Elsevier

Ecological Modelling

Volume 430, 15 August 2020, 109056
Ecological Modelling

Modeling cyanobacteria biomass by surface sediment diatoms in lakes: problems and suggestions

https://doi.org/10.1016/j.ecolmodel.2020.109056Get rights and content

Highlights

  • DI-TP was informative for modeling cyanobacteria biomass in lakes.

  • DI-TP performed better on modeling potential N2-fixing cyanobacteria biomass.

  • DI-TP performed better on modeling cyanobacteria biomass in deep man-made lakes.

Abstract

Cyanobacteria dominance threats ecological integrity and diminishes ecological service of lakes throughout the world. Variety kinds of chemical and physical environmental factors have been used to model cyanobacteria condition in lakes except biological environmental factors. Essential resources competition plays a central role in shaping algae community in lakes. Cyanobacteria and diatoms are assumed to have conserved species’ functional traits and use resources and habitats in similar ways. Therefore, diatom metric, inferred-total phosphorus based on diatoms (DI-TP), could be a valuable predictor for cyanobacteria biomass in lakes. With data sets from the 2007 National Lakes Assessment (NLA) of the US Environmental Protection Agency, we compared the performance of DI-TP with TP and other predictors on predicting cyanobacteria biomass (CBB) by boost regression tree analysis in around 1,000 lakes. In light of effects of lake types and diverse cyanobacteria functional groups on model performance, we did a priori classification on lakes based on lakes types (deep/shallow, natural/man-made lakes) and cyanobacteria functional groups (bloom-forming, potential toxigenic, heterocyst-producing and potential N2-fixing cyanobacteria). Our results showed: (1) DI-TP was informative for modeling CBB in different lake types or various cyanobacteria functional groups, but its importance was not as significant as traditional TP; (2) Performance of DI-TP on modeling CBB was better in deep man-made lakes than in shallow natural lakes; (3) DI-TP performed better on modeling the biomass of potential N2-fixing cyanobacteria than other functional groups of cyanobacteria. The lower importance of DI-TP could be caused by different sampling locations of diatom and cyanobacteria from a same site. The uneven distribution of number of shallow lakes along TP gradient could contribute to a better performance of DI-TP on predicting CBB in deep lakes than in shallow lakes. In man-made lakes, a shorter water residence helped diatoms coexist with cyanobacteria and contributed to a better performance of DI-TP on predicting CBB. We conclude that DI-TP performed better on modeling CBB in deep man-made lakes and potential N2-fixing cyanobacteria biomass. Further studies are needed to thoroughly assess the valuableness of DI-TP on predicting CBB in lakes.

Introduction

Cyanobacteria dominance threats ecological integrity of lakes and reduces ecosystem goods and services provided by lakes, such as cultural, aesthetic and provisions (Fristachi and Sinclair, 2008; Griffiths and Saker, 2003). Nutrient input reduction is the most direct, simple and economically feasible cyanobacteria management strategy. Nutrient predictors, either phosphorus/nitrogen concentration (Downing et al., 2001; Håkanson et al., 2007) or ratio of phosphorus to nitrogen (Schindler, 1977; Smith, 1983, 1986), were found to have a great influence on cyanobacteria dominance in lakes. Meanwhile, harmful cyanobacterial blooms in eutrophic waters were favored by rising temperatures, enhanced vertical stratification of aquatic ecosystems, and alterations in seasonal and interannual weather patterns (Paerl and Huisman, 2009). Beaulieu et al. (2013) found increases in temperature and nitrogen were as important as phosphorus on predicting cyanobacterial biomass in lakes. Therefore, cyanobacteria dominance typically is controlled by a suit of chemical, physical which interact synergistically and/or antagonistically (Paerl, 2014). However, rarely were biological factors used to model cyanobacteria condition in lakes. For management purpose, it is of great importance to accurately evaluate and refine the relative importance of environmental variables on modeling cyanobacteria biomass in lakes.

Species’ functional traits of diatoms and cyanobacteria are assumed to be conserved across phylogeneies and they use resources and habitats in similar ways, so they compete more strongly with one another than with more distant relatives (Tilman et al., 1982; Mittelbach, 2012). Therefore, diatom metric might be a good indicator for modeling cyanobacteria condition in lakes. R* is the minimum amount of resource R that a consumer species requires to maintain a stable population at a given mortality rate (Mittelbach, 2012). Fogg (1973) found blue-green species has a lower R* on nitrogen, while diatom has a lower R* on phosphorus (Tilman et al., 1982; Grover, 1989) (Figure 1a). Diatoms and cyanobacteria cannot coexist on one limiting resource (either phosphorus or nitrogen), and the one that can maintain positive growth at the lowest resource concentration (lowest R*) will win in competition. If phosphorus/nitrogen concentration fluctuates over time, spatial heterogeneity in the supply of resources in lakes can allow diatoms and cyanobacteria to coexist on two limiting resources. Strong resource competition between diatoms and cyanobacteria could be helpful for modeling cyanobacteria with diatom metric in lakes. Secondly, diatom can integrate effects of variety kinds of chemical and physical environmental factors, so diatom-inferred total phosphorus (DI-TP) can more accurately assess bioavailable phosphorus (orthophosphate) than direct measurement of TP (Smol and Stoermer, 2010a). DI-TP can better characterize environmental conditions than TP and can better reflect Cladophora accumulation conditions than averaged measured TP in streams of Michigan and Kentucky of the US (Smol and Stoermer, 2010a). Thirdly, DI-TP had a much lower root mean square error than the commonly observed TP concentration range in streams (Pan et al., 1996). The 2007 National Lakes Assessment (NLA) project of the USEPA, which collected empirical data for more than 1, 000 lakes, provides a unique opportunity to evaluate the performance of DI-TP on modeling cyanobacteria biomass in lakes. To the best of our knowledge, we have not seen any earlier works on evaluating DI-TP on predicting cyanobacteria biomass in around 1000 lakes.

A priori classification on general lake types (deep versus shallow, natural versus man-made) could be helpful for modeling cyanobacteria biomass in lakes. Firstly, nutrients concentrations in lakes are not only affected by human disturbance but also by hydrogeomorphic features and lakes origin related factors (Wetzel, 2001; Soranno et al., 2008; 2010). Internal nutrient loading from sediment into water column between deep lakes and shallow lakes is generally different (Wetzel, 2001). TP in lakes was found to have a close relationship with maximum depth of lakes and bedrock geology variables, such as percent of carbonate, hard rock, iron etc. (Soranno et al., 2008; 2010). On the other hand, Taranu et al. (2012) and Beaulieu et al. (2013) found cyanobacteria biomass is more predictable in deep lakes than in shallow lakes. One possible explanation is that cyanobacteria is sensitive to water-level changes which are larger in shallow lakes (Kalff, 2002). Differences in water residence time of lakes (e.g. natural lakes versus reservoir) can also affect phytoplankton community structure (Carvalho et al., 2011). Our results based on a long sediment core of Muskegon Lake, MI USA, which receives 95% of its input from one river and is with only a 23-d average retention time, showed diatom species composition (especially planktonic diatoms) in the core was significantly affected by the upstream river (Liu et al., 2018).

Compared with benthic cyanobacteria (e.g. Lyngbya), planktonic cyanobacteria (e.g. Microcystis and Aphanizomenon) are supposed to experience different nutrient concentrations in lakes. Moreover, some cyanobacteria (mainly heterocyst-producing cyanobacteria) have lower R* on nitrogen and higher compensation point of phosphorus than other cyanobacteria because they can fix nitrogen themselves (Figure 1b). Therefore, different algae groups are supposed to respond to nutrients in different ways and classifying cyanobacteria before the data analysis could allow response patterns to be seen more clearly. On the other hand, stakeholders are more concerned on those cyanobacteria which are most related with their benefits, such as toxigenic, bloom-forming blue-green algae etc. (Griffiths and Saker, 2003; Fristachi and Sinclair, 2008). In addition, allelopathic substances on the cell surface of microalgae could affect interaction among different algae (Liu et al., 2007) and may further influence the prediction ability of using one group to model another. All these factors call for a priori classification on cyanobacteria for an accurate evaluation on the performance of diatom metrics on modeling cyanobacteria condition in lakes. After cyanobacteria functional groups classification, Beaulieu et al. (2013) found total phosphorus (TP) and total nitrogen (TN) had a better performance on modeling biomass of cyanobacteria which can fix N2 than those which cannot.

In the present study, we evaluated the importance of DI-TP on predicting cyanobacteria biomass (CBB) for general lake types (shallow vs. deep lake and natural lakes vs. reservoirs) and different cyanobacteria functional groups with the 2007 NLA data sets of the USEPA. Based on an extensive literature survey, our three hypotheses are: (1) Compared to the performance of DI-TP in streams/rivers condition assessment, inferred TP based on surface sediment diatoms in lakes performs better on modeling CBB than TP; (2) DI-TP performs better on predicting CBB in deep lakes than in shallow lakes, and (3) DI-TP can explain larger amount of biomass variation of potential N2-fixing cyanobacteria than other functional groups of cyanobacteria . To do this, firstly we reconstructed DI-TP using the 2007 NLA surface sediment diatom data. Then we classified lakes according to their depth and origin. Thirdly, we calculated the biomass and grouped counted cyanobacteria (bloom-forming, potential toxigenic, heterocyst-producing and potential N2-fixing). Finally, a suit of natural predictors (water temperature, pH, DI-TP, total nitrogen, total phosphorus, ratio of total nitrogen to total phosphorus, conductivity and water depth) were selected to model all CBB and biomass of each site group to thoroughly evaluate the performance of DI-TP on modeling CBB.

Section snippets

Data sets

Water chemistry condition estimates, soft algae and diatom count data, and sampled lake information were downloaded from the NLA database (http://water.epa.gov/type/lakes/NLA_data.cfm). The 2007 NLA was conducted by the United States Environment Protection Agency (USEPA). The NLA provides a nationwide dataset and analysis in which the same standardized field and laboratory protocols and the same data analyses were used for individual biological assemblages (//water.epa.gov/type/lakes/lakessurvey_index.cfm

DI-TP reconstruction

The resulting diatom inference model for ln (TP) had a coefficient of determination (r2) of 0.50 between measured and model-inferred ln (TP) concentrations (Figure 2). The proportion of the variance in the TP that is predictable from surface diatoms was acceptable. After deshrunk, natural transformed DI-TP and TP had a good positive linear relationship with zero intercept. The weighted average TP model had a bootstrapped r2 of 0.49 between measured and model-inferred ln (TP) concentrations.

Modeling CBB in all 998 lakes

Although DI-TP was informative for modeling CBB in different lake types or various cyanobacteria functional groups, its importance was inferior to TP in the present study. The performance of DI-TP on modeling CBB in all 998 lakes was poor probably because the diatom samples were not exactly comparable to the cyanobacteria samples of the 2007 NLA project. First, diatoms and the corresponding cyanobacteria sample were from the same sampling site, but one was from settled sediment and the other

Acknowledgements

This paper is supported by the Green Channel Project of Natural Science Foundation of Hebei Province (D2020201003) and the Advanced Talents Incubation Program of the Hebei University (Grant No. 521000981302). We thank for the anonymous reviewers’ valuable comments on the manuscript which greatly improved the quality of this paper.

References (50)

  • S. Brothers et al.

    Empirical models for describing recent sedimentation rates in lakes distributed across broad spatial scales

    J Paleolimnol

    (2008)
  • C. Butterwick et al.

    Diversity in the influence of temperature on the growth rates of freshwater algae, and its ecological relevance

    Freshwater Biol

    (2005)
  • G. Cronberg et al.

    Manual on aquatic cyanobacteria: A photo guide and a synopsis of their toxicology

    (2006)
  • J.A. Downing et al.

    Predicting cyanobacteria dominance in lakes

    Can. J Fish. Aquat. Sci.

    (2001)
  • J. Elith et al.

    A working guide to boosted regression trees

    J Anim. Ecol.

    (2008)
  • D.J. Griffiths et al.

    The Palm Island mystery disease 20 years on: a review of research on the cyanotoxin cylindrospermopsin

    Environ. Toxicol.

    (2003)
  • J.P. Grover

    Phosphorus‐dependent growth kinetics of 11 species of freshwater algae

    Limnol. Oceanogr.

    (1989)
  • S.J. Guildford et al.

    Total nitrogen, total phosphorus, and nutrient limitation in lakes and oceans: Is there a common relationship?

    Limnol. Oceanogr.

    (2000)
  • H. Hillebrand et al.

    Biovolume calculation for pelagic and benthic microalgae

    J phycol.

    (1999)
  • J. Huisman et al.

    Changes in turbulent mixing shift competition for light between phytoplankton species

    Ecology

    (2004)
  • J. Kalff
    (2002)
  • Komárek, J. 2012. Süßwasserflora von Mitteleuropa, Cyanoprokaryota, 2. Teil/Part 3: Heterocytous Genera. Spektrum...
  • Komárek, J., and Aangnostidis, K. 2008. Süßwasserflora von Mitteleuropa, Cyanoprokaryota, 2. Teil/Part 2:...
  • Komárek, J., and Aangnostidis, K. 1998. Süßwasserflora von Mitteleuropa, Cyanoprokaryota, 1. Teil/Part 1:...
  • S. Kosten et al.

    Warmer climates boost cyanobacterial dominance in shallow lakes

    Glob. Change Biol.

    (2012)
  • Cited by (3)

    • Improving biological condition assessment accuracy by multimetric index approach with microalgae in streams and lakes

      2021, Science of the Total Environment
      Citation Excerpt :

      Currently, ecological assessment on freshwater ecosystems mainly relies on single assemblage (mostly benthic diatoms) and most studies are focused on streams and rivers rather than on lakes (Kelly et al., 2016; Poikane et al., 2016). Although soft-bodied algae share many ecological characteristics with diatoms (Liu et al., 2020b), only a few studies have used both of them to develop MMI in an assessment. Three disadvantages of soft-bodied algae metrics could be: dominant of diatoms in most samples, hard to identify and poor knowledge on autecology of soft-bodied algae (Liu et al., 2020b).

    • Ecological impacts of freshwater algal blooms on water quality, plankton biodiversity, structure, and ecosystem functioning

      2021, Science of the Total Environment
      Citation Excerpt :

      Besides that, resource acquisition and utilization are similar between Cyanobacteria and diatoms, which justify their coexistence pattern (Tilman et al., 1982). So, diatoms can act as a good indicator of cyanobacterial blooms, especially in man-made lakes (Liu et al., 2020), as in the present study. The same coexistence pattern was observed for other phytoplankton taxa (mixotrophic organisms) during Cyanobacteria and mixed blooms (Fig. 4).

    View full text