A machine-learning approach to predict success of a biocontrol for invasive Eurasian watermilfoil reduction

Myriophyllum spicatum, more commonly known as Eurasian watermilfoil (EWM), is one of the most invasive aquatic plants in North America, causing negative ecological and economic impacts in ecosystems where it proliferates. Many control strategies have been developed and implemented to mitigate EWM growth and spread, although the results are mixed and there is no consensus on lake-specific strategies. Here, we describe the development of a predictive model using a support vector technique, that predicts the success of biological pest control using Euhrychiopsis lecontei (the milfoil weevil), a milfoil specialist, to reduce EWM in lakes. Such a model is informed by lake characteristics (limnological and landscape) and augmentation strategies. To develop our predictive model, we performed a metadata analysis from 133 published peer-reviewed literature and professional reports of milfoil weevil augmentation field experiments that contained information on lake characteristics. The predictive model's algorithm uses a support vector machine (SMV) to learn patterns among lake characteristics, along with the recorded augmentation strategy and the reported success of each study, where success is a measure of EWM change over a season and is recorded in a variety of ways (e.g., EWM biomass change, EWM percent change, EWM visual change, etc.,). Overall, the model results suggests that shallower lakes, more frequent weevil augmentations, and larger weevil overwintering habitat are the most important predictors for EWM reduction success by weevil augmentation. Although watermilfoil weevil augmentation is a promising mitigation strategy, it may not work for all lakes. However, in terms of suggesting weevil augmentation, our model is a valuable tool for lake stakeholders and resource managers, who can use it to determine whether milfoil weevil augmentation, which can be very costly due to the difficulties in finding and raising milfoil weevils, will be a useful and sustainable approach to control EWM in their lake community.

Winter road management effects on roadside soil and vegetation along a mountain pass in the Adirondack Park, New York, USA

In 2003-2005, we resurveyed and expanded plots surveyed in 1985 to examine the cumulative impact of road salt (sodium chloride) and sand along a two-lane highway in the Adirondack State Park in New York State (USA). Annual salt applications in the period 1985-2005 ranged from 50 tonnes per centerline-km (1985) to 140 tonnes (2005) and sand applications ranged from nearly zero tonnes (2005) to 325 tonnes (1985). Roadside soils and vegetation were significantly impacted by salt deposition compared to soils and vegetation 30 m and 150 m from the road. Roadside soil contained more sand, less organic matter, had a lower cation exchange capacity, was denser, and retained less water than soils 30 m and 150 m from the road. The concentration of sodium in roadside soils was elevated (103 vs. 44 ppm in soil 150 m from the roadside), and roadside concentrations of plant-nutritive cations were lower than 150 m from the road (roadside Mg, Ca and K concentrations were 0.2, 5, and 1 ppm respectively vs. 23,168, and 30 ppm at 150 m from the road). Along the roadside, paper birch trees (Betula papyrifera) and other woody vegetation present in 1980 were absent in 2004, suggesting that survival and recruitment of paper birch trees was impacted by degradation of soil fertility, deposition of road salt and aerosolization of salt from the roadway. Roadside environmental degradation caused by winter road management has worsened since 1980; revegetation with native salt-tolerant plants may provide some mitigation of the most severe effects. Overall, we conclude that the full extent of roadside environmental degradation caused by winter road management can take decades to manifest, and this may be the case more generally along cold-climate montane roadways.

Ice cover extent drives phytoplankton and bacterial community structure in a large north-temperate lake: implications for a warming climate

Mid-winter limnological surveys of Lake Erie captured extremes in ice extent ranging from expansive ice cover in 2010 and 2011 to nearly ice-free waters in 2012. Consistent with a warming climate, ice cover on the Great Lakes is in decline, thus the ice-free condition encountered may foreshadow the lakes future winter state. Here, we show that pronounced changes in annual ice cover are accompanied by equally important shifts in phytoplankton and bacterial community structure. Expansive ice cover supported phytoplankton blooms of filamentous diatoms. By comparison, ice free conditions promoted the growth of smaller sized cells that attained lower total biomass. We propose that isothermal mixing and elevated turbidity in the absence of ice cover resulted in light limitation of the phytoplankton during winter. Additional insights into microbial community dynamics were gleaned from short 16S rRNA tag (Itag) Illumina sequencing. UniFrac analysis of Itag sequences showed clear separation of microbial communities related to presence or absence of ice cover. Whereas the ecological implications of the changing bacterial community are unclear at this time, it is likely that the observed shift from a phytoplankton community dominated by filamentous diatoms to smaller cells will have far reaching ecosystem effects including food web disruptions.

Seasonal changes in microbial community structure and activity imply winter production is linked to summer hypoxia in a large lake

Carbon and nutrient cycles in large temperate lakes such as Lake Erie are primarily driven by phototrophic and heterotrophic microorganisms, although our understanding of these is often constrained to late spring through summer due to logistical constraints. During periods of > 90% ice cover in February of 2008, 2009, and 2010, we collected samples from an icebreaker for an examination of bacterial production as well as microbial community structure. In comparison with summer months (August 2002 and 2010), we tested hypotheses concerning seasonal changes in microbial community diversity and production. Bacterial production estimates were c. 2 orders of magnitude higher (volume normalized) in summer relative to winter. Our observations further demonstrate that the microbial community, including single-celled phototrophs, varied in composition between August and February. Sediment traps deployed and collected over a 3 year period (2008-2011) confirmed that carbon export was ongoing and not limiting winter production. The results support the notion that active primary producers in winter months export carbon to the sediments that is not consumed until the warmer seasons. The establishment of this linkage is a critical observation in efforts to understand the extent and severity of annual summertime formations of a zone of regional hypoxia in Lake Erie.

Crustacean zooplankton in aerated wastewater treatment lagoons as a potential feedstock for biofuel

Zooplankton biomass productivity was estimated for two 64,000 m3 (1.7 ha) facultative aerated wastewater treatment lagoons to evaluate potential biodiesel production from zooplankton biomass. Lagoons were monitored bi-weekly during summer 2010. Lipid accumulated by crustacean zooplankton was considered the most efficient means by which to collect lipid produced by phytoplankton owing to the greater ease in the collection of these organisms (>0.153mm) compared with unicellular algae (size <0.06 mm). In terms of biomass, the cladoceran Daphnia and the copepod Cyclops were the dominant zooplankton in these hypereutrophic lagoons, while unicellular chlorophytes dominated the phytoplankton community. Secondary productivity of these lagoons (250 g of dry weight m(-2) yr(-1)) is comparable to the secondary productivity of other sewage lagoons. The potential biodiesel production for one lagoon was estimated to be 0.04 +/- 0.02 L m(-2) yr(-1), which results in a total of 1120 +/- 560 L from two lagoons. This study showed that there are organisms present in wastewater lagoons, besides algae, that can serve as a biofuel feedstock. Additionally, this research expands the current knowledge of facultative aerated wastewater lagoon ecology and waste stream-derived biofuel. Future research should include complete life cycle and economic analyses to determine if harvesting zooplankton from wastewater lagoons is a sustainable endeavour.

Patterns of mercury accumulation among seston in lakes of the Adirondack Mountains, New York

Mercury (Hg) concentrations in seston (biotic and abiotic particles < 200 microm) were measured during summer thermal stratification in nine lakes in the Adirondack Mountains of New York State that exhibit a range of physical and chemical characteristics. Fractionation of Hg bound to particulate matter was conducted to identify the influences of seston density and water chemistry on Hg accumulation among various ecologically important seston size classes (0.2-2, 2-20, 20-200, and > 153 microm). Total dissolved Al (< 0.45 microm) in lake water was significantly related to Hg concentrations in the 20-200 microm size class, and was positively correlated with Hg concentrations in the other seston size classes. Seston density was negatively correlated with seston Hg concentrations and significantly related to Hg bound by seston in the 0.2-2 microm size class. The results suggest that surface water Al concentrations and seston density influence Hg accumulation at the base of the aquatic food chain.

Application of the biotic ligand model to explain potassium interaction with thallium uptake and toxicity to plankton

Competitive interaction between TI(I) and K was successfully predicted by the biotic ligand model (BLM) for the microalga Chlorella sp. (Chlorophyta; University of Toronto Culture Collection strain 522) during 96-h toxicity tests. Because of a greater affinity of T1(I) (log K = 7.3-7.4) as compared to K (log K = 5.3-6.3) for biologically sensitive sites, an excess of 40- to 160-fold of K is required to suppress T1(I) toxic effects on Chlorella sp., regardless of [T1(I)] in solution. Similar excess of K is required to suppress T1(I) toxicity to Synechococcus leopoliensis (Cyanobacteria; University of Texas Culture Collection strain 625) and Brachionus calyciflorus (Rotifera; strain AB-RIF). The mechanism for the mitigating effect of K on T1(I) toxicity was investigated by measuring 204T1(I) cellular uptake flux and efflux in Chlorella sp. Potassium shows a competitive effect on T1(I) uptake fluxes that could be modeled using the BLM-derived stability constants and a Michaelis-Menten relationship. A strong T1 efflux dependent only on the cellular T1 concentration was measured. Although T1 efflux does not explain the effect of K on T1(I) toxicity and uptake, it is responsible for a high turnover of the cellular T1 pool (intracellular half-life = 12-13.5 min). No effect of Na+, Mg2+, or Ca2+ was observed on T1+ uptake, whereas the absence of trace metals (Cu, Co, Mo, Mn, Fe, and Zn) significantly increased T1 uptake and decreased the mitigating effect of K+. The importance of K+ in determining the aquatic toxicity of T1+ underscores the use of ambient K+ concentration in the establishment of T1 water-quality guidelines and the need to consider K in predicting biogeochemical fates of T1 in the aquatic environment.

Bioavailability of iron sensed by a phytoplanktonic Fe-bioreporter

This study describes a short-term (12 h) evaluation of iron (Fe) bioavailability to an Fe-dependent cyanobacterial bioreporter derived from Synechococcus PCC 7942. Several synthetic ligands with variable conditional stability constants for Fe(lll) (K* of 10(19.8) to 10(30.9)), in addition to several defined natural Fe-binding ligands and a fulvic acid of aquatic origin (Suwannee River), were used to elucidate the forms of Fe that are discerned by this phytoplanktonic microbe: Fe-HEBD (log conditional stability constant, K*, = 28.1, HEBD = N,N'-di(2-hydroxybenzyl)ethylenediamine-N,N'-diacetic acid monohydrochloride hydrate), Fe-HDFB (K* = 30.9, DFB = desferroxamine B), Fe-ferrichrome (K* = 23.2), Fe-DTPA (K* = 21.1, DTPA = diethylenetrinitrilopentaacetic acid), Fe-(8HQS)2 (K* = 20.4, 8HQS = 8-hydroxyquinoline-5-sulfonic acid), Fe-CDTA (K* = 19.8, CDTA = trans-1,2-cyclohexylenedinitrilotetraacetic acid), and Fe-EDTA (K* = 19.2). Iron bioavailability sensed by the bioreporter was related to diffusion limitation and activity of high-affinity transporters rather than by siderophore secretion. Iron complexed with a K* < 23.2 contributes to the bioavailable pool; bioavailability could be explained by disjunctive ligand exchange considerations and fully, partially, and nonbioavailable complexes could be distinguished according to their conditional stability constant. The use of Fe-bioreporters provides a relevant measurement of bioavailability to an important group of primary producers in freshwaters (cyanobacteria) and is thus a promising technique for understanding Fe cycling in aquatic systems.

Quantification of toxic Microcystis spp. during the 2003 and 2004 blooms in western Lake Erie using quantitative real-time PCR

In August of 2003 and August of 2004, blooms of potentially toxic cyanobacteria Microcystis spp. persisted in western Lake Erie. Samples collected from the bloom were analyzed for the cyanobacterial toxin microcystin and the presence of Microcystis spp. cells. Estimates of microcystin toxicity exceeding 1 microg L(-1) (microcystin-LR activity equivalents), the safety limit set by the World Health Organization, were found from the samples in both 2003 and 2004. The presence of Microcystis spp. in water samples was confirmed through standard polymerase chain reaction (PCR) using a combination of four primer sets. Quantification of Microcystis was accomplished by a real-time PCR assay utilizing specific primer-Taq-man probe sets targeted on a conserved, Microcystis-specific 16S rDNA fragment and a microcystin toxin synthetase gene mcyD. This approach allowed us to specifically study the distribution and abundance of toxic Microcystis in the lake in contrast to previous studies that have assessed Microcystis populations with less refined methods. On the basis of quantification by quantitative real-time PCR analysis, the total abundance of Microcystis cells in the bloom area varied from 4 x 10(8) to 2 x 10(3) cells L(-1). The results of this study provide novel insight regarding the distribution and abundance of Microcystis spp. in the western basin of Lake Erie, a region plagued in recent years by large-scale (>20 km2) blooms. Our results suggest that the Maumee River and Bay may serve as a source for Microcystis to western and central Lake Erie.

Evaluation of primary production in Lake Erie by multiple proxies

Direct measurements of rates of primary production in Lake Erie are few and uncertainties surround rate measurements based on radiocarbon and the light-dark bottle incubation methods. For these reasons, we conducted a series of simultaneous primary productivity measurements in Lake Erie in July and August of 2003, based on incubation with [14C]-NaHCO3, the light-dark bottle method, and incubation with (18)O enriched water. Significant differences in the rates of primary production obtained by incubations with [(18)O]-H2O (0.19-34.60 mmol-O2 m(-3) h(-1)), [14C]-NaHCO3 (0.03-90.50 mmol-C m(-3) h(-1)), and light-dark bottles (0.06-60.78 mmol-O2 m(-3) h(-1)) were evident in six out of nine comparisons. Within the epilimnion, [(18)O]-H2O rates of primary production were significantly different from rates based on [14C]-NaHCO3 and light-dark bottles in all four comparisons and lower rates were obtained in three out of four comparisons. Eutrophic conditions in Sandusky Bay, Lake Erie were evident from the high primary production rates of 20.50-34.60 mmol-O2 m(-3) h(-1) ([(18)O]-H2O), 34.39-90.50 mmol-C m(-3) h(-1) ([14C]-NaHCO3), and 46.66-60.78 mmol-O2 m(-3) h(-1) (light-dark bottle). The photosynthetic quotient (PQ), or ratio of O2 production to CO2 consumption during photosynthesis, averaged 0.64+/-0.33 and 1.93+/-1.93, respectively, based on a comparison of [(18)O]-H2O to [14C]-NaHCO3 rates or light-dark bottle to [14C]-NaHCO3 production rates, respectively, demonstrating that photosynthesis in Lake Erie communities primarily follows expected stochiometric trends. The average of the ratio of production rates based on incubation with [(18)O]-H2O relative to those obtained by the light-dark incubation method was 0.66+/-0.33, indicating a tendency for the [(18)O]-H2O method to provide slightly lower estimates of production in Lake Erie. Lower estimates of primary production based on [(18)O]-H2O incubation relative to the other two approaches is most likely a consequence of consumption of labeled O2 within the cell or dilution of label by the release of O2 from supersaturated cells. This latter effect may be particularly characteristic of eutrophic environments.

Bioconcentration of inorganic and organic thallium by freshwater phytoplankton

Uptake of inorganic Tl(I) and dimethylthallium, (CH3)2Tl+, by Chlorella spp. (Chlorophyta) and the diatom Stephanodiscus hantzschii (Heterokontophyta) were measured using radio-tracer techniques in water from Lakes Erie and Superior (North America). Uptake of both Tl(I) and dimethylthallium was bioactive. Uptake of [204Tl]-Tl(I) was greater in Lake Superior water than in Lake Erie water due to the greater K content in Lake Erie that inhibits Tl(I) uptake by phytoplankton but not that of [204Tl]-dimethylthallium. Volume-based bioconcentration factors for Tl(I) after 72 h of exposure were 5 x 10(4) and 1.1 x 10(4) for Chlorella sp. and S. hantzschii; for dimethylthallium they were 7.8 x 10(2) and 8.3 x 10(3). Both Tl(I) and Tl(III) were concentrated similarly by Chlorella spp. These results suggest that chlorophytes, but not diatoms, accumulate Tl(I) to a greater extent than dimethylthallium. Greater bioaccumulation factors of inorganic Tl are possible in waters containing low amounts of K+; water quality guidelines seeking to protect biota from deleterious effects of Tl should consider the role of K.

Oxidation of thallium by freshwater plankton communities

Thallium is a toxic metal that is of emerlI(I) or Tl(III), and its oxidation state affects its complexation and subsequent bioavailability and toxicity. We conducted lab and field incubations with 204Tl(I) and natural plankton assemblages to study the occurrence and mechanism of Tl oxidation. We observed that Tl(III) comprised 74% of total dissolved Tl after a 60 h incubation in surface waters from Lake Ontario, revealing a maximum specific oxidation rate of 0.014 h(-1). No Tl(I) oxidation was observed in sterile-filtered control treatments, indicating that solar radiation alone does not oxidize Tl(I) to Tl(III). Additional incubations with pond water revealed that Tl(I) oxidation is mediated by microbial activity and is not related to the presence of abiotic particles or phytoplankton or protozoa. We also identified a minor fraction (5-13%) of nonion-exchangeable (Chelex-100 resin; pH 1.5) Tl that may represent dimethylthallium or complexed thallium. This study demonstrates that planktonic bacteria are responsible for oxidizing the thermodynamically stable Tl(I) to the more abundant Tl(III).

Construction and initial characterization of a luminescent Synechococcus sp. PCC 7942 Fe-dependent bioreporter

A Synechococcus sp. PCC 7942 bioreporter strain capable of sensing bioavailable Fe was constructed by fusing the Fe-responsive isiAB promoter to the Vibrio harveyi luxAB genes. Monitoring luxAB-dependent luminescence through the growth curve demonstrated that in Fe-replete media, transcription from the isiAB promoter was induced transiently in the mid-exponential phase of growth. The initiation of transcription was the functional response to a 10-fold depletion of intracellular Fe to approximately 12 amol Fe per cell. Constitutive isiAB-dependent transcription was observed in Fe-depleted growth media. A dose-response relationship of the bioreporter was generated using trace metal-buffered Fraquil medium and was best represented by a sigmoidal curve having a linear component extending between pFe 21.1 (Fe3+=10(-21.1) M) and pFe 20.6 (Fe3+)=10(-20.6) M). Initial field trials conducted using water sampled from Lake Erie demonstrate that the bioreporter can serve as a quantitative tool to assess Fe deficiency in natural freshwater environments.

Comparison of copper speciation in coastal marine waters measured using analytical voltammetry and diffusion gradient in thin-film techniques

The diffusion gradient in thin-film hydrogel (DGT) probe is a promising tool for metal speciation work. Based on a passive sampling principle, it provides the potential for large data sets in complex regimes. DGT probes were deployed in waters characterized independently using competitive ligand exchange-adsorptive cathodic stripping voltammetry (CLE-ACSV). The CLE-ACSV used benzoyl acetone as the competitive ligand in discrete water samples collected during the deployment of the DGT probes. The DGT probes used a 15% polyacrylamide/0.4% bis-acrylamide cross-linker hydrogel and a Na-form of Chelex-100 to complex metal that fluxed into the probe through the hydrogel. Probes were deployed in locations characterized by the degree of pollution impact: the relatively pristine Vineyard Sound, MA, [Cu]total approximately 6 nM, small seasonally active harbors on Cape Cod, MA, [Cu]total = 12-64 nM, as well as a large polluted estuary, the Elizabeth River, VA, [Cu]total = 44-58 nM, and a large polluted port, San Diego Harbor, CA, [Cu]total = 23-103 nM. This is the first study where DGT probes have been compared with an independent speciation technique in marine systems and used to establish the diffusion coefficient of Cu-complexing ligands in situ. Results showed that the probes produced highly precise data sets, with substantial differences in copper accumulation between contaminated and pristine waters. Comparison of DGT results with CLE-CSV indicate that at least 10-35% of the organically complexed copper derived by CLE-ACSV measurements was DGT-labile. Diffusion coefficients (corrected to 25 degrees C) of organically complexed DGT-labile Cu through the hydrogel ranged from 0.77 x 10(-6) cm2 s(-1) in Vineyard Sound to 2.16 x 10(-6) cm2 s(-1) in the Elizabeth River estuary. Accumulation rates of copper were substantially higher in contaminated waters than in pristine waters, suggesting that the probes in their current form may be useful as tracking tools to detect episodic sources of contamination.