Are Archaea inherently less diverse than Bacteria in the same environments?


  • Editor: Gary King

Correspondence: Josephine Y. Aller, School of Marine and Atmospheric Sciences, Stony Brook University, Stony Brook, NY 11794-5000, USA. Tel.: +1 631 632 8655; fax: +1 631 632 3066; e-mail:


Like Bacteria, Archaea occur in a wide variety of environments, only some of which can be considered ‘extreme’. We compare archaeal diversity, as represented by 173 16S rRNA gene libraries described in published reports, to bacterial diversity in 79 libraries from the same source environments. An objective assessment indicated that 114 archaeal libraries and 45 bacterial libraries were large enough to yield stable estimates of total phylotype richness. Archaeal libraries were seldom as large or diverse as bacterial libraries from the same environments. However, a relatively larger proportion of libraries were large enough to effectively capture rare as well as dominant phylotypes in archaeal communities. In contrast to bacterial libraries, the number of phylotypes did not correlate with library size; thus, ‘larger’ may not necessarily be ‘better’ for determining diversity in archaeal libraries. Differences in diversity suggest possible differences in ecological roles of Archaea and Bacteria; however, information is lacking on relative abundances and metabolic activities within the sampled communities, as well as the possible existence of microhabitats. The significance of phylogenetic diversity as opposed to functional diversity remains unclear, and should be a high priority for continuing research.


Once thought to only inhabit a limited number of extreme environments or specialized ecological niches, prokaryotes of the domain Archaea are now thought to be globally widespread (Delong, 1992; McInerney et al., 1995; Preston et al., 1996; Bintrim et al., 1997; Munson et al., 1997; Marc et al., 1998; Murray et al., 1998; Massana et al., 2000; Church et al., 2003; Zhang et al., 2005; Teske, 2006) and a significant component of microbial assemblages in a broad range of habitats (Fuhrman et al., 1992; DeLong et al., 1994; Karner et al., 2001; Herndl et al., 2005), including man-made environments (Godon et al., 1997; Sekiguchi et al., 1998; Zhu et al., 2003; Leclerc et al., 2004). Archaea are now recognized to be phylogenetically (Fuhrman et al., 1993; Ovreas et al., 1997; Bowman & McCuaig, 2003; Kormas et al., 2003; Mills et al., 2003; Elshahed et al., 2004) and physiologically (Ouverney & Fuhrman, 2000; Wuchter et al., 2003; Könneke et al., 2005; Robertson et al., 2005) diverse.

In a previous paper, we reviewed what is known of the phylogenetic diversity of bacterial assemblages in aquatic and other environments (Kemp & Aller, 2004a), and in a companion paper (Kemp & Aller, 2004b) asked how well bacterial diversity in the source environment is represented by 16S rRNA gene libraries, by far the most commonly used method to assess phylogenetic diversity. In this paper, we turn to the Archaea and assess what is known of archaeal phylogenetic diversity in different environments, again as represented by 16S rRNA gene libraries. Very few 16S rRNA gene libraries sample microbial diversity exhaustively and most published studies do not include an evaluation of how well the library represents diversity in the source environment (Kemp & Aller, 2004a). Therefore, we apply an assessment procedure described in Kemp & Aller (2004b) and available online ( to determine whether the library is large enough to yield a stable estimate of total phylotype richness using the abundance-based richness estimators SChao1 (Chao, 1984, 1987) and SACE (Chao et al., 1993). We compare archaeal to bacterial diversity, and then consider the implications of the observed differences and similarities to address two very basic questions: (1) is it important to measure and understand phylogenetic diversity and (2) are there unique characteristics of both Archaea and Bacteria which can explain the observed diversity patterns?

We note that studies of phylogenetic diversity have fallen somewhat out of favor, as techniques to explore functional diversity have been developed. While functional diversity cannot be mapped directly onto rRNA gene-based phylogenetic diversity, a phylogenetically diverse community certainly is more likely to be functionally diverse than a community with few phylotypes. A phylogenetically diverse community also allows the possibility of functional redundancy, in which more than one phylotype can perform a given transformation. Arguably, an understanding of functional diversity requires an understanding of phylogenetic diversity.

Materials and methods


Consistent with our earlier treatment of the diversity of Bacteria (Kemp & Aller, 2004a), we use the term ‘clones’ to denote the identified products of PCR amplification; ‘library’ to denote a collection of such identified PCR products derived from a given environmental source; ‘library size’ to denote the number of PCR products that were actually characterized; and ‘phylotype’ to denote a group of PCR products judged by the original authors to be essentially identical, regardless of the method or criteria they used to assess phylogenetic similarity. As with our previous examination of bacterial libraries, most studies used sequence similarity as the criterion for distinguishing among phylotypes, and in most published studies, phylotypes are considered identical if their 16S rRNA gene sequences are ≥97% identical. We make no assumptions regarding the relationship of ‘phylotype’ to traditional concepts of genera and species.

Source data

Most of what is known of the phylogenetic diversity of Archaea in aquatic and other environments is based on distinguishing among different organisms using PCR amplification of 16S rRNA gene, without actually culturing them or having any direct knowledge of their morphology, physiology, or ecology. While 16S rRNA gene libraries are ideally representative of the diversity present in the source environment, technically they are only samples of the PCR-amplified material from which clones are made. Another limitation in phylogenetic studies is that likely none of the primers commonly used to amplify 16S rRNA gene are truly ‘universal’ and therefore no single set will successfully target all Archaea in a given environment. Many others have commented on potential PCR biases and their effect on any PCR-based representation of diversity. While this concern is appropriate, rRNA gene libraries are still the primary means by which phylogenetic diversity is assessed, and we note that most researchers recognize and take precautions to minimize such biases.

We have located 173 libraries from at least 12 different types of environments (Moyer et al., 1994, 1995, 1998; Bintrim et al., 1997; Bowman et al., 1997, 2000a, b; Fuhrman & Davis, 1997; Godon et al., 1997; Munson et al., 1997; Chandler et al., 1998; Dojka et al., 1998; Sekiguchi et al., 1998; Vetriani et al., 1998, 1999; Crump et al., 1999; Tajima et al., 1999, 2001; Takai & Sako, 1999; Bond et al., 2000; Cifuentes et al., 2000; Crump & Baross, 2000; Cytryn et al., 2000; Jurgens et al., 2000; Lueders & Friedrich, 2000; Massana et al., 2000; Orphan et al., 2000; Reysenbach et al., 2000; Skirnisdottir et al., 2000; Watanabe et al., 2000, 2002; Brambilla et al., 2001; Casamayor et al., 2001, 2002; Hjorleifsdottir et al., 2001; Jackson et al., 2001; Karner et al., 2001; Lanoil et al., 2001, 2005; Madrid et al., 2001a, b; Marteinsson et al., 2001; Rudolph et al., 2001; Takai et al., 2001; Bano & Hollibaugh, 2002; Huang et al., 2002, 2004; Huber et al., 2002, 1986, 1998; Moissl et al., 2002; Oren, 2002; Pesaro & Widmer, 2002; Reed et al., 2002; Stein et al., 2002; Teske et al., 2002; Bowman & McCuaig, 2003; Chen et al., 2003; Church et al., 2003; Elshahed et al., 2003, 2004; Gonzalez-Toril et al., 2003; Itoh, 2003; Kormas et al., 2003; Mounéet al., 2003; Nercessian et al., 2003; O'Connell et al., 2003; Schrenk et al., 2003, 2004; Zhu et al., 2003; Chachkhiani et al., 2004; Gallagher et al., 2004; Higaski et al., 2004; Leclerc et al., 2004; Lysnes et al., 2004; Rutz & Kieft, 2004; Shin et al., 2004; Grabowski et al., 2005; Heijs et al., 2005; Høj et al., 2005; Knittel et al., 2005; Kvist et al., 2005; Meyer-Dombard et al., 2005; Mills et al., 2005; Pašićet al., 2005; Snell-Castro et al., 2005; Zhang et al., 2005; Gihring et al., 2006; Karr et al., 2006; Lin et al., 2006; Schwarz et al., 2006; Siering et al., 2006; Clementino et al., 2007; Perreault et al., 2007; Zeng et al., 2007) for which source data met the same criteria used in our previous analyses of bacterial diversity (Kemp & Aller, 2004a). In 79 cases both bacterial and archaeal libraries were constructed from the same source environment.

We excluded studies that did not use archaeal-specific primers (Casamayor et al., 2002); that intentionally selected amplification products (e.g. based on intensity and frequency of bands: Heck et al., 1975; Demergasso et al., 2004); or did not provide sufficient data regarding the frequency distribution of phylotypes (e.g. McInerney et al., 1995; Munson et al., 1997; Marc et al., 1998; Piñar et al., 2001; Furlong et al., 2002; Lysnes et al., 2004; Zhang et al., 2005; Galand et al., 2006; Li et al., 2007). We excluded pooled data derived from several libraries.

Large enough library assessment

The approach described in Kemp & Aller (2004b) was used to determine whether libraries were large enough to yield stable and unbiased richness estimates with application of either SACE (Chao et al., 1993) or SChao1 (Chao, 1984, 1987), or both. SChao1 and SACE estimators are highly correlated when most phylotypes are present only once or twice, as is true of many prokaryotic libraries. The assessment procedure is similar to rarefaction analysis (Heck et al., 1975). For each library, 1000 data subsets are derived by random sampling with replacement. These derived data subsets range in size ni from i=1–100% of the total size N of the library, in increments of 1% of N rounded to the nearest integer value. For each subsample size ni, 10 replicate data subsets are drawn with replacement from the model library; at size ni=N, all 10 data subsets are identical to the complete library. SChao1 and SACE are calculated for each derived data subset, and plotted against subsample size ni to determine whether an asymptote has been reached. If the estimated phylotype richness reached an asymptote (for our purposes usual examination was sufficient), we infer that the library was large enough to yield a stable estimate of phylotype richness. The stable estimates can be compared between libraries. In this study, the numbers of clones in each phylotype from a given library were entered into the spreadsheet found at The form processor subsampled the library, calculated values of SChao1 and SACE, and plotted richness estimates against subsample size to predict whether a library adequately represents the diversity in the source environment.

Nonparametric abundance-based estimators of coverage

Two nonparametric abundance-based estimators are useful when sampling moderately even communities (K.A. Keating and J.F. Quinn, pers. commun.): Good's coverage (Good, 1953) and Chao et al.'s CACE (Chao et al., 1993; Colwell & Coddington, 1994).

Estimating the number of unseen phylotypes

The most basic estimate of phylotype richness is the observed number of different phylotypes in a library, which provides a minimum estimate of the total number of phylotypes present in the source assemblage. The estimates of SChao1 and SACE are usually higher than the number of observed phylotypes, and the difference is the number of unseen phylotypes; that are believed to occur in the source environment, but were not actually collected or amplified.


Large enough library determination

One hundred of the 151 archaeal libraries from aquatic systems, and 14 of 22 libraries from nonaquatic systems were determined to be large enough to provide stable phylotype richness estimates (Table 1). Among the 79 companion bacterial libraries, 31 of the 62 libraries from aquatic systems and 14 of the 17 from nonaquatic libraries were determined to be large enough.

Table 1.   Comparison of coverage and phylotype richness estimates for large enough archaeal and bacterial libraries from different environments
Archaeal Libraries
173 (114 large enough)
 n (large
Library size
Good's C
Aquatic systems
 Plankton37 (23)40 (6–129)10 ± 80.831 ± 0.1530.638 ± .31016 ± 1614 ± 15
 Gas Hydrates11 (10)72 (35–125)8 ± 70.885 ± 0.1490.973 ± 0.02612 ± 1310 ± 10
 Groundwater23 (15)58 (21–181)9 ± 70.799 ± 0.1610.928 ± 0.06815 ± 1315 ± 19
 Hyperthermal51 (30)59 (17–201)9 ± 70.866 ± .1540.946 ± 0.10012 ± 1711 ± 17
 Sediment21 (15)41 (5–190)11 ± 60.787 ± 0.2000.808 ± 0.19722 ± 1818 ± 15
 Suspended Particles3 (2)34 (9–83)11 ± 40.293 ± 0.2170.293 ± 0.1002218
 Microbial Mats-nonhyperthermal437 (14–76)7 ± 50.679 ± 0.2220.837 ± 0.16714 ± 1711 ± 11
 All aquatic systems observed/predicted (%)151 (100)52 ± 429 ± 7 (2–40)0.827 ± 0.1830.895 ± 0.16315 ± 1613 ± 16
Other Systems
 Digestive5 (4)30 (19–45)6 ± 40.894 ± 0.1510.894 ± 0.15113 ± 1214 ± 19
 Soils9 (3)54 (30–190)12 ± 100.878 ± 0.1290.942 ± 0.0747 ± 17
 Bioreactors6 (5)28 (19–44)6 ± 30.805 ± 0.1780.935 ± 0.0449 ± 38 ± 3
 Liquid Natural gas226 (21–31)18 (16,19)0.5610.56132 (30–32)43 (36–52)
 All other systems observed/predicted (%)22 (14)39 ± 358 ± 5 (2–19)0.784 ± 0.3260.832 ± 0.26813 ± 1015 ± 17
Bacterial Libraries 79 (44 large enough)
 nSizePhylotypesCACE large enoughCoverage Good's CSACESchao1
  1. Phylotype estimates are for libraries judged to be large enough to yield a stable prediction of richness where n=number in parentheses from n column.

Aquatic systems
 Plankton13 (4)33 (7–72)16 ± 140649 ± 0.2290.733 ± 0.18337 ± 5628 ± 44
 Biofilms1 (0)29141.01.044
 Gas Hydrates7 (5)60 (34–127)16 ± 60.879 ± 0.1200.897 ± 0.12425 ± 1723 ± 14
 Groundwater4 (2)123 (46,174)40 (28,57)0.639 ± 0.1310.760 ± 0.128109 ± 39113 ± 64
 Hyperthermal16 (10)73 (20–171)19 ± 170873 ± 0.1400.735 ± 0.24039 ± 4933 ± 43
 Sediment11 (4)130 (7–347)60 ± 680.653 ± 0.2040.717 ± 0.246157 ± 193136 ± 170
 Suspended Particles6 (2)46 (16–86)19 ± 130.5880.37887,4964,34
 Microbial Mats-nonhyperthermal4 (3)112 (39–301)36 ± 210.705 ± 0.0920.753 ± 0.12589 ± 3670 ± 23
 All aquatic systems observed/predicted (%)62 (31)76 ± 7129 ± 370.714 ± 0.1980.781 ± 0.19863 ± 8556 ± 76
Other Systems
 Digestive2 (1)44340.4780.4787872
 Soils371 (23–167)42 ± 550.693 ± 0.1850.718 ± 0.14889 ± 12582 ± 117
 Bioreactors10 (8)78 (19–202)29 ± 290.816 ± 0.1090.843 ± 0.12557 ± 7362 ± 100
 Liquid Natural gas220 (15–24)12,140.4710.47144,3429,21
 All other systems17 (14)64 ± 6029 ± 290.711 ± 0.1720.731 ± 0.18263 ± 7562 ± 88

Library size and sources

The average size of the 173 archaeal libraries was 52±42 (mean±SD) and ranged from five clones in sediments from Guanabara Bay between Rio de Janeiro and Niteroi, Brazil (Clementino et al., 2007) to an Icelandic hot spring with 201 clones (Kvist et al., 2005) (Table 1 and Fig. 1). Eight libraries had more than 150 clones, six from hyperthermal environments and two from cold briny springs. Bacterial libraries from the same source environments varied widely in size averaging 76±71 phylotypes (mean±SD, n=79) with several libraries from anoxic sediments containing over 275 clones (Fig. 1).

Figure 1.

 Total reported diversity of Archaea at a glance compared with bacterial diversity which is illustrated by the ellipse (from Kemp & Aller, 2004a). We provide Table 1 as a service to readers who might like the details.

Library composition

The average number of observed phylotypes for archaeal libraries from aquatic systems was 9±7 (mean±SD, n=151, range 2–40; Table 1). The most diverse libraries were from a microbial mat in an acidic thermal spring (40 unique phylotypes in 84 clones) (Jackson et al., 2001), and from sediments at the source of an artesian sulfide and hydrocarbon-rich spring (39 unique phylotypes in 83 clones) (Elshahed et al., 2003). In nonaquatic systems, the average number of unique phylotypes was 8±5 (mean±SD, n=22, range 2–19; Table 1). The number of observed phylotypes in archaeal libraries was not correlated with library size (r2=0.11, n=173).

In contrast, bacterial libraries from the same source environment in aquatic systems contained on average 29 phylotypes in both aquatic systems (29±37, n=62, range 2–227), and nonaquatic systems (29±29, n=17, range 7–106). The number of observed phylotypes in bacterial libraries was correlated with library size (r2=0.63, n=79).

Rare species are defined operationally by various authors as those species occurring once, up to two times, or up to 10 times in samples of a community. We defined rare phylotypes as those appearing only once or twice in a library. Figure 2 compares rare phylotypes among archaeal and bacterial libraries across all environments. Rare phylotypes contributed 66% of the phylotypes observed in all archaeal libraries (Fig. 2a; linear regression slope=0.66, r2=0.80, n=173) even though 26 had no phylotypes which occurred only once, and in 17 of these, all phylotypes appeared three or more times. When libraries without singletons and doubles are excluded, rare phylotypes contributed a majority [67% (linear regression slope=0.67, r2=0.79, n=156)] of phylotypes. Library size was unrelated to the percentage of rare phylotypes in archaeal libraries (linear regression slope=0.013, r2=0.0071, n=173).

Figure 2.

 (a) Comparison of the number of phylotypes which occur only once or twice to the total number observed in an archaeal library. Libraries which fall below the 1 : 1 line tend to have uniformly abundant phylotypes, while libraries above the line have a skewed distribution with many rare phylotypes. Rare phylotypes constituted 76% of the total observed phylotypes although 17 of 173 libraries had no rare phylotypes. (b) In bacterial libraries 63% of the observed phylotypes were rares with only two of the 79 libraries having no rare phylotypes. (c) Rare phylotypes contributed at least 50% of all phylotypes observed in libraries from all environments except liquefied natural gas. Rare phylotypes were more abundant in archaeal than bacterial libraries from soils, microbial mats, sediments, groundwater, and bioreactors. Data from three large bioreactor libraries (13 rare/166 observed, 202 rare/275 observed and 93 rare/118 observed) were not plotted to allow for more clear comparisons.

Among the bacterial libraries from the same source environments (Fig. 2b), rare phylotypes contributed 63% of the observed phylotypes (Fig. 2b; linear regression slope=0.63, r2=0.72, n=79). Only four libraries had no phylotypes that occurred only once and of those, only two lacked any rare phylotypes. In three source environments: microbial mats, bioreactors, and lithotrophic biofilms, rare phylotypes were more common in archaeal than in bacterial libraries (Fig. 2c). There was no significant relationship between library size and the percentage of rare phylotypes (linear regression slope=1.44, r2=0.42, n=79).


The two coverage estimators summarized in Table 1, suggest that on average archaeal libraries captured more than half of the estimated total number of phylotypes in their source environments. In aquatic environments, capture rates were close to 75%, and for large-enough libraries, capture rates were still greater (Cace=83±18%; C=89±16%). The highest coverage was found for gas hydrates and the lowest for liquid natural gas libraries. Good's (1953) varied widely (column 5 in Table 1), with the extremes found for an archaeoplankton library composed entirely of rare phylotypes (occurring once or twice) (Moissl et al., 2002), to 17 archaeal libraries that had no rare phylotypes. Both coverage estimators were highest in libraries from gas hydrates, groundwater, hyperthermal and bioreactor environments, and were lowest for natural gas and suspended particles (Table 1). They were unrelated to library size across all environments although the largest libraries also have the highest coverage.

For the ‘large-enough’ libraries, Good's C and Chao et al.'s CACE estimators were generally lower for bacterial than for archaeal libraries derived from the same environment (Table 1). They were not correlated with library size for either index. As in the archaeal libraries, coverage estimates for bacterial libraries varied highest in libraries from gas hydrates and hyperthermal environments and lowest among libraries from digestive systems and liquid natural gas (Table 1).

Phylotype richness

In 26 instances both bacterial and archaeal libraries from the same source environment were considered large enough, so that a valid comparison of the estimated phylotype richness values can be made. The observed diversity of Archaea was greater than that of Bacteria in only five cases (Table 1), and the estimated total phylotype richness estimated by SACE or SChao1 was nearly always lower for Archaea than for Bacteria.

Libraries of all sizes were found to be large enough for both Archaea and Bacteria and library size was unrelated to the number of observed phylotypes (P<0.05) (Fig. 3). For the large enough archaeal libraries from aquatic systems, we estimate the source community contains 29±22% additional unique phylotypes than were actually observed, based on SACE, and 20±21% based on SChao1. For nonaquatic libraries, the corresponding values were 23±19% (SACE) and 16±20% (SChao1; Fig. 4). In contrast, for bacterial libraries both estimators predicted that the source community actually contains 44±25% (SACE estimator) or 38±25% (SChao1 estimator) additional unique phylotypes than were actually observed in the large enough libraries from aquatic systems, and 49±19% (SACE) and 43±23% (SChao1; Fig. 4) from nonaquatic systems.

Figure 3.

 Archaeal libraries determined to be large enough to estimate phylotype richness compared with corresponding bacterial libraries from the same source environment. Libraries of all sizes for both Archaea and Bacteria were determined to be large enough. Library size was unrelated to the number of observed phylotypes (P<0.05).

Figure 4.

 Comparison estimates of the total number of phylotypes in a source environment predicted by the SChao1 and SACE richness estimators for large enough archaeal and corresponding bacterial libraries. Both estimators predict greater bacterial than archaeal diversity for most source environments. Similar bacterial diversity is only predicted for biofilms and liquefied natural gas environments. See Table 1 for numbers of libraries.


Diversity of Archaea vs. Bacteria

rRNA gene libraries are miniscule samples of extremely abundant organisms, and it is highly likely that all but the most abundant phylotypes would be missed, or found only rarely (see Kemp & Aller, 2004b). As Colwell & Coddington (1994) note, even exhaustive sampling can leave many species represented by only a few specimens or even a single specimen. Although some archaeal libraries have few or no rare phylotypes, approximately two-thirds of both archaeal and bacterial libraries of all sizes from many different kinds of environments are composed of predominantly rare phylotypes. This would suggest that a priori, researchers might expect to spend equal effort exploring archaeal vs. bacterial diversity. In reality, archaeal libraries are rarely as large as corresponding bacterial libraries from the same source environments (Table 1 and Fig. 1). We found that the number of phylotypes observed is not correlated with library size for Archaea (Figs 1 and 2), suggesting that larger is not necessarily better when sampling Archaea in a new environment. In contrast, the number of observed phylotypes is correlated with library size for bacterial libraries, suggesting that few bacterial libraries had exhaustively sampled diversity in the source environment, and larger is indeed better.

Among the 173 libraries examined, there were only five libraries where the observed archaeal diversity was greater than bacterial diversity at the same sampling location: two sets of plankton assemblages (Fuhrman & Davis, 1997; Bano & Hollibaugh, 2002), an arsenite-oxidizing acidic thermal spring (Fig. 5d) (Jackson et al., 2001), subterranean hot spring (Marteinsson et al., 2001), and methane-rich sediments associated with a hydrocarbon seep (Fig. 5c) (Mills et al., 2005). When libraries were judged large enough and richness estimators could be used for a valid comparison, we find that the diversity of Archaea and Bacteria may be similar in some environments including suspended particles, biofilms, and liquid natural gas. In all other environments predicted bacterial diversity appears to be much greater than archaeal (Figs 3 and 4).

Figure 5.

 Examples of rarefaction curves for microbial diversity in different environments, and predicted total phylotype richness in the source community. (a) Pig manure digester (Snell-Castro et al., 2005) where bacterial diversity is extensive and archaeal limited; (b) sediment at 20–21 cm depth on Antarctic Continental Shelf with a diverse bacterial community and moderate archaeal diversity predicted (Bowman & McCuaig, 2003); (c) sediments and gas from methane hydrate seep mounds in Gulf of Mexico where archaeal diversity is predicted to be somewhat higher than bacterial (Mills et al., 2003); (d) arsenite-oxidizing acidic thermal spring where the predicted archaeal diversity is predicted to be three times higher than the bacterial (Jackson et al., 2001); (e) liquified natural gas in pipelines where observed diversity would suggest similar or higher archaeal diversity, but richness estimators predict higher bacterial diversity (Zhu et al., 2003); and (f) microbial mat from Lake Fryxell, McMurdo Dry Valleys, Antarctica with significantly higher bacterial diversity (P<0.05) and a 72 clone archaeal library had two phylotypes and richness estimators would predict four (Brambilla et al., 2001).

The limitations of rarefaction analyses and usefulness of our ‘large-enough’ library assessment procedure (31) are demonstrated by the study, by Mills et al. (2005), of methane-rich sediments associated with a hydrocarbon seep in the Gulf of Mexico. While their published rarefaction analyses did not indicate that a sufficient number of clones had been screened to estimate the potential diversity of Archaea and Bacteria, our assessment procedure suggests that their libraries actually are large enough to yield stable phylotype richness estimates. We predict that archaeal diversity is 30% greater than bacterial diversity in their source environment (Fig. 5c). In another example, in liquid natural gas pipelines (Zhang et al., 2005) the size and diversity of the archaeal and bacterial libraries were almost identical (15 phylotypes/24 clones for Bacteria, 16/21 for Archaea). Because both libraries were judged to be large enough to estimate total diversity, we can apply richness estimators to predict that bacterial diversity is actually greater than archaeal diversity in their source environment (Fig. 5e).

Why does the diversity of Archaea differ from that of Bacteria?

Like Bacteria, Archaea occur in a wide variety of environments only some of which would be considered ‘extreme’ in some manner. In the majority of environments, Archaea tend to have lower diversity than bacteria in the same environment. However, Archaea do contribute a large proportion of the total prokaryotic phylotypes present in most environments studied to date (Table 2). It is uncertain how this view will change as additional data is acquired from understudied environments such as the marine plankton, previously not thought of as typical for Archaea.

Table 2.   Relative proportion of archaeal and bacterial rRNA phylotypes predicted by the SACE estimator in large enough companion libraries from various environments
Mean ± SD
Mean ± SD
Hyperthermal366 ± 4724 ± 15
Liquified Natural Gas232 ± 339 ± 7
Groundwater226 ± 3109 ± 40
Non Hyperthermal Mats220 ± 22110 ± 9
Suspended Particles11815
Gas Hydrates511 ± 1225 ± 17
Bioreactors59 ± 372 ± 93

Why are Archaea less diverse than Bacteria in most environments where they co-occur? This is an obvious and important question that bears upon any assessment of the ecological roles of Archaea and Bacteria, but which may not be answerable with conventional sampling approaches. We speculate that Archaea may perceive and make use of the environment in ways that are more restrictive than for Bacteria. For example, Archaea might live in microniches while Bacteria live more broadly or in different microniches in the same macroenvironment. More generally, nearly any sample collected for analysis of prokaryotic diversity will contain a multitude of microenvironments; in a sense, Archaea and Bacteria may not truly coexist even if they are collected in the same sample. We would question whether co-occurrence in a macroscale sample is a sufficient justification for comparisons of the diversity of prokaryotes found in that sample. We note that the same comments could be made regarding comparisons within the Archaea or Bacteria, among groups occupying different microenvironments within a single microenvironment.

The energetic costs of metabolic processes carried out by Archaea in at least some environments may be so great that phylogenetic diversification is limited, in comparison to that of Bacteria. For example, Oren (1999) suggested that the diversity of halophilic Archaea was directly limited by the energetic costs of maintaining themselves in extreme environments compared with the small free-energy change associated with dissimilatory reactions.

More broadly, we question whether Archaea possess less physiological flexibility than Bacteria in environments that are not considered to be extreme. Although only limited information is available on metabolic diversification among Archaea beyond methane and hydrogen-based energy metabolisms, recent studies find Crenarchaeota that actively assimilate amino acids, protein, and diatom extracellular polymers in Arctic waters and at depths of 100–200 m in the spring, accounting for up to >40% of dissolved organic matter assimilation (Kirchman et al., 2007). Other studies report isolation of a nitrifying Crenarchaeon (Könneke et al., 2005) and Archaea that can function as chemorganotrophs (Wuchter et al., 2003) and assimilate free amino acids (Ouverney & Fuhrman, 2000). Much more information is needed to assess whether Archaea are typically more, less or equally physiologically diverse than Bacteria in the same environment, but certainly these studies suggest that Archaea are physiologically diverse as a whole.

The relative abundance and metabolic activities of Bacteria and Archaea within a specific community are not necessarily related to their diversity. Abundance estimates of microbial cells based on FISH labeling with Archaea- and Bacteria-specific probes, suggest that Archaea generally account for only a minor fraction (<5%) of the total prokaryotic community (Gonzalez-Toril et al., 2003; Wobus et al., 2003; Koizumi et al., 2004; van der Wielen et al., 2005; Kirchman et al., 2007) but can account for 10% (Watanabe et al., 2002), 20% (Bowman & McCuaig, 2003) even 40% (Koizumi et al., 2004; Kirchman et al., 2007) of some communities. Even if they are not overwhelmingly abundant, the activities of Archaea particularly those forming metabolic alliances with Bacteria, can be critical to the formation of functional microbial communities (e.g. Meyer-Dombard et al., 2005; Daffonchio et al., 2006; Ariesyady et al., 2007; Yakimov et al., 2007) and even minor changes in biotic and/or abiotic factors can alter the diversity of both Archaea and Bacteria, the community structure and ecosystem operation (e.g. Meyer-Dombard et al., 2005; Fuhrman et al., 2006).

Finally, the diversity of Archaea may be linked to that of Bacteria in quite another way. Studies of tundra wetlands (Hoj et al., 2005); waste water sludge digesters (Ariesyady et al., 2007), and methane hydrate outcrops (Orphan et al., 2001; Teske et al., 2002; Kormas et al., 2003; Knittel et al., 2005) suggest that factors influencing bacterial communities and consequently the availability of substrates for methanogenic Archaea may indirectly limit archaeal diversity.

Implications of abundance, diversity, and activity measurements

Studies of engineered and natural ecosystems allow us to consider whether there are fundamental differences in diversity, dominance, and the extent of functional redundancy sensuWalker (1992) in archaeal and bacterial communities living in the same environments. Utilizing bioreactors as model ecosystems, Briones & Raskin (2003) drew upon theoretical concepts in higher ecological organization (Walker, 1992; Peterson et al., 1998) to examine diversity, trophic interactions, and process efficiency. They proposed that the performance stability of an ecosystem dominated by microorganisms, even one facing frequent disturbances, is the result not of phylogenetic diversity per se, but of functional redundancy, which is ensured by the presence of a reservoir of species able to perform the same ecological function. In the mixed-methanogenic reactor systems studied by Fernández et al. (1999) and Zumstein et al. (2000) and considered by Briones & Raskin (2003), fermenting Bacteria, syntrophic Bacteria, and methanogenic Archaea coexist. The archaeal and synthrophic bacterial phylotypes were less phylogenetically diverse and lacked functional redundancy, and varied little in abundance over time although they did exhibit patterns of dominance and succession. The far more phylogenetically diverse fermenting bacteria varied temporally in response to substrate changes, but system stability was maintained by functional redundancy among the oscillating populations. Variations in the methanogens and syntrophic bacteria actually controlled the overall operation of the system.

In frequently disturbed microbially dominated systems such mobile muds (e.g. Madrid et al., 2001a, b; Aller & Aller, 2004) that are characterized by extremely diverse and abundant Bacteria and few Archaea, system stability may result from a high degree of functional redundancy. High diversity with multiple organisms capable of the same metabolic functions allows efficient remineralization of any and all organic matter in spite of population oscillations.

If most microbial diversity represents functionally interchangeable taxa as suggested by models like that of Curtis et al. (2002) then phylogenetic diversity data will provide little information regarding microbial community function, unless that functional redundancy is recognized and understood. However, in California coastal waters Fuhrman et al. (1992) found repeating and predictable subsets of bacteria within the microbial community that were not functionally interchangeable and that occurred predictably under specific combinations of biotic and abiotic variables. The significance of phylogenetic diversity as opposed to functional diversity remains unclear, and should be a high priority for continuing research.


This project was supported by National Science Foundation OCE Grants 9818574 (to J.Y.A.) and 0083193 (to P.F.K. and J.Y.A.) and CICEET Grant 35757 (to J.Y.A. and P.F.K.). P.F.K. was supported by NSF grant EF-0424599. Contribution No. 1328 from the School of Marine and Atmospheric Sciences, Stony Brook University. Contribution No. 45 of the Center for Microbial Oceanography: Research and Education at the University of Hawaii.