An ecological perspective on microbes and immune defences in avian eggs Grizard, Stephanie

Transcription

1 University of Groningen An ecological perspective on microbes and immune defences in avian eggs Grizard, Stephanie IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record Publication date: 2015 Link to publication in University of Groningen/UMCG research database Citation for published version (APA): Grizard, S. (2015). An ecological perspective on microbes and immune defences in avian eggs. [Groningen]: University of Groningen. Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons). Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/UMCG research database (Pure): For technical reasons the number of authors shown on this cover page is limited to 10 maximum. Download date:

2 Chapter 4 ON THE ORIGINS OF BACTERIAL SPECIES ASSOCIATED WITH PASSERINE EGGS Stéphanie Grizard, Francisco Dini-Andreote, Henry K. Ndithia, B. Irene Tieleman, Joana Falcão Salles Unpublished manuscript

3 Chapter 4 ON THE ORIGINS OF EGGSHELL BACTERIAL SPECIES ABSTRACT Eggshell-dwelling microorganisms may be at the origin of egg infections and potentially impact the embryo survival. Depicting this microbiome represents therefore an essential step to define its actual role in the avian life history. Yet, little is known about these microbial communities, and more specifically whether certain bacteria are consistently selected on eggshells and which source - either the environment or the host - mostly drives their composition. Here, we examined the eggshell bacterial communities from two related passerines distributed over three habitats: the red-capped lark (Calandrella cinerea) in two tropical locations (South and North Kinangop in Kenya) and the skylark (Alauda arvensis) in a Dutch temperate-zone one. Although the two lark species live far apart and under distinct environments, eggshell communities from North Kinangop and the Netherlands exhibited particularly close community structures and diversities and shared up to three-fifth of their bacterial species. This noticeable core microbiome comprised mainly Ralstonia sp. and, to a lower extent, Herbaspirillum, Rhodococcus, Propionibacterium, Arthrobacter, and Mesorhizobium genera. Moreover, while targeting three microbial sources - feathers and cloaca of the incubating females and nest lining materials - we found that plumagerelated bacterial species contributed the most to the eggshell microbiome, and highlighted the dominance of microbial horizontal instead of vertical transmission structuring this microbiome. Overall, we advocate that a better understanding of the eggshell microbiome may help fulfilling gaps in the evolutionary and ecological processes underlying its selection, which are of major importance to further elucidate its functional roles in the avian life history strategies. 84

4 INTRODUCTION It is commonly entrenched that microorganisms often live in close relationship with their hosts. Interactions between animals and microbes lead to complex associations, from beneficial and commensal to pathogenic, which are likely to vary at both ecological and evolutionary scales (Lee and Mazmanian 2010; McFall-Ngai 2002). In this context, there are rising evidences that host-associated microbiomes shape life history traits and that environmental conditions are important factors mediating this relationship (Archie and Theis 2011; Ezenwa et al. 2012; McFall-Ngai et al. 2013; Russell et al. 2014). As trans-shell invading microbes may cause embryonic death (Beissinger et al. 2005; Cook et al. 2003, 2005b; but see Ruiz-de-Castañeda et al. 2011c; Wang et al. 2011b), characterising the singular eggshell microbiome represents an essential step forward for understanding its potential role in the avian life history. However, little is known about how these microbial communities are differentially influenced by ecological processes and environmental conditions and to which extent microbes are acquired from different sources, albeit those questions are of great relevance from a bird perspective. Microorganisms are well-known for their astonishing diversity, ubiquity, and biogeographical distribution (Fierer 2008; Martiny et al. 2006). Yet, exploring how and to what degree the environment sculpts the microbial communities associated with eggshells has been overlooked. Considering that experimental designs demonstrated that high temperature and humidity affected microbial loads (Cook et al. 2003, 2005a, 2005b), and that eggs from distinct climatic locations exhibited substantial differences in community abundance, composition, and structure (chapter 5 of this thesis), it is reasonable to assume that habitat exerts a strong selection on eggshell microbiome. Apart from possible environmental effects, microbial selection on eggshells may be driven by additional nonmutually exclusive factors (e.g. nest characteristics and parental inputs). If such selection operates on eggshells, specific bacteria would be consistently found among habitats and would constitute a core microbiome (e.g. Shade and Handelsman 2012). Supporting this contention, a cross-checking references on eggshell microorganisms highlighted a potential core microbiome composed of certain bacterial genera; Pseudomonas, Staphylococcus, Enterococcus, and Stenotrophomonas genera have been frequently retrieved from different bird species and across diverse habitats (Brandl et al. 2014; Grizard et al. 2014, 2015; Lee et al. 2014; Potter et al. 2013; Shawkey et al. 2009). Examining eggshell microbial communities among environments by delving deep in taxonomical composition might bring new perspectives on the selection of avian eggshell microorganisms. Primarily knowledge about the eggshell microbiome - mostly provided by poultry - indicates that microbes are transmitted during shell formation or through the cloaca at laying (Barrow 1994; Board 1966; Gantois et al. 2009) and afterwards transferred by environmental sources surrounding the eggs (Gantois et al. 2009; Nowaczewski et al. 2011; Szablewski et al. 2010). In an ecological context, despite cumulative evidences that parental 85

5 Chapter 4 ON THE ORIGINS OF EGGSHELL BACTERIAL SPECIES - feathers/skin of the incubating parent(s) (Soler et al. 2010, 2012a) - and environmental sources - nest types and lining materials (Godard et al. 2007; Peralta-Sánchez et al. 2010, 2014; Walls et al. 2012) - may greatly affect eggshell microbiome, only few studies have actually concomitantly characterised the bacterial communities associated with eggshells and their potential sources. A recent study in reed warbler (Acrocephalus scirpaceus), so far the single one applying molecular tools, showed an overlap in bacterial community assemblages from eggshells and nesting materials (Brandl et al. 2014). Such studies are primordial because they bring evidence for the relative importance of horizontal over vertical microbial transmissions and for the potential evolutionary role of host-associated microbes. As little is known about the extent to which eggshell bacteria are acquired from close sources, targeting several sources at the time may help to fulfill this gap. Here we made use of the broad biogeographical distribution of two close-related passerines to characterise their eggshell microbiomes among three habitats. We examined eggs from red-capped larks (Calandrella cinerea) in two tropical locations and the ones of skylarks (Alauda arvensis) in a temperate-zone one. These two Alaudidae species are groundnester passerines and eat similar food limiting confounding effects from behaviour or diet (Alström et al. 2006, 2013; Tieleman et al. 2003) on the characteristics of the eggshell microbial communities. First, we investigated the possible existence of a core microbiome by examining whether bacteria were either shared among birds or at least between two habitats, or were unique per habitat. Relying upon molecular tools, we described eggshell community diversity, phylogenetic and taxonomical composition, and determined the proportion and the identity of the most common Operational Taxonomic Units (OTUs). Moreover, in order to determine the origin of the eggshell microbiome, we characterised bacterial communities associated with three sources that potentially mould the eggshell communities - feathers and cloaca of the incubating females and nest lining materials - as well as their contribution to the eggshell communities. MATERIALS & METHODS Bird species and study sites We studied two larks species, the red-capped lark (Calandrella cinerea) and the skylark (Alauda arvensis) during their breeding season, in tropical and temperate-zone habitats, respectively. In Kenya, our study took place in two geographically close fields: South Kinangop (S 0 42 E ) and North Kinangop (S 0 35 E ) on the Kinangop Plateau where red-capped larks are common, and in northern Netherlands (Aekingerzand, province of Drenthe, N E ) where a population of skylarks is yearly present. Red-capped larks and skylarks belong to the Alaudidae family. They are both groundnesting passerines building open-cup nests in open grassland fields and occasionally in heath lands. Females commonly laid one egg per day. The clutch size of red-capped larks 86

6 typically consists of two eggs while skylarks produce three to five eggs per clutch. Incubation generally starts with the last egg and chicks hatch synchronously twelve days after clutch completion (Del Hoyo et al. 2004; Horrocks et al. 2014a). Sample collection and processing We monitored breeding activity, nest construction, and egg laying daily. We first captured females at their nest using clap traps. Upon capture, we handled them wearing gloves previously sterilized with 70% ethanol. We collected feathers from the female brood patch region by cutting them at about half of their length, using sterile scissors and tweezers. We ensured that enough feathers remained to cover the brood patch. We sampled cloacae by introducing and carefully rotating a swab (ClassiqSwabs, Copan Flock Technologies, Brescia, Italy) in the cloaca on each female for 10s. Shortly after female was released, we sterilely collected eggs and nest lining materials. In the Kenyan sites, we collected the two eggs per nest on the day of clutch completion (day 1; n=6 nests) and during following days (days 2-3: 4 nests; day 5: 2 nests) (Table S4.1). When laying date was unknown, we assessed egg age using yolk shape as it quickly changes during the first days of embryonic development and enables determining the clutch completion day (Grizard et al. 2015). In the Netherlands, we collected the two first eggs of a clutch on the day where the second egg was laid (day 1). We collected lining material immediately after egg collection, by carefully removing a few twigs from each nest center using sterile tweezers. Feathers, cloacal swabs, and nest lining were all individually kept into sterile screwed tubes. Eggs were individually stored in sterile bags (Whirl-Pack Write-On Bags, Nasco, Fort Atkinson, WI). Samples were kept on ice during fieldwork (<7h) then frozen at -20 C until processing. We performed egg dissections following Grizard et al. (2014). In South Kinangop, we collected seven red-capped lark associated samples - seven feather units, seven cloacal swabs, fourteen eggs and lining of seven nests - and five red-capped lark ones in North Kinangop, between March and April In the Netherlands, we collected five skylark associated samples between May and June 2012 (Table S4.1). Assessing bacterial communities from eggshells and their three associated sources DNA extraction protocols were carried out using the Fast DNA SPIN kit (MP Biomedicals LLC, Solon, OH) following manufacturer s instructions, except that cell disruption was achieved by bead beating (Grizard et al. 2014). We implemented preliminary steps to optimise extractions depending on sample types. Briefly, after crushing each entire eggshell into liquid nitrogen, we extracted DNA from the eggshell powder. We followed this crush protocol except that the final elution step was done in 150µL. For feathers, they were initially sonicated for 15min in Sodium Phosphate and MT buffers (provided with the Fast DNA SPIN kit), followed by 10min vortexing (Vortex-Genie2, MoBio Laboratories Inc, 87

7 Chapter 4 ON THE ORIGINS OF EGGSHELL BACTERIAL SPECIES Carlsbad, CA) to dislodge the attached bacteria prior to DNA extraction (Bisson et al. 2007; Saag et al. 2011). Cloacal swabs were wetted using two buffers (Sodium Phosphate and MT buffers) before removing the cotton part from the metal stick. We proceeded with nest lining materials without any preliminary step. In total, we extracted DNA from thirty-four eggshells, and from seventeen feather samples, cloacal swabs, and lining materials. Extracted DNA concentrations were measured by fluorescent quantification using Quant-iT PicoGreen dsdna kit (Molecular Probes Incubation., Eugene, OR) (see Grizard et al. 2014). We assessed bacterial community composition by 454-Roche multitag pyrosequencing of the V4-V6 region of the 16S rrna gene using the primer set 16s-515F/16s-1061R. Details about reactions, thermal cycling conditions, purification, quantification, and pooling of the amplicons, are described in Grizard et al. (2015). Amplicons from the eighty-five samples were run on a Roche GS-FLX 454 automated pyrosequencer (Titanium chemistry) at Macrogen (Korea). Pyrosequencing raw data was processed using the Quantitative Insights Into Microbial Ecology (QIIME) toolkit (version 1.7.0) (Caporaso et al. 2010a). Sequence quality trimming and OTU assignment were done as in Grizard et al. (2015). After trimming, 82,485 sequences from the eighty-five samples were retrieved. All sequencing data have been deposited in the MG-RAST database ( To minimize the effects of sampling effort on α-diversity metrics and β-diversity analyses, we rarefied the number of sequences to 320 per sample. In this process, twenty-six samples were discarded and we retrieved fifty-nine samples (twenty eggshells, fourteen feathers, twelve cloacal, and thirteen lining material samples; Table S4.1). The chosen cutoff ensured a good coverage of the OTU diversity across all samples (Figure S4.1). We estimated four α-diversity metrics: Shannon s diversity, Species richness (number of OTUs), Faith s phylogenetic diversity, and Chao1 indices. Rarefied OTU tables were used for calculating community phylogenetic β-diversity visualized by nonmetric multidimensional scaling (NMDS) plots, for plotting taxonomical histogram, and for drawing OTU networks. We generated NMDS graphs by exporting OTU tables into PRIMER-E v0.6 (Clarke and Gorley 2006). Tables were modified by a fourth-root transformation and resemblance matrices obtained via Bray-Curtis similarity. We performed statistical analyses based on the Analysis of Similarity (ANOSIM; one-way analysis; 5000 permutations). The associated global R described the percentage of permutations related to P-value and the stress value indicated how faithful the relationship among samples is on the ordination plot. We generated networks in QIIME and visualised them with Cytoscape v2.8.3 (Smoot et al. 2011). Only OTUs from which the total number of sequences, while counting all samples, was equal or superior to ten reads were considered. The associated numbers of OTUs were further assigned into a Venn diagram. We analysed the α-diversity metrics with linear mixed-effects models (package nlme, Pinheiro et al. 2011) where we included nest as a random factor, as we frequently had two 88

8 or several observations per nest. We simplified models using backward elimination based on log-likelihood ratio tests and using P<0.05 as selection criterion. The normality of the residuals of final models was tested using Shapiro tests. Only Chao 1 index values occasionally deviated from Gaussian distribution and were log-transformed when appropriate. We used R for all statistical analyses (R Development Core Team 2011). RESULTS Eggshell core bacterial community Bacterial communities associated with red-capped lark eggshells in South Kinangop strikingly clustered apart from the ones associated with red-capped larks in North Kinangop and skylarks in the Netherlands. South Kinangop showed low similarities with North Kinangop (3.4% ±0.36, R = 0.940, P < 0.001) and with the Netherlands (1.3% ±0.18, R = 1.00, P < 0.001), while those two latter ones shared up to ten-fold more similarities among their communities (14.0% ±1.31, R = 0.434, P = 0.02). Additionally, eggshells-associated microbial communities were more similar within South Kinangop (65.2% ±1.82) than they were within North Kinangop (16.7% ±2.12) and within the Netherlands (30.0% ±2.08) (Figure 4.1). All four α-diversity metrics significantly differed among sites (Table S4.2). South Kinangop communities showed the lowest community diversity when compared with the two other sites which held similar values. For instance, red-capped larks in South Kinangop counted 16.0 OTUs (±1.62) and significantly differed from red-capped larks in North Kinangop (Z = -3.44, P = 0.002) and skylarks in the Netherlands (Z = -3.39, P = 0.002). North Kinangop and the Netherlands exhibited higher species richness (77.6 OTUs ±16.86, 78.0 OTUs ±9.36, respectively; Z = 0.03, P = 0.99) (Table S4.3, Figure S4.2). Similar trends were obtained for Shannon's diversity, phylogenetic diversity, and Chao 1 indices (Table S4.3, Figure S4.2). While considering the bacterial communities associated with eggshells from South and North Kinangop and the Netherlands only 7 OTUs were in common among these habitats. Remarkably, when restricting our comparison to red-capped lark in North Kinangop and skylark eggshells in the Netherlands, we observed that 29 OTUs - more than a half of their OTUs - were shared, showing therefore an enlarged eggshell core microbiome among these two particular habitats (Figure 4.2). Among the 7 OTUs that characterised the eggshell core community of all habitats, Ralstonia sp. affiliated OTU dominated red-capped lark in North Kinangop (49.0%) and skylark communities (49.0%) but was quasi-absent from South Kinangop (0.2%). The remaining 6 OTUs, affiliated to Herbaspirillum, Rhodococcus, Propionibacterium, Arthrobacter, and Mesorhizobium genera, were present in lower abundances (<2.0%). While only considering North Kinangop and the Netherlands, aside from one Ralstonia sp. affiliated 89

9 OTU, each of the 28 remaining OTUs that constitute the eggshell core community were present at lowered abundances (mean: 0.5% ±0.08; range: 0.1%-5.2%; Table S4.4). While Ralstonia sp. dominated each of the North Kinangop and the Netherlands communities, one OTU affiliated to Herbaspirillum sp. represented 85.1% of the overall red-capped lark eggshell communities in South Kinangop (Table S4.4). NKes1a 2D Stress: 0,1 NKes1b NLes3a NLes3b NLes2 SKes1b SKes4 SKes5 SKes3 SKes1a SKes2b SKes2a NLes4 NLes1a NKes3b NLes1b NKes3a NKes4b NKes2 NKes4a Chapter 4 ON THE ORIGINS OF EGGSHELL BACTERIAL SPECIES Figure The community structure of the eggshell microbiome. The non-metric multi-dimensional scaling plot shows the bacterial communities associated with eggshells (es) plotted per habitat: skylarks in the Netherlands (NL, red circles) and red-capped larks in South Kinangop (SK, blue circles) and North Kinangop (NK, green circles). The NMDS plot is based on a Bray-Curtis similarity matrix calculated from the OTU table associated with the β-diversity of all samples retrieved from pyrosequencing. Figure The core microbiome associated with eggshells from two larks in three habitats. In the network, OTUs shared among eggshell bacterial communities are plotted (dark grey dots). Eggshells are grouped per habitat: skylarks in the Netherlands (pink circles) and red-capped larks in South Kinangop (blue circles) and in North Kinangop (green circles). The Venn diagram summarizes the number of OTUs present in the network. Only OTUs counting more than ten sequences were taken into account (see Table S4.4 for a list of OTUs). SK NK NL 22 90

10 Bacterial communities from skylarks in the Netherlands The community structure analysis revealed that eggshell communities clustered apart from all sources (feathers: R = 0.77, P = 0.005; nests: R = 0.63, P = 0.006; cloacae: R = 0.94, P = 0.004; Table 4.1). Nonetheless, eggshells shared higher similarity with feathers (21.4% ±1.21) and nests (23.1% ±1.88) when compared with cloacae (10.4% ±3.46), as visualized on the NMDS plot (Figure 4.3). In fact, of the 64 OTUs identified on eggshells, 58 were uniquely shared with feathers, 51 with nests, and only 8 with cloacae. Moreover, of those 64 OTUs, 44 were found in both feathers and nests (Figure 4.4A). In addition, two OTUs were unique to eggshells: uncultured Cyanobacteria and Rasltonia sp. affiliated OTUs (OTU72 and OTU75, respectively, in Table S4.4). Skylark eggshells were largely represented by Beta- and Alphaproteobacteria which encompassed 48.2% (±11.51) and 28.6% (±9.47) of the overall communities, respectively. These two classes were also associated with feathers and nests: Betaproteobacteria was the fourth most abundant class in feathers (11.3% ±4.58) and the third one in nests (9.3% ±3.32); Alphaproteobacteria was the second one in feathers (18.1% ±3.99) and the first one in nests (43.2% ±9.17). Beta- and Alphaproteobacteria represented however minor percentages in cloacae (1.1% ±1.09, 0.9% ±0.94, respectively), which mostly encompassed Gammaproteobacteria (29.5% ±0.06). The Actinobacteria class, single representative of the Actinobacteria phylum, accounted for 11.4% (±2.41) of the overall eggshell communities. It was the main class of feathers (37.8% ±7.52) but constituted a small proportion of lining material (5.5% ±1.16). On the two cloacal samples, this class represented 60.6% of one sample while none was retrieved from the second one (Figure 4.5, Figure S4.3). The analyses of α-diversity metrics showed that bacterial communities associated with eggshells and sources differed from each other (Table S4.5). For instance, eggshell species richness counted significantly less OTUs (78.0 ±9.36) than feathers (114.0 OTUs ±11.04; Z = 2.95, P = 0.02) but significantly more than cloacae (23.6 OTUs ±6.85; Z = 3.39, P = 0.004), and exhibited comparable richness with nest lining materials (72.8 OTUs ±6.45; Z = 0.41, P = 0.98) (Figure 4.6A, Table S4.4). Similar patterns were observed for Shannon s diversity (except for cloacae that did not significantly differ from eggshells), phylogenetic diversity, and Chao 1 indices (Table S4.5, Figure S4.4A). Bacterial communities from red-capped larks in South Kinangop (Kenya) The community structure analysis in South Kinangop revealed that red-capped lark eggshells significantly clustered apart from sources (feathers: R = 1, P = ; nests: R = 1, P < 0.001; cloacae: R = 0.99, P < 0.001), presenting very low similarity with feather (1.1% ±0.19), cloaca (2.1% ±0.37), and nest communities (2.4% ±0.16) (Table 4.1, Figure 4.3). Of the 20 eggshell-related OTUs, 8 were shared with feathers, 9 with cloacae, and 4 with nests, but about a half (9 OTUs) was unique to eggshells (Figure 4.4B). Moreover, one single OTU - affiliated to Herbaspirillum sp. - dominated the eggshell community (85.1%, OTU81 in Table S4.4). 91

11 Table Similarities among bacterial communities associated with eggshells and sources. For each comparison (eggshell vs one source), R statistic and P-value are based on the Analysis of Similarity (ANOSIM; one-way analysis; 5000 permutations). In addition, the mean similarity (%) shared between the communities of eggshells and of each source tested is given per habitat (and species) with its standard error. Comparison R statistic P Similarity (%) ± SE Skylarks (Netherlands) Eggshell - nest Eggshell - feather Eggshell - cloaca Red capped larks (South Kinangop) Eggshell - nest 1 < Eggshell - feather 1 < Eggshell - cloaca 0.99 < Red capped larks (North Kinangop) Eggshell - nest Eggshell - feather Eggshell - cloaca Chapter 4 ON THE ORIGINS OF EGGSHELL BACTERIAL SPECIES Figure The community structure of the microbiome associated with eggshells and sources. The nonmetric multi-dimensional scaling plot shows the bacterial communities plotted per sample type - eggshells (es, open circles), feathers (fe, full circles), cloacae (cl, squares), and nest lining materials (nst, triangles) - and per habitat - skylark in the Netherlands (NL, red) and red-capped larks in South Kinangop (SK, blue) and North Kinangop (NK, green). The NMDS plot is based on a Bray-Curtis similarity matrix calculated from the OTU table associated with the β-diversity of all samples retrieved from pyrosequencing. 92

12 Beta- and Gammaproteobacteria were the two major classes of eggshells accounting for 85.1% (±3.39) and 9.1% (±1.90), respectively. Betaproteobacteria constituted 28.8% (±3.38) of feathers, 23.0% (±6.72) of cloacae, and 36.7% (±6.53) of nests. While Gammaproteobacteria represented 15.4% (±6.07) of the cloacae, it was quasi-absent from feathers (0.63% ±0.33) and nests (0.63% ±0.21). Additionally, Actinobacteria was the third most abundant class (2.4% ±0.90) in eggshells, and a main class of cloacae (46.6% ±7.52) and feathers (29.7% ±3.30), and to a lesser extent of nests (11.3% ±0.68) (Figure 4.5). Eggshell bacterial communities exhibited particularly low α-diversity values making them always significantly different from all three sources (Table S4.6). For instance, eggshell communities counted only 16.0 OTUs (±1.62), while feathers (126.5 OTUs ±10.17; Z = 8.63, P < 0.001), cloacae (99.7 OTUs ±10.24; Z = 11.29, P < 0.001), and nest materials (58.5 OTUs ±4.51; Z = 4.45, P < 0.001) comparatively possessed significant higher species richness values (Figure 4.6B, Table S4.6). Similar patterns were observed for the three other indices, Shannon s diversity, phylogenetic diversity, and Chao 1 (Table S4.6, Figure S4.4B). Bacterial communities from red-capped larks in North Kinangop (Kenya) In North Kinangop, eggshell community structure was significantly different from sources (feather: R = 0.53, P = 0.02; cloacae: R = 0.69, P = 0.008; nests: R = 0.70, P = 0.03). Eggshells were closer to feathers (10.4% ±1.41) rather than to cloacae (5.1% ±1.32) or lining materials (5.5% ±0.93) (Table 4.1, Figure 4.3). Of the 61 eggshell OTUs, more than one-fourth (17 OTUs) was unique to eggshell communities. The remaining OTUs were shared: 10 between all sources, 12 between nests and feathers, 6 between cloacae and feathers, and 1 between cloacae and nests; 37 OTUs were in common with feathers, 25 with nests, and 21 with cloacae (Figure 4.4C). Eggshells were mainly represented by Beta- and Gammaproteobacteria (43.1% ±10.52; 20.0% ±7.70; respectively). Betaproteobacteria was the second main class of feathers (30.8% ±8.66) and nests (24.5% ±5.47) and Gammaproteobacteria the first of cloacae (36.8% ±19.48). The Actinobacteria class was the third most abundant (19.2% ±8.36) in eggshells and a major class for all sources (feathers: 41.4% ±3.00; cloacae: 30.52% ±6.15; nests: 37.7% ±12.03) (Figure 4.5). By examining the four α-diversity metrics, we did not find any significant differences between eggshells and sources (Table S4.7). Regarding species richness, eggshells counted 77.6 OTUs (±16.86), feathers exhibited the largest richness (146.4 OTUs ±19.65), cloacae (68.6 OTUs ±16.76) and nest materials (83.2 OTUs ±14.00) had similar number of OTUs than eggshell communities (Figure 4.6C). Strikingly however, metric values followed similar patterns than the ones associated with skylarks in the Netherlands (Figure 4.6C, Figure S4.4C). Nonetheless, the high standard errors observed in North Kinangop might, at least partially, explain the absence of significant differences. 93

13 (A) The Netherlands Eggshells Cloacae Nests Feathers (B) South Kinangop Eggshells ON THE ORIGINS OF EGGSHELL BACTERIAL SPECIES - 27 Nests Chapter Cloacae 18 Feathers (C) North Kinangop Eggshells Nests 4 1 Cloacae Feathers Figure Networks and Venn diagrams of shared Operational Taxonomic Units (OTUs) among eggshell and source bacterial communities. In networks, OTUs shared among bacterial communities associated with eggshells and sources, and the ones that are unique to each sample are plotted (dark grey dots). Samples are grouped by per sample type: eggshells (es, green), feathers (fe, red), cloacae (cl, yellow), and nest lining materials (nt, violet). Each network represents one habitat: (A) skylarks in the Netherlands (80 OTUs) and (B) red-capped larks in South Kinangop (110 OTUs) and (C) North Kinangop (82 OTUs). Number associated with samples corresponds to a single set per habitat: they all came from the same female (e.g. es1 is the eggshell associated with feathers fe1, cloaca cl1, and nest materials nt1 ). Venn diagrams summarize the number of OTUs within each network. Only OTUs counting more than ten sequences were taken into account (see Table S4.4 for a list of OTUs). 94

14 Relative abundance 100% 80% 60% 40% Acidobacteria(ph) Actinobacteria(cla) Bacteroidetes(ph) Chloroflexi(ph) Cyanobacteria(ph) Bacilli(cla) Clostridia(cla) Alphaproteobacteria(cla) Betaproteobacteria(cla) Gammaproteobacteria(cla) other phyla 20% 0% NL.es SK.es NK.es NL.cl SK.cl NK.cl NL.fe SK.fe NK.fe Bacterial source NL.nst SK.nst NK.nst Figure Taxonomical distribution of eggshell and source bacterial communities. Relative abundances of Operational Taxonomic Units (OTUs) are represented within phyla (ph)/classes (cla). Lanes are ordered by sample type (eggshells (es), cloacae (cl), feathers (fe), and nest lining materials (nest)) and by habitat: skylarks in the Netherlands (NL) and red-capped larks in South Kinangop (SK) and North Kinangop (NK). Species richness (number of OTUs) (A) Skylarks in the Netherlands Eggshell Feather Cloaca Nest Species richness (number of OTUs) (B) Red-capped larks in South Kinangop Eggshell Feather Cloaca Nest Species richness (number of OTUs) (C) Red-capped larks in North Kinangop Eggshell Feather Cloaca Nest Bacterial source Bacterial source Bacterial source Figure Species richness associated with eggshell and source communities. The number of Operational Taxonomic Units OTUs is represented by grey open circles and their mean by a black diamond (±SE). Data are independently plotted for habitat (and species): (A) skylarks in the Netherlands and (B) red-capped larks in South Kinangop and (C) North Kinangop. 95

15 DISCUSSION Chapter 4 ON THE ORIGINS OF EGGSHELL BACTERIAL SPECIES Our two main goals were to determine the possible existence of an eggshell core microbiome (shared OTUs) and to examine to what extent three potential sources contribute to eggshell bacterial communities, among two close-related lark species, distributed over three distinct climatic habitats. We showed that, although red-capped larks live under tropics and skylarks under temperate climate, their eggshells shared up to threefifth of their bacterial species. Furthermore, our results point out the important contribution of female feathers (from brood patch) and nest lining materials as bacterial sources of eggshell communities comparatively to the maternal reproductive/digestive tracts, independently of habitats. For the first time, our study highlights the possibility for a microbial selection in structuring avian eggshell microbiome. The analysis of the eggshell-related bacterial communities from red-capped larks, including both South and North Kinangop locations, and skylarks in the Netherlands, revealed a limited core microbiome composed of 7 OTUs. However, communities associated with the North Kinangop red-capped lark and skylark eggshells shared up to 60% of their OTUs (29 OTUs in common). Moreover, we observed that eggshell microbiome associated with the latter had similar patterns of diversity whereas South Kinangop rather exhibited notably low one. South Kinangop suffers from extreme wet conditions (high precipitations and relative humidity and periods of standing water) that may deeply affect the characteristics of the eggshell-associated microbial communities (see chapter 5 of this thesis for additional discussion). As eggs are immobile, bacterial sources susceptible to mould the shell communities are restricted to the surrounding environment. Surprisingly, in South Kinangop, all three sources shared very little similarities in community structure and diversity with eggshells, explaining their limited contribution to the shell microbiome. As communities from feathers and nest lining, and to a lesser extent from cloacae, associated with red-capped larks in South and North Kinangop clustered together (NMDS) and had similar α-diversity values, it is unlikely that sources in South Kinangop would contribute differently to eggshells than the ones in North Kinangop toward their own eggs. However, eggshells in South Kinangop were dominated by Herbaspirillum sp. (85%) known for its production of siderophores which may outcompete other bacteria (Rosconi et al. 2013). Combined with its high abundance (chapter 5 of this thesis), Herbaspirillum may impede the establishment of foreign microbial species. Additional research are needed in order to experimentally establish its competitiveness. Particularly interesting are the similarities in the community composition of the core microbiome we found between eggshells from red-capped larks in North Kinangop and skylarks in the Netherlands though these two bird species live in two well-distinct climatic regions. While environmental conditions are likely to influence the presence of the unique OTUs observed within each of the two habitats, our results suggested that at least another 96

16 mechanism may select for the particular bacterial species found in the core microbiome. The nest microclimate has been hypothesised to play a major role into this microbial selective process (Baggott and Graeme-Cook 2002). As previous experimental studies solely relied on correlations based on bacterial loads (with nest relative humidity (Ruiz-De- Castañeda et al. 2011a) or nest types (Godard et al. 2007)), and culture-dependent tools, it hampers further comparisons with our results. Nevertheless, as these two larks are closerelated passerines exhibiting similar nest structure and incubation pattern, microbial selection among nests across habitats might be analogous. Future work comparing eggshell communities from unrelated/related birds and from different nest types (hole- vs opennests; Godard et al. 2007; Peralta-Sánchez et al. 2012; Wang et al. 2011b) may provide additional information on this theme. In particular, focusing into the selective process at the origin of the core formation (Boon et al. 2014) and identifying its functional roles (Shade and Handelsman 2012; Shafquat et al. 2014) will yield new perspectives into the importance of the microbiome associated with avian eggshells. While considering red-capped larks in North Kinangop and skylarks, our study revealed that nest lining materials and feathers contributed more to the eggshell microbiome than the cloacae. This outcome is surprising as bacteria were originally assumed to be notably vertically transmitted (Barrow 1994; Board 1966; Gantois et al. 2009). Although this phenomenon was indirectly confirmed in the free-living pied flycatcher (Ruiz-de-Castañeda et al. 2011b), this was no longer true when focusing on early incubation stages (Ruiz-de-Castañeda et al. 2011c). Moreover, and in accordance with our results, Potter et al. (2013) suggested that environmental origins preferentially shape eggshell microbiomes; this was recently experimentally confirmed with nest lining materials (Brandl et al. 2014). We acknowledge that the small cloacae sample size led to broad variability in community structure and diversity therefore restricting the strength of our conclusion. Nonetheless, the vertical transmission mechanism is still misconceived and required further investigation. Several factors may shape the relationship between eggshell and cloacae; for instance, this depends on microbial features in the ovary, oviduct, digestive tract, and cloacae (Gantois et al. 2009), on local environmental conditions (Lucas and Heeb 2005), and on the female immune status (Soler et al. 2011b; Matson et al. 2015). Feathers contributed more to the eggshell microbiome than nest lining in North Kinangop and the Netherlands. Within each habitat, eggshell communities preferentially clustered with the communities associated with feathers; this was particularly verified for skylarks in the Netherlands. As far as we know, it is the first time that the contribution of feather-related bacteria upon eggshells is quantified, although it has been previously suggested that feathers influence the eggshell microbiome (Peralta-Sánchez et al. 2010, 2014). Body feathers of most birds (including passerines) are daubed with preen gland secretions which are rich in antibiotic agents and/or bacteria producing chemical compounds (Shawkey et al. 2003; Soler et al. 2008) which are likely to promote harmless bacteria (Soler et al. 2010, 2012a) (but see Giraudeau et al. 2014) and thus affect the 97

17 characteristics of the eggshell microbial communities. Although we collected eggs at early incubation stages, we showed that plumage had the major contribution to the shell microbiome. Early inoculation might be favoured by empty spaces in eggshell pores and/or by an efficient antibiotic production. What yet remains unknown is why red-capped larks and skylarks, while living geographically so far apart, exhibit similar plumage-related bacterial assemblages. The hypothesis of the maintenance of feather microbiome would support this idea (Bisson et al. 2007; Burtt and Ichida 1999; Shawkey et al. 2003) but remains to be further investigated. The relevance of the microbial communities associated with eggshells is still in its infancy but delving deeper into their characteristics, in particular into the formation of a core microbiome, may bring new perspectives upon any pathogenic, beneficial, or commensal functions of those egg-related microorganisms. Moreover, understanding the ecological and evolutionary processes underlying the formation of eggshell microbiomes is essential to comprehend avian laying strategy, in particular how females transfer immune defences into eggs, which remains a central research question in ecological immunology (Horrocks et al. 2011a; Norris and Evans 2000). Besides characterising community composition, we also advocate future work to adjoin the description of their functional roles. These new ecological insights will undoubtedly deeply enrich the comprehension of the eggshell microbiome. Chapter 4 ON THE ORIGINS OF EGGSHELL BACTERIAL SPECIES ACKNOWLEDGEMENTS We are grateful to Peter K. Gachigi, Abrahim M. Kuria, Paul M. Kimani, and Susan V. Cousineau, who contributed to the sampling effort in Kenya. We thank Rob Voesten for fieldwork in the Netherlands and a few volunteers for their punctual help. In the Netherlands, Staatsbosbeheer Drents-Friese World kindly allowed working in its area. Maaike A. Versteegh provided advice on the statistical analyses. Financial support for this study was provided by a VIDI grant from the Netherlands Organisation for Scientific Research (NWO) (to BIT). 98

18 Table S4.1 - Sampling collection. Set of two eggs, maternal cloaca and feathers, and nest lining materials were retrieved from red-capped larks (Calandrella cinerea) in South Kinangop and North Kinangop (Kenya) and from skylarks (Alauda arvensis) in Aekingerzand (the Netherlands), during their breeding season in Lark species Country Location Nest code Collection date Clutch age (day) Egg ID.1 Egg ID.2 Cloaca ID Feather ID Nest lining ID Red-capped lark Kenya South Kinangop SK /03/ SKes2b SKes2a SKcl2 SKfe2 SKnt2 Red-capped lark Kenya South Kinangop SK /03/ SKesX4 SKes4 SKcl4 SKfe4 SKnt4 Red-capped lark Kenya South Kinangop SK /03/ SKes1a SKes1b SKcl1 SKfe1 SKnt1 Red-capped lark Kenya South Kinangop SK /03/ SKes3 SKesX3 SKcl3 SKfe3 SkntX3 Red-capped lark Kenya South Kinangop SK /04/ SKesX6a SKesX6b SKcl6 SKfe6 SKnt6 Red-capped lark Kenya South Kinangop SK /03/ SKesX7a SKesX7b SKcl7 SKfe7 SKnt7 Red-capped lark Kenya South Kinangop SK /03/ SKes5 SKesX5 SKcl5 SKfe5 SKnt5 Red-capped lark Kenya North Kinangop NK /03/ NKes3a NKes3b NKclX3 NKfe3 NKnt3 Red-capped lark Kenya North Kinangop NK /04/ NKes1b NKes1a NKcl1 NKfe1 NKnstX1 Red-capped lark Kenya North Kinangop NK /04/ NKes4a NKes4b NKclX4 NKfe4 NKnt4 Red-capped lark Kenya North Kinangop NK /04/ NKesX2 NKes2 Nkcl2 NKfeX2 NKnstX2 Red-capped lark Kenya North Kinangop NK /04/ NKesX5a NKesX5b NKcl5 NKfeX5 NKnstX5 Skylark Netherlands Aekingerzand SL /05/ NLesX5a NLesX5b NLcl5 NLfe5 NLnt5 Skylark Netherlands Aekingerzand SL /05/ NLes3a NLes3b NLclX3 NLfeX53 NLnt3 Skylark Netherlands Aekingerzand SL /05/ NLes1a NLes1b NLcl1 NLfe1 NLnt1 Skylark Netherlands Aekingerzand SL /05/ NLesX4 NLes4 NLclX4 NLfe4 NLnt4 Skylark Netherlands Aekingerzand SL /06/ NLes2 NLesX2 NLclX2 NLfe2 NLnt2 Sample names in italics have been collected but not used for further analysis (due to limited sequences retrieved after trimming for quality in pyrosequencing data). SUPPLEMENTARY INFORMATION 99

19 Table S4.2 - Linear mixed-effect models examining variation in α-diversity metrics associated with eggshell microbiomes. Site corresponds to the three habitats: South Kinangop and North Kinangop in Kenya and the Netherlands. F tests and P-values are reported for each model; P-values are marked up in bold when significant (P<0.05). Model outputs* Explanatory variables df F P Shannon's diversity Julian day 1, clutch age 1, site 2, Species richness (number of OTUs) Julian day 1, clutch age 1, site 2, Phylogenetic diversity Julian day 1, clutch age 1, site 2, (log) Chao 1 index Julian day 1, clutch age 1, site 2, <0.001 Chapter 4 ON THE ORIGINS OF EGGSHELL BACTERIAL SPECIES 100

20 Table S4.3 - α-diversity metrics and associated Tukey s post-hoc tests from eggshell microbiomes. Four α- diversity indices are calculated: Shannon s diversity index, species richness (number of OTUs), phylogenetic diversity, and Chao 1. For each index, the mean, the standard error (±SE), and the minimal (min) and maximal (max) values are given. The post-hoc test aims at testing for differences among pairs of site, South Kinangop versus North Kinangop (SK vs NK), South Kinangop versus the Netherlands (SK vs NL), and North Kinangop versus the Netherlands (NK vs NL). Z tests and P-values are reported for each model; P-values are marked up in bold when significant (P<0.05). Metric values & post-hoc tests Shannon's diversity index South Kinangop North Kinangop Netherlands 1.2 (±0.20) 4.5 (±0.60) 4.3 (±0.51) min: max: 2.0 min: max: 6.7 min: max: 5.6 Post-hoc test SK vs NK SK vs NL NK vs NL Z=-4.38, P< Z=-3.86, P= Z=-0.43, P=0.90 Species richness (number of OTUs) South Kinangop North Kinangop Netherlands 16.0 (±1.62) 77.6 (±16.86) 78.0 (±9.36) min: max: 21.8 min: max: min: max: Post-hoc test SK vs NK SK vs NL NK vs NL Z=-3.44, P=0.002 Z=-3.39, P=0.002 Z=0.03, P=0.99 Phylogenetic diversity South Kinangop North Kinangop Netherlands 1.5 (±0.12) 4.9 (±0.84) 5.3 (±0.49) min: max: 1.9 min: max: 9.0 min: max: 7.4 Post-hoc test SK vs NK SK vs NL NK vs NL Z=-3.66, P= Z=-4.02, P= Z=0.43, P=0.90 Chao 1 index South Kinangop North Kinangop Netherlands 24.9(±4.71) (±43.85) (±25.93) min: max: 51.1 min: max: min: max: Post-hoc test SK vs NK SK vs NL NK vs NL Z=-5.31, P< Z=-6.07, P= Z=0.96, P=

21 Chapter 4 ON THE ORIGINS OF EGGSHELL BACTERIAL SPECIES Table S4.4 - Main Operational Taxonomic Units (OTUs) associated with eggshell bacterial communities among habitats. Relative abundance (%) Affiliation (2) #OTU ID (1) SK NK NL Class (3) Closest hit (4) Accession number Shared NL/SK/NK Similarity OTU Beta- Ralstonia sp. EF OTU Beta- Herbaspirillum sp. AF OTU Actinobacteria Rhodococcus erythropolis AJ OTU Actinobacteria Propionibacterium sp. AY OTU Actinobacteria Rhodococcus sp. AY OTU Actinobacteria Arthrobacter sp. AY OTU109 < Alpha- Mesorhizobium plurifarium DQ Shared NL/SK None Shared NK/SK OTU Deinococci Thermus scotoductus Y OTU Gamma- Serratia sp. DQ OTU Beta- Herbaspirillum sp. AY Shared NK/NL OTU Bacilli Bacillus sp. AJ OTU Beta- Ralstonia sp. EU OTU Beta- Pelomonas saccharophila AB OTU Acidobacteria(Gp2) unc. Acidoibacteriales JQ OTU Gamma- Pantoea sp. GQ OTU Beta- unc. Massilia sp. GQ OTU Beta- Ralstonia sp. AY OTU Gamma- Pantoea agglomerans AB OTU Alpha- Methylobacterium sp. HQ OTU Beta- Ralstonia sp. AY OTU Actinobacteria Arthrobacter globiformis FR OTU Alpha- Sphingomonas sp. EU OTU Cyanobacteria(Gp1) unc. Cyanobacteria FJ OTU Gamma- Moraxella sp. FJ OTU Alpha- Methylocella sp. HQ OTU Acidobacteria(Gp2) unc. Solibacteres KJ OTU Beta- Burkholderia sp. AY OTU Beta- Curvibacter sp. GU OTU Beta- Ralstonia sp. AY OTU Actinobacteria Arthrobacter oxydans JF OTU Beta- Burkholderia sp. AJ OTU Actinobacteria Mycobacterium sp. JX Unique to SK OTU Gamma- Stenotrophomonas maltophilia JN OTU Alpha- Ochrobactrum sp. AF OTU Alpha- Phyllobacterium myrsinacearum AY OTU Gamma- Stenotrophomonas maltophilia U OTU Alpha- Ochrobactrum sp. AF OTU Gamma- Pseudomonas sp. JN OTU Gamma- Pseudomonas sp. AF OTU Beta- Herbaspirillum sp. AY (%) 102

22 Table S4.4 continued. Relative abundance (%) Affiliation (2) #OTU ID (1) SK NK NL Class (3) Closest hit (4) Accession number Unique to NK Similarity OTU Gamma- Pantoea sp. KC OTU Bacilli Planococcus sp. HM OTU Bacilli Bacillus cereus AF OTU Beta- Acidovorax sp. KF OTU Clostridia Veillonella sp. AJ OTU Gamma- Pseudomonas sp. HM OTU Gamma- Pantoea agglomerans AB OTU Gamma- Dyella sp. KF OTU Gamma- Pseudomonas sp. KF OTU Gamma- Acinetobacter radioresistens AY OTU Cyanobacteria(Gp1) unc. Cyanobacteria FJ OTU Bacilli Streptococcus sp. HQ OTU Gamma- Pantoea sp. JX OTU Gamma- Acinetobacter sp. AB OTU Gamma- Pseudomonas sp. EF Unique to NL OTU Alpha- Acidisphaera sp. HQ OTU Actinobacteria Nakamurella sp. HE OTU Alpha- unc. Rhizobiales GU OTU Alpha- Sphingomonas sp. DQ OTU Alpha- Sphingomonas sp. KJ OTU Alpha- Beijerinckia sp. GU OTU Actinobacteria Pseudonocardia sp. HQ OTU Actinobacteria Mycobacterium sp. AY OTU Alpha- Sphingomonas sp. DQ OTU Actinobacteria Frondihabitans australicus DQ OTU Gamma- Luteibacter sp. FR OTU Alpha- Sphingomonas sp. HE OTU Alpha- Methylobacterium sp. FR OTU Alpha- Sphingomonas sp. JF OTU Alpha- unc. Acetobacteraceae FJ OTU Armatimonadia Armatimonas rosea AB OTU Alpha- unc. Rhizobiales JF OTU Alpha- unc. Rhizobiales JF OTU Alpha- unc. Acetobacteraceae JF (%) (1) Only the main Operational Taxonomic Units (OTUs) which included at least ten sequences over the three habitats are shown: South Kinangop (SK), North Kinangop (NK), and the Netherlands (NL). (2) The taxonomic affiliation is presented at the class and genus/species levels and was based on a single representative sequence from each OTU clustered at 97% of nucleotide identity. (3) The phylogenetic classification was based on a single representative sequence from each OTU clustered at 97% of nucleotide identity. In addition, Proteobacteria classes were abbreviated such as Alpha- stands for Alphaproteobacteria, Beta- for Betaproteobacteria, and Gamma- for Gammaproteobacteria. (4) The representative sequence was compared with RDP database allowing establishing similarity shared (in percentage) with a reference sequence. 103