contribute to structuring among populations via matching habitat choice or spatial sorting (Edelaar et al. 2008, Shine et al. 2011a, Berggren et al. 2012, Calboli et al. 2016). However, population divergence in cases where individuals co-occur throughout the majority of their life cycle and only separate briefly during breeding has rarely been studied, and the processes involved in divergence among sympatric populations remain largely unknown (Schluter 2001). There is thus a need for a better understanding of how phenotypic diversity emerges and is retained in such systems, especially at relatively fine spatial scales (Wennerström et al. 2013).

Intraspecific variation exists to varying degrees within populations due to genetics, developmental plasticity of irreversible traits, and intra-individual phenotypic flexibility of reversible traits (Bradshaw 1965, Pigllucci 1996, Bolnick et al. 2003, West-Eberhard 2003, Forsman 2015). This can have

(22)

consequences for the success of both populations and individuals. For example, differences in the use of microhabitats and resources among individuals can reduce competition, broaden the niche of the populations (and species), and affect their dynamics and spatiotemporal distributions (Van Valen 1965, Bolnick et al. 2003, Svanbäck and Bolnick 2007, Forsman 2015). Individual-based studies have the ability to uncover how variation in phenotypic characteristics and behaviours affect performance and fitness-related traits (e.g., growth, survival) (Endler and May 1986).

Studies of population structure, local adaptations, and the consequences of phenotypic variation and flexibility for the success of individuals and populations, can benefit our understanding of genetic and phenotypic diversity among individuals, populations and species. It can further help explain their distributional patterns in space and time. However, the implications go beyond the species level. Large, long-lived and mobile fish harbour communities of small and short-lived microbes, such that one can study whether and how the consequences of intraspecific variation affect yet another level of biological organization. Host individuals can be regarded as islands with different properties according to intrinsic (e.g., host genetic and phenotypic variation) and extrinsic factors (e.g., external environmental conditions in the habitat), and their associated microbiomes are exposed to constantly changing, and possibly contrasting, environmental conditions. What are the consequences of the environmental variation induced by fish for their associated microbiomes?

Community assembly, composition and dynamics

To identify the assembly rules that impact the species composition of communities has been a longstanding challenge in ecology (Rosenzweig 1995, Keddy 2005, Kraft et al. 2007, Vellend 2010). Community ecology seeks to explain changes in abundance and distribution of species through space and time. Microbial communities seem to be able to colonize any imaginable niche (Bohannan and Hughes 2003, Lopez-Garcia and Moreira 2008) and have important functional roles within biogeochemical and ecological processes in both terrestrial and aquatic environments (Horner-Devine et al. 2004, Prosser et al. 2007). Microbial diversity is of practical importance for humans within several areas, such as medicine, agriculture and industry (Horner-Devine and Bohannan 2006). Microbial organisms are also involved in many of the environmental processes that sustain life on Earth, most notably nutrient recycling (Horner-Devine et al. 2004). However, microbes can also be detrimental to other organisms. They act for example as pathogens on individuals, and on larger scales they can be harmful to whole ecosystems, for instance colony-forming cyanobacteria causing toxic blooms. That microbes comprise the majority of the species on earth, that they are of conisderable ecological importance, and that the regulation of microbial populations is not

(23)

8

yet fully understood together emphasize the need to better understand the processes, patterns and consequences of their diversity (Horner-Devine et al. 2004, Prosser et al. 2007, Lopez-Garcia and Moreira 2008, Nemergut et al. 2013).

Animals are inhabited by different species of prokaryotes (i.e., bacteria and archaea), viruses and eukaryotic microorganisms that are either commensals, symbionts or opportunistic pathogens, which are collectively called the

microbiota (defined by marker genes e.g., 16S or 18S) or microbiome (including

genomic and environmental information) (Larsen and Arias 2014, Marchesi and Ravel 2015). Host-associated microbiomes have received considerable scientific attention since the beginning of the 21st century (Figure 2) and their

Figure 2. Publications per year on skin and gut microbiomes in aquatic and terrestrial habitats. Since the beginning of the 21st century, the number of studies on skin- and gut

microbiomes has vastly increased. However, the number of studies focusing on skin (dashed lines) is far less in comparisons with gut (solid lines) microbiome studies – in both aquatic (grey) and terrestrial (black) organisms.

ecological role is now undisputable (Vorburger and Perlman 2018, Compant et al. 2019, Moran et al. 2019, Davidson et al. 2020, McLaren and Callahan 2020, Popkes and Valenzano 2020). The gut microbiome of fish hosts is crucial for the development of gastrointestinal functioning and the immune system (Perez et al. 2010), and the microbiomes associated with skin forms a protective barrier together with the mucosal surfaces (Shephard 1994, Tort et al. 2003, Larsen and Arias 2014). Hence, both external (skin) and internal (gut) microbiomes seem to be of great significance for wellbeing and fitness. While gut microbiomes are well studied, factors that shape the skin-associated microbiome composition are less understood (Figure 2). For instance, little is known about temporal

0 250 500 750 1000 1925 1950 1975 2000 Publication Year

Absolute research output

(24)

dynamics of skin microbiome composition within individual hosts, the rate at which microbiomes respond to environmental shifts, the roles of dispersal between hosts, and whether they are resilient to disturbances (Costello et al. 2009, McKenzie et al. 2012, Shade et al. 2012, Ross et al. 2019). Microbial communities associated with fish skin are exposed to a range of factors that influence their assembly and dynamics. What happens when host-associated microbial communities are exposed to environmental changes – do the governing processes change? The role of ecological filtering in community ecology is a puzzling area.

To simplify, there are two opposing theories of community assembly, as derived from macro-ecology. According to the unified neutral theory of biodiversity, many species occupy comparable niches such that they are more or less exchangeable, and species composition is primarily influenced by stochastic events (Hubbell 2001). If random processes dominate the assembly of host-associated microbiomes, one would expect their community compositions to vary greatly among hosts (Burns et al. 2016), and diversity patterns of host associated microbial communities that depend on host species or population affinity would be negligible.

In contrast, the competition theory posits that species composition is governed by deterministic processes, and strongly influenced by differences among species in niche utilization, capabilities and tolerances (Diamond 1978). This is based on the notion that two species utilizing the exact same realized niche cannot coexist, and if they do, selection for reduced competition would result in modifications of resource utilization and reduced niche overlap (e.g., Diamond 1978, Philippot et al. 2010, Koskella et al. 2017). In the context of host microbiomes, this assembly process would lead to convergence in microbiome community composition among hosts exposed to similar selection/filtering. Accordingly, one can hypothesize that microbiome community composition should vary according to host species, and perhaps even to populations and individuals. Under what conditions neutral or deterministic assembly processes dominate (and how to properly uncover it) is still an open question in microbial ecology (Langenheder and Szekely 2011, Stegen et al. 2012, Stegen et al. 2013).

Theories originating from population and community ecology can be applied to formulate hypotheses about the relative impact of host intrinsic and extrinsic factors for the assembly, composition and dynamics of microbiomes. Leibold et al. (2004) suggested that species interactions occurred on a larger scale than in the local community, and proposed that a meta-population view should be considered also in community ecology (i.e., metacommunity). This idea has been proposed previously by Wilson (1992) who suggested that a set of local communities in a mosaic of patches – metacommunities – are connected by complex interactions that introduce a new source of variation at the patch level. Recently, metacommunity theory has been proposed to aid in understanding the

(25)

10

distribution and dynamics of microbial communities associated with hosts (Miller et al. 2018). If fish hosts are regarded as islands, or their microbiomes as metacommunities, this might benefit the understanding of how host-associated microbial communities are regulated (Vellend 2010, Miller et al. 2018). Using both fish and their associated microbiomes as study system thus offers a unique opportunity to study and understand how different processes contribute to spatiotemporal variation in biodiversity at different levels of organization.

(26)

Aims and questions addressed

The studies in this thesis have investigated the impact of environmental variation on diversity at different levels of biological organization.

Effects of coarse- and fine-scale environmental variation on the performance of fish populations and individuals

¨ Test for local adaptation to spawning habitat in reproductive traits in two fish (Esox lucius) populations exhibiting spawning site fidelity to different breeding habitats (Paper I).

¨ Study effects of individual variation in sun-basking behaviour on growth in carp (Cyprinus carpio) (Paper II).

Spatiotemporal dynamics and ecological drivers of microbial communities associated with fish hosts

¨ Investigate how host (Perca fluviatilis) body site, individual and population contribute to variation in fish skin microbiome composition (Paper III).

¨ Examine how assembly processes, environmental conditions, host (Rutilus rutilus) identity and connectivity influences the composition and dynamics of fish skin microbiomes (Paper IV).

¨ Test for associations of host (C. carpio) phenotypic dimensions (sex, colour, growth, and phenotypic flexibility) with variation in skin associated microbiome composition among individuals (Paper V).

(27)

12

Methods – an overview

This thesis includes both observational and experimental studies on macro- (fish) and microorganisms. Accordingly, a wide range of methods have been used and only the most important are briefly presented here, for more details the reader is kindly referred to each specific paper. Furthermore, several fish species have been used in the different papers and their different ecology are just briefly presented here. All studies included in this thesis were approved by the Ethical Committee on Animal Research in Linköping, Sweden (see individual papers for the relevant approval numbers).

Studied fish species

Pike (Esox lucius), is a long-lived, iteroparous, large keystone predator fish species that is widespread in the northern hemisphere (Craig 1996, 2008) and inhabits both fresh and brackish water environments. The pike populations included in Paper I are genetically distinct subpopulations that spawn in different freshwater wetlands separated by a short geographic distance (~10 km) relative to the dispersal capacity of pike (Larsson et al. 2015, Tibblin et al. 2015, Nordahl et al. 2019, Sunde et al. 2020). Individuals spend only a few weeks in the freshwater spawning habitats during the reproductive and larval periods; hence, subpopulations are only physically separated for a fraction of the life cycle (Engstedt et al. 2010, Nilsson et al. 2014, Larsson et al. 2015). This study system thus offers good opportunities to study divergence and local adaptation in sympatric subpopulations that only encounter different environmental conditions and divergent selection for a short period.

Carp (Cyprinus carpio), is a benthic omnivorous cyprinid species often found in shallow, warm and eutrophic waters. It is distributed around the globe, but its native range include east Europe and Asia (Zambrano et al. 2006). Carp is a temperature generalist, but thrive in warm waters (Réalis-Doyelle et al. 2018). Accordingly, carp should seek out warmer water (micro-)habitats to benefit bodily functions when water temperature is below optimum (Reynolds and

(28)

Casterlin 1980, Forsman 2000). Moreover, the colour variation within the species (Balon 2004) offers an opportunity to investigate the effect of colouration on heat gain, which is an important factor in terrestrial environments (Clusella Trullas et al. 2007). These traits make carp an excellent species for studying sun-basking behaviour, and the effects of host characteristics for microbiome variation (Papers II & V).

Perch (Perca fluviatilis), is a medium-sized predatory and iteroparous fish species widely distributed in fresh- and brackish waters in the northern hemisphere. In Paper III, we sampled microbiomes of perch at two locations along the southeast Baltic coast of Sweden (Kalmar and Figeholm) separated by a swimming distance of approximately 80 km which exceeds the general dispersal pattern and home ranges (~20 km) of Baltic Sea perch (Craig 2000, Ahlbeck Bergendahl et al. 2017, Hansson et al. 2019). Thus, the two locations likely harbour distinct populations with non-overlapping home ranges, which is also supported by microsatellite data of perch from different regions of the Baltic Sea, including our study area, that show genetic clusters at a finer spatial resolution than our two locations (Bergek and Björklund 2009, Olsson et al. 2011). Therefore, perch are well suited for studying potential population structure in microbiome composition.

Roach (Rutilus rutilus), is a small (15-30 cm) cyprinid species geographically distributed all over Europe (except the Mediterranean area) and parts of Asia. The distribution range includes freshwater habitats but also brackish environments like the Baltic Sea. Roach is an omnivore that feed on zooplankton, benthos, detritus, and algae. Here (Paper IV), we have studied a Baltic Sea (southeast Sweden) roach population that migrates from foraging grounds in a coastal brackish environment to spawn in the freshwater stream Oknebäcken and adjacent inundated floodplains. The ecology of this species thus offers opportunity to study drivers of spatiotemporal variation in microbiome composition.

Field studies and experiments

Field manipulation experiments combine the strengths of studying the organisms in their natural environment with the benefits of experimental approaches that allow for inferring causality, and potential effects caused by interacting factors are more likely to be discovered under natural settings (Forsman 2014, Voelkl et al. 2020). The overall aim of this thesis was to assess the impact of natural environmental variation on diversity at different levels of biological organization. Therefore, all included studies involved natural settings in the field, combining observational and experimental studies with longitudinal sampling of individuals, to gain in-depth understanding of the processes shaping diversity at different organizational levels in situ.

(29)

14

Figure 3. Schematic pictures of experimental set-ups in the field. a) In Paper I we conducted a reciprocal translocation experiment with 2500 pike embryos to test for local adaptation to spawning habitat. We used a split-brood approach. b) In Paper IV we conducted an experiment on 80 wild-caught roach individuals that involved several manipulations such as rebooting the microbiome by disinfectant (indicated by T, controls with C), social or solitary housing, and environmental translocations. Microbiomes were sampled weekly (week 0-3), thus comprising a three week time-series with four sampling occasions.

(30)

Manipulation studies

We used environmental translocation experiments to infer how individuals,

populations and communities responded to contrasting environmental conditions (Papers I, IV & V) (Figure 3).

In Paper IV, we disinfected the skin of roach with benzalkonium chloride (henceforth BKC) to reboot the microbial communities in order to study the

assembly processes involved in subsequent colonization and succession. BKC

is a disinfecting agent that lyses bacterial cells without harming the host (Anderson and Conroy 1969, Bullock and Conroy 1971). To evaluate the effect of bathing fish in BKC on viable bacteria, a laboratory experiment was performed on roach using a culture dependent technique to evaluate the effect of BKC in viable cells. Results showed that a 10 min bath in 1% solution of BKC reduced the number of colony-forming units with an average of 96% (median = 97%, range = 85-99%). For more details about this procedure, see Supplementary information in Paper IV.

We also manipulated the social settings for fish hosts in Paper IV to study

the role of inter-host dispersal for microbiome richness and composition.

Observational studies

We performed longitudinal sampling of the same individuals to evaluate how

different host genotypes interact with environmental conditions, uncover spatiotemporal patterns in distribution of individuals, and to study the effects of intrinsic versus extrinsic environment on microbiome richness and composition

(Papers I, II, IV & V).

In Paper III, we studied general patterns of microbial community diversity

on different spatial scales by conducting repeated sampling of the same

individuals to evaluate measurement repeatability (see below).

We used biologgers (i.e., data storage tags) to gain detailed information

about natural behaviours, activity and movement patterns of fish (Papers II &

V). The biologgers were equipped with two temperature sensors to measure

both the internal body temperature of fish and the external temperature of the ambient water. The biologgers were also equipped with a pressure sensor that enabled us to monitor the depth (vertical migration) of the fish. This method provided high resolution data by measuring temperature and pressure every 5th minute, but also demands the recapture of individuals to acquire the data.

Sampling of microbial communities

The microbial sampling procedures of fish and water respectively, were similar across all microbiome studies (Papers III-V). Before sampling of the microbiome, touching of the dorsal areas of fish was strictly avoided, each fish was rinsed with MilliQ-water, and subsequently sampled on a 2´2 cm skin area with a sterile cotton swab twirled four times. The swabs were put in separate

(31)

16

Eppendorf tubes (1.5 ml) with 750 µl TE-buffer (Tris-EDTA, 10:1), which were stored on ice until sampling was completed and after returning to the lab, the samples were subsequently stored in a -80° C freezer. To avoid cross-contamination of microbiome samples from different fish we changed gloves and sterilized all equipment with 70% ethanol between each fish.

At each sampling occasion and location, surface water samples (0.5-1 L) were taken to enable comparisons of the microbial communities on the fish skin with those present in the water. Water samples were vacuum-filtered through a Supor® membrane filter (Pall Corporation, pore size 0.22 µm, Ø 47 mm). The filter was transferred to a sterile tube containing 1.8 mL TE-buffer (Tris-EDTA, 10:1) and stored at -80°C until DNA extraction.

Characterization of microbial communities

To characterize the microbial communities on fish skin and in the surrounding water (Papers III-V), we used amplicon sequencing with primer pair 341f-805r targeting V3-V4 region on the bacterial 16S rRNA gene (Herlemann et al. 2011, Hugerth et al. 2014). This region has been shown to detect microbial diversity comparable with that of the whole 16S region (Youssef et al. 2009, Robinson et al. 2010).

Samples were sequenced on the Illumina MiSeq platform (Illumina, USA) with 2×300 paired-end settings at Science for Life Laboratory (SciLifeLab, Stockholm, SWEDEN). Data from each sequencing run were processed independently with the DADA2 package (Callahan et al. 2017) implemented in QIIME2 (version 2018.8) (Bolyen et al. 2019), using the default settings except for the parent over abundance parameter that was set to 4. After correcting for Illumina sequencing errors, DADA2 produces sequences with single-nucleotide resolution called ‘amplicon sequence variants’ (henceforth ASVs).

Taxonomic composition was assigned using a naïve Bayesian classifier trained on the V3-V4 region of the 16S rRNA gene with reference sequences from the SILVA database (SILVA 132; (Quast et al. 2013) in QIIME2 (version 2019.10).

Measurement repeatability of microbiomes

When studying fish skin microbiome with molecular methods, biases are generated in every possible step such as DNA extractions and protocols for PCR, for example annealing temperature, and how many cycles to mention a few. We therefore evaluated measurement repeatability of replicated samples from the same individuals (Paper III) to investigate the reliability of microbiome estimates. Knowledge of measurement consistency and how it influences the partitioning of the total variance can inform sampling design, with potential to increase the reliability and improve reproducibility of future studies.

(32)

We found that estimates of microbial species richness (number of ASVs), based on samples taken from same body site of the same individual, were highly correlated. With this knowledge we can be more confident that the observed patterns (e.g., high variation among samples or individuals) are not due to measurement error.

(33)

18

Results and Discussion

Effects of environmental variation on the

performance of populations and individuals

Population structures typically arise due to reproductive isolation mediated by physical separation (allopatry) (Endler 1977). Pike-populations in the Baltic Sea coastal areas spend most of their life-time in sympatry, but during the reproductive period they migrate to freshwater spawning areas, exhibiting natal spawning site fidelity (Engstedt et al. 2014, Tibblin et al. 2015, Tibblin et al. 2016b, Nordahl et al. 2019). If environmental conditions vary among spawning localities, and those conditions prevail over time, this short-term allopatry may lead to local adaptation in traits important for reproductive performance, such as early life-history traits (Jensen et al. 2008), or reproductive strategies (Roff 1992). Adaptations to local environment can be inferred if individuals have higher fitness in their native compared to alternative spawning habitats, and higher relative fitness in their native spawning habitat than immigrant genotypes (Kawecki and Ebert 2004, Hereford 2009).

In Paper I, we performed a reciprocal translocation experiment (Figure 3) to investigate whether two subpopulations of pike, Lervik and Okne (separated by approximately 10 km), had evolved local adaptation to spawning habitats. The outcome indicated local adaptation in one of the two subpopulations; the Okne population had higher hatching success in the native spawning habitat, both when compared to foreign spawning habitat and when compared to translocated immigrant genotypes from Lervik (Figure 4).

It is known that adaptations related to early life stages can impact on life-history traits expressed later in life, for example parental reproductive investment strategies (Roff 1992). For instance, if mortality is high during early life stages, any adaptations that reduce mortality have fitness advantages. Hence, we further investigated whether maternal investment strategy differed between the two populations by comparing the reproductive effort and size of eggs.

(34)

Figure 4. Comparisons of hatching success for each population in the different spawning habitats. Hatching success represents percentage of eggs that hatched. Box colour indicates population origin: a) grey = Okne (n = 18); b) white = Lervik (n = 22). Box-plot elements: center line: median; box limits: upper and lower quartiles; whiskers:

1.5×interquartile range.

Results uncovered that females from Lervik invested more resources into reproduction and produced more (but smaller) eggs compared to females from Okne (Figure 5). Size of pike eggs have been positively associated with size of fry, and negatively correlated with hatching success and oxygen transfer between the egg and the surrounding environment (Vandenberghe and Gross 1989, Murry et al. 2008). Depending on the environmental conditions experienced during early life stages, the optimal reproductive allocation strategies might differ, and we propose that the observed differences in egg size might reflect adaptations to the abiotic and biotic conditions in the different spawning habitats (e.g., Sunde et al. 2018, 2019).

Figure 5. Comparisons of pike reproductive investment in Paper I. a) Egg size measured as dry mass per egg as a function of female body size. b) Reproductive effort measured as dry gonad mass as a function of female body size. Each dot represents one individual and the same individuals are represented in both graphs. Colour of dots indicate population origin: grey = Okne (n=23); white = Lervik (n=23).

Native

habitat Non-native habitat

% h at ch ed la rv ae a) b) Native

habitat Non-native habitat

Okne Lervik 0 50 100 150 200 250 300 35 45 55 65 75 85 Gon ad d ry mass (g) 0 1 2 3 4 35 45 55 65 75 85 Eg g dr y mass (mg)

Total length (cm) Total length (cm)

(35)

20

There is evidence, based on analyses of neutral and functional genetic markers, that anadromous subpopulations of pike, separated by short spatial distances, are differentiated in both types of genetic regions (Nordahl et al. 2019, Sunde et al. 2020). Common garden experiments have further demonstrated that the genetic differentiation manifest in fitness related traits (i.e., vertebral count and growth rate), and also point to local adaptations to salinity and temperature (Tibblin et al. 2015, Tibblin et al. 2016a, Sunde et al. 2018, Sunde et al. 2019). Yet, we cannot determine, based on the available evidence in Paper I, whether the differences in reproductive effort and egg size were of genetic origin or reflected a plastic response to different environmental conditions. However, past investigations have shown that these subpopulations co-exist in a common foraging habitat during the majority of their life (~90%, Tibblin et al. 2015). It is therefore unlikely that the observed differences are due to plasticity induced by differential habitat use.

In addition to environmental variation in spatially separated habitats, as studied in Paper I, environmental conditions may vary within habitats, both on a temporal and spatial basis. Fish individuals can cope with such varying conditions by choosing those micro-habitats that best meet their specific requirements and thereby potentially benefit from broader niches resulting in improved performance and fitness. Temperature is important for physiological functions and therefore a good example of an environmental variable that may vary within habitats and thus contribute to spatial distributions and temporal activity patterns of individuals (Reynolds and Casterlin 1980, Gillooly et al. 2002, Johnston and Bennett 2008, Forsman et al. 2016). Fish are generally ectotherms, but some species of sharks (Lamnidae) and tuna (Scombridae) can regulate their body temperature via physiological adaptations (Madigan et al. 2015, Watanabe et al. 2015). Other fish species can optimize growth and development through behavioural regulation of body temperature by utilizing temperature variation in their environment (Reynolds and Casterlin 1980, Gillooly et al. 2002, Helfman et al. 2009). Ectotherms in terrestrial environments can obtain higher body temperature by sun-basking (Stevenson 1985, Huey and Kingsolver 1989, Peterson et al. 1993, Forsman 2000, Forsman et al. 2002). There exists anecdotal evidence that sun-basking occurs also in fish, but it has not previously been systematically evaluated whether such behaviour enables fish to become warmer than ambient water.

In Paper II, we wanted to assess i) whether carp individuals displayed sun-basking behaviour, ii) whether variation in sun-basking behaviour among individuals was associated with their personality (bold/shy) and colour morph (dark brown or pale orange), iii) how this impacted body temperature, and finally, iv) whether the excess in body temperature gained by sun-basking resulted in superior physiological performance manifested as growth. To this end, we performed a longitudinal study of free-ranging carp under natural

(36)

conditions where we observed behaviour and monitored the body and ambient water temperature for each individual.

Results revealed that carp individuals varied considerably in movement patterns, specifically frequency of vertical migration. Carp that spent time near the water surface during sunny conditions attained body temperatures that were up to 4°C higher than ambient water. Notably, the temperature excess gained by basking was larger in dark than in pale individuals, increased with behavioural boldness, and was positively associated with growth, thus indicating a possible fitness gain of the sun-basking behaviour.

To sum up, the results so far show that spawning migration and sun basking behaviour can contribute to spatiotemporal distributions within and among populations of fish – and perhaps also impact the microbial communities associated with their skin? In the next part of the thesis, I explore whether and how fish skin-associated microbial communities are affected by phenotypic properties of the fish hosts, and by exposure to spatiotemporal variation in environmental conditions associated with fish body sites, behaviours and population affinity.

Composition and dynamics of fish-associated

microbial communities

Fish skin-associated microbial communities are distinct from

bacterioplankton in the ambient water

If fish are considered habitat islands that vary according to the properties of the individual host and the environment that the host utilize, it can be hypothesized that the bacterioplankton community constitute an important source of potential colonizers to the fish skin microbiome. However, among the studies included in this thesis, the community composition of fish skin microbiomes was always distinct from bacterioplankton communities in the water (Papers III-V), with little overlap of ASVs between fish and water (approximately 2.5%, see Papers

IV & V). This indicates that microbial colonization was not directly correlated

with their presence in the water, and points to the conclusion that fish constitute important habitat for epibiotic microorganisms in aquatic environments. Together with findings from previous studies we can be confident in concluding that fish skin microbiome composition is different from the surrounding bacterioplankton, pointing to an important role of ecological filtering (Horsley 1977, Wang et al. 2010, Stevens and Olson 2013, Chiarello et al. 2018, Chiarello et al. 2019, Krotman et al. 2020, Sylvain et al. 2020, Uren Webster et al. 2020).

Across all studies, the taxonomic composition of fish skin microbiomes was dominated by Proteobacteria (Papers III-V, Figure 6), Actinobacteria and Bacteroidetes (Papers III & V, Figure 6). These three phyla are commonly

(37)

22

recognized as the most dominant taxonomic groups in skin microbiomes of freshwater fish (Boutin et al. 2013, Boutin et al. 2014, Webster et al. 2018, Chiarello et al. 2019, Uren Webster et al. 2020), and have been proposed to include important symbiotic species for the function of fish skin microbiomes (i.e., direct or indirect pathogen resistance), or commensal bacterial species that thrive on the fish skin (Merrifield and Rodiles 2015, Tarnecki et al. 2019). Moreover, at phylum level, the most abundant groups were same within studies, despite environmental switch (Papers IV & V), which infers a role of ecological filtering associated with the host.

Figure 6. Prevalence among samples of the 5 most abundant phyla within each study. a) perch, these phyla represented 72% of all sequences, b) roach, these phyla represented 81% of all sequences and c) carp, these phyla represented 79% of all sequences.

Diversity of fish-associated microbial communities across spatial

scales and hierarchical levels

In the literature, there are signs of species dependent factors for the composition of fish skin microbiomes i.e., core microbiome (Larsen et al. 2013, Larsen et al. 2015, Schmidt et al. 2015, Chiarello et al. 2018, Webster et al. 2018, Tarnecki et al. 2019, Sylvain et al. 2020, Uren Webster et al. 2020). If the fish hosts exert filtering due to intrinsic and genetic properties, this can lead to a core microbiome among individuals of the same species (Shade and Handelsman 2012, Sullam et al. 2012, Risely 2020). This is common in the digestive tract among a large number of animal species where it is attributed to a strong selective pressure due to its importance for the host (e.g., immune system and gastrointestinal functions) (Roeselers et al. 2011, Sullam et al. 2012). Furthermore, fish host-microbiome interactions create niches that potentially make the skin surface less accessible for free-living microorganisms and thus

0.00 0.25 0.50 0.75 1.00 Actinobacter ia

BacteroidetesEuryarchaeotaProteobacter ia Tener icutes Relativ e ab undance a) Perch 0.00 0.25 0.50 0.75 1.00 Actinobacter ia BacteroidetesCyanobacter ia Planctom ycetes Proteobacter ia b) Roach 0.00 0.25 0.50 0.75 1.00 Actinobacter ia BacteroidetesFir micutes Planctom ycetes Proteobacter ia c) Carp

(38)

more resistant to variation in abiotic factors along environmental gradients (e.g., pH, temp, nutrients) than free-living bacterioplankton communities (Pinhassi et al. 2003, Martiny et al. 2006, Stevens and Olson 2015). However, neither of the studies included in this thesis identified a core microbiome, instead community composition was highly variable owing to factors such as host environment, population affinity, individuals, and sometimes even body sites (Papers III-V). Spatially separated host populations can be hypothesized to harbour dissimilar microbial communities owing to both genetic and environmental differences (Stevens and Olson 2015, Webster et al. 2018, Chiarello et al. 2019, Sylvain et al. 2020). In Paper III, we sampled dorsal and ventral body parts of perch originating from two putatively genetically differentiated populations along the Baltic Sea Swedish coast (Bergek and Björklund 2009, Olsson et al. 2011). We found that microbiome composition differed according to both population and individual, and that host individual accounted for a larger proportion of the variation in microbiome composition than did population. If individual-specific factors override extrinsic environmental variation, this indicates that there is strong filtering associated with individual host properties. To further evaluate the relative importance of intrinsic and extrinsic factors for microbiome variation we performed repeated longitudinal sampling in combination with environmental translocation, as reported in Paper IV (see below).

Individual-specific microbiome composition was evident in all studies (Papers III-V). However, individual-specific patterns were not as pronounced in captive and laboratory conditions (Papers IV & V), as in the wild-caught fish that were exposed to more variable environmental conditions and had the opportunity to express natural behaviours (Papers III & V). In samples representing wild-caught specimens, microbiome richness and composition varied greatly among individuals (Figure 7).

Variation among individuals in microbiome richness and composition can reflect individual-specific filtering mediated by different inherent properties, such as host secretion and auto-immune molecules (Boutin et al. 2013, Chen et al. 2018), but also habitat choice resulting in varying exposure to extrinsic factors (e.g., temperature, pH, sediments, UV-light) (Shephard 1994, Wotton 2004, Grice and Segre 2011, Wahl et al. 2012, Beck and Peatman 2015, Hess et al. 2015). Such host-specific variation could potentially also arise if contrasting body sites are affected differently by abiotic and biotic factors, as suggested by previous studies (Ley et al. 2008, Costello et al. 2009, Chiarello et al. 2015, Lowrey et al. 2015). Given that patchy environments are expected to harbour higher species richness than homogeneous environments (Johnson 1974, Tews et al. 2004), this could ultimately contribute to inter-individual variation.

To investigate the role of body site, we evaluated intra-individual variation in microbiome species richness and composition in Papers III and V and found somewhat different results. In Paper III, the microbiome composition was not

(39)

24

Figure 7. Alpha- and beta-diversity comparisons between free-ranging and captive individuals of carp and roach. Data represents longitudinal samples from same individuals

that were either translocated to natural environment (carp, n=27, represented by two samples at each location, Paper V) or caught in the wild and then kept in captivity in the field (roach, n=44, Paper IV). Captive samples represent two groups – placed in fresh- or brackish water - week 1 and 2, wild samples from week 0 (see Figure 3). a) Richness as estimated with function “breakaway” (package breakaway in R). b) Beta diversity, derived from distance to Euclidean centroid within each group. Box-plot elements: center line, median; box limits, upper and lower quartiles; whiskers, 1.5×interquartile range; points, outliers.

distinct between dorsal and ventral body sites within perch individuals. This can either reflect that connectivity and dispersal of microbes is high between bodily regions within hosts, or that similar assembly processes are governing the community composition within hosts. However, we also found that the difference between dorsal and ventral body parts were larger in some individuals than in others: in Paper V, fish skin microbiomes became strikingly more diverse on dorsal compared to ventral body sites of carp, in both richness and composition, after two months of free-ranging in the pond (Figure 8). The inconsistency of findings in Papers III and V could reflect differences between the two studied species in behaviour and life style. For instance, carp displayed sun-basking behaviour (Papers II & V) and it is possible that exposure to UV-radiation during sun-basking caused disturbances (Hijnen et al. 2006) that affected the colonization-extinction dynamics (Connell 1978), which led to increased variation among microbiomes at dorsal sites. Ecological theory posits that variation in community composition among habitat patches should increase with disturbance frequency (Warwick and Clarke 1993, Leibold et al. 2004, Vanschoenwinkel et al. 2013). Frequent disturbances may cause higher extinction rate and increase niche availability which allow for higher immigration rates from the surrounding regional species pool (Miller and Bohannan 2019). It is thus likely that the sun-basking is a major cause of these different diversity patterns associated with body sites in carp (Figure 8).

captive wild captive wild 0 500 1000 1500 2000 α− div ersity a)

captive wild captive wild 0 100 200 300 β− div ersity b) carp roach

(40)

It is possible that sun-basking behaviour exists also in perch. As mentioned previously, it has been indicated in other species than carp. For instance, Nordahl et al. (2020) found signs of sun-basking behaviour in pike with higher prevalence during summer months. Many behaviours can be associated with seasons and limited to specific time of the seasonal cycle (Nelson et al. 1990). Accordingly, the different outcomes may depend on the season, and not on differences in utilization of sunlight per se. The study reported in Paper II and

V was conducted during the summer, when sun-basking is prominent in many

animals, whereas Paper III was conducted in the autumn.

Figure 8. Diversity gain in dorsal and ventral skin microbiomes of carp following translocation from laboratory to a natural environment in Paper V. a) Alpha diversity (species richness). b) Beta diversity (distance to Euclidean centroid). Dorsal (teal) microbiomes diverged more than ventral (green) microbiomes following translocation. Box-plot elements: center line, median; box limits, upper and lower quartiles; whiskers, 1.5×interquartile range; points, outliers.

Taken together, the results reported above point to the conclusion that individual host and the environment it is experiencing (i.e., spatially separated populations) jointly determines the microbiome composition. This raises the question whether the microbiome composition is dynamic in response to contrasting and changing environmental conditions? We investigated the interacting effects of host individual, succession stage, and environment on microbiome composition and dynamics, and evaluated whether there were any signs of a homogenizing effect of dispersal among hosts in Paper IV (Figure

3). The degree of heterogeneity varied among hosts which could potentially

reflect that individual-specific behaviour is an important driver of microbiome richness and composition – an objective we evaluated in Paper V.

pre translocation post translocation

0 500 1000 1500 2000 α− div ersity

a) pre translocation post translocation

0 100 200 300 β− div ersity b) dorsal ventral

(41)

26

Spatiotemporal environmental variation drives variation among

skin-associated microbial communities

Water is an intrusive environment, thus the external factors most likely also affect the colonization-extinction dynamics of fish skin microbiomes to an important degree (Llewellyn et al. 2014, Chiarello et al. 2018, Sylvain et al. 2020, Uren Webster et al. 2020). Anadromous fish spend the major part of their life cycle in the sea and then migrate to freshwater to spawn. During spawning migration, the fish and its associated skin microbiome are exposed to several changes in the external environment such as changes in salinity, temperature, pH, and to different surrounding microbial communities (Caporaso et al. 2011, Wang et al. 2012, Schmidt et al. 2015). Moreover, both changing abiotic conditions and reproductive period can impose physiological modifications of the host, with the potential to influence the structure and dynamics of the associated microbiomes (Tort et al. 2003, Ángeles Esteban 2012, Boutin et al. 2014). In Paper IV, we wanted to encapsulate the joint effects of host-microbe interactions and environmental factors. To this end, we conducted longitudinal repeated sampling of wild-caught individuals of anadromous roach during the spawning migration, followed by in situ manipulations (Figure 3).

The results from the environmental translocation in combination with repeated longitudinal sampling on the same host individuals demonstrated that the external environment had a major impact on the microbiome composition (Figure 8 & 9) (Papers IV & V). In Paper IV, skin microbial community composition shifted repeatedly within a week, in response to environmental translocation of roach individuals, demonstrating that skin microbiomes of fish can be highly dynamic.

Figure 9. Spatiotemporal dynamics of roach skin microbiome composition in Paper IV. Environmental translocation between brackish and freshwater habitats in combination with repeated longitudinal sampling on the same host individuals demonstrated that the external environment had a major impact on the microbiome composition.