Relative Diversity and Community Structure of Ciliates in Stream Biofilms According to Molecular and Microscopy Methods

Abstract

Ciliates are an important component of aquatic ecosystems, acting as predators of bacteria and protozoa and providing nutrition for organisms at higher trophic levels. Understanding of the diversity and ecological role of ciliates in stream biofilms is limited, however. Ciliate diversity in biofilm samples from four streams subject to different impacts by human activity was assessed using microscopy and terminal restriction fragment length polymorphism (T-RFLP) analysis of 18S rRNA sequences. Analysis of 3′ and 5′ terminal fragments yielded very similar estimates of ciliate diversity. The diversity detected using microscopy was consistently lower than that suggested by T-RFLP analysis, indicating the existence of genetic diversity not apparent by morphological examination. Microscopy and T-RFLP analyses revealed similar relative trends in diversity between different streams, with the lowest level of biofilm-associated ciliate diversity found in samples from the least-impacted stream and the highest diversity in samples from moderately to highly impacted streams. Multivariate analysis provided evidence of significantly different ciliate communities in biofilm samples from different streams and seasons, particularly between a highly degraded urban stream and less impacted streams. Microscopy and T-RFLP data both suggested the existence of widely distributed, resilient biofilm-associated ciliates as well as ciliate taxa restricted to sites with particular environmental conditions, with cosmopolitan taxa being more abundant than those with restricted distributions. Differences between ciliate assemblages were associated with water quality characteristics typical of urban stream degradation and may be related to factors such as nutrient availability and macroinvertebrate communities. Microscopic and molecular techniques were considered to be useful complementary approaches for investigation of biofilm ciliate communities.

Heterotrophic microeukaryotes such as ciliates are thought to be of considerable importance in aquatic ecosystems, as they are major predators of bacteria and constitute a nutritional resource for other protozoa, invertebrates, and probably fish larvae (9, 22, 36, 52, 62, 63, 71). In addition, protozoan bacterivory contributes to enhanced decomposition of leaf detritus—a vital nutrient resource in streams—by increasing turnover of bacterial populations through predation (57). It is not well understood, however, how ciliate diversity and community structure in streams are affected by changing environmental conditions, or how ciliate communities affect other stream biota and processes. The effects of various physical, chemical, and biological factors on freshwater protozoan communities have been considered by a number of studies, but most of these have focused upon planktonic organisms in lentic habitats (for example, see references 2, 11, and 44). However, the complex microbial communities in biofilms have been recognized as important contributors to critical ecological processes, such as auxotrophic primary production, nitrogen fixation, and nutrient cycling, and may underpin the function of stream food webs (31, 45, 61). The few studies which have investigated benthic habitats in lotic systems have found evidence of the existence of diverse communities of abundant ciliates (3, 20, 56) and shifts in community structure in response to ecophysiological parameters (30, 42, 43). With one exception, however, these investigations were based on aquatic sediments, and the organisms within epilithic biofilms have continued to receive little attention.

Most studies of ciliate diversity and ecology have utilized microscopy-based methods of identification (for example, see references 3 and 56), as ciliate cells are relatively large and morphologically diverse. Such methods demand a high level of taxonomic expertise, however, and are difficult and time-consuming—for example, many ciliates are fragile and fast moving, and they often require difficult fixing and staining protocols for reliable identification. Molecular biological tools offer the possibility of more accurate and efficient methods for protozoan study and may provide a useful complement to traditional approaches (12, 18, 28, 65), yet we know of only a few molecular studies of environmental ciliate diversity (18, 20, 37). A series of recent investigations used culture-independent analysis of 18S rRNA gene sequences to reveal the existence of diverse microeukaryote communities in assorted marine, anoxic, and extreme environments (40, 48, 66, 69, 70, 72). Furthermore, a growing body of evidence suggests the existence of significant genetic diversity among various ciliate taxa which has escaped detection by microscopy (14, 18, 23, 34, 60, 64, 78), pointing to the potential for molecular techniques to generate new insights into ciliate diversity and ecology, and suggesting a need for comparison of the effectiveness of these different techniques in environmental samples.

Terminal restriction fragment length polymorphism (T-RFLP) analysis provides an efficient, inexpensive, and semiquantitative means for comparing microbial molecular diversity between different samples and has been widely used to investigate bacterial communities, although only a few studies have applied T-RFLP methods to the analysis of microeukaryote diversity (6, 16, 17). In this study, ciliate diversity and community structure were investigated in biofilm samples from streams representing a range of levels of anthropogenic degradation, with the objective of testing the null hypothesis that human impacts have no effect upon this important heterotrophic component of stream ecosystems. To achieve this, ciliate-targeted PCR primers were used in conjunction with T-RFLP and multivariate statistical analyses. Additionally, ciliate diversity measures obtained using molecular techniques were compared with those derived from microscopy-based methods in order to assess the relative effectiveness of these approaches.

MATERIALS AND METHODS

Sampling sites.

Biofilm samples were collected from each of four differently impacted streams in Auckland, New Zealand. Site 1 (Cascade Stream) is a largely unimpacted stream, located in an undeveloped native forest catchment (36°53′32" S, 174°31′07" E). Site 2 (Stoney Creek) is mildly impacted, located in a partially developed native forest catchment with nearby houses and roads (36°54′24" S, 174°34′06" E). Site 2 is a lower-order tributary of site 3 (Opanuku Stream), which is proximate to rural agricultural development and is moderately impacted (36°53′42" S, 174°35′44" E). Site 4 (Pakuranga Stream) is located in a highly developed urban catchment (36°53′50" S, 174°54′21" E) and is highly impacted. Sites 1, 2, and 3 all have natural stony substrates, while site 4 consists of a concrete channel at the sampling location.

Sites 1 and 3 are ranked as having the best and fifth-best water quality, respectively, of 25 streams throughout the Auckland region based on monthly monitoring between 1995 and 2005; three locations in the site 4 stream catchment are ranked in the worst five (4, 5). Physical and chemical attributes of the streams are presented in Table 1.

TABLE 1.

Physical and chemical characteristics of streams included in this study throughout 2005^a

Measurement	Result for:
	Site 1	Site 2e	Site 3	Site 4
Catchment area (ha)b	233	375	2,652	275
Land useb (%) (native, forestry, agriculture, urban)	100, 0, 0, 0	98.9, 0, 1.1, 0	55, 2, 25.4, 17.6	1, 0.3, 0, 98.7
Stream widthc (m)	5.8	4.1	6.3	0.7
Water depthc (cm)	11 (10-15)	20 (15-24)	23 (19-28)	13 (14-18)
Water velocityc (ms−1)	0.31 (0.15-0.61)	0.46 (0.18-0.6)	0.71 (0.54-0.88)	0.4 (0.15-0.63)
Temperaturec (°C)	13.6 (11.9-14.4)	13.8 (10.9-16.2)	14.2 (10.8-17.2)	18.1 (14.0-25.5)
pHc	7.6 (7.2-7.9)	7.3 (7.0-7.6)	7.5 (7.2-7.6)	7.5 (7.3-7.8)
Turbidityd (NTU)	3.8 (1.1-13.0)		7.7 (1.9-18.0)	11.9 (3.8-46.9)
Dissolved oxygend (gm−3)	9.9 (8.8-11.1)		9.4 (7.3-10.8)	8.9 (5.8-12.3)
Conductivityd (μS cm−2)	166.4 (134.7-187.9)		142.2 (124.2-168.1)	311.3 (208.9-411.3)
Ammoniacal nitrogend (gm−3)	0.01 (0.01-0.02)		0.03 (0.01-0.05)	0.09 (0.03-0.18)
Nitrate/nitrited (gm−3)	0.01 (0.00-0.03)		0.2 (0.01-0.83)	0.60 (0.13-1.63)
Total Kjeldahl nitrogend (gm−3)	0.31 (0.21-0.93)		0.33 (0.21-1.14)	0.61 (0.20-1.80)
Total nitrogend (gm−3)	0.22 (0.20-0.92)		0.41 (0.20-0.84)	0.96 (0.42-2.24)
Dissolved reactive phosphorusd (gm−3)	0.021 (0.013-0.031)		0.016 (0.010-0.024)	0.021 (0.017-0.045)
Total phosphorusd (gm−3)	0.032 (0.023-0.046)		0.045 (0.024-0.064)	0.100 (0.044-0.336)

^aValues are means and ranges. NTU, nephelometric turbidity units.

^bData were generated using the Land Cover database 2 (Terralink International Ltd., Wellington, New Zealand).

^cData were recorded in this study.

^dData are from reference 5 and are unavailable for site 2.

^eSite 2 is an upstream tributary of site 3. It is presumed that water quality at site 2 is the same as or better than that at site 3.

Sample collection.

Quantitative methods for sampling biofilm material and associated protozoa from submerged surfaces were developed. There are no clearly established protocols for sampling protozoa associated with epilithic biofilms in lotic systems, and for this reason two methods were tested in this study. For both methods, stream biofilm was collected from substrate surfaces while submerged, to avoid the potential loss of material upon removal of stones from the water column (29). The first method involved the use of sterile Speci-Sponges (Nasco, Fort Atkinson, WI) to thoroughly swab submerged surfaces (rocks or concrete channel) within a 55-cm² area defined by a circular neoprene template. Dislodged biofilm material was then squeezed from the collecting sponges into sterile Whirl-Pak bags (Nasco).

The second biofilm collection method involved a syringe sampler based on devices recommended for subsurface sampling of epilithic periphyton (1, 39, 54, 67). The syringe sampler, illustrated in Fig. S1 in the supplemental material, consisted of a 60-ml syringe with its end removed to create a wide opening and a toothbrush head glued to the end of the syringe plunger. A rubber ring was attached to the end of the syringe to seal the sampler against the rock surface and to minimize the loss of dislodged material due to water currents. Biofilm material was removed from a 4.91-cm² area by pressing down and rotating the syringe plunger. Loosened material was drawn up into a 10-ml collection syringe attached to the base of the larger syringe with plastic tubing. Samples were then decanted into sterile Whirl-Pak bags (Nasco).

Biofilm sampling was carried out during January (summer), May (autumn), August (winter), and November (spring) of 2005. On each sampling occasion, biofilm material was collected from two 20 m reaches of each stream. In general, the exposed surfaces of 4 to 10 randomly selected rocks (220 to 550 cm² in total) were sampled using the sponge method, and the surfaces of 10 rocks (about 50 cm² in total) were sampled using the syringe method, from within each 20-m reach of each stream. Similarly, 10 samples were collected from each of two 20-m reaches along the concrete channel at site 4, using each sampling method. The 10 samples obtained using each method at each sample point were combined, giving a total of four composite samples per stream (one sponge sample and one syringe sample from each of two sampling points in each stream). Samples were chilled on ice for transport.

Assessment of ciliate diversity by microscopy-based analysis.

Samples were stored at 4°C and analyzed within 4 to 10 hours of collection. For the enumeration of ciliates, subsamples of 1 ml were transferred to a Sedgewick Rafter cell and scanned at a magnification of ×25 to generate preliminary lists of taxa. Subsamples were then examined at magnifications of ×200 to ×630. Due to the low density of biofilm material and associated ciliates, concentration of samples from site 1 prior to examination was typically required, as follows: 25- to 100-ml samples were concentrated by filtering through 25-μm nylon mesh and backwashing the retentate into a graduated 15-ml tube using filtered water (typical final volume, ∼3 ml). Aliquots (1 ml) were then transferred to a Sedgewick Rafter cell and examined at magnifications of ×200 to ×630. Ciliate cells were identified to at least genus level, where possible, using criteria described in taxonomic keys (25, 53). Photographs were used to ensure that identifications were consistent. The relative abundance of different taxa was scored on a scale of 1 to 8, corresponding to approximate abundances of 1 to 5, 5 to 10, 10 to 15, 15 to 20, 20 to 50, 50 to 100, 100 to 200, and over 200 cells per ml, respectively.

DNA extraction and PCR amplification.

Subsamples (30 ml) of each combined biofilm sample were transferred to preweighed sterile 35-ml centrifuge tubes and centrifuged at 6,000 × g for 10 min at 4°C. Supernatants were removed and pellets resuspended in 15% glycerol to achieve final concentrations of 100 to 200 mg biofilm ml⁻¹. Samples were then frozen at −80°C until required.

DNA was extracted from biofilm samples as previously described (20). Following extraction, the concentration of DNA in each extract was assessed using a Quant-iT PicoGreen double-stranded-DNA kit (Invitrogen, Auckland, New Zealand) according to the manufacturer's directions, in combination with absorbance measurements at 260 nm using a Nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA), and electrophoresis on 1% agarose gels stained with Sybr Safe (Invitrogen). Based on the combined results of these procedures, the concentration of DNA in extracts was standardized to approximately 20 ng μl⁻¹ before use as templates in PCRs.

PCR primers 384F (5′-YTB GAT GGT AGT GTA TTG GA-3′) (20) and 1147R (5′-GAC GGT ATC TRA TCG TCT TT-3′) (20), targeting an ∼700-bp fragment of the ciliate 18S rRNA gene, were labeled at their 5′ termini with 6-carboxyhexachlorofluorescein (HEX) and 6-carboxyfluorescein (FAM) fluorophores (Invitrogen), respectively. DNA was amplified from biofilm extracts using these primers in 50-μl PCR mixes (25 μl GoTaq Green master mix [Promega, In Vitro Technologies, Auckland, New Zealand], 0.5 μM forward and reverse primers, 0.4% BSA [Invitrogen] and 2 μl of template DNA). The following PCR protocol was used: initial incubation for 5 min at 94°C, then 30 amplification cycles of 45 s at 94°C, 60 s at 55°C and 90 s at 72°C, followed by a final extension step of 7 min at 72°C. PCR products were purified using a Purelink PCR purification kit (Invitrogen) according to the manufacturer's instructions. The concentration of fluorescently labeled PCR products was determined by absorbance measurements at 260 nm using a Nanodrop ND-1000 spectrophotometer.

T-RFLP analysis.

T-RFLP analysis is a semiquantitative molecular fingerprinting technique which provides an efficient method of comparing populations. Fluorescently labeled PCR products are digested with one or more restriction enzymes, resulting in the production of fluorescently labeled terminal fragments, the length (in base pairs) and abundance of which can be automatically detected. This results in the generation of profiles in which the number of peaks indicates the number of different terminal fragments present, while the height and area of peaks indicate their relative abundance. As terminal fragment length varies across taxa, these data can provide a profile of community structure within each sample.

For T-RFLP analysis in this study, the DNA concentration in each purified PCR product was adjusted to 20 ng μl⁻¹. PCR products were digested with the restriction endonucleases HaeIII and RsaI (Invitrogen) in 10-μl reaction mixtures, incubated overnight at 37°C. Each digestion reaction mixture contained 1 U of each enzyme, 1 μl of reaction buffer (Invitrogen), and approximately 175 ng of purified amplicon. Digested samples were electrophoresed alongside a size standard with markers at 20-bp intervals up to 1,200 bp (LIZ1200; Applied Biosystems, Melbourne, Australia). Terminal restriction fragments were detected using a 3130XL genetic analyzer (Applied Biosystems). This resulted in generation of peak profiles representing the abundance of HEX- and FAM-labeled terminal fragments, which were analyzed using GeneMapper 4.0 (Applied Biosystems), which automatically calculates the number, height, and area of peaks and their corresponding fragment lengths (in base pairs). Profiles from two runs of each sample were compared to check for consistency, and any inconsistent results were discarded. T-RFLP peaks with a height of ≤50 relative fluorescence units and a length of less than 10 bp were excluded from analysis to eliminate background interference. Each remaining peak was presumed to represent a different terminal restriction fragment and hence a different ciliate-derived 18S rRNA gene sequence. Peaks representing terminal fragments in excess of about 650 bp in length were assumed to represent PCR products which were not cut during restriction digestion.

The length (in base pairs) and area of HEX- and FAM-labeled peaks in each T-RFLP profile were imported into Microsoft Excel. Peak positions were rounded to the nearest whole number, and the overall area of each profile was standardized to 1, to ensure comparability between samples.

Statistical analyses.

Analysis of variance (ANOVA) was used to test for significant differences among numbers of ciliate taxa detected in samples from different streams in each month. Significant ANOVA results were further analyzed with post-hoc Tukey-Kramer honestly significant difference multiple comparison tests. Differences between numbers of taxa in samples obtained using the two different sampling methods and according to the two different analysis methods were investigated using t tests. These analyses were carried out in JMP 7.0 (SAS Institute Inc., Cary, NC). As each different PCR product can be expected to produce both a HEX-labeled fragment (primer 384F) and a FAM-labeled fragment (primer 1147R), the numbers of HEX-labeled and FAM-labeled T-RFLP peaks were averaged to provide a composite diversity estimate for each sample.

For multivariate analysis, data from both HEX-labeled and FAM-labeled T-RFLP peaks were combined into a single data set for each sample and subjected to a square-root transformation to moderate the influence of large peaks in subsequent analyses. Relative abundance data obtained by microscopy were not transformed. Multivariate analyses were carried out in Primer v6.1.6 (Primer-E Ltd., Plymouth, United Kingdom). Bray-Curtis similarity between all pairs of biofilm samples was calculated. Similarities and differences among ciliate communities were visualyzed by nonmetric multidimensional scaling (MDS), which clusters samples with higher levels of pairwise similarity more closely than samples with lower pairwise similarity. Analysis of similarities (ANOSIM) was used to test the null hypotheses of no significant differences between ciliate communities from different streams, seasons, and sampling methods. ANOSIM compares within-group similarity and between-group similarity; R values around 0 indicate that within-group and between-group similarities are the same, while R values approaching 1 indicate that samples within groups are more similar to each other than to samples from different groups, allowing the null hypothesis to be rejected (15).

RESULTS

Sampling methods.

The effectiveness of sponge- and syringe-based biofilm sampling methods for detecting ciliates was compared, but there was no clear evidence of greater efficacy of either method in terms of ciliate diversity detected, according to either microscopy or T-RFLP. Little evidence for significant differences in ciliate community structure was found between sponge-derived and syringe-derived samples according to ANOSIM (Table 2), and furthermore, sponge- and syringe-derived samples from the same stream and sampling date were typically grouped very closely in MDS plots, suggesting very similar composition of the communities sampled by each method. Sponge- and syringe-derived samples from each site and date were therefore pooled for subsequent analyses.

TABLE 2.

Comparison of ciliate diversity in sponge and syringe samples according to microscopy and T-RFLP analysis (ANOSIM)

Month of sponge-syringe sample comparison	Difference according to microscopy		Difference according to T-RFLP
	Global R statistic	P	Global R statistic	P
Januarya	0.438	0.07
May	0.031	0.48	−0.106	0.67
August	0.167	0.44	−0.086	0.59
Novembera	−0.313	1.00

^aMolecular analysis of January sponge and November syringe samples was unsuccessful.

Ciliate diversity in stream biofilm samples (microscopy and T-RFLP methods).

Our methodology targeted a standard area of substrate for all samples, and our diversity results represent the number of taxa detected per area sampled. However, biofilm biomass was particularly sparse at site 1, especially in winter (G. Lewis and S. Tsai, unpublished data). Similarly, the density of ciliates at site 1 was generally very low, necessitating concentration of these samples for analysis. Even after concentration, the number of ciliate cells detected and identified in site 1 samples was low, and this may have affected the level of diversity detected.

The number of ciliate taxa detected in stream biofilm samples using microscopy-based methods ranged from 0, for samples from site 1 in May and August, to 17, for a site 4 sample from May (Fig. 1). Site 1 typically had the fewest biofilm-associated ciliate taxa, while samples from site 3 and site 4 contained the highest numbers of ciliate taxa. Significant differences were detected among microscopy results from May and November (ANOVA; P < 0.05) but not from January or August.

FIG. 1.

Numbers of ciliate taxa detected by microscopy and numbers of ciliate 18S rRNA gene sequences indicated by T-RFLP analysis of stream biofilm samples. The numbers on the x axis denote samples from two different reaches within each stream. Each bar shows the mean of counts from two samples (microscopy data) or the mean of HEX-labeled and FAM-labeled terminal fragment counts from two samples (T-RFLP data). Error bars show one standard deviation. Significant differences were found between streams during each month according to T-RFLP analysis and in May and November according to microscopy data (ANOVA; P < 0.05). Within each month, samples not linked by the same letter (A to C) are significantly different (Tukey-Kramer honestly significant difference; P = 0.05). *, samples from site 1 were typically concentrated due to low biomass and ciliate abundance; there was insufficient biofilm biomass at site 1 in August for T-RFLP analysis.

The number of different peaks present in a T-RFLP profile is assumed to reflect the number of different ciliate 18S rRNA gene sequences—and therefore the number of ciliate taxa—present in the stream biofilm sample. The number of T-RFLP peaks detected ranged from 5 (site 1, November) to 61 (site 3, January) (Fig. 1). Significant differences were detected in all months (ANOVA; P < 0.005). The pattern of ciliate diversity across different streams according to T-RFLP analysis was broadly similar to that derived from microscopic investigations, with the lowest ciliate diversity being detected in the most pristine stream and higher diversity at the more impacted sites.

Overall, in each stream the average number of peaks detected in T-RFLP profiles exceeded the average number of ciliate taxa detected by microscopy (t test; P < 0.0001) (Fig. 2). In total, 183 different HEX-labeled terminal fragments and 191 different FAM-labeled terminal fragments were detected among all samples, compared with 68 different ciliate taxa identified by microscopy.

FIG. 2.

Average numbers of ciliate taxa detected by microscopy (white bars) and average numbers of ciliate 18S rRNA gene sequences indicated by T-RFLP analysis (gray bars) in stream biofilm samples throughout 1 year. Each bar shows the mean of counts from eight samples (microscopy data) or the mean of HEX-labeled and FAM-labeled terminal fragment counts from eight samples (T-RFLP data). Error bars show one standard deviation. The average number of T-RFLP peaks detected significantly exceeded the average number of taxa identified by microscopy in all cases (t test; P < 0.0001).

Differences in ciliate diversity and community structure between streams and seasons according to microscopy and T-RFLP.

Microscopy-based analysis of biofilm samples found evidence of ciliate taxa common to multiple stream environments as well as many taxa restricted to individual sites (Fig. 3). Taxa common to all four streams were detected only in January and November, although these taxa were always present in at least one stream throughout the year. In January, a greater number of unique taxa were found in site 2 than in site 3. In all other months, taxa unique to site 3 and site 4 together accounted for the majority of taxa detected, although these two streams also had generally higher overall levels of diversity than the other sites. Taxa unique to site 4 were particularly frequent in August. In November, taxa unique to site 2 were not detected, and taxa unique to site 1 were evident only in May. Fewer than one in five of the taxa unique to particular streams were detected on more than one sampling date.

FIG. 3.

Comparison of diversities of ciliate taxa occurring in biofilms from different stream environments, according to microscopy and T-RFLP analysis. *, there were no T-RFLP data for site 1 in August due to insufficient biofilm biomass for analysis.

Overall, only 7% of ciliate taxa identified throughout the year using microscopy were common to all four streams (Fig. 3). Forty-four percent of the detected taxa were each found in only one stream, most commonly site 3 or site 4, and no taxa were unique to the least-impacted stream, site 1.

According to microscopy, ciliates common to all four streams were generally small species from Oligohymenophorea or Phyllopharyngea, such as Glaucoma spp., Trochilia spp., or Cyclidium spp. These ubiquitous species were generally more abundant in the more impacted streams. Several unidentified hypotrichs were typically characteristic of site 2 biofilm samples, while taxa unique to site 3 included Actinobolina spp., Aspidisca lynceus, and sessile peritrich species such as Epistylis spp. and Vorticella spp. A large number of taxa were found only at site 4, including Spirostomum spp., Stylonychia spp., Euplotes spp., Strombilidium spp. and predatory taxa, including Monodinium spp. and several species of Litonotus.

Visual inspection of T-RFLP profiles shows that some peaks are present in profiles from multiple streams (although these peaks are often markedly different in magnitude), while other peaks appear to be unique to particular stream biofilms (Fig. 4). Overall, 17% of the different T-RFLP peaks detected throughout the year were found in all four streams. Forty-five percent of the T-RFLP peaks occurred in only one stream (Fig. 3), consistent with the microscopy data. The proportion unique to each stream ranged from 5% (site 1) to 16% (site 3).

FIG. 4.

T-RFLP profiles derived from stream biofilm samples in November 2005 using ciliate-targeted PCR primers. Sizes of gray and black peaks, respectively, indicate the abundance of HEX-labeled and FAM-labeled terminal restriction fragments.

The proportions of T-RFLP peaks found in different streams showed a higher degree of consistency between months than the microscopy data (Fig. 3). Peaks unique to each of the four streams were detected in all months except August, when biofilm biomass at site 1 was insufficient for molecular analysis. Compared with the microscopy data, the proportion of T-RFLP peaks unique to site 3 and site 4 together accounted for less of the total diversity detected, except for in January samples. Peaks unique to site 1 were more frequently detected, however, as were peaks unique to site 2 in November samples. As for the microscopy data, T-RFLP peaks unique to site 4 were most frequent in August samples, although the proportion of peaks unique to site 3 was highest in January.

The T-RFLP peaks found in all four streams throughout the year (17% of all peaks) together accounted for 75% of total profile area, indicating that the corresponding terminal fragments were relatively abundant (Fig. 5). Conversely, the peaks that were each detected in only one stream (45% of all peaks) accounted for less than 6% of the total T-RFLP profile area, and as for the microscopy data, few of these unique T-RFLP peaks were detected on more than one sampling date. These findings suggest the existence of populations of abundant and cosmopolitan ciliate taxa in stream biofilms from different environments, together with rarer taxa, which had low abundance and restricted spatial and temporal distributions.

FIG. 5.

Abundance and diversity of ciliates in stream biofilms according to T-RFLP profiles. Bars represent the combined area of T-RFLP peaks found in profiles from individual streams only and peaks found in profiles from multiple streams, as a proportion of total profile area for each sampling date. Numbers of peaks contributing to each bar are indicated. *, there were no T-RFLP data from site 1 for August.

Microscopy relative abundance data and T-RFLP peak area data were used to generate nonmetric MDS plots in which each data point represents the assemblage of ciliate taxa or T-RFLP peaks detected in one sample. The proximity of the data points to each other reflects the relative similarity of their ciliate assemblages. MDS plots based on microscopy data show separation of samples from highly impacted site 4 and moderately impacted site 3 from each other and from samples from less impacted site 2 and site 1 (Fig. 6). Similarly, MDS plots based on T-RFLP data show site 4 samples forming a clearly separated group, while site 3 and site 2 samples form overlapping clusters and site 1 samples are scattered widely. Temporal patterns are less clear, with little discernible grouping of data points by sampling month according to microscopy data. According to T-RFLP data, November samples and May samples from site 1 are more dispersed than January and August samples.

FIG. 6.

MDS grouping of ciliate assemblages in stream biofilm samples based on microscopy data (left; two-dimensional stress = 0.15) and T-RFLP profiles (right; two-dimensional stress = 0.19).

ANOSIM analysis of microscopy and T-RFLP data provided evidence of significant differences between ciliate communities in different streams and between samples from different seasons (Table 3). For the microscopy data, significant and generally large differences were found between ciliate assemblages from all streams, and moderate differences between assemblages from different sampling dates. Similarly, significant differences were found between ciliate assemblages from all streams according to T-RFLP analysis, with the exception of site 1 and site 2. The largest differences according to T-RFLP were found between site 4 and both site 2 and site 3, while the largest differences according to microscopy were found between site 1 and site 3 and between site 3 and site 4. The largest difference between sampling months according to T-RFLP analysis was between January and August, but these months showed relatively little difference according to microscopy. Significant differences were not detected between January and May or between January and November T-RFLP results.

TABLE 3.

ANOSIM comparison of ciliate assemblages in samples from different streams and sampling dates according to microscopy and T-RFLP data

Comparison	R	Pa
Stream comparison
Microscopy datab
Global	0.719	0.001*
Site 1-Site 2	0.414	0.001*
Site 1-Site 3	0.988	0.001*
Site 1-Site 4	0.798	0.001*
Site 2-Site 3	0.748	0.001*
Site 2-Site 4	0.621	0.001*
Site 3-Site 4	0.842	0.001*
T-RFLP datac
Global	0.49	0.001*
Site 1-Site 2	0.269	0.102
Site 1-Site 3	0.497	0.01*
Site 1-Site 4	0.508	0.006*
Site 2-Site 3	0.332	0.037*
Site 2-Site 4	0.769	0.002*
Site 3-Site 4	0.678	0.001*
Sampling date comparison
Microscopy datab
Global	0.383	0.001*
January-May	0.487	0.002*
January-August	0.284	0.011*
January-November	0.469	0.001*
May-August	0.257	0.011*
May-November	0.428	0.001*
August-November	0.418	0.001*
T-RFLP datac
Global	0.395	0.001*
January-May	0.201	0.105
January-August	0.722	0.002*
January-November	0.125	0.222
May-August	0.502	0.001*
May-November	0.433	0.006*
August-November	0.376	0.020*

^a*, statistically significant result (P < 0.05).

^bn = 13 to 16 samples per stream or month.

^cn = 8 to 12 samples per stream and 8 to 15 samples per month.

Links between environmental data and ciliate assemblage MDS data.

Links between ciliate community assemblage data and environmental trends can be visualized as bubbles overlaid on MDS ordination plots, with the size of bubbles representing the magnitude of environmental parameters at sites and dates corresponding to biofilm sampling occasions. The resulting figures suggest that observed MDS ordination patterns are associated with a combination of environmental factors (Fig. 7). Separation of site 4 samples from others appears to be associated with factors typically associated with urban stream degradation, such as higher levels of nitrogenous compounds and lower levels of dissolved oxygen, in addition to very low levels of forest cover. Turbidity, temperature, and phosphorus levels are similarly elevated at site 4 on certain sampling occasions, while water velocity and pH do not follow this trend. Conversely, site 1 samples are associated with high native forest cover, low levels of nitrogenous compounds, total phosphorus and turbidity, and elevated oxygen. Grouping of site 3 samples in a cluster adjacent to site 4 samples appears to be related to intermediate levels of nitrogenous compounds and total phosphorus, although pH and levels of dissolved reactive phosphorus are generally lower than for site 1 samples.

FIG. 7.

MDS grouping of ciliate assemblages in stream biofilm samples, based on T-RFLP analysis, showing magnitude of various environmental parameters associated with samples (two-dimensional stress = 0.19). Bubble sizes are scaled to reflect the ranges of values indicated in Table 1. Samples for which environmental data were unavailable were omitted.

DISCUSSION

Ciliate diversity according to microscopy and T-RFLP analysis.

Ciliates have been considered very amenable to microscopic study due to their high level of morphological diversity and relatively large size. However, it has been suggested that a current list of described ciliate morphospecies may contain 5 to 10 times as many biological species (13, 24, 26). Furthermore, ciliate morphospecies have been shown to include organisms with clearly different ecophysiological characteristics (75). This suggests that the morphospecies concept may substantially underestimate ciliate species diversity and ecosystem complexity (26, 75). Morphologically identical but genetically, physiologically, or biochemically divergent ciliates are likely to occupy separate ecological niches, and measurement of this functional ciliate diversity is relevant to ecological studies. While molecular techniques have recently contributed to great insights into protistan diversity in various inaccessible and extreme environments, few studies have applied these techniques to specific phyla, such as Ciliophora.

This investigation found that the number of ciliate taxa suggested by T-RFLP analysis was more than double that indicated by microscopy. This suggests the existence of a significant component of genetic ciliate diversity in stream biofilms which is not evident upon microscopic examination of morphology. A similar finding was made in a recent study of oligotrich ciliate diversity in seawater, with diversity according to molecular analysis being about 10-fold higher than that according to morphological observations (18). In the present study, ciliates were identified using simple light microscopy methods, and although most species can be detected at relatively low magnification, it is possible that some organisms may have been overlooked due to small size or inconspicuousness, or due to their being present in encysted form. Silver staining procedures and electron microscopy can improve taxonomic discrimination of ciliate taxa based on morphological and morphometrical analysis. However, even when these more complex methods have been used, ciliate diversity based on molecular analysis has still been found to exceed diversity according to morphology (34). A growing number of studies provide evidence of cryptic molecular diversity exceeding apparent morphological diversity in various ciliate taxa, including Carchesium (78), Cyclidium (23), Halteria (34), Oxytricha (60), Strombidium (34), Tetrahymena (41, 49, 64), and Zoothamnium (14). The consistency of these findings suggests little reason to expect any difference in the great majority of ciliate taxa which have not yet been subjected to genetic analysis. Cyclidium and rRNA gene sequences closely matching those of Oxytricha, Tetrahymena, and Zoothamnium have all previously been detected in these Auckland streams (20) and therefore may have contributed to the cryptic genetic diversity detected using T-RFLP in this study.

Although our molecular analysis indicates a high level of genetic diversity underlying ciliate morphospecies, the PCR primers used in this study, while highly ciliate specific, are not perfectly so (20). It is possible that a limited number of nonciliate sequences and resulting terminal restriction fragments may be represented in our results. However, this effect may be outweighed by the tendency of T-RFLP analysis to underestimate diversity of closely related taxa, due to conservation of restriction sites and consequent generation of terminal fragments of identical length (16). Additionally, ciliate taxa with indistinguishable 18S rRNA gene sequences may be discriminated by examination of other genes (41), suggesting that the sequences targeted in this study may not provide complete resolution of different species. It thus seems likely that the assessments of ciliate diversity provided by T-RFLP analyses in this study are conservative.

Limitations and complementarity of methods.

Although T-RFLP analysis is an efficient method of obtaining and comparing microbial genetic diversity data, interpretation of T-RFLP information, in isolation, remains challenging. T-RFLP analysis lacks a straightforward means of reliably assigning taxonomic identities to observed peaks in complex samples, particularly for groups of organisms for which availability of DNA sequence data is limited, such as ciliates. The value of diversity measures derived from T-RFLP analyses has been questioned, on the bases that different organisms may contribute to single T-RFLP peaks, different restriction enzymes will produce different results, and the use of thresholds to eliminate background noise from T-RFLP profiles means that terminal fragments (and organisms) of low abundance will be excluded from the resulting analysis (10). Multivariate statistical analysis of T-RFLP data is considered reliable, however, with conclusions being little affected by the exclusion of minor T-RFLP peaks or the choice of restriction enzyme (8, 77). T-RFLP is thus a useful method for comparing complex microbial community structures.

Microscopy-based analysis of morphology does permit identification of ciliates, subject to sufficient taxonomic expertise being available, and can allow the classification of ciliates into functional categories, such as feeding groups, which can be used to examine the ecological role of protozoa (55). The level of taxonomic resolution used can affect whether significant differences between protozoan communities will be detected, with identification of protozoa to taxonomic levels higher than genus being less effective in discriminating surface-associated protozoan communities (35). Microscopy does have the advantage of quantitative power—cells can be counted, albeit tediously—which may be lacking in PCR-based assays. Clearly, morphological identification is possible only within the constraints of the morphospecies concept, which has recognized limitations (26).

In this study, both T-RFLP and microscopy-based analyses showed broadly similar overall trends of diversity in the different streams, and both methods produced evidence of significant differences between ciliate communities in differently impacted stream biofilms. It seems, therefore, that T-RFLP and microscopic analyses may be considered complementary methods, the former providing a robust and efficient method for comparing ciliate community structure, and the latter allowing attribution of differences between microbial assemblages and systems to particular ciliate taxa or functional groups. Of course, group-targeted PCR primers such as those used in this study can be used for cloning and sequencing, which—if sequence and morphological data have been reconciled—does allow reliable identification of microbes in environmental samples, thus avoiding one of the limitations of T-RFLP. Fluorescence in situ hybridization-based techniques offer a useful means of linking molecular sequence data with microscopy-derived morphological information (68).

Further studies combining molecular and microscopic methods are necessary for the expansion of currently limited sequence database coverage of microeukaryotes (21). Assuming that the availability of microeukaryote DNA sequence information will improve, it seems probable that in future, molecular identification methods may prove more straightforward and accurate than morphology-based methods. However, it is likely that combined approaches may prove more informative than either molecular or microscopy-based techniques alone (68, 73). Molecular profiling methods allow efficient and robust comparisons of community structure, while identification and description of taxa using sequencing, microscopy, and fluorescence in situ hybridization-based techniques can provide additional insights and links to ecological, phenotypic, and physiological information and may allow pinpointing of ecologically important organisms.

Differences between biofilm ciliate communities in different streams.

Both microscopy and T-RFLP analysis methods have provided clear evidence of differences in ciliate assemblages in biofilms from differently impacted streams. These differences can be associated with environmental parameters typical of urban stream degradation, suggesting that our null hypothesis, that human impacts have no effect upon biofilm-associated ciliate communities, can be rejected.

The fewest ciliate taxa were found at site 1, a relatively pristine stream, while the most ciliate taxa were detected in samples from moderately impacted site 3 and highly impacted site 4. This trend of greater diversity at the more impacted sites seems contrary to the generally accepted tendency for ecological perturbation to lead to a simplification of community structure. A number of contrasting biotic and abiotic factors suggest possible reasons for this finding. Site 1 and site 2 are characterized by little exposure to sunlight or anthropogenic pollutants. Site 3 and site 4, in contrast, are exposed to elevated nitrogen loads, derived from nearby areas of agricultural and urban land use, respectively. Reduced density and height of riparian vegetation expose site 2 and site 3 to more sunlight than site 1, and site 4 receives virtually no shade whatsoever. Nutrient enrichment and sunlight have been shown to promote periphyton growth in lotic systems (27, 32, 74), suggesting that the more impacted sites are likely to have communities of more abundant phototrophic organisms than the less impacted streams. This is consistent with observations of increased biofilm biomass and more diverse bacterial and algal (especially diatom) communities at sites 3 and 4 (Lewis and Tsai, unpublished). The three less impacted sites in this study also receive significant amounts of allochthonous debris from surrounding vegetation, while site 4 does not, suggesting a difference in the relative importance of detritus-derived nutrients in these streams. Elevated nutrient availability can increase benthic ciliate abundance in rivers and streams (19, 50, 58) and may affect ciliate abundance, biomass, and community composition in lentic habitats (2, 33, 51, 76). It seems possible that the abundant biofilms in the more impacted streams in this study may provide more resources and a wider variety of feeding niches for heterotrophic protozoan organisms. The greater variety and abundance of bacterivorous, algivorous, and predatory ciliates detected in samples from the two more impacted sites is consistent with this suggestion.

Site 1 is home to diverse and abundant benthic macroinvertebrates, while site 4 has a macrobenthic invertebrate fauna of very low diversity, consisting almost entirely of chironomid larvae (38). Biofilms at site 1 may therefore be subjected to very different grazing pressures than biofilms at site 4, which is likely to further affect the nutrient resources available in these streams. Macroinvertebrates may also negatively affect protozoa through predation (51, 71). In addition, macroinvertebrates consume meiofauna, such as rotifers (59), which may also predate upon protozoa (47). Studies of the effects of top-down predation pressures on ciliates in lakes and ponds have had mixed results (2, 51, 76). There is very little information available on the nature of trophic interactions between ciliates and invertebrates in stream biofilms, although one study found evidence of invertebrate predation and/or competition negatively affecting biofilm-associated ciliates (46). Nevertheless, it can be speculated that the homogeneity of the invertebrate community at site 4 may mean that ciliates of only certain types and sizes are subjected to predation pressures, resulting in the selective proliferation of nontarget taxa. In contrast, the diverse invertebrates at site 1 may represent a broad competitive and predatory factor, perhaps contributing to the lower abundance and diversity of ciliates at this site.

Different catchment land uses may cause development of different biofilm-associated ciliate assemblages by favoring tolerant taxa while eliminating sensitive organisms. Being surrounded by extensive urban development, site 4 is probably exposed to various pollutants in addition to increased levels of nitrogenous compounds and lower levels of dissolved oxygen. Furthermore, the artificial substrate in site 4 may lack refuges for flow-sensitive or light-sensitive organisms. A previous investigation suggested that site 4 may be home to fewer sessile peritrich taxa and a higher frequency of predatory ciliates such as Litonotus spp. and Loxophyllum spp. than the other streams investigated in this study (20). Similarly, in this study, predatory ciliates such as Monodinium spp. and Litonotus spp. were detected by microscopy-based analysis only in the site 4 biofilm. These, and other characteristic taxa identified in site 4, are typically found in β-α mesosaprobic or polysaprobic waters, indicating that they can tolerate reasonably high organic loads and hence more heavily polluted environments (25). This suggests that physicochemical conditions in site 4 influence the development of a very different ciliate community than those in the less impacted streams included in this study. How this different ciliate community affects the ecological processes and interactions occurring in this stream awaits further investigation.

Conclusion.

Ciliates and other protozoa are major predators of bacteria and provide an important trophic link in aquatic habitats, such as stream biofilms (52). Understanding protozoan community diversity and abundance is therefore important for gaining insights into the function of these hot spots of microbial activity, which contribute substantially to ecosystem processes in streams (7). Both molecular and microscopy-based analyses provided evidence of diverse biofilm-associated ciliate communities, with greater diversity being seen in the more impacted streams and with significant differences between ciliate assemblages in streams in different states of degradation being observed. These observations may be related to a variety of differences in environmental parameters characteristic of urban stream degradation, such as elevated nutrient and sunlight availability, as well as different assemblages of autotrophic biofilm organisms and communities of benthic macroinvertebrates. The discrepancy between numbers of taxa suggested by T-RFLP analysis and those obtained by microscopy-based analysis in this study adds to evidence that ciliate diversity has been underestimated by traditional microscopic approaches. Microscopic analysis allowed identification of ciliate taxa characteristic of different stream environments, however. Future application of these complementary techniques will improve our understanding of the causes and effects of stream degradation at the microbial level, leading to development of more effective stream monitoring and remediation strategies.