Distinctive Phyllosphere Bacterial Communities in Tropical Trees
- First Online:
- Cite this article as:
- Kim, M., Singh, D., Lai-Hoe, A. et al. Microb Ecol (2012) 63: 674. doi:10.1007/s00248-011-9953-1
- 793 Views
Recent work has suggested that in temperate and subtropical trees, leaf surface bacterial communities are distinctive to each individual tree species and dominated by Alpha- and Gammaproteobacteria. In order to understand how general this pattern is, we studied the phyllosphere bacterial community on leaves of six species of tropical trees at a rainforest arboretum in Malaysia. This represents the first detailed study of ‘true’ tropical lowland tree phyllosphere communities. Leaf surface DNA was extracted and pyrosequenced targeting the V1–V3 region of 16S rRNA gene. As was previously found in temperate and subtropical trees, each tree species had a distinctive bacterial community on its leaves, clustering separately from other tree species in an ordination analysis. Bacterial communities in the phyllosphere were unique to plant leaves in that very few operational taxonomic units (0.5%) co-occurred in the surrounding soil environment. A novel and distinctive aspect of tropical phyllosphere communities is that Acidobacteria were one of the most abundant phyla across all samples (on average, 17%), a pattern not previously recognized. Sequences belonging to Acidobacteria were classified into subgroups 1–6 among known 24 subdivisions, and subgroup 1 (84%) was the most abundant group, followed by subgroup 3 (15%). The high abundance of Acidobacteria on leaves of tropical trees indicates that there is a strong relationship between host plants and Acidobacteria in tropical rain forest, which needs to be investigated further. The similarity of phyllosphere bacterial communities amongst the tree species sampled shows a significant tendency to follow host plant phylogeny, with more similar communities on more closely related hosts.
Molecular study of bacteria, concentrating mainly on the 16S rRNA gene, has already identified at least some tentative broad trends in bacterial diversity and community composition in soils [17, 24, 27]. However, there has been little study of bacterial communities on the surfaces of plants, for example leaf surfaces—an environment known as the ‘phyllosphere’ . The phyllosphere is one of the largest microbial habitats on earth, with a total surface area estimated at over 4 × 108 km2 . Although a wide variety of microbes, such as fungi, archaea, and algae are found on leaves, bacteria have long been known to be the dominant member inhabiting plant leaves . Phyllosphere bacteria have adapted to the leaf surface, interacting with plants both positively and negatively or neutral (e.g. commensal bacteria). Some bacteria such as Pseudomonas and Xanthomonas cause serious diseases in plants; in contrast, others may be beneficial to plants by producing plant growth hormones and suppressing the colonization of bacterial pathogens [22, 29]. Several functional roles have been described so far, including nitrogen fixation, nitrification, and methanol degradation, although the importance of these processes for the host plants and for the ecosystems they inhabit are still poorly understood [5, 26, 28].
Although phyllosphere bacteria might have an important role in biogeochemical cycles as well as affecting their host plants, little is known about the extent of diversity and ecology of this phyllosphere community. Studies of phyllosphere bacteria extend well back in time before culture-independent methods were available . Early work was based on culturing, but this has since proven to be a very incomplete way of assessing bacterial communities . Cultivation-independent studies largely performed on the 16S rRNA gene have shown that the phyllosphere bacterial diversity is far greater than was previously recognized. To date, the level of diversity of phyllosphere bacteria has been found to be rather lower than that of other environments such as soil and seawater, and similar to that of human gut . Previous culture-independent approaches have shown that Alpha- and Gammaproteobacteria are generally the dominant groups of bacteria colonizing leaf surfaces and Betaproteobacteria, Bacteroidetes, Firmicutes, and Actinobacteria are also found in a high proportion, although their abundance level varies depending on plant species and circumstances [7, 11, 32]. On the other hand, Acidobacteria and Cyanobacteria are considered almost absent or infrequently present inhabitants of leaves [16, 32].
There has been considerable work on the phyllosphere of crops, revealing for example that bacterial community composition varies between species and even between different cultivars of the same species [2, 6, 29]. Fewer studies have focused on trees, and the phyllosphere of tropical forest trees has remained largely unknown despite the rainforests being regarded as the acme of diversity. The closest that molecular studies have come is work on the subtropical Atlantic rainforest of Brazil , at 23° N falling outside the conventional definition of tropical rainforest . That study revealed a high percentage of uncultured bacterial species and bacterial communities that clustered according to host tree species. To date, there has been no study which tried to understand the whole diversity of phyllosphere bacteria dwelling on leaves of tropical trees, or trees in Southeast Asia—so this remains a large unknown area.
From a functional viewpoint, there is a priori evidence that phyllosphere bacteria in tropical rainforest might be more important than in cooler areas, in that rates of nitrogen fixation on leaves by phyllosphere bacteria were higher in tropical forest than temperate forest .
Recent work by Redford et al.  on a range of temperate tree species has further suggested that tree species each have their own distinctive assemblages of bacteria, which may remain fairly constant even when they are planted in different regions of the world. The study by Redford et al.  also raised the interesting observation that the degree of similarity of bacterial communities on different tree species depends upon the taxonomic affinities of the host trees.
We set out to study the phyllosphere of tropical rainforest trees in Southeast Asia, to investigate two questions: (1) Is the phyllosphere community as species specific as in subtropical trees , temperate trees , and crop plants ? (2) Are the same groups of bacteria dominant in the tropical tree phyllosphere as in temperate and subtropical trees? In this study, six tree species which are common in the rainforests of the Malay Peninsula were selected and compared in terms of the composition and diversity of bacteria in each sample, using 16S rRNA gene-based pyrosequencing.
Materials and Methods
Site Description and Sample Preparation
We chose trees growing in a tropical forest arboretum at the Forest Research Institute of Malaysia (FRIM), at Kepong near Kuala Lumpur (101 37′34″ E to 101°37′34″ E). The arboretum was started in the 1930s and features over 100 species of trees native to SE Asia, many as multiple specimens, each identified by a name tag. Most trees are widely spaced (>10 m) with the ground kept clear of any understory. Shade and leaf litter are sufficient to eliminate most grassy cover. Nevertheless, in many ways the arboretum provides a rainforest-like environment. Tree crowns are touching or almost touching, temperature and humidity levels are similar to tropical rainforest (Lai-Hoe, personal observation), and the site is surrounded on all sides by intact rainforest. The FRIM site has a true tropical equatorial climate, with rainfall (ca. 2,600 mm) exceeding potential evapotranspiration in all months of the year, and peaks in rainfall in March and October. The mean annual temperature is 26.5°C, with only 0.1°C variation in mean temperature during the months of the year. The FRIM arboretum is on gently sloping ground covering around 50 ha.
The wide spacing and ease of tree species identification of the trees facilitates standardized sampling. Whereas in a true rainforest environment, trees of different species and growth stages grow closely intermingled, dripping water from one to another, in the arboretum, the trees are spaced widely enough that this cross-contamination is not a problem. Individual trees of each species were randomly chosen in the arboretum to exclude a possible effect of the spatial distance on phyllosphere bacterial community. We chose multiple specimens of six species of trees to sample, all of them native to the rainforests of the Malay Peninsula. Our choice of species was intended to give a taxonomic spread of species, including the gymnosperm Gnetum and the monocot giant bamboo Schizostachyum as outliers from the main eudicot group of the other trees. All the trees were sampled on a day in February 2010, during the period of transition to the first rainfall peak of the year. Rain showers had occurred on each of the five preceding days before sampling.
Low-hanging branches within 2 m of the ground were sampled. To standardize conditions as much as possible, we chose only green healthy-looking leaves with less than 20% herbivory damage, growing as shade leaves in a position under the canopy in which they were unlikely to be in direct sunlight during the middle hours of the day (9 a.m.–5 p.m.). Shoots of around 30–40 leaves were cut off by field workers wearing sterile surgical gloves (changed after each sampling) and cut with a sterilized knife. The branch was placed in a sterile plastic bag and kept at 20–25°C for up to 12 h until processing for bacterial DNA. About 40 g of fully developed leaf samples collected from each tree was placed in 50-m sterile tubes (100-ml tubes for relatively bigger leaves) and filled with 30 ml (60 ml for 100 ml tube) of sterile TE buffer (pH 7.5, 10 mM Tris, 1 mM EDTA) supplemented with 0.2% Silwet L-77 (GE Bayer Silicones) . To wash the bacterial cells off the leaves, shaking (350 rpm), vortexing, and sonicating every 30 s were conducted with alternations for 3 min. The cell suspension was filtered through a nylon mesh (pore size, 200 μm, Spectrum Europe BV) to separate microbial cells from plant material and then centrifuged for 15 min at 3,150×g at 4°C. Cell pellets from multiple tubes were pooled and washed twice with TE buffer. The resulting pellets were stored at −80°C.
In order to compare the composition and the diversity level of bacteria between phyllosphere, rhizoplane, and rhizosphere, we used datasets of 11 rhizoplane and 27 rhizosphere samples collected from FRIM (work in preparation).
Pyrosequencing and Data Processing
DNA was extracted from each of the collected leaf samples using Power Soil DNA extraction kit (MO BIO Laboratories, Carlsbad, CA, USA) as directed by manufacturer’s instructions, with an additional incubation step at 70°C for 10 min followed by 5 min of bead beating to recover more cells, which are difficult to lyse, and reduce DNA shearing. Isolated DNAs were stored at −20°C. DNAs isolated from each sample were amplified using primers (9F and 541R) targeting the V1 to V3 regions of the bacterial 16S rRNA and PCR reactions were carried out as described previously . The DNA sequencing was performed by Chunlab Incorporation (Seoul, Korea) using 454 GS Junior Sequencing System (Roche), according to the manufacturer’s instructions. Sequences were processed and analysed following the procedures described previously . All sequences were clustered using CD-Hit at 97% sequence similarity cut off to do further analyses such as rarefaction curve and general description of diversity estimates . All sequences were classified using EzTaxon-extended database (http://eztaxon-e.ezbiocloud.net/) and sequences belong to Acidobacteria were reclassified using RDP II classifier, which provides a standard nomenclature scheme by dividing all Acidobacterial sequences into 26 subgroups. All 454 sequences were deposited in Short Read Archive (SRA) at NCBI with an accession number of SRA045953.
To test whether there is a significant difference in bacterial community composition among groups of sampling units, we used the analysis of similarity (ANOSIM) with 999 permutations in PRIMER v6 . We performed ordination studies to determine the relative similarities of the samples within and between tree species using the UniFrac algorithm , which provides a phylogenetic metric of community distance by measuring the fraction of branch length unique to each sample across constructed phylogentic trees. UniFrac distances were calculated based on a phylogenetic tree of randomly chosen subsets (n = 1,000/subset) of 20 samples. Sequences were aligned against SILVA-aligned reference sequences (http://www.arb-silva.de/) using Mothur software and then the maximum likelihood (ML) tree was inferred using RAxML . RAxML (v.7.2.7) with GTR + CAT model was done on CIPRES Portal 2 . Non-metric multidimensional scaling (NMDS) was generated using Bray–Curtis index of similarity, which is based on traditional ecological species/sample matrix, and then compared with the result of UniFrac-based ordination.
We used a Mantel test to test if there is an association between bacterial community and plant phylogeny. The pairwise UniFrac distances between bacterial communities and phylogenetic distances between tree species were compared. Plant phylogenetic distance matrix of five different tree species (Dyera samples were not included in the analysis due to low sequencing reads) was generated using Phylocom . All graphs were generated using R packages (http://www.R-project.org).
For the 20 tree samples spanning six tree species, we generated a total of 21,930 sequencing reads which affiliated to bacteria only, with an average length of 431 bp after quality filtering and trimming processes. The average number of sequences per sample was 1,327—ranging from 553 to 2,281 (four additional Dyera samples were not counted due to low reads less than 300). An average of 296 operational taxonomic units (OTUs) per sample was identified at the level of 97% similarity, corresponding to the standard definition of a bacterial ‘species’. Abundant bacterial phyla include Alphaproteobacteria (27%), Acidobacteria (17%), Gammaproteobacteria (13%), Actinobacteria (9%), Bacteroidetes (8%), and Deltaproteobacteria (5%) (Supplementary Figure S1). Some phyla such as Betaproteobacteria, Planctomycetes, Firmicutes, Cyanobacteria, Chloroflexi, OP10, and TM7 are consistently present across all samples (ranging from 1% to 4%), even though they were not dominant inhabitants of the phyllosphere. Relative abundance of bacterial phyla did not vary much between tree species. Only two phyla showed distinctive patterns between tree species. Planctomycetes are more common on Gnetum sp. (10.3%, P < 0.01) and Gammaproteobacteria are found more abundantly on Shorea (23.1%, P < 0.05). At the order level, Gnetum harbours abundant Flavobacteriales (Bacteroidetes) and Arytera has more Pseudomonadales (Gammproteobacteria). An unknown bacterial phylotype, which belongs to Edaphobacter, turned out to be the most abundant OTU across all samples (3.4% on average). Interestingly, another genus (Elizabethkingia) was found only on Gnetum leaves in a high proportion (4.9%).
Composition of abundant members of Acidobacteria on tropical tree leaves
75.8 ± 19.0
97.8 ± 2.7
93.2 ± 5.0
76.5 ± 21.0
86.3 ± 5.7
0.9 ± 1.5
0.4 ± 0.5
23.2 ± 18.4
0.6 ± 0.7
6.8 ± 5.0
22.2 ± 20.7
13.6 ± 5.5
0.9 ± 0.8
0.4 ± 0.3
0.4 ± 0.3
0.1 ± 0.2
0.1 ± 0.2
0.4 ± 0.3
0.1 ± 0.1
0.5 ± 0.4
The phyllosphere community is also very distinct in terms of composition from the soil and rhizoplane communities at FRIM, with 0.5% OTU (defined at 97% sequence similarity cutoff) overlap between phyllosphere and soil, and 1.0% between the phyllosphere and rhizoplane communities (work submitted for publication). By contrast, members of the same tree species show around 10–18% OTU overlap in phyllosphere community between samples, and different tree species show around 3–8% overlap.
To test the generality of the pattern found in previous studies that bacterial community similarity follows host tree phylogenetic relatedness (and the implied conjecture that bacterial communities have co-evolved with host species and, as a result, follow plant evolutionary history), we compared the host tree phylogeny and bacterial community phylogenetic structure using a Mantel test. There is a significant but weak association between host phylogeny and phyllosphere community structure (r = 0.458, P < 0.001), which suggests that bacterial communities on plant leaves depend somewhat on host phylogenic affinities.
The results of this study emphasize the diversity of the phyllosphere bacterial community in tropical trees. The level of diversity of phyllosphere bacteria was rather lower than the mean for the surrounding soils at the same site, but it placed around the lower limit of the soil bacterial diversity, suggesting that its diversity level is within the range of variation of total soil bacterial diversity.
Differences from the Temperate Phyllosphere Community
In terms of bacterial taxonomic composition, the character of the phyllosphere community in the tropics did not differ greatly from temperate or subtropical trees, including relatively high abundances of Alphaproteobacteria, Gammaproteobacteria, and Bacteroidetes, as generally observed from previous studies [16, 32]. Additionally, TM7, Planctomycetes, Firmicutes, and Cyanobacteria are consistently found across all tree species, indicating that these minor lineages are also consistent residents on leaves and might have significant effects on host plant physiology or ecosystem functioning.
However, a distinctive pattern in the tropical phyllosphere community composition seen in this study is the fact that the assemblage on all the tropical tree species sampled here is dominated by Acidobacteria (17% on average), whereas members of this group are generally absent or found at very low abundance on trees from other regions, or on crop plants [16, 29, 32]. For example, a study of 56 temperate tree species in Colorado showed that Acidobacteria accounted for less than 1% of phyllosphere bacteria across all the tree species sampled . Lambais et al. , who investigated subtropical forest in Brazil, reported that there is no Acidobacteria population, though it might not have been detected due to a small number of clones per sample. However, in an exception to the general pattern, a study of seasonal variation of cottonwood leaves in the USA showed Acidobacteria as one of dominant members of leaf surface communities . Moreover, another recent study of Magnolia sampled in Mississippi, USA revealed that Acidobacteria are one of the important components of the phyllosphere . In both cases, the abundance of Acidobacteria varied by season, indicating that seasonal variation could be the reason for the absence of particular bacterial groups in some studies. It is evident that Acidobacteria are a consistent feature of tree leaf surface communities in Malaysia, and in this almost aseasonal climate, it is unlikely that this varies fundamentally during the year. The abundance of Acidobacteria on tree leaves corresponds with the pattern of the taxonomic composition of soil bacteria in the same area, with Acidobacteria dominant in the soils. The relative abundance of Acidobacteria in the soil from FRIM and other rainforest areas in Malaysia accounts for more than 30% on average (work currently submitted for publication). However, there is just 0.5% OTU overlap between phyllosphere and surrounding soils from the FRIM site, which suggests that distinctive members of Acidobacteria reside on leaves. At a lower taxonomic level, almost all the Acidobacteria of the phyllosphere consisted of subgroup 1 (84.0%) and subgroup 3 (15.2%), which are usually found in soil in a high proportion . Their ecological functions in nature are unknown except for the strong relationship with pH [8, 13]. The reasons why these two subgroups dominated the tropical phyllosphere need to be investigated further.
It is interesting to see that the closest relatives of dominant Acidobacteria OTUs found in the phyllosphere in this study were also found in other habitats, such as soil, freshwater, human skin, and indoor air. Moreover, several OTUs (OTUs 2, 6, 19, 20, and 26 in Supplementary Table S2) turned out to dwell on plant leaves of other geographical regions, indicating that they are widely distributed despite geographic distance and have an unique association with plant leaves, which might be caused by certain characteristics of the leaf surface.
Another interesting result is that a large amount of Elizabethkingia found only on Gnetum leaves (4.9% on average), and not other tree species. It might be insufficient to generalize that this genus is Gnetum specific because we only targeted a small subsection of the tree species diversity of Malaysia. However, its distinctive abundance is evidence enough to conclude there is a strong relationship between Gnetum and Elizabethkingia. Two species, Elizabethkingia meningoseptica and Elizabethkingia miricola, which are water-borne pathogens causing meningitis and sepsis in humans, were found on the leaves in relative proportions of 84% and 9% of the genus, respectively. They are known to be widely distributed in nature (e.g. freshwater, salt water or soil) and reported to be present in fish and frogs [10, 19]. Interestingly, they were also found in Gnetum rhizoplane samples (data in preparation for publication) and not on other tree species, indicating that these species are particularly associated with Gnetum species, not because of random chance or contamination. Chemical and physical characteristics of Gnetum together with the evolutionary history of the tree could be the reason, but this remains speculative.
Overall, it is evident that in tropical trees, as in other plant communities sampled by other groups, each species has its distinctive phyllosphere community [16, 32, 36, 38]. We also confirmed that in tropical trees, the variability of bacterial community structure between tree species is higher than that within the same species, as was also the case in temperate trees .
Location is also one of main determinants of the phyllosphere bacterial community in Methylobacterium population . However, in this study, other factors such as distance, leaf senescence stage, and leaf position apart from species difference are unlikely to influence bacterial community because samples were randomly collected in an arboretum, and consisted only of fully developed and non-senescent leaves. Overall, it is unclear what the origin of these distinctive host species-specific communities might be. Possibilities include texture and topography of the leaf surface, and secondary compounds from the plant. It has been reported that in cultivation studies, bacteria are found more abundantly on succulent herbaceous plants than grasses or waxy broad-leaved plants [14, 20]. Water and phosphorus contents of leaves are also known as major factors determining the abundance of phyllosphere bacteria on plants . Biotic factors such as other microbes, insects, and birds specific to each tree species also could be a cause, although this needs to be shown. The relationship between physiochemical properties of leaves of tropical trees and bacterial community structure as well as interaction with other organisms will be investigated in future studies.
A recent study by Redford et al.  suggested a significant correlation between plant species and bacterial community phylogeny, although this was not viewed by them as statistically strong enough to be generalized. We also observed a significant but weak association between tree species and phyllosphere community phylogeny. It is unclear whether leaf surface factors linked to host phylogeny merely select from a pre-existing pool of bacterial OTUs, or whether there has actually been ongoing evolution of bacterial lineages in parallel with the phylogeny of their host trees.
This work was in part funded by a Malaysian government ‘Brain Gain Fellowship’ granted to Jonathan Adams December 2008 to August 2010. We thank Y-M. Oh and B. Tripathi of SNU for their help with sample processing.