With the increasing rate of species extinctions following anthropogenic perturbation, there is a growing interest in biodiversity research. Although productivity and species richness relationships have been tested and applied in contemporary aquatic ecological studies, none have been applied to paleoecology with contrasting trophic states. The present study explores the applicability of a contemporary production and species richness relationship in high-resolution paleoecological records with low, intermediate and mid to high productive aquatic systems. Results from our study reveal that diatom species richness was positively correlated in low to intermediate productive lakes. In contrast, the relationship was hump shaped (unimodal) in a mid to high productive system concurrent with the species diversity analyses. Contrasting relationships between diatom species richness and stable isotope records (δ13C and δ15N) suggested that the nutrient biogeochemical cycle might play an important role in controlling species richness. From fossil pigment records we show that the variations in algal functional group signatures were highest in intermediate state. Collectively, these results suggest that the hump shaped (unimodal) relationship between diatom species richness and production might be limited to high productive systems with maximum richness and diversity in intermediate states, which is also supported in contemporary studies. Moreover, fossil pigment records as proxies for algal functional groups reveal that in a mid to high productive system with intense watershed scale disturbances, community composition of algal functional groups declined favoring certain diatoms. Our results demonstrate the applicability of production and diversity relationship theory in paleo-perspective and that recent watershed scale land use changes might affect the species diversity in aquatic systems.
Ecosystem-level disturbances and changes in species and/or community structure are emerging issues in modern ecological studies. Species composition can be influenced by both natural factors (Shurin, 2001; Bruno et al., 2003) and human activities (Proulx et al., 1996; Dodson and Lillie, 2001). Lakes are model systems to study such types of effects on aquatic community structure. This is partly because most of lakes are isolated with defined boundary and drainage areas and easily accessible (Hoffmann and Dodson, 2005). Thus, lake production and species richness and often used as a measure of aquatic environmental health. Two different approaches have been demonstrated to determine the relationship between productivity and species richness. One approach has examined how contrasting ecosystem processes can affect species number (Tillman and Pacala, 1993; Huston, 1994) and community composition. The other focused on the effect of community composition on ecosystem functioning (Tillman et al., 1997). The purpose of this study is to explore the relationship between algal production and diversity in a century-scale timeframe among contrasting aquatic systems.
Hypothesized production-diversity relationships can be positive (Abrams, 1995), negative (Rosenzweig, 1971) or unimodal (Tilman and Pacala, 1993) and are applicable to both species and functional group level (Weithoff et al., 2001; Weithoff, 2003). Species richness in lakes can be influenced by primary production; initially it increases but eventually decreases with productivity, producing a unimodal relationship (Tilman, 1982). For example, a highly productive system with stressed environment typically has a lower number of species with one or two species (those adapted to the stress) having many more individuals than the other species (Dodson et al., 2000). Additionally, other factors including the lake morphology can also affect the aquatic community composition (Dodson, 1992). A meta-analysis by Proulx and Mazumder (1998) demonstrated that species richness is influenced by grazing activity and level of productivity. Diversity and species richness may also depend on the disturbance type, frequency and its generation time (Gaedeke and Sommer, 1986). It has also been demonstrated that, (a) species diversity is reduced in the absence of very intensive disturbances, and (b) maximum diversity is achieved at intermediate disturbance (Kondoh, 2001). Recently, Rusak et al. (2004) demonstrated the richness-production relationship over a millennial scale timeframe. Although Rusak et al. (2004), considered to be the pioneer in the application of contemporary diversity-production theory on a millennial-scale, had the study period was limited to higher production regimes without comparing varying trophic states. Fossil algal remains, specifically diatoms and pigments, are confidently and consistently used in reconstructing community structure and paleoproduction in lentic systems (Leavitt and Hodgson, 2001; Das et al., 2005). Additionally, diatom and pigment records in lake sediments can be used to track changes in species richness and functional group structures, respectively, responding to contrasting watershed scale land use changes (Quinlan et al., 2002; Itoh et al., 2003; Saros et al., 2003). The current study is based on the contemporary species richness and production relationship, to see whether similar responses can be reconstructed from fossil remains in lake sediments responding to different anthropogenic events. In particular, we focused on the following three questions in a century-scale timeframe: (1) Is the relationship between diatom productivity and species richness linear (positive or negative) or quadratic (unimodal)? (2) Do these relationships vary with contrasting trophic states? (3) Can fossil pigments be used to reconstruct the temporal changes in algal functional groups?
Study lakes were selected based on their trophic status (Table 1) and disturbance histories. Sediment cores were obtained from the deepest depth of Sooke Reservoir (48°33′N latitude and 123°42′W longitude), Shawnigan Lake (48°37′N latitude and 123°27′W longitude) and Elk Lake (48°31′N latitude and 123°23′W longitude), located on the Vancouver Island, British Columbia, Canada. Sooke Reservoir is an ultra-oligotrophic lake (Table 1). The water residence time of the lake is 1.4 yr. The Devonian and Carboniferous bedrock of the Sooke watershed is principally composed of Metchosin volcanic materials including basalt flows, tuffs and agglomerates (CRD, 1999). The catchments are characterized by Western Hemlock stands interspersed with Coastal Douglas Fir and Western Red Cedar (CRD, 1999; Barraclough, 1995). The climate in this area is characterized by mild winters and cool summers owing to the moderating influence of the ocean. The area is in a rain shadow created by the Olympic Mountains (Tuller, 1979). The watershed receives about 1,226 mm of precipitation per year, with maxima occurring during winter months (MacKay, 1966; Nowlin et al., 2004). The lake water is characterized by low conductivity of ∼45 μ S·cm− 1 and low nutrient concentrations. The first dam on Sooke Lake (hereafter, Sooke Reservoir) was constructed in 1910 to supply drinking water to the city of Victoria. Since initial dam construction, the water level in Sooke Reservoir has been raised twice, once in 1970 by 5 meters and once in 2002 by 6 meters.
Shawnigan Lake is classified as oligotrophic to mesotrophic (Table 1). It is monomictic in nature, typically becoming thermally stratified by May. It is fed through the major inflow of Shawnigan Creek and has a moderate residence time (∼2.0 yrs). The bedrock of most of the watershed is of Devonian origin with part of the Wark and Coquitz Gneiss Complexes (CRD 1999; Barraclough 1995). Catchment vegetation types and climate are similar to Sooke Reservoir. The first human settlement began on the shoreline of Shawnigan Lake around 1910 and, since then, the population has been increasing steadily.
Elk Lake is a mesotrophic lake (Table 1) influenced by different types of anthropogenic activities and under constant pressure from the recreational activities undertaken on and in both the lake and watershed. It has high water residence time, varying between 4.4 (McKean, 1992) to 7.5 (Nordin, 1981) years. Its drainage basin lies within the coastal Douglas fir biogeoclimatic zone, which has mild wet winters and dry summers (McKean, 1992). The most important developments influencing the lake system were the construction of the Victoria waterworks between 1873 and 1879 and the Patricia Bay Highway in the 1950.
Sampling and sediment chronology
Sediment cores were taken from the deepest part of the lakes using a modified gravity corer (Kliza and Telmer, 2001) and were extruded on site. Sampling resolution was 0.5 cm. After extrusion, samples were kept in a cooler with ice. After transportation to the laboratory, samples were preserved in a freezer (−80°C) until further analysis. All cores were dated using 210Pb dating techniques by α-spectroscopy (Appleby and Oldfield, 1978). The constant rate of supply (CRS) model was applied to the 210Pb data.
Sediment stable isotope characterization
The isotopic compositions (δ13C and δ15N) of sediment samples were analyzed using a continuous flow system high temperature elemental analyzer coupled to a DELTAplus Advantage mass spectrometer. Reproducibility of duplicate analyses was ± 0.1‰. Measured values for δ13C are dependent of the historic isotopic signature of dissolved inorganic C, when organic C is produced photosynthetically. An acidification test confirmed that inorganic C was of minor importance in the bulk sediments. Data for δ13C were normalized to account for historic depletion of δ13C in atmospheric CO2, owing to fossil fuel burning as recorded by fossil air trapped in ice cores (Friedli et al., 1986); this effect is termed the Suess effect. We applied the following polynomial equation (Friedli et al., 1986) to remove the Suess effect, where t is time (in yr):
Sediment samples were digested and fossil diatom frustules were identified and counted following standardized paleolimnological protocols. A brief description of the digestion of sediment samples and diatom identification is as follows. To oxidize carbonate from homogenized sediment sub-samples (wet: 0.700–1.200 g), 10% HCl (∼ 10 ml) was added and left undisturbed for 24 hours before the remaining HCl was removed by aspiration. The vials were then placed in a support rack and ∼ 15ml of a 1:1 mixture of concentrated nitric and sulphuric acid was slowly added up to the half or three quarter mark on each vial. The samples were stirred and left for at least 24 hours before being placed in a water bath (∼ 90°C) for 3 hours. On each of 6 consecutive days, the samples were stirred and aspirated down to 1 cm depth above the sediment, topped up with distilled water, stirred, and allowed to settle for at least 24 hours, until all acid was removed. A portion of the resulting slurry was then pipetted, in a series of dilutions, onto cover slips and allowed to evaporate on a hot plate. The dried cover slips were then mounted using Hyrax® mounting medium (refractive index = 1.71). Between 300 and 500 diatoms were counted using an Olympus IMT-2 inverted microscope, at 1000× and 1500× magnification on each of the prepared slides. During the counts, the diatoms were also identified, primarily following Patrick and Reimer (1966) and Cumming et al. (1995). Consecutively, the relative abundance of each species was determined by dividing the number of valves of the species encountered by the total number of valves counted on the slide. Species richness was counted as the total number of individual taxa identified in a sample.
The Shannon-Weaver index is a distribution-free measure of diversity. A narrow range of diversity index values implies minor changes in distribution, which also supports the hypothesis that diversity is maximal when all species have equal relative abundance (i.e., maximum number of niches occupied). The Shannon-Weaver diversity index, which accounts for both relative abundance as well as the total number of species in a population, has been used to estimate diatom diversity in study lake sediments (Wetzel, 2001). The following equation was used for the Shannon-Weaver diversity estimate (H′).
Pigments were analyzed following the methods of Leavitt and Hodgson (2001) and Leavitt (1989) using high pressure liquid chromatography (HPLC). Pigment extractions were performed by soaking ∼ 100 mg of freeze-dried sediment in 5 ml of degassed mixture (bubbling He through the solvent mixture for 5–10 min) of acetone:methanol:water (80:15:5 by vol.) in a 10–20 ml glass tube. Samples were then flushed with N2 and stored in the dark at −20°C for 12 hours. Four ml of the extract were transferred into clean 20 ml glass tubes and dried under N2. Dried extracts were dissolved into 500 μ L of injection solution (Solvent A: 10% ion-pairing reagent in methanol by volume) and 500 μL of 3.2 ppm solution of Sudan II as an internal reference. Sediment samples were thoroughly soaked with extraction solution and washed with solvent after each decanting. The ion-pairing reagent was a mixture of 0.75 g tetrabutyl ammonium acetate and 7.7 g ammonium acetate in 100 ml of distilled de-ionized water. One ml of dissolved extract was transferred into a brown 2ml vial, flushed with N2 and cupped using Dionex cups. Vials were kept in a freezer until HPLC analysis. The HPLC system consisted of a DIONEX gradient pump, an AS50 Autosampler, a C-18 column (5 μ m particle size; 15 cm length), and a PDA detector at 435 nm. Flow rate was 1.5 ml · min− 1 and the separation gradient was (i) isocratic for 1.5 min in 10% IPR in methanol (Solvent A), (ii) a linear ramp to 100% of a mixture of 27% acetone in methanol (Solvent B) over 7 min and an isocratic hold for 15.5 min, and (iii) a linear return to 100% Solvent A over 3 min with isocratic hold for 12.5 min for column equilibration.
Pigment concentrations were quantified using equations derived from commercially available standards (DHI Water Environment). Diatoxanthin pigments representing diatom communities were used as a proxy of diatom production. All pigment concentrations were normalized to their mass per gram of organic matter (mg · g− 1 OM) where organic matter content in the sediment samples were measured as loss on ignition (LOI) at 550°C for 2 hours (Heiri et al., 2001).
Before regression analysis, all variables except species richness were log transformed to normalize distributions. Because the quantified pigment concentrations were too low and δ13C values were negative, a constant (log10[x + 1] for pigments and log10[x + 30] for δ13C) were added, respectively before transformation. A list of pigments detected in both study lakes representing different algal functional groups have been compiled in Table 2.
Changes in diatom species richness and production
Historical profiles of diatom species richness, production and δ13C and δ15N signatures in all three lakes illustrate noticeable changes in recent years (Figure 1). In Sooke Reservoir samples, an increase in species numbers was observed around 1860 (Figure 1, Figure 2). This increasing continued until 1900, coinciding with diatom production and δ13C and δ15N signatures. In Shawnigan Lake samples, the species numbers were unchanged in pre-1920 periods, whereas an increase was observed post-1920. Species richness also coincides with the diatom production and δ13C signatures in Shawnigan Lake samples. Interestingly, a decrease in diatom production and δ13C signatures, with decrease in species richness, was observed in surface samples. Abrupt changes in species richness (Figure 1, Figure 2), diatom production, and δ15N signatures were found during post-1920 periods in Elk Lake samples.
Regression analyses indicated that the species richness-productivity relationships differed among the three systems (Figure 3). A unimodal relationship was observed for recent samples in Elk Lake. Samples from Sooke Reservoir showed a non-significant linear relationship whereas in Shawnigan Lake a significant linear relationship was observed. When the species richness values were plotted against δ13C and δ15N (Figure 3), contrasting responses were evident in the three different systems. Species richness was positively correlated with both δ13C and δ15N in all three lakes. This relationship was significant in Shawnigan Lake and Elk Lake.
Changes in diatom species diversity
Although there were no differences in the Shannon-Weaver diversity index values during pre-disturbance regimes (i.e., pre-1900), they differ over time (Figure 4). Accordingly, the Shannon index values increased over time in each lake. The highest values were reported both in Shawnigan and Elk Lake sediments (Figure 4). Diatom diversity decreased after 1970 in the Elk Lake sediment profile.
Changes in algal functional groups
In this study, the production of different algal functional groups was reconstructed from the sedimentary fossil pigment records and were restricted to the following algal functional groups: alloxanthin (cryptophyceae), diatoxanthin (mainly diatoms), lutein-zeaxanthin (chlorophyceae and cyanophyceae), canthaxanthin (nostocales cyanophyceae), echinenone (total cyanophyceae), chlorophyll a and pheophytin a (all algae). Fossil pigment records indicate an increasing trend in historical production, from low (Sooke Reservoir), moderate (Shawnigan Lake) to high (Elk Lake) that also justifies the selection of sampling locations. The temporal profile of percent changes from natural background levels (bottom of each core and based on 210Pb dates and disturbance histories) of different algal functional groups differs markedly among lakes and groups. When comparing all three lakes, the most noticeable changes were detected for chlorophyceae, filamentous cyanophyceae and overall algal productions and these changes were clearly detectable in Shawnigan Lake sediment samples (Figure 5). For most of the algal functional groups, the greatest change occurred at intermediate disturbance (transition period between pre-and post-1900) levels, while some change at high disturbance levels was also detected. To compare the relative production contribution of algal functional groups to overall algal production, log10 transformed production of each of the functional groups were plotted against log10 overall algal production as inferred from chlorophyll a and all derivatives (Figure 6). Noticeable differences can also be seen in these plots. For both oligotrophic lakes (Sooke Reservoir and Shawnigan Lake), the relationships were positively linearly related for the most cases. However, for the mesotrophic system (Elk Lake), relationships were mostly scattered with the exception of diatoms and cyanophyceae, where significant correlations were present.
Lake sediments preserve fossil algal remains (Leavitt and Hodgson, 2001; Das et al., 2005), and thus may provide a unique opportunity for evaluating the effects of production on species richness. This study concentrated on demonstrating the relationship between paleo-production and a) diatom species richness, and b) inference of algal functional groups. Our results demonstrate that the contemporary ecological hypotheses of species richness-production relationships can be applied to plaeoecology with varying trophic states inferred from fossil diatom records and pigment remains in lake sediments.
Species richness responded to contrasting trophic states in a significant linear or hump-shaped (unimodal) fashion. In Sooke Reservoir, seasonal succession following changes in water residence time may have played a key role in controlling diatom diversity and algal functional groups. Several studies have shown that algal communities during pre-disturbance periods with short residence time are generally dominated by diatoms. Cyanobacterial communities often dominate in post-distrurbance periods with variations in flushing rates (Stockner and Benson, 1967; Peterson and Stevenson, 1989; Perry et al., 1990). Additionally, changes in phytoplankton diversity can be dependent on the availability of growth limiting resources. The basic idea is that increasing resources can support additional species in aquatic systems. However, the rapid colonization and resource competition by pioneer oligotrophic taxa, primarily cryptophyceae, and the highly variable immigration of other weak competitors following inundation, likely masked a peak in average richness along the flooding frequency gradient (Tilman, 1982; Interland and Kilham, 2001). The strategy that best allows species to exist under oligotrophic conditions may be one that allows for tolerance of low nutrients (Interland and Kilham, 2001). This is also evident in the relationship between species richness and δ13C and δ15N, where a weak but positive linear response was observed. This may be due to two different reasons. Firstly, carbon and nitrogen inputs from different sources (e.g., decomposed plant litter and catchment soil) can influence stable isotope signatures. Secondly, contributions from denitrifying cyanobacterial communities following watershed impoundments may result in higher values of δ15N (Figure 3). Growth limiting resource availability with contrasting flooding durations (i.e., time) might also have occurred in Sooke Reservoir (Tilman et al., 1982).
In Shawnigan Lake, a linear relationship between species richness and production was observed. This result supports the diversity-productivity hypotheses (Dodson et al., 2000; Mittelbach et al., 2000). Species richness reached a maximum in mid-productive states. In the mid to high productive system (Elk Lake), a significant linear relationship was observed initially from the core samples. Although the species diversity increased linearly with production, the relationship between production of algal functional groups and overall algal production was only found to be significant for diatoms and cyanophytes as reconstructed from fossil pigment records. Such relationships could be related to several factors associated with either changes in food web structure and productivity (Proulx and Mazumder, 1998) or competitive exclusion due to dominance by few algal functional groups (Tilman, 1982). In recent years, when disturbance levels are assumed to be higher due to increased human settlement and increased use for recreational purposes that may lead to elevated nutrient loadings (e.g., N loading), some functional algal groups cannot keep pace with the exploitive competitors, resulting in declining species diversity (Huston, 1994). The spatial structure may also play an important role in shaping community structure in the Shawnigan Lake ecosystem (Tilman et al., 1994; Comins and Hassell, 1996; Pacala and Levin, 1997). However, selective grazing pressure and exploitative competition over resources may eventually lead to loss of weak competitors resulting in loss of diversity. It is quite problematic to explain such processes from paleoecological data. If one considers longer time scales, nutrient depletion often induces shifts in community structure (e.g., from diatom dominated systems to picoplankton), favoring species (e.g., cyanophytes) that are more efficient at scavenging the limiting nutrient or do not require it (Tilman et al., 1994).
With intense disturbances, the high productive state in Elk Lake changed the linear relationship between species richness and productivity into a hump-shaped curve. Responses to changes in higher trophic levels and grazing activity may also have played a key role in controlling species richness in Elk Lake (Proulx and Mazumder, 1998). As predicted by food web theory, the responses to increased productivity levels at the base of the food chain are manifested most strongly in population densities, biomass and size structure of the top trophic level (Abrams, 1995; Proulx and Mazumder, 1998). Relative changes in available nutrient concentrations may create conditions conducive to further proliferation of cyanobacteria along with other algal communities (Smith, 1983; Kotak et al., 1995; Watson et al., 1997; Downing et al., 2001). When considering the low, mid and mid to high productive systems demonstrated in this study, it is evident that resource competition may play a key role in contributing to changes in the composition of phytoplankton communities (Leibold, 1997).
Additional analyses indicate that diversity differed in contrasting trophic states. The greater variability in diversity during post-1900 periods in the study lakes might be related to alterations in nutrient loading in the water bodies. In all three cases during post-1900 periods, the relative abundance of both nutrient rich species Tabellaria fenestrata and Asterionella formosa increased (data not shown) compared with other oligotrophic species (e.g., Cyclotella sp.). Over time this resulted in changes in diversity. Additionally, the narrow range of diversity values in Sooke Reservoir samples implies that the abundance of diatom species might have changed over time, but species evenness values were always in approximately the same proportions. In contrast, higher trophic states resulted in higher diversity ranges, which is evident in the Elk Lake sediment profile. Overall, the elevated trends are evident among Shawnigan and Elk Lake sediment diatom diversity. A decrease in the diversity in recent sediment samples of Elk Lake reflects the rapid growth rate of both T. fenestrata and A. formosa. Although a decrease in abundance of Cyclotella sp. was detected, the growth rate of T. fenestrata and A. formosa, along with other eutrophic species was more rapid, resulting in decreased diversity.
Overall, the results from this study confirm that the contemporary diversity-production hypothesis can be applied to paleoecological research (Rusak et al., 2004). However, questions remain as to whether these relationships could be related to relative disturbance levels (i.e. watershed development). Here we demonstrate that changes in algal diversity could be related to contrasting watershed land use histories. In ecosystem ecology, the idea that disturbance can help maintain species coexistence is largely based on the assumption that, without disturbance, competitively dominant species tend to monopolize resources such as space and food, and consequently drive less competitive species to local extinction (Tilman et al., 1994). Thus, disturbance might be regarded as a mediator in competitive assemblages, operating as a reset mechanism whereby populations of competitive dominants are periodically curtailed and resources are released for less dominant species to maintain their populations (Tilman et al., 1994; Pacala and Levin, 1997). This is supported in our study. Temporal profiles illustrate that species richness in all three studied lakes were likely to follow natural variability during low disturbance periods. At low levels, watershed disturbance might enhance richness by contributing limiting nutrients to lakes (Rosenzweig, 1971). On the other hand, at high levels, watershed disturbance reduces richness by eutrophication and contamination (Wilson et al., 2003).
The results of our study differ from previous reports where it has been demonstrated that temporal profiles of diversity-production relationship would differ over millennia scale compared with smaller scale studies (Rusak et al., 2004). This could be due to two different reasons: Firstly, the reported study was limited to lakes with high primary productivity regimes. Secondly, the study lakes were saline in nature. Therefore, more study lakes combining different trophic states and limnological regimes would help to determine whether the diversity-production relationship holds over spatial and temporal scales.
One of the greatest threats to biodiversity is the outright loss of habitat due to human activity. Habitat destruction is also a major threat to biodiversity in aquatic regions. However, changes in biodiversity may differ in contrasting trophic states. We compared our results at three different limnological regimes: low (ultra-oligotrophic), mid (oligotrophic to mesotrophic), and mid to high (mesotrophic) productive systems. Conclusion of overall historical species richness-production relationships from our study is that a hump shaped or unimodal relationship may be limited to mid to high productive systems at high disturbance levels. Additionally, with increased disturbance and primary production, diversity of algal functional groups declines favoring certain diatoms and filamentous cyanobacterial growth suggesting that the functional diversity governed the overall primary production.
The authors acknowledge Simon Thomson, Shane Edmison, and Kelly Field, Water and Watershed Research program, Department of Biology, University of Victoria for helping with sediment core collection; Sergei Verenitch, Jutta Kolhi and Nikolas Desilet, Water and Watershed Research Laboratory for helping with the pigment analysis; Shapna Mazumder for stable isotope analysis; for diatom records Cori Laurine Barraclough (Sooke Reservoir); Laura Greenway (Shawnigan Lake) and Roel Groeneveld (Elk Lake); Christopher Lowe and Niki Eyding, Department of Biology, University of Victoria, Rick Espie, Saskatchewan Environment for editorial correction and two anonymous reviewers for helpful comments on the manuscript. This research has been supported by the NSERC Industrial Research Chair Grant and by partnership support from the CRD Water Services to A. M. and the NSERC CGS D scholarship to B. D.