Rates of Species Accumulation and Taxonomic Diversification during Phototrophic Biofilm Development Are Controlled by both Nutrient Supply and Current Velocity

Abstract

The accumulation of new and taxonomically diverse species is a marked feature of community development, but the role of the environment in this process is not well understood. To address this problem, we subjected periphyton in laboratory streams to low (10-cm · s⁻¹), high (30-cm · s⁻¹), and variable (9- to 32-cm · s⁻¹) current velocity and low- versus high-nutrient inputs. We examined how current velocity and resource supply constrained (i) the rates of species accumulation, a measure of temporal beta-diversity, and (ii) the rates of diversification of higher taxonomic categories, defined here as the rate of higher taxon richness increase with the increase of species richness. Temporal biofilm dynamics were controlled by a strong nutrient-current interaction. Nutrients accelerated the rates of accumulation of new species, when flow velocity was not too stressful. Species were more taxonomically diverse under variable than under low-flow conditions, indicating that flow heterogeneity increased the niche diversity in the high-nutrient treatments. Conversely, the lower diversification rates under high- than under low-nutrient conditions at low velocity are explained with finer resource partitioning among species, belonging to a limited number of related genera. The overall low rates of diversification in high-current treatments suggest that the ability to withstand current stress was conserved within closely related species. Temporal heterogeneity of disturbance has been shown to promote species richness, but here we further demonstrate that it also affects two other components of biodiversity, i.e., temporal beta-diversity and diversification rate. Therefore, management efforts for preserving the inherent temporal heterogeneity of natural ecosystems will have detectable positive effects on biodiversity.

INTRODUCTION

Fundamental objectives in ecology are to understand patterns of biodiversity and turnover through space and time in response to environmental changes. One of the simplest measures of biodiversity is species richness. However, observations of species richness are influenced by the size of area sampled, the time frame of sampling, and the joint effects of the two (1, 2). Illustrating these points are the well-known and extensively studied species-area relationship (SAR), or the increase in species richness with the area sampled, and the less-studied temporal companion to SAR, i.e., the species-time relationship (STR), or the increase in species richness with the time span of observation (3, 4). Several factors influence STR or temporal beta-diversity, such as the observation of rare species previously unsampled, new species dispersing into the area, or the emergence of new species from dormant propagules with ecological succession (5). Consequently, factors which control ecological succession in a community will likely also affect the rate of species accumulation, yet experimental evidence is extremely limited and only recently have environmental conditions been shown to influence relationships between taxa and time (4, 6–8). Environmental stress has been shown to influence STR, e.g., in bacterial communities growing in bioreactors, STR decreased with increasing industrial wastewater concentrations (6); in large mammals, STR increased with temperature (8). Nonetheless, the response of STR to variation in environmental conditions has received considerably less attention than that of SAR (5), despite suggestions that similar underlying mechanisms generate the two patterns (9).

In addition to species accumulation rates, the extent to which communities experiencing different environmental conditions vary taxonomically is also largely unresolved. Depending on whether environmental conditions change throughout community development, newly added species may have either similar or very different niche requirements compared to those of species already present, which should influence the rates of increase of higher taxa with the rise in species richness. This will affect the taxonomic composition because closely related species are more likely to share similar traits and respond similarly to environmental conditions (10). For example, harsh environments and stress can lead to environmental filtering such that taxonomically similar species accumulate through time (10, 11), yet consistent patterns have not been observed (12). Alternatively, biotic modification of the environment by the growing community may allow only species with new strategies, e.g., for coping with crowding, and consequently with different taxonomic affiliations to establish, creating a taxonomic disparity between resident species and new recruits. Temporal variability in the form of disturbance or seasonally changing environmental conditions is an important source of heterogeneity (e.g., intermediate disturbance hypothesis [13]). It may provide new niches through time, allowing greater temporal coexistence and cumulative biodiversity than those of more temporally homogenous environments (14, 15).

Microbial communities, having a polyphyletic nature and high growth and turnover rates, could provide valuable systems for examining how temporal beta-diversity and taxonomic dissimilarity vary under different environmental conditions. In particular, attached algae or periphyton is a cohesive community encompassing photosynthetic organisms spanning nine phyla and two kingdoms (16). Periphyton is highly responsive to environmental conditions, which explains its broad use in bioindication (17). Coexistence in the periphyton is a function of resource supply and disturbance, which control the growth of species, broadly classified as tolerant versus sensitive to these influences. Under high resource availability but low disturbance, the two groups of species coexist and form thick biofilms, while mainly tolerant species, organized in thin biofilms, survive under nutrient and/or disturbance stress (18–20). Consequently, biofilms growing under resource-replete conditions may also contain a greater assortment of species exhibiting a wider variety of traits (both trophic and morphological), while under conditions of stress, only stress-resistant species from a smaller set of tolerant genera may co-occur. Therefore, we predict that in general, higher rates of species accumulation or temporal beta-diversity will be observed in nutrient-replete communities than in nutrient-limited communities. High nutrient concentrations result in thick biofilms with long internal resource gradients, allowing finer resource partitioning and considerable niche overlap (21). This translates into coexisting species having a higher likelihood of being taxonomically related; therefore, we expect lower rates of diversification of higher taxa (DHT) under high-nutrient conditions.

In addition to nutrient concentrations, periphyton communities in streams and rivers are also highly influenced by current velocity, such that high current reduces the species pool through elimination of loosely aggregated species or those without firm attachment, while intermediate, subscouring flows can stimulate biofilm growth by increasing algal metabolism and nutrient uptake (22). However, rarely is flow constant in natural streams and rivers and spatial heterogeneity in flow has been shown to increase diversity (23, 24), while temporal variability in flow has an impact on species composition (25) in biofilms. Flow variability can enhance coexistence in comparison to constant-flow conditions by providing opportunities for growth of both flow-sensitive and flow-tolerant species. Community composition and rates of biomass accumulation and nutrient uptake vary with changes in current velocity and nutrient concentrations in periphyton communities (26–29), yet relatively few studies have examined experimentally the effects of both factors on temporal trends in community dynamics. It is likely that environmental conditions, such as nutrient abundance and current velocity, interact to influence the rates of both species accumulation and diversification in periphyton communities, which are explored here across six current-nutrient treatments.

We generated three current treatments, consisting of a low- and a high-constant-flow treatment, as well as a variable-flow treatment, crossed with two nutrient treatments and tested the following hypotheses: (i) rates of species accumulation in periphyton communities would differ between treatments varying in current velocity and nutrient abundance such that higher rates would be observed in the high-nutrient treatments and in the low- and variable-flow treatments but low rates would be observed in all high-current treatments; (ii) the rate of diversification of higher taxa would differ significantly across treatments, where lower taxonomic diversification would be observed under high-nutrient conditions due to greater diversity and length of environmental gradients within the thick biofilms allowing the accumulation of closely related species (i.e., species from the same genera, families, etc.).

MATERIALS AND METHODS

Artificial stream flumes.

Experiments were conducted in six donut-shaped laboratory streams, each with an experimental trough measuring 80 cm in length, 12 cm in width, and 13 cm in depth (see Fig. S1 in the supplemental material). Eighty liters of modified Guillard's WC medium (see below) was recirculated in each stream channel at a uniform current velocity (±1 cm · s⁻¹) or variable flow (see below) measured by a Marsh-McBirney model 2000 flowmeter (Marsh-McBirney Inc., Frederick, MD). Current velocity in each stream was maintained by adjusting a belt and multiple drive step pulleys attached to a 1.5-hp motor and a water pump. For the variable-flow treatments, current velocity was adjusted every 3 days for the duration of the experiments. Drop-in chillers (1/5 hp; TradeWind Chillers, Escondido, CA) maintained stream water at room temperature (∼20°C) in the high- and variable-velocity channels, while the water in the low-velocity channels was at room temperature. In each experimental trough, 4.9- by 4.9-cm unglazed porcelain tiles were placed equidistant from one another. A 250-watt metal halide lamp, positioned above each experimental trough, provided sufficient light for attached algae, i.e., ∼200 μmol · m⁻² · s⁻¹ (16) on a 14:10 daily light/dark ratio.

Temporal trends within algal communities were examined under different current and nutrient regimes. Streams were subjected to constant flows of either 10 or 30 cm · s⁻¹ or a variable-flow treatment (mean flow of 20 cm · s⁻¹; standard deviation [SD], ±8; see Fig. S2 in the supplemental material) to simulate temporal heterogeneity in flow. In the variable-flow treatments, flows were randomly chosen from current velocities ranging between 9 and 32 cm · s⁻¹. Additionally, nutrient concentration was varied across current regimes with either high (11,210-μg · liter⁻¹ N-NO₃ and 1,550-μg · liter⁻¹ P-PO₄) or low (290-μg · liter⁻¹ N-NO₃ and 40-μg · liter⁻¹ P-PO₄) levels in modified Guillard's WC medium (30). Other than the manipulation between treatments of nitrate and phosphate, modified WC medium consisted of all constituents in their normal concentrations. Nutrient analyses of water samples, collected at the time of algal sampling, were performed with AutoAnalyzer III (Seal Analytical Inc., Mequon, WI). In our low-nutrient treatments, average concentrations (μg · liter⁻¹) of NO₃⁻ and PO₄³⁻ were 208 to 407 and 5.95 to 6.53, respectively, which were within the ranges shown to limit algal communities (31, 32). Conversely, concentrations in the high-nutrient treatments, averaging 7,730 to 9,123 for NO₃⁻ and 779 to 1,165 for PO₄³⁻, greatly exceeded these values. Hence, there were six treatments: low nutrients at 10- and 30-cm · s⁻¹ constant flow and a variable flow, referred to as 10-low, 30-low, and variable-low, respectively, and high nutrients at 10- and 30-cm · s⁻¹ constant flow and a variable flow, referred to as 10-high, 30-high, and variable-high, respectively. The limited number of channels did not allow for replication in each flow × nutrient treatment, so the experiment itself was replicated 3 times in November and December 2006 and February 2007, with each experimental run lasting 35 days, when sloughing occurred in some channels. For one of the experimental runs, equipment failure resulted in the termination of the variable-high treatment after day 18. For the duration of each run, 24 liters of water was replaced every third day with new medium.

All artificial streams were seeded once, at the beginning of the experiment, with epilithic algae from several small streams in the Dallas-Fort Worth area encompassing diverse physicochemical conditions. Seed algae were suspended in carbon-filtered water, and 2 liters of this mixture was dispensed to each stream. Not unexpectedly, taxonomy varied between the three seed communities, yet species which became abundant during the course of community development were either observed at very low numbers or not encountered in the seed algae. Furthermore, the procedures that we employed in PRIMER (see below) accounted for variability between replicates.

Sample preparation and analysis.

After allowing an initial period of 7 days for biofilm colonization, two tiles were randomly retrieved from each stream channel. Tiles were taken from the same locations within each stream on days 7, 11, 14, 18, 21, 25, 28, and 35. According to procedures outlined in the work of Larson and Passy (33), five random fields on each tile were examined with a Zeiss Axioplan 2 LSM 510 Meta confocal microscope (Zeiss, Jena, Germany) using a 40- by 0.80-numerical aperture (NA) water-immersion objective. Biofilm thickness for each field was measured, and a mean value was calculated across the 10 measured fields (33) from the two sampled tiles. Following quantification of biofilm thickness, the biomass on the surface of each tile was removed with a razor blade and a toothbrush until visibly clean. The tiles were then returned back into the streams but never retrieved again for the duration of the experiment. Biomass from the two tiles was consolidated, suspended in carbon-filtered water, and preserved in 4% buffered formalin solution. Samples were uniformly mixed by pulse sonication, and a subsample was placed into a Palmer-Maloney counting cell and observed under a light microscope at an ×400 magnification. Algal community composition was assessed by counting 20 random fields and a minimum of 500 algal units, where an algal unit was an individual cell for unicellular organisms, a 25-μm length for filaments, and a 10-μm by 10-μm area for colonies. Soft algae were identified in this count, and diatoms were lumped into a single taxonomic category. For diatom species identification, material was acid digested, washed with carbon-filtered water, and mounted in Naphrax (Brunel Microscopes Ltd., Chippenham, Wiltshire, United Kingdom) mounting medium. Three hundred diatom units (unit = one frustule or two valves) were counted and identified for each sample at an ×1,000 magnification. Counts were converted to density of units per surface area of tiles (number of units · cm⁻²).

Determination of temporal beta-diversity.

Species accumulation rates were calculated with the PRIMER-E software application (version 6.1; Plymouth Marine Labs, Plymouth, United Kingdom). Species accumulation curves were generated for each experimental treatment using average values across the three experimental runs for each sampling date. Values were obtained using observed species counts with samples permutated 999 times and plotted against increasing time periods. The relationships for all treatments were best explained by an exponential model (semilog rather than log-log) as determined by corrected Akaike information criterion values obtained in GraphPad Prism for Windows (version 5; GraphPad Software, La Jolla, CA). The slopes of the exponential models were used as measures of temporal beta-diversity in each treatment. The normality of the residuals for each treatment was confirmed with a D'Agostino and Pearson test. An F test was performed to test for overall differences in slopes, followed by pairwise examination of 95% confidence intervals of slopes.

Determination of taxonomic diversification rates.

Taxonomic identifications for all species in our experiments were checked with AlgaeBase.org (http://www.algaebase.org) to ensure that taxonomic classifications were current. The relationship between taxonomic richness at the species level and that at all higher categories from genus to order across the six current-nutrient treatments was examined by regression analyses using an exponential model. We limited our analyses to levels below phylum due to a lack of variability at higher levels. From the permuted results, slope estimates and 95% confidence intervals were computed for each experimental treatment. For each treatment, the normality of residuals was confirmed with a D'Agostino and Pearson test and an F test was used to formally test for differences in slopes.

Determination of taxonomic dissimilarity across treatments.

For each treatment, a composite taxonomy was compiled and six taxonomic levels were considered (species, genus, family, order, class, and phylum) and a taxonomic dissimilarity index (gamma+) derived from taxonomic distance was computed using presence/absence data (34, 35). Briefly, gamma+ is the mean of all path lengths between each species in one sample and its closest relation in the other sample and is a measure of the relative taxonomic distance between species in different treatments.

The changes in taxonomic distinctness within and between treatments through time were captured by nonmetric multidimensional scaling (NMDS). Permutation multivariate analysis of variance (PERMANOVA) was performed across all sampling dates to test for differences in taxonomic distinctness between treatments. The test statistic for PERMANOVA is the pseudo-F ratio, which is tested for significance by using a permutation test which randomly shuffles sample labels within and among treatment groups using 999 permutations and the pseudo-F ratios of the randomly assigned communities compared to the pseudo-F ratio of the observed communities (36). Biofilm thickness was analyzed with a repeated-measure ANOVA using SYSTAT version 12 (Systat Software Inc., Richmond, CA).

RESULTS

Temporal beta-diversity.

Slopes of species accumulation through time differed among treatments (F_5,36 = 63.78, P < 0.0001), i.e., temporal beta-diversity was greater in high-than in low-nutrient treatments and this difference was significant in low and variable flows (Fig. 1a and b). Across nutrient treatments, temporal beta-diversity was significantly higher in variable and low flow than in high flow.

Fig 1.

(a) Rates of species accumulation through time (eight time points) for the six current-nutrient treatments. (b) Slope values for species accumulation curves. Error bars represent 95% confidence interval for slope values.

Diversification rates.

Slopes of taxonomic diversification through time varied across taxonomic levels and among treatments (i.e., P < 0.0001 at each taxonomic level), where high nutrient supply promoted diversification at variable velocity but suppressed it at low velocity at a genus and family level (Fig. 2a to c). In the high-velocity treatments, rates were of low to intermediate values across nutrient regimes, with differences between nutrient treatments observed at the order level. These trends persisted even when differences in gradients of species richness were removed and accumulation of higher taxonomic levels was examined across similar species richness gradients.

Fig 2.

Slope values for the relationship between species richness and the average richness of all higher-order taxonomic categories across the six current-nutrient treatments. (a) Species richness and genus richness; (b) species richness and family richness; (c) species richness and order richness. Error bars represent 95% confidence intervals for slope values.

Taxonomic dissimilarity.

Taxonomic dissimilarity through time, as shown with NMDS (Fig. 3), was pronounced in all treatments other than the 10-high treatment. Furthermore, a clear distinction between the high- and low-nutrient treatments was evident and increased with time. PERMANOVA revealed that taxonomic dissimilarity varied through time (pseudo-F_7,47 = 2.09, P = 0.001, 998 unique permutations), with significant treatment effects for current (pseudo-F_2,47 = 3.31, P = 0.001, 997 unique permutations), nutrients (pseudo-F_1,47 = 15.89, P = 0.001, 998 unique permutations), and their interaction (pseudo-F_2,47 = 2.47, P = 0.001, 999 unique permutations). With the exception of the 10-low versus variable-low, pairwise comparisons revealed significant differences (P < 0.05) in taxonomic dissimilarity among all other treatments. Additionally, average abundances across several higher taxonomic levels (i.e., genus, family, order, and class) for the different current-nutrient treatments are found in the supplemental material (see Tables S1 to S3 in the supplemental material).

Fig 3.

Multivariate-analysis NMDS plot of average taxonomic dissimilarity (gamma+) through time within and between treatments. Numbers next to symbols represent days of sampling.

Biofilm thickness.

Biofilm thickness increased with time across all treatments (Fig. 4). Biofilm thickness in the high-nutrient treatments was significantly greater than that in the low-nutrient treatments, with a mean thickness across current treatments of 364 μm for the high-nutrient treatments versus a mean of 152 μm for the low-nutrient treatments (Fig. 4; see also Fig. S3 in the supplemental material). Repeated-measure ANOVA revealed significant between-subject effects for nutrients (F_1,10 = 24.48, P = 0.00058) but not for flow or their interaction (P > 0.05), while the within-subject effect of day was significant (F_7,70 = 4.08, P = 0.00083), with all other interactions being not significant (P > 0.05).

Fig 4.

The average response of biofilm thickness over the course of these experiments to experimental treatments varying in current velocity and nutrient concentration. Different letters correspond to statistical significance.

DISCUSSION

Individually, current velocity and nutrient abundance have been shown to influence rates of various processes (i.e., immigration and emigration, biomass accumulation, and species composition) in periphyton communities (26–28), but here we demonstrated that their combined effect also controlled the rates of species accumulation and diversification in developing biofilms. High nutrient supply increased the rate at which new species were added to the periphyton communities for two of the three current treatments that we tested. Species richness has generally been shown to increase with fertilization in freshwater communities (37), and results from our study suggest that this trend may in part be due to increased rates of species accumulation under conditions of high resource availability. However, this was not evident in our high-current treatments (30 cm · s⁻¹), with overall low rates of accumulation and no considerable differences between nutrient treatments. These results suggest a synergistic effect of current velocity and fertilization on the rates of species accumulation.

Current velocity, similar to this in our high-current treatments, has been shown to reduce richness (38). Here, we demonstrate that high current velocity also decreased the species accumulation rates. High flow velocity selects for growth morphologies, which allow species to reduce or avoid current stress (39), and at the conclusion of our experiments, communities in the high-current treatments were dominated by stress-resistant low-profile (30-low) or stress-avoiding motile (30-high) growth forms (19, 39). Bacterial communities can be negatively impacted by high current velocity (24); therefore, it is also possible that the high-current treatments reduced bacterial accumulation, thus delaying or inhibiting surface adherence by algae. Notably, our variable-flow treatments showed the highest rates of species accumulation compared to their constant-flow counterparts. Intermediate frequency in temporal disturbance led to the greatest richness in marine intertidal algal communities (40), and temporal variability in flow is an important source of heterogeneity in many natural streams with generally positive consequences for diversity (41–43). Our investigation revealed that the rates of species accumulation too increased with temporal heterogeneity in flow, which creates new niches throughout community development and promotes coexistence of species with diverse adaptations and requirements.

Further mechanistic understanding of the nature of the species accumulation curves was achieved through analyses of the diversification rates. In the 10-cm · s⁻¹ treatments, slopes were higher in the low- than in the high-nutrient treatments for genus and family, meaning that each new species added to the community had a greater probability of being a member of a different genus and family. At the level of order, the diversification rates in the 10-cm · s⁻¹ treatments did not differ significantly between the two nutrient treatments. Conversely, in the 30-cm · s⁻¹ treatments, the diversification rates varied between the two nutrient treatments only at the level of order. These results can too be explained with the synergistic effect of resource gradients and disturbance. Low nutrient levels support thin biofilms, as shown here, with short internal resource gradients (18), forcing stronger niche differentiation among the resident species, which is achieved through greater taxonomic disparity. In algae, considerable differences in the characteristics that impact niche differentiation, such as those pertaining to species morphology, physiology, and life history, typically correspond to higher taxonomic levels (21). Indeed, in 10-low, with an average biofilm thickness of 133 μm throughout the experiment, the dominant species, i.e., Microcoleus sp., Ankistrodesmus braunii, Lyngbya vanderberghenii, Chamaesiphon fuscus, Gloeocystis ampla, and Achnanthidium minutissimum, belonged to different genera, families, and orders. High resource supply, on the other hand, produces thick biofilms with multiple long internal gradients (i.e., light, nutrient availability, and disturbance), allowing greater opportunities for coexistence. It has been suggested that in such environments, large niche overlaps among species are possible as long as these species have complementary positions along at least one gradient (21). In agreement with this hypothesis, we measured an average biofilm thickness of 416 μm at 10-high and numerous species from only a few genera, i.e., Scenedesmus, Navicula, Nitzschia, and Gomphonema.

The role of nutrient supply was diminished under high disturbance, e.g., in the 30-cm · s⁻¹ treatments, where species without modes of attachment or ability to escape current stress were selected against. This reduced not only temporal beta-diversity but also the rates of DHT in both high- and low-nutrient treatments. Overall low rates of diversification in the 30-cm · s⁻¹ treatments suggest that the ability to withstand current stress was conserved within closely related species across the taxonomic levels examined. Disturbance has been shown to produce communities composed of more similar species than those in undisturbed communities (10). However, the greatest rate of DHT was observed in the variable-flow–high-nutrient treatments at the genus and family level, likely due to the diverse physical conditions and resource gradients present in those treatments.

In our experiments, we show that several important components of biodiversity, STR, or temporal beta-diversity, and taxonomic diversification, were impacted by environmental factors, i.e., nutrient and flow regime. White (48) discussed the importance of studying STRs, as they provide insight into the ecological processes underlying temporal turnover and richness in communities. Our results also highlight similarities in one of the mechanisms regulating both STR and SAR, i.e., greater environmental heterogeneity. Increased spatial heterogeneity in larger areas is one explanation for SAR, but results from our study indicate that greater temporal heterogeneity influences STR. In general, our experimental results suggest that under low and variable current, which represent low- and intermittent-flow disturbance regimes, respectively, new species are added to communities at higher rates under conditions of high resource supply than under conditions of low resource supply. Notably, resource supply was less important under constant-current stress, which led to overall low rates of species accumulation and DHT. The effects of fertilization on producer richness have been studied across aquatic and terrestrial systems (37). Our study reveals that fertilization also accelerates the rates of species turnover, leading to a greater temporal biodiversity in stream producers. However, greater temporal biodiversity did not necessarily translate into greater DHT, as the taxonomic relatedness between species added to the developing communities and those already present depended upon the interaction of current flow with nutrients. The greatest DHT was observed in the treatments with high nutrients and temporal variability in flow. Therefore, in the absence of temporal fluctuations in disturbance, fertilization produced rich but comparatively taxonomically homogeneous communities. These results emphasize the need for a more holistic approach in biodiversity studies, considering not only species richness but also the level of taxonomic dissimilarity among coexisting species, which are not always correlated, as clearly demonstrated here.

Temporal variability is an important source of heterogeneity with influence on biodiversity (14, 15). Human modifications to ecosystems are increasingly leading to temporal and/or environmental simplification, with consequences for biodiversity (44). A particular problem facing many streams and rivers is human modifications to the natural flow regime through damming and channelization (44, 45), and efforts are increasingly under way to help restore some of the natural variability in flows lost because of these practices (46). As algal communities form the base of the food web in streams and provide important ecosystem services (47), understanding the factors which control biodiversity in these communities should be an important endeavor for stream ecologists. The physical heterogeneity intrinsic to streams has been shown to promote diversity in biofilms (24), and results from our study suggest that temporal flow variability is important as well. It influences not only rates of species accumulation but also rates of taxonomic diversification, and therefore, efforts should be made to preserve the inherent temporal variability of natural ecosystems.