Exploring Biodiversity in the Bacterial Community of the Mediterranean Phyllosphere and its...

PLANT MICROBE INTERACTIONS

Exploring Biodiversity in the Bacterial Communityof the Mediterranean Phyllosphere and its Relationshipwith Airborne Bacteria

Despoina Vokou & Katerina Vareli & Ekaterini Zarali &Katerina Karamanoli & Helen-Isis A. Constantinidou &

Nikolaos Monokrousos & John M. Halley & Ioannis Sainis

Received: 25 October 2011 /Accepted: 4 April 2012 /Published online: 29 April 2012# Springer Science+Business Media, LLC 2012

Abstract We studied the structure and diversity of thephyllosphere bacterial community of a Mediterranean eco-system, in summer, the most stressful season in this envi-ronment. To this aim, we selected nine dominant perennialspecies, namely Arbutus unedo, Cistus incanus, Lavandulastoechas, Myrtus communis, Phillyrea latifolia, Pistacialentiscus, Quercus coccifera (woody), Calamintha nepeta,and Melissa officinalis (herbaceous). We also examined theextent to which airborne bacteria resemble the epiphyticones. Genotype composition of the leaf and airborne bacte-ria was analysed by using denaturing gradient gel electro-phoresis profiling of a 16S rDNA gene fragment; 75 bandswere cloned and sequenced corresponding to 28 taxa. Ofthese, two were found both in the air and the phyllosphere,eight only in the air, and the remaining 18 only in the

phyllosphere. Only four taxa were found on leaves of allnine plant species. Cluster analysis showed highest similar-ity for the five evergreen sclerophyllous species. Aromaticplants were not grouped all together: the representatives ofLamiaceae, bearing both glandular and non-glandular tri-chomes, formed a separate group, whereas the aromatic andevergreen sclerophyllous M. communis was grouped withthe other species of the same habit. The epiphytic commu-nities that were the richest in bacterial taxa were those of C.nepeta and M. officinalis (Lamiaceae). Our results highlightthe remarkable presence of lactic acid bacteria in the phyllo-sphere under the harsh conditions of the Mediterraneansummer, the profound dissimilarity in the structure of bacterialcommunities in phyllosphere and air, and the remarkabledifferences of leaf microbial communities on neighbouringplants subjected to similar microbial inocula; they also point tothe importance of the leaf glandular trichome in determiningcolonization patterns.

Introduction

The living leaf acts as a landing stage for spores and otherpropagules in the air, whether gravity, boundary-layer ex-change or water splash deposits them [13, 20]. Exposed topronounced variation of environmental conditions, the leafis a highly dynamic habitat [1]. This has important conse-quences for its microbial colonizers, which vary consider-ably not only between but also within species, at the levelsof individual plants and leaves, and over long and short timescales [11, 22, 32].

Although the leaf habitat is considered nutrient- andwater-limited [23], it is usually inhabited by high numbersof bacteria, yeasts and other fungi. More than 85 species of

D. Vokou (*) :N. MonokrousosDepartment of Ecology, School of Biology, Aristotle University,54124 Thessaloniki, Greecee-mail: vokou@bio.auth.gr

K. Vareli : J. M. HalleyDepartment of Biological Applications and Technology,University of Ioannina,45110 Ioannina, Greece

E. ZaraliMedical School, University of Ioannina,45110 Ioannina, Greece

K. Karamanoli :H.-I. A. ConstantinidouSchool of Agriculture, Aristotle University,54124 Thessaloniki, Greece

I. SainisInterscience Molecular Oncology Laboratory, Human CancerBiobank Center, University of Ioannina,45110 Ioannina, Greece

Microb Ecol (2012) 64:714–724DOI 10.1007/s00248-012-0053-7

microorganisms belonging to 37 genera have been reportedin the phyllosphere of rye, olive, sugar beet, and wheat [36].More recently, culture-independent methods have shownthat leaf bacterial communities are much more diverse thanpreviously thought [7, 15, 36]. The full extent of bacterialdiversity in the phyllosphere remains unknown [30].

Major questions that have driven ecological research inthe field are whether different plant species harbour differentepiphytic communities and whether the latter can be pre-dicted. To answer these questions, bacterial communities ofdifferent plant species have been described and factors thatmight influence colonization of the leaf habitat have beenanalysed [8, 10, 17, 22, 32–34]. Leaf age, leaf traits, season,and phylogeny are found to play major roles. A correlationwas found between tree phylogeny and bacterial communitycomposition, whereas dispersal constraints were estimatedas less important in structuring the biogeography of micro-bial communities than the biogeography of most plant andanimal communities [22].

Although approaches at the ecosystem level can give abetter understanding of the factors determining and differ-entiating phyllosphere colonization, they have rarely beenapplied. This has been done in a number of studies con-ducted in a Mediterranean ecosystem of northern Greece;these studies examine the epiphytic microbial communitiesof perennial species that grow wild and co-occur and, hence,they are subjected to the same climatic and other macro-environmental conditions and to similar microbial inocula[32–35]. From these studies, conducted with culture-dependent methods, it was found that bacteria on leavesare lognormally distributed; the average size of the leafmicrobial community ranged from the non-detectable to amaximum of 1.4×107 CFU g−1 of leaf tissue [32], but withconsiderable seasonal variation and with inter-species ex-ceeding intra-species variability. With a focus on summer,the most distinctive season, but also the most restrictive togrowth and survival in the Mediterranean environment, itwas shown that leaf water content is the primary explanatoryattribute of epiphytic bacterial abundance followed by phos-phorus content [34]. It was also found that aromatic plantsare on average more highly colonized and sustain morefunctionally diverse epiphytic microbial communities [35].

Using culture-independent methods, the current studyis designed to provide an insight into the compositionand taxa diversity of the leaf bacterial community, underthe microbe-unfriendly conditions of the Mediterraneansummer, and further examine how much the airbornebacteria resemble the epiphytic ones. It also aims atexamining the hypothesis that the aromatic plants ofthis system sustain not only more functionally diverse[35], but also more taxonomically diverse microbialcommunities than the other plant species.

This study addresses an important gap in our knowledge.Published information about the diversity of epiphytic mi-crobial communities, particularly with culture-independentmethods, is scarce, with research in the area being dominat-ed by a few, mainly cultivated species. Comparative studiesof plants coexisting in their natural habitat are especiallyscarce [11, 22]. Also, information about epiphytic microbeshas been limited mainly to annuals or deciduous perennials[6]; much less is known about epiphytes on the leaves ofevergreen species, as most of the species here studied are.

Materials and Methods

Study Area

The study site is located in a Mediterranean-climate area, innorthern Greece (40° 9 N, 23° 54 E). There was a meteoro-logical station operating in the study area only during theyears 1968–1975. To have a better estimate of the area’sclimatic character, we compared data from this station tothose from the meteorological station of Thessaloniki, theclosest major city, for the same years [32]. We found tem-perature data highly correlated (R00.99, P<0.001); similar-ly correlated were rainfall data (R00.54, P<0.001). As theclimatic situation in the study area is adequately describedby the Thessaloniki station, we subsequently assessed datafrom this station over a 20-year period; they showed that themonth of July is the hottest and one of the driest of the year,with mean, maximum, and minimum monthly air temper-atures equalling 26.4°C, 32.1°C, and 21.1°C, respectively,relative humidity 0.59 %, and total solar radiation588.6 mWh cm−1, over this 20-year period.

Plant Material, Leaf, and Air Sampling

We examined the epiphytic microbial communities of nineperennial plant species representing seven plant families(Table 1). Of these species, two are herbaceous and sevenare woody; of the latter, five are evergreen sclerophyllous,dominant in the Mediterranean environment, whereas theremaining two are seasonally dimorphic (Table 1). Fourspecies are clearly aromatic producing essential oils in sub-stantial quantities, on average 0.2–2.0 ml 100 g−1 [35]; ofthese, three belong to the family of Lamiaceae, character-ized by the existence of glandular hairs on their leaf surface,and one to the family of Myrtaceae.

In July 2009, we took samples from each of the abovespecies. Samples consisting of mature leaves of approxi-mately 0.3 g were collected at random from three individu-als per species. Sampled individuals were the same as in theprevious studies of the same system [32–35]. The sampling

Bacterial Diversity of Mediterranean Habitats 715

protocol is described in detail in [35]; the sample size wasrepresentative of the species’ local populations. Two sets ofsamples were taken: (a) for assessing the composition of theepiphytic microbial community with culture-independentmethods and (b) for estimating the epiphytic microbialabundance with culture-dependent methods. The seasonalvariation of the leaf bacterial abundance of these specieswas extensively studied in the past [32] with special atten-tion being given to summer abundance [35]. In the currentwork, we sampled for (b) only to provide a comparative,general estimate of abundance and, therefore, we kept to aminimum (three) the number of samples taken; these sampleswere analysed within 24 h. For (a), five samples were collect-ed at random per sampled individual; all 15 samples for eachspecies (5 per individual × 3 individuals) were pooled.

Air samples were also taken on the spot. We used anAndersen six-stage microbial impaction sampler (Andersen2000 Inc., Atlanta, GA) deployed on a tripod at a height1.5 m above ground. Sampling was repeated twice. Air-borne bacteria were captured on two sets of Andersen platesvia vacuum filtration for 15 min at a flow rate of 28.3 Lmin−1 (a) on sterile cellulose filters (0.22 μm pore size,90 mm diameter, MF-Millipore) placed on top of the platesand (b) on plates containing nutrient agar (NAG) supple-mented with 2.5 % (v/v) glycerol and amended with30 μg ml−1 natamycin to prevent fungal contamination.The Andersen plates were placed in sterile plastic bags inan icebox and transferred to the lab. The filters from the firstset of plates were aseptically transferred to test tubes con-taining sterile phosphate buffer and used for DNA

extraction; the other, media-containing Andersen plates,were incubated under controlled conditions.

To estimate bacterial deposition, four Petri dishes, equalin size to the plates used in the Andersen sampler (88 mminternal diameter) and containing the same media, wereplaced in different microhabitats in the sampling area andremained open for 30 min. They were then closed, placed insterile plastic bags in an icebox, and transferred to the lab forincubation.

Bacteria deposited on both the Petri dishes and theAndersen plates containing NAG were enumerated asColony Forming Units (CFU) following a 2–5-day in-cubation at 24°C. Sampling of both airborne and leafbacteria started 1.5 h after dawn on a practically wind-less day. During sampling, air temperature ranged be-tween 20°C and 23°C.

Estimation of the Leaf Bacterial Populations

Leaf samples were weighed and placed in 10 ml of 0.1 Mpotassium phosphate buffer (pH 7.0) amended with 0.1 %bacto-peptone. Bacteria were recovered from leaves by a10-min sonication in an ultrasonic bath (Transsonic 460,Elma GmbH and Co) with water temperature maintainedbelow 20°C. Serial dilutions of leaf washings were thenplated onto the same media as for airborne bacteria (NAGwith glycerol and natamycin). Bacterial populations wereenumerated as Colony Forming Units (CFU) following a 2–5-day incubation at 24°C.

Table 1 Plant species studiedand their traits, and size of theepiphytic bacterial population(average±standard error), insummer

aAfter [31]

Plant species Plant family Plant habit Leafglandulartrichomea/essential oilcontent(ml 100 g−1)a

Log(CFU+1) g−1

Arbutus unedo L. Ericaceae Woody, evergreensclerophyllous

No/Absence 3.55±0.09

Calamintha nepeta (L.)Savi

Lamiaceae Non-woody perennial Yes/2.0 4.96±0.33

Cistus incanus L. Cistaceae Woody, seasonallydimorphic

No/Traces 2.13±1.07

Lavandula stoechas L. Lamiaceae Woody, seasonallydimorphic

Yes/1.6 4.20±0.08

Melissa officinalis L. Lamiaceae Non-woody perennial Yes/0.2 4.24±0.30

Myrtus communis L. Myrtaceae Woody, evergreensclerophyllous

No/0.3 4.63±0.29

Phillyrea latifolia L. Oleaceae Woody, evergreensclerophyllous

No/Absence 2.77±1.39

Pistacia lentiscus L. Anacardiaceae Woody, evergreensclerophyllous

No/Traces 2.68±1.41

Quercus coccifera L. Fagaceae Woody, evergreensclerophyllous

No/Absence 4.07±0.23

716 D. Vokou et al.

Extraction of DNA from Phyllospheric and AirborneBacteria

Leaf samples for DNA extraction were initially washed in25 ml sterile phosphate buffer (1× PBS, 137 mM NaCl,10 mM phosphate, 2.7 mM KCl, pH 7.4) and then trans-ferred to Erlenmeyer flasks containing 50 ml 1× PBS. Flaskswere sonicated in an ultrasonic cleaner for 10 min; thetemperature of water did not exceed 20°C. Buffer suspen-sions were centrifuged at 9,500 rpm at 4°C for 20 min.

We initially tested three different methods for their abilityto efficiently extract DNA: a method described by Vareli etal. [25], which had proved efficient in lysing tough cyano-bacterial cells and two commercially available kits, the stoolDNA extraction kit (Qiagen) and the ZR soil microbe DNAkit (Zymo research). The three methods were evaluated bycomparing (a) the quantity of extracted DNA, (b) the qualityof DNA (intact or fragmented), (c) the ability of DNA to beamplified by the primers used, and (d) the denaturing gra-dient gel electrophoresis (DGGE) profiles of PCR productscorresponding to differentially extracted DNA samples.Based on these comparisons, we selected the ZR soil mi-crobe DNA kit, which proved overall the most efficient.

DNA was extracted from the cell pellets (phyllosphericbacteria) or from bacteria immobilized on filters (airbornebacteria). Filters containing airborne bacteria were directlysubmerged in the extraction tube of the ZR soil microbeDNA kit. Phosphate buffers, in which filters were sub-merged during transportation to the Lab, were centrifugedat 9,500 rpm at 4°C for 20 min; the resulting pellets werealso added to the same tube. Finally, DNA was extractedaccording to the manufacturer’s instructions. For furtherpurification of the eluates, the Wizard column (Promega,Madison WI, USA) was used according to the manufac-turer’s recommendations. The final eluates were directlyused as template for PCR.

PCR Amplification of 16S rDNA Genes and DGGEAnalysis

Α 100-ng quantity of target DNA was used in each PCRreaction to amplify a fragment of the 16S ribosomal DNAcoding region. Two primers, 341F-GC and 907R, were usedto amplify a 550-bp rDNA fragment, as described earlier[19]. In all cases, a proofreading DNA polymerase was used(Expand High Fidelity DNA polymerase, Roche). PCR am-plified products were loaded onto 6 % (w/v) polyacrylamidegels, 1 mm thick, in 1× TAE [20 mM Tris–acetate (pH 7.4),10 mM acetate, 0.5 mM EDTA] with a denaturing gradientcontaining 20–60 % denaturant [100 % denaturant corre-sponds to 7 M urea and 40 % (v/v) formamide]. Electropho-resis was performed for 16 h at 75 V and the temperaturewas set at 60°C [25].

Cloning and Sequencing

DGGE bands from the fingerprints of all plant species andair were excised and incubated in 50 μl sterile MilliQ waterovernight at 4°C. The eluates were reamplified by using theoriginal primer set and run on a new DGGE gel to confirmtheir identity. After this, the PCR products were subjected todA addition by incubating them for 30 min in the presenceof Taq polymerase (Platinum Taq, Invitrogen). This step iscrucial, since proofreading polymerases, such as Expand-high-fidelity DNA polymerase (Roche) used in this study,do not add dA overhangs at the end of their PCR products.The new PCR products were purified using a Macherey-Nagel DNA clean-up kit (Nucleospin Extract) and weresubsequently cloned using a TOPO TA cloning kit (Invitro-gen) according to the manufacturer’s instructions.

After cloning, ten randomly picked bacterial recombinantclones (white colonies) from each sequence were screenedfor their inserts by EcoR1 digestion. Clone inserts werefurther screened by HaeIII digestion in order to identifydifferent Restriction Fragment Length Polymorphism(RFLP) patterns among them. In all cases, all clones fromthe same band provided a single RFLP pattern. Inserts werefully determined by sequencing both strands. Sequencingwas performed by Macrogen Inc. Seoul, Korea.

Nucleotide Sequences and Accession Numbers

The sequences were deposited at GenBank and wereassigned accession numbers JF274009-JF274036.

Data Analysis

Abundance data were log transformed (log10). Since nobacterial population was detected in three leaf samples,integer 1 was added to all CFU counts before log transfor-mation [32].

The presence or absence of bacterial taxa from the air andthe phyllosphere of the nine species was converted to abinary (0/1) matrix. The Sørensen similarity index (SS)representing similarities between pairs of habitats was cal-culated using the following equation:

SS ¼ 2C

Aþ B;

Table 2 Density of airborne bacteria and deposition rate in the Med-iterranean ecosystem studied

Estimated parameter Bacteria counts

Bacterial deposition rate(average±S.E.)

12.8±5.5 CFU cm−2 h−1 or 12.8×104±5.5×104 CFU m−2 h−1

Bacterial density in the air 1,018 CFU m−3

Bacterial Diversity of Mediterranean Habitats 717

Table 3 Composition of the microbial communities on the summer leaves of nine co-occurring Mediterranean plant species and in the air, asrevealed by 16S rDNA DGGE and sequencing analyses

DGGEbandnumber

Base pairssequenced

Identicalto excisedDGGE band

Closest matching organism Base pairscompared

Similarity (%) Occurrence

1 591 7,12,17,23,31,43,55,61

Leuconostoc citreum HM058995 585 99 A. unedo, C.incanus, C.nepeta, L.stoechas, M.communis,M. officinalis,P. latifolia, P.lentiscus, Q.coccifera

2 590 18,32,56,69 Lactococcus lactis HM462401 584 99 Air, A. unedo,C. netepa, P.lentiscus, Q.coccifera

3 591 8,13,19,24,33,45,57,62

Gut bacterium of Coptotermes formosanus AY533172 585 99 A. unedo, C.incanus, C.nepeta, L.stoechas, M.communis,M. officinalis,P. latifolia, P.lentiscus, Q.coccifera

4 589 9,14,20,25,34,45,58,63

Lactococcus sp. GU272382 582 99 A. unedo, C.incanus, C.nepeta, L.stoechas, M.communis,M. officinalis,P. latifolia, P.lentiscus, Q.coccifera

5 591 10,15,21,26,35,46,59,64

Uncultured bacterium (Lactic acid bacterium yak milkTibet) GQ267983

585 99 A. unedo, C.incanus, C.nepeta, L.stoechas, M.communis,M. officinalis,P. latifolia, P.lentiscus, Q.coccifera

6 589 Lactococcus fujiensins AB485959.1 582 97 A. unedo

11 568 Uncultured bacterium (BBD seasonal changes) GU471982 565 99 M. communis

16 589 Lactococcus fujiensins AB485959.1 584 98 C. incanus

22 568 28,37,50,60,65,73

Uncultured bacterium (BBD seasonal changes) GU471982 567 99 Air, C. nepeta,L. stoechas,M. officinalis,P. latifolia, P.lentiscus, Q.coccifera

27 591 Uncultured Lactococcus clone 42-9BM HQ143303 585 99 L. stoechas

29 564 41 Methylobacterium sp. 602 (phyllosphere of Arabidopsisthaliana) FN868956

558 100 C. nepeta, L.stoechas

30 590 42,53 Erwinia sp. HQ154566.1 583 100 C. nepeta, L.stoechas, M.officinalis

718 D. Vokou et al.

where A is the number of phyllosphere bacteria of one plantspecies, B is the number of phyllosphere bacteria of theother species or of bacteria in the air, and C is the numberof common bacteria of the two plant species or of one plantspecies and air. Habitat relationships and groupings werefurther identified by applying the clustering algorithmUPGMA (unweighted pair-group method using arithmeticaverage) and multidimensional scaling (MDS) ordination[24]. For all statistical analyses, the Statistica 7 software(Statsoft, Tulsa, USA) was used.

Results

Phyllosphere and Airborne Bacterial Abundance

The total size of the epiphytic bacterial populations rangedon average from 1.3×102 CFU g−1 leaf tissue, in Cistusincanus, to 9.1×104, in Calamintha nepeta (Table 1). Air-borne microbes were on average deposited at a rate of97.3 CFU per 30 min on a Petri dish of 60.8 cm2. Thiscorresponds to an hourly bacterial deposition rate of12.8 CFU cm−2 h−1 (or 12.8×104 CFU m−2 h−1). Regardingthe Andersen sampler, in 424.5 l of air sampled (flow rate of28.3 l min−1×15 min), there were a total of 432 CFU; this

corresponds to a concentration of bacteria in the air ofapproximately 103 CFU m−3 (Table 2).

Denaturing Gradient Gel Electrophoresis Profiles of Leafand Airborne Bacteria and Diversity of the BacterialCommunities

From the genotype composition analysis of the leaf and theairborne microbial communities (Table 3), 28 bacterial taxawere detected (Fig. 1). Of these, two were found both in theair and the phyllosphere, eight only in the air, whereas18only in the phyllosphere. The epiphytic communities thatwere the richest in bacterial taxa were those of C. nepeta andMelissa officinalis with 12, followed by that of Lavandulastoechas with 8 taxa. In all other epiphytic communities,only six or five bacterial taxa were detected (Fig. 2).

The profiles of the bacteria participating in the airborneand the phyllosphere microbial communities are shown inFig. 3. Following the denaturing gradient gel electrophoresisof a 16S rDNA gene fragment, both prominent and faintbands were excised, re-amplified, cloned, and sequenced, intotal 75 bands. Results of the sequencing analyses andfollowing BLAST searches are presented in detail in Table 3.In brief, Leuconostoc citreum, Lactococcus sp., an uncul-tured lactic acid bacterium and a gut bacterium were

Table 3 (continued)

DGGEbandnumber

Base pairssequenced

Identicalto excisedDGGE band

Closest matching organism Base pairscompared

Similarity (%) Occurrence

36 581 49 Uncultured bacterium (apple phyllosphere) HM450033 575 98 C. nepeta, M.officinalis

38 565 Sphingomonas paucimobilis (Lake Honghu)GU204960

559 100 C. nepeta

39 589 Aquitalea sp. EU589414 583 99 C. nepeta

40 590 52 Lactobacillus delbrueckii HM058018 583 99 C. nepeta, M.officinalis

47 583 Uncultured bacterium (kerosene BTEX) FJ669110 577 99 M. officinalis

48 591 Streptococcus lutetiensis HQ293091 585 100 M. officinalis

51 582 Uncultured bacterium (Lake Honghu) HM487995 574 98 M. officinalis

54 565 Roseomonas stangi (pond water, Japan) AB369258 560 97 M. officinalis

66 562 Uncultured Acidobacterium sp. (Termite I) AM491117 550 97 Air

67 565 Uncultured Firmicutes sp. (on pages of Leonardo daVinci’s Atlantic Codex) HM231345

556 97 Air

68 589 Delftia tsuruhatensis HM587796.1 583 99 Air

70 567 Uncultured bacterium (phyllosphere) GU225979 563 97 Air

71 569 Uncultured bacterium (phyllosphere) GU225979 563 97 Air

72 588 Uncultured bacterium (from compost MFC non-polarized anode) ONC012 JN795213

582 99 Air

74 564 Uncultured Firmicutes sp. (on pages of Leonardo daVinci’s Atlantic Codex) HM231345

559 98 Air

75 572 Brevibacterium sp. (from Lechuguilla Cave) LC486JN863519

566 99 Air

Bacterial Diversity of Mediterranean Habitats 719

detected on all nine plant species. Interestingly, these verycommon taxa were not found amongst the airborne bacteria.The two genotypes, detected from both leaf extracts and air,correspond to Lactococcus lactis and an unculturedbacterium.

Ten genotypes were very narrowly distributed, only onleaves from one species (Fig. 1, Table 3). The identified onesare a Roseomonas stangi strain from M. officinalis, Sphingo-monas paucimobilis and Aquitalea strains from C. nepeta, aLactococcus fujiensis strain from Arbutus unedo, and anotherLactococcus strain from L. stoechas. A genotype isolatedfromM. communis is homologous to the uncultured bacteriumGU471982, but slightly different from other genotypes alsohomologous to this bacterial taxon (Table 3).

Four genotypes were found on leaf surfaces of only a fewspecies (Fig. 1, Table 3). These were a Methylobacteriumgenotype from L. stoechas and C. nepeta, Lactobacillusdelbruckii and an uncultured bacterium from C. nepeta

and M. officinalis, as well as an Erwinia strain from L.stoechas, C. nepeta, and M. officinalis.

Airborne genotypes are 97–99 % homologous to alreadydeposited sequences in GenBank/EMBL/DDBJ. Apart fromthe two taxa that were also detected in the phyllosphere,they correspond to Delftia, Acidobacterium, and Brevibac-terium strains, two uncultured bacteria and one Firmicutesrepresentative (Table 3).

Similarity of Bacterial Communities

The Sørensen similarity index of the epiphytic bacterialcommunities ranges between 0.44 and 1.0 (Table 4). Therewas less than 50 % similarity for six pairs of species allmade from representatives of different plant habit groups.Full or almost full similarity (>90 %) was recorded for thepairs of Quercus coccifera and Pistacia lentiscus, Q. cocci-fera and Phillyrea latifolia, and P. lentiscus and P. latifolia,i.e. between evergreen sclerophyllous species. The similar-ity of airborne bacteria to epiphytic ones was always low(<30 %); there was no similarity at all with C. incanus, A.unedo, and M. communis.

Results of MDS ordination and UPGMA clustering [24]of the nine plant species/habitats and the air, based on theSørensen similarity matrix of their microbial communities(Table 4), are presented in Fig. 4, respectively. The airbornemicrobial community, very dissimilar from the epiphyticones, is ordinated to the far-right end of the graph, whereasphyllosphere communities are all ordinated to the left, along

Figure 1 Number of different habitats colonized by each of the 28bacteria taxa detected in the Mediterranean system studied; taxa areranked according to the number of habitats they colonized

Figure 2 Number of bacteria taxa harboured by each Mediterraneanhabitat examined, i.e. the air and the leaves of the nine native perennialspecies

Figure 3 Denaturing gradient gel electrophoresis (DGGE) profiling of a16S rDNA gene fragment of the microbial communities on the leaves ofnine perennial plant species and of airborne bacteria in a Mediterraneanecosystem. All dotted bands were excised, reamplified, and sequenced.Dotted bands are numbered consecutively from the top to the bottom andfrom left to right; at the bottom of the profiles’ pictures, given are the bandnumbers corresponding to each specific habitat, i.e. plant species or air

720 D. Vokou et al.

the y axis (Fig. 4b). UPGMA clustering resulted in threemajor groups (Fig. 4a); air makes a distinct group, allevergreen sclerophyllous species are grouped together alongwith the drought-semideciduous, C. incanus, whereas thethree Lamiaceae aromatic plants form the third group.

Discussion

In agreement with previous reports [34, 35], the size of thebacterial colonization of the Mediterranean phyllosphere, insummer, was low; in all cases, it was lower than105 CFU g−1. Of the phyllosphere bacteria detected, eightare members of the lactic-acid group, of which five belongto the genus Lactococcus. This makes lactic acid bacteriathe most abundant group (∼30 %) and Lactococcus the mostabundantly represented taxon. A gut bacterium, first isolatedfrom the termite Coptotermes formosanus, was found on theleaves of all nine perennial species. Since the first time thatthe truly epiphytic relationship of an Enterococcus specieswas revealed [18], the use of the leaf habitat by naturalintestine inhabitants has been verified in several cases. Suchbacteria have been identified from different leaf surfaces;from the leaves and also twigs, drupes and brines of olive[12], 193 bacteria have been isolated and identified asEnterococcus, Leuconostoc spp., and Lactobacillus planta-rum. As olive is a typical evergreen sclerophyllous Medi-terranean plant, results concerning microbial colonization ofits different parts including leaves are in agreement with thehigh participation of lactic acid bacteria in the Mediterra-nean phyllosphere reported here. The fact that these bacteriaare the major microbial colonizers of the harsh environmentof the summer Mediterranean phyllosphere suggests partic-ular survival skills and hence the need for closer attention tothis group of bacteria and their ability to persist in stressfulenvironments.

Erwinia sp. was detected in the epiphytic community ofonly three species, all belonging to the group of aromaticplants; of these, only L. stoechas is woody. This finding is in

agreement with previous studies showing that ice-nucleation-active bacteria are not common on perennial woody species inMediterranean-climate areas [9, 32].

When Bowers et al. [3] collected aerosol samples fromdifferent land-use types, assessed the composition of theairborne bacterial communities, and meta-analysed previ-ously published data, they found composition to be relatedto land-use type and to be dominated by sequences assignedto Firmicutes, Proteobacteria and Actinobacteria. In ourstudy area, representatives of Firmicutes were the mosthighly represented phylum. Proteobacteria and Actinobac-teria were also represented and we detected, in addition, arepresentative of Acidobacteria, which is a novel phylum,consistently detected by 16S rDNA-based molecular sur-veys, in many different habitats around the globe [21].

Only 20 bacterial taxa were detected from the nineplant species. This is a low number for epiphytic bac-teria even taking into consideration the fact that theDGGE analysis cannot detect the full microbial diversitybut only the portion corresponding to rather abundantlyrepresented taxa. This low number could be attributed tothe harsh conditions prevailing. Summer is the leastmicrobe-friendly season in the Mediterranean environ-ment; under hot and dry conditions, microbes cannotproliferate and survival is confined to tolerant strains.This is reflected both in the recorded abundance and thediversity of the epiphytic communities.

Sharing only two of the 28 taxa detected, airborne andphyllosphere bacteria differed profoundly. This shows thatbacterial colonization of the phyllosphere is not alwaysrelated to bacterial abundance in the air. Previous researchersreported that the atmospheric bacterial communities in thedifferent environments are more similar among themselvesthan to the bacterial communities of their corresponding po-tential sources, the terrestrial habitats of leaves and soil [3],and suggested distant sources of bacteria [5]. Given the lowpopulation sizes of bacteria on plants in the middle of thesummer, it is not entirely surprising that they might not havebeen a major source of the bacteria found in the atmosphere:

Table 4 Sørenson similarity matrix generated from the DGGE fingerprints of the nine plant species studied and the air in a Mediterranean ecosystem

M. communis P. latifolia P. lentiscus Q. coccifera C. incanus L. stoechas C. nepeta M. officinalis Air

A. unedo 0.73 0.73 0.83 0.83 0.73 0.57 0.44 0.44 0

M. communis 0.80 0.73 0.73 0.80 0.62 0.47 0.47 0

P. latifolia 0.91 0.91 0.80 0.77 0.59 0.59 0.13

P. lentiscus 1 0.73 0.71 0.67 0.56 0.25

Q. coccifera 0.73 0.71 0.67 0.56 0.25

C. incanus 0.62 0.47 0.47 0

L. stoechas 0.70 0.60 0.11

C. nepeta 0.67 0.18

M. officinalis 0.09

Bacterial Diversity of Mediterranean Habitats 721

the portion of bacteria expected to leave a leaf at a given timemight be relatively too small in their case.

The estimated deposition rate of bacteria from the atmo-sphere onto foliage was quite high, despite the low abundanceand diversity of the epiphytic communities. This indicateslittle multiplication of bacteria on leaves and suggests stresssurvival as a major factor determining epiphytic microbialcommunities; in fact, it was found that the major determinantof epiphytic microbial abundance in this Mediterranean eco-system, in summer, is leaf water content [34]. This alsosuggests that only few of the immigrants can establish andpersist in this environment or, alternatively, that incomingbacteria survive on the leaf habitat at such low sizes that theymay become non-detectable. In a previous work [32] regard-ing the same system, it was found that for some species, 40 %

or more of the summer samples do not bear detectable pop-ulations. As the number of ‘zeroes’ (absences) was consistentwith the picture that would be expected with a lognormaldistribution of the epiphytic bacterial population, it was ar-gued that zeroes are more a consequence of the inability of themeasurement process to detect very low populations ratherthan an indication of complete absence of bacteria from leaves[32]. The sampling protocol for the current work does notallow us to assess the variability underlying diversity, asit was done for variability underlying abundance, and so,we cannot make inferences about the complete absence ofbacteria from air on leaves and vice versa or presence invery low sizes. Nevertheless, it allows adequate portrayalof the substantive members of the airborne and epiphyticbacterial communities.

Figure 4 Relationships of theair and the nine co-occurringperennial plant species afterUPGMA clustering (a) andMDS ordination (b), on thebasis of the composition of themicrobial communities usingthem as habitats; the Sørensensimilarity matrix was used torun UPGMA and MDS

722 D. Vokou et al.

According to Lindow [14], most leaves in the field aresubject to bacterial immigration of about 104 cells per monthand, therefore, a substantial proportion of epiphytic bacteriaare just temporary immigrants and not real colonizers. Wefound a deposition rate of 12.8 CFU cm−2 h−1 equalling to9,216 CFU cm−2 per month. As the leaf surface of the ninespecies ranges from one to much less than a hundred squarecentimetres [31], this value is very close to the bacterialimmigration that leaves are subjected to, as estimated byLindow [14]. He also suggested that plant species withepiphytic bacterial populations less than 104 CFU peraverage-sized leaf should be considered as only incidentalhosts. The size of the bacterial populations on the leaves ofthe perennial Mediterranean species is of similar or evenlower magnitude. More specifically, we report here bacterialcolonization on a leaf weight basis. If we take into consid-eration the specific leaf mass of these species ranging from6.1 mg cm−2 for M. officinalis to 15.6 for Q. coccifera [31],the resulting colonization values per surface unit wouldclearly place these species into the group of incidental hosts[14]. However, the remarkable difference between the com-position of the air inoculum and the epiphytic communitiessuggests that these plant species could be real hosts. Fur-thermore, the fact that the majority of bacteria taxa detected(65 %) was found on only one or two plant species could beregarded as an indication of host specificity.

Only 20 % of the detected phyllosphere bacterial taxawas common to all nine coexisting perennial species, where-as in 30 % of the species pairs, the similarity was lower than0.6. Clustering analysis showed that the group of evergreensclerophyllous species was the most homogeneous. In con-trast, aromatic plants were split; three were grouped togeth-er, whereas the evergreen-sclerophylous M. communis wasgrouped along with all other species of the same habit. In astudy of nine tree species of the Atlantic forest in Brazil[11], a much lower similarity of the epiphytic communitieswas recorded, with only 0.5 % of the bacteria detected beingcommon to all nine trees. Such results support the argumentthat phyllosphere bacteria are not passive inhabitants of theleaf surface where they are deposited and that plant specieshave a strong direct or indirect influence on the structure andcomposition of their leaf-associated bacterial communities[22].

The epiphytic communities of the aromatic, non-woodyperennials C. nepeta and M. officinalis, each consisting of12 taxa, were the most diverse, followed by that of L.stoechas; all three belong to Lamiaceae. The leaves of M.communis (Myrtaceae) differ from those of the Lamiaceaerepresentatives in that they are waxy and devoid of trichome[31]; hairs are observed on C. nepeta, M. officinalis, L.stoechas, and C. incanus, but glandular ones only on thefirst three [31]. Structures like hairs, where microbes canadhere, are regarded as microhabitats that give colonization

opportunities to different bacteria taxa [2, 15, 16]. If thesestructures are associated with different food sources, habitatdifferentiation may become even more pronounced withconsiderable effects on the epiphytic bacterial community.

As aromatic plants with leaf surface glands harbour morediverse microbial communities than the other componentsof the same ecosystem, we argue that the glandular trichomemay act as a microbial-diversity generating mechanism,adding to the spatial heterogeneity of the leaf habitat andoffering an opportunity to more bacterial strains to establishand coexist.

Despite the well-known antibacterial activity of essentialoils, bacteria not only tolerant to the presence of essentialoils and their individual constituents but also able to usethem as carbon and energy sources are found in soils fromseveral different ecosystem types [4, 26–29]. This could bealso the case for phyllosphere bacteria. As essential oils aremixtures of many different compounds, glandular hairscould function as reservoirs of different food sources grant-ing the opportunity to food specialists to establish andpersist. If this is so, competition among such bacteria wouldbe weaker than for generalists competing for commonresources. In turn, this would allow for a higher diversityof the epiphytic microbial community. Experiments will bedesigned to test this hypothesis involving aromatic plantsthat differ in essential oil content and composition. Furtherto that, studies at different times of the year can showwhether a high microbial diversity on the gland-bearingleaves of aromatic plants is a permanent feature or only aseasonal response to the stress imposed by the Mediterra-nean summer.

Acknowledgements We thank Prof. S. E. Lindow, University ofCalifornia, Berkeley, for his constructive comments on an earlierversion of our manuscript.

References

1. Andrews JH (1992) Biological control in the phyllosphere. AnnuRev Phytopathol 30:603–635

2. Baldotto LEB, Olivares FL (2008) Phyloepiphytic interaction be-tween bacteria and different plant species in a tropical agriculturalsystem. Can J Microbiol 54:918–931

3. Bowers RM, McLetchie S, Knight R, Fierer N (2011) Spatialvariability in airborne bacterial communities across land-use typesand their relationship to the bacterial communities of potentialsource environments. ISME J 5:601–612

4. Chalkos DG (2010) Effects of secondary metabolites of aromaticplants on the structure and metabolism of the soil microbial com-munities. PhD thesis. Aristotle University of Thessaloniki: The-ssaloniki, Greece (in Greek with English summary)

5. Constantinidou ΗΑ, Hirano SS, Baker LS, Upper CD (1990)Atmospheric dispersal of ice nucleation-active bacteria: the roleof rain. Phytopathology 80:934–937

6. de Jager ES, Wehner FC, Korsten L (2001) Microbial ecology ofthe mango phylloplane. Microb Ecol 42:201–207

Bacterial Diversity of Mediterranean Habitats 723

7. Delmotte N, Knief C, Chaffron S, Innerebner G, Roschitzki B,Schlapbach R, von Meringb C, Vorholta JA (2009) Communityproteogenomics reveals insights into the physiology of phyllo-sphere bacteria. Proc Natl Acad Sci USA 106:16428–16433

8. Fierer N, McCain CM, Meir P, Zimmermann M, Rapp JM, SilmanMR, Knight R (2011) Microbes do not follow the elevation diver-sity patterns of plants and animals. Ecology 92:797–804

9. Karamanoli K, Vokou D, Menkissoglu O, Constantinidou HΙ(2000) Bacterial colonization of the phyllosphere of Mediterraneanaromatic plants. J Chem Ecol 26:2035–2048

10. Knief C, Ramette A, Frances L, Alonso-Blanco C, Vorholt JA(2010) Site and plant species are important determinants of theMethylobacterium community composition in the plant phyllo-sphere. ISME J 4:719–728

11. Lambais MR, Crowley DE, Cury JC, Bull RC, Rodrigues RR(2006) Bacterial diversity in tree canopies of the Atlantic forest.Science 312:1917–1917

12. Lavermicocca P, Gobbetti M, Corsetti A, Caputo L (1998) Char-acterization of lactic acid bacteria isolated from olive phylloplaneand table olive brines. Ital J Food Sci 10:27–39

13. Lindemann J, Constantinidou HΑ, Barchet WR, Upper CD (1982)Plants as sources of airborne bacteria, including ice nucleation-active bacteria. Appl Environ Microbiol 44:1059–1063

14. Lindow SE (1996) Role of immigration and other processes indetermining epiphytic bacterial populations: implications for dis-ease management. In: Morris CE, Nicot PC, Nguyen C (eds) Aerialplant surface microbiology. Plenum, New York, pp 155–168

15. Mariano RLR, McCarter SM (1993) Epiphytic survival of Pseu-domonas viridiflava on tomato and selected weed species. MicrobEcol 26:47–58

16. Mew TW, Vera Cruz CM (1986) Epiphytic colonization of hostand non-host plants by phytopathogenic bacteria. In: Fokkema NJ,van den Heuvel J (eds) Microbiology of the phyllosphere. Cam-bridge University Press, New York, pp 269–282

17. Muller T, Behrendt U, Muller M (1996) Antagonistic activity inplant-associated lactic acid bacteria. Microbiol Res 151:63–70

18. Mundt JO, Coggins JH, Johnson LF (1962) Growth of Streptococcusfaecalis var. liquefaciens on plants. Appl Microbiol 10:552–555

19. Muyzer G, Teske A, Wirsen CO, Jannasch HW (1995) Phyloge-netic relationships of Thiomicrospira species and their identifica-tion in deep-sea hydrothermal vent samples by denaturing gradientgel electrophoresis of 16S rDNA fragments. Arch Microbiol164:165–171

20. Pugh GJF, Buckley NJ (1971) The leaf surface as substrate forcolonization by fungi. In: Preece TF, Dickinson CH (eds) Ecologyof leaf surface micro-organisms. Academic, London, pp 431–445

21. Quaiser A, Ochsenreiter T, Lanz C, Schuster SC, Treusch AH, EckJ, Schleper C (2003) Acidobacteria form a coherent but highlydiverse group within the bacterial domain: evidence from environ-mental genomics. Mol Microbiol 50:563–575

22. Redford AJ, Bowers RM, Knight R, Linhart Y, Fierer N (2010)The ecology of the phyllosphere: geographic and phylogeneticvariability in the distribution of bacteria on tree leaves. EnvironMicrobiol 12:2885–2893

23. Schönherr J, Baur P (1996) Cuticle permeability studies: a modelfor estimating leaching of plant metabolites to leaf surfaces. In:Morris CE, Nicot PC, Nguyen C (eds) Aerial plant surface micro-biology. Plenum, New York, pp 1–23

24. Sokal R, Michener C (1958) A statistical method for evaluatingsystematic relationships. Univ Kans Sci Bull 38:1409–1438

25. Vareli K, Briasoulis E, Pilidis G, Sainis I (2009) Molecular con-firmation of Planktothrix rubescens as the cause of intense, micro-cystin-synthesizing cyanobacterial bloom in Lake Ziros, Greece.Harmful Algae 8:447–453

26. Vokou D (2007) Allelochemicals, allelopathy and essential oils: afield in search of definitions and structure. Allelopathy J 19:119–135

27. Vokou D, Chalkos D, Karamanoli K (2006) Microorganisms andallelopathy: a one-sided approach. In: Reigosa MJ, Pedrol N,Gonzalez L (eds) Allelopathy: a physiological process with eco-logical implications. Springer, Dordrecht, pp 341–371

28. Vokou D, Margaris NS (1988) Decomposition of terpenes by soilmicroorganisms. Pedobiologia 31:413–419

29. Vokou D, Margaris NS, Lynch JM (1984) Effects of volatile oilsfrom aromatic shrubs on soil microorganisms. Soil Biol Biochem16:509–513

30. Whipps JM, Hand P, Pink D, Bending GD (2008) Phyllospheremicrobiology with special reference to diversity and plant geno-type. J Appl Microbiol 105:1744–1755

31. Yadav RKP, Bosabalidis AM, Vokou D (2004) Leaf structuralfeatures of Mediterranean perennial species: plasticity and lifeform specificity. J Biol Res 2:21–34

32. Yadav RKP, Halley JM, Karamanoli K, Constantinidou HI, VokouD (2004) Bacterial populations on the leaves of Mediterraneanplants: quantitative features and testing of distribution models.Environ Exp Bot 52:63–77

33. Yadav RKP, Karamanoli K, Vokou D (2011) Bacterial populationson the phyllosphere of Mediterranean plants: influence of leaf ageand leaf surface. Front Agr China 5:60–63

34. Yadav RKP, Karamanoli K, Vokou D (2005) Bacterial colonizationof the phyllosphere of Mediterranean perennial species as influ-enced by leaf structural and chemical features. Microb Ecol50:185–196

35. Yadav RKP, Papatheodorou EM, Karamanoli K, ConstantinidouH-IA, Vokou D (2008) Abundance and diversity of the phyllo-sphere bacterial communities of Mediterranean perennial plantsthat differ in leaf chemistry. Chemoecology 18:217–226

36. Yang CH, Crowley DE, Borneman J, Keen NT (2001) Microbialphyllosphere populations are more complex than previously real-ized. Proc Natl Acad Sci USA 98:3889–3894

724 D. Vokou et al.