Biodegradability and Molecular Composition of Dissolved Organic ...

68 downloads 3939 Views 2MB Size Report
Mar 11, 2016 - Soil and Water Quality Laboratory, Gulf Coast Research and Education Center, University of Florida, Institute of Food and. Agricultural Sciences ...
Article pubs.acs.org/est

Biodegradability and Molecular Composition of Dissolved Organic Nitrogen in Urban Stormwater Runoff and Outflow Water from a Stormwater Retention Pond Mary G. Lusk and Gurpal S. Toor* Soil and Water Quality Laboratory, Gulf Coast Research and Education Center, University of Florida, Institute of Food and Agricultural Sciences, 14625 CR 672, Wimauma, Florida 33598, United States ABSTRACT: Dissolved organic nitrogen (DON) can be a significant part of the reactive N in aquatic ecosystems and can accelerate eutrophication and harmful algal blooms. A bioassay method was coupled with Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) to determine the biodegradability and molecular composition of DON in the urban stormwater runoff and outflow water from an urban stormwater retention pond. The biodegradability of DON increased from 10% in the stormwater runoff to 40% in the pond outflow water and DON was less aromatic and had lower overall molecular weight in the pond outflow water than in the stormwater runoff. More than 1227 N-bearing organic formulas were identified with FT-ICR-MS in the stormwater runoff and pond outflow water, which were only 13% different in runoff and outflow water. These molecular formulas represented a wide range of biomolecules such as lipids, proteins, amino sugars, lignins, and tannins in DON from runoff and pond outflow water. This work implies that the urban infrastructure (i.e., stormwater retention ponds) has the potential to influence biogeochemical processes in downstream water bodies because retention ponds are often a junction between the natural and the built environment.



INTRODUCTION Dissolved organic nitrogen (DON) is a dynamic participant in aquatic ecosystems and a potential source of reactive N to the phytoplankton and bacteria that cause water quality degradation.1−6 Various harmful algal species are known to use organic N for some or all of their N needs.7−9 A growing number of studies have addressed the ecological significance of DON in marine3,10−12 and freshwater systems.13−15 These studies collectively report that DON is a structurally complex mixture of materials that vary in chemical structure and composition and thus in bioavailability and ecological functioning. The bulk of DON present in water bodies may include thousands of molecules ranging from simple compounds (e.g., sugars, proteins, and amino acids) readily used by plants and microbes to more complex compounds (e.g., polyphenols and tannins) that are not generally readily metabolized.16 Recently, high-resolution mass spectrometry using high magnetic fields has been used to characterize DON in soils and in marine and stream samples.17−19 The very high resolving power of mass spectrometry with high magnetic fields allow us to make compound-level assessments of DON and identify individual DON compounds on the basis of their molecular formulas. This technique has shown promise for investigating potential relationship between DON chemical composition and ecological functioning (i.e., bioavailability). For example, Osborne et al.19 used ultra-high-resolution Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) to assess changes in the DON at the compound level in © 2016 American Chemical Society

streamwater samples before and after 5 days of bioassay experiment. They were able to determine the molecular formulas of DON compounds in initial streamwater samples and then identify the DON compounds consumed and formed by microbial transformations during the bioassay. Compared to studies in marine and stream systems, a few studies have investigated DON in urban stormwater runoff, even though DON is a significant and often bioavailable form of dissolved N in urban stormwater.20−22 Seitzinger et al.2 reported that 59 ± 11% of DON in stormwater runoff in urban watersheds was bioavailable. Wiegner et al.23 added that DON from an urban area was bioavailable on the order of days, so it may be expected to cause water quality degradation quickly after export from land to water bodies. We are aware of no studies that have investigated DON in water outflow from urban retention ponds. Stormwater retention ponds may number in the thousands in a single metropolitan area24 and represent a new type of aquatic ecosystem growing in importance as global urbanization continues to expand.25 In particular, urban stormwater ponds serve as the junction between the natural and built environment because they capture, store, and transform runoff from urban surfaces before conveying the water onward to a receiving water body or Received: Revised: Accepted: Published: 3391

November 19, 2015 March 9, 2016 March 11, 2016 March 11, 2016 DOI: 10.1021/acs.est.5b05714 Environ. Sci. Technol. 2016, 50, 3391−3398

Article

Environmental Science & Technology wetland system.26 In particular, these ponds can support conditions necessary to convert inorganic N or organic N from plant materials in stormwater into labile DON.26,27 Our objectives in this study were to characterize the molecular composition and assess the biodegradability of DON in urban stormwater runoff and water outflow from an urban stormwater retention pond. We hypothesized that (1) both stormwater runoff and pond outflow water would contain a labile DON pool and N-bearing biomolecules known to be readily bioavailable and (2) DON in the pond outflow water would be more labile than the stormwater runoff. We coupled bioassays with FT-ICR-MS to investigate these hypotheses. To our knowledge, we are the first to report on the molecular composition and biodegradability of DON from urban stormwater retention ponds.

using a discrete analyzer (AQ2+, Seal Analytical Inc., Mequon, WI) with EPA method 353.2. For total dissolved N (TDN) analysis, a subsample of the filtered sample was oxidized with the alkaline persulfate oxidation method using sample/reagent ratio of 2:1 and autoclaving at 110 °C for 30 min,30 followed by NOx-N analysis as above. The difference between TDN and total dissolved inorganic N (DIN; NOx-N and NH4‑N) was determined to be DON. Quality assurance and quality control (QA/QC) of analyses during instrument running included external standards prepared monthly from 1000 mg/L certified NO3 and NH3 standards (Sigma-Aldrich, St. Louis, MO), reagent blanks, analytical duplicates, and continued control verification (CCV). Bioassay Experiments. To determine the biodegradability of DON in stormwater runoff and pond outflow water, samples were inoculated using streamwater collected from the Alafia River at a site downstream of the stormwater retention pond. The bioassay studies were similar to the methods described by Osborne et al.19 and were initiated by mixing 400 mL of stormwater or pond outflow water and 125 mL of inoculum in 1 L glass Erlenmeyer flasks. Triplicate samples of the initial stormwater runoff and pond outflow water (designated as T0) were filtered through 0.45 μm filters and refrigerated (4 °C) until analysis. A second set of flasks (designated T5) were incubated for 5 days on a 12:12 light/dark cycle. After 5 days, the incubated samples were filtered and analyzed as described for the T0 samples. Molecular Characterization by FTICRMS. For FT-ICRMS analysis, the DON was first concentrated by using 75 mL of 0.45 μm-filtered and acidified (pH ∼2.0) samples (both T0 and T5) and extracted with 1 g (6 mL) of Varian Bond Elut PPL solid-phase extraction (SPE) cartridges in the manner described by Osborne et al.19 Cartridges were first rinsed with two cartridge volumes (12 mL) of HPLC-grade methanol; then, acidified samples were loaded and allowed to flow through the sorbent bed at a flow rate not exceeding 20 mL/min. The sorbent was rinsed with 12 mL of deionized water adjusted to pH ∼2 with HCl and then dried by applying a light vacuum for approximately 5 min. Finally, the sorbent bed was eluted with 6 mL of methanol, which was collected into 20 mL borosilicate scintillation vials and frozen until analysis. Prior to the analysis, samples obtained from SPE were diluted 9:1 with toluene (JT Baker, Phillipsburg, NJ) to increase ionization efficiency through dopant-assisted photoionization. Positive-mode atmospheric pressure photoionization (+APPI) was selected as the mode of ionization because it has been shown to be highly effective at ionizing nonpolar as well as polar molecules and to generate more N-bearing molecules than electrospray ionization.19,31 However, +APPI may select against some high-molecular-weight compounds;32 thus, our results should be used with the understanding that we may not be detecting the entire DON pool in the samples. Analysis was performed on a modified ThermoFisher APPI source (ThermoFisher Corp., San Jose, CA)33,34 at the National High Magnetic Field Laboratory (NHMFL) in Tallahassee, FL. A fused silica capillary was used to introduce samples at a rate of 50 μL/min. The samples were mixed with a nebulization gas (N2) inside the heated vaporizer of the source (∼250 °C) and passed under a krypton ultraviolet lamp (Syagen Technology Inc., Tuscan, CA) producing 10 eV photons (120 nm), thus allowing photoionization to occur at atmospheric pressure. The NHMFL used a custom-built 9.4 T FT-ICR-MS35 and a modular ion cyclotron resonance data



METHODS Study Site. The research site is a low-density residential neighborhood of single-family homes in Hillsborough County, FL, located along Florida’s subtropical Gulf Coast in the Tampa Bay watershed. Within the neighborhood, we identified a stormwater wet retention pond and used Arc GIS to delineate the pond’s “pipeshed,” or the area of homes, lawns, and impervious surfaces that drain stormwater runoff to the pond via piped conveyances. The total area of the pipeshed (catchment) is 0.11 km2, of which impervious area is 37%. The dominant vegetation in the catchment is live oak (Quercus virginiana) and St. Augustine turfgrass (Stenotaphrum secundatum). Soils at the site are predominantly Seffner fine sand series (sandy, siliceous, hyperthermic aquic humic dystrudepts).28 Data mined from the NOAA Plant City, FL climate station (station ID GHCND: USC00087205) showed that average annual rainfall in the 10 years preceding the study (2003− 2013) for the area was 133 cm, of which 66% occurred during the wet seasons (June to September).29 Sample Collection and Preparation. A single-stormevent-driven stormwater runoff and a single pond outflow water sample were collected on June 13, 2014 from the pipe draining the residential neighborhood and pond outfall, respectively. This storm produced 2.7 cm of rainfall over approximately 2 h. We focused on one storm event because of the time-consuming nature of FT-ICR-MS data acquisition and analysis. We acknowledge that study of a single storm event precludes development of conclusions about seasonal or temporal trends. Our study is therefore a case study of one point in time, but it provides the first case of which we aware to use bioassay and FT-ICR-MS to study the nature of DON in urban stormwater runoff and pond outflow water. The samples were collected into sterile 5 L polycarbonate carboys and stored on ice to limit biological activity during transport to the lab. Samples were filtered through 0.45 μm filter papers (Pall Life Sciences, Ann Arbor, MI) and stored in the dark at 4 °C until used in the molecular characterization or bioassay experiments, which was within 24 h of sample collection. On the same day as the pond outflow collection, we also collected a 5 L sample of streamwater from the mainstem of the Alafia River, at a site downstream from the pond. This sample was used as inoculum in the bioassay experiments and was filtered through a 150 μm mesh to remove zooplankton that might consume the autotrophic organisms that use N.19 Nitrogen Forms and Concentrations Analysis. For NH4-N and NOx-N analysis, filtered aliquots were analyzed 3392

DOI: 10.1021/acs.est.5b05714 Environ. Sci. Technol. 2016, 50, 3391−3398

Article

Environmental Science & Technology

Table 1. Mean Nitrogen Concentrations (± SD) in Initial (T0) Samples and 5 Day Bioassay (T5) Samples in Stormwater Runoff and Pond Outflow Watera stormwater runoff (mg/L) T0 T5 mean % reactive DON a

pond outflow water (mg/L)

TDN

DIN

DON

TDN

DIN

DON

1.48 ± 0.31 1.33 ± 0.57

0.39 ± 0.29 0.36 ± 0.11

1.09 ± 0.04 0.98 ± 0.26 10.3 ± 0.85

2.01 ± 0.33 1.25 ± 0.39

0.17 ± 0.11 0.13 ± 0.24

1.85 ± 0.15 1.12 ± 0.35 39.3 ± 0.65

SD: standard deviation; TDN: total dissolved N; DIN: dissolved inorganic N; DON: dissolved organic N.

acquisition system (PREDATOR).36 Multiple (100) individual time-domain transients were added, Hanning-apodized, zerofilled, and fast Fourier transformed prior to frequency conversion to mass-to-charge ratio to obtain the final mass spectrum.37 Transient length was 6 s. All observed ions were singly charged, as evident from unit mass to charge ratio (m/z) spacing between species that differ by 12Cc versus 13C112Cc−1. Data were analyzed and peak lists generated with custom-built software (MIDAS) and processed with PetroOrg software (Corilo, Y. E. PetroOrg Software, Florida State University, www.petroorg.com).36 Internal calibration of the spectrum was based on homologous series whose elemental compositions differ by integer multiples of 14.01565 Da (i.e., CH2).38 Compounds with the same heteroatom content (i.e., same n, o, and s in CcHhNnOoSs) but differing in degree of alkylation were grouped together to allow for molecular formula assignment to each homologous series.39 This analysis technique of separating structurally homologous compounds into categories on the basis of their number of CH2 groups was first developed by Kendrick38 and is now a common means of visually inspecting ultra-high-resolution mass spectrometry data.40

microbial exudates.44 Some stormwater retention ponds support intense microbial metabolism, such that there can be a shift from DIN to DON as inorganic N is consumed and converted to organic microbial metabolites.25,26 McEnroe et al.25 reported that the dissolved organic matter (DOM) in 45 urban stormwater ponds in Ontario, Canada was predominantly autochthonous, likely from an in-pond algal source. In this way, the urban stormwater ponds are differentiated from stream ecosystems, in which allochthonous DOM dominates.15,45 At our study site, although not quantified, an established tree canopy in the residential neighborhood contributes fresh supplies of PON to stormwater throughout the year in leaf and acorn litter. In addition, lawn grass clippings and in-pond vegetation also likely contribute year-round PON to the pond, and we hypothesize that some of the DON carried from the pond is derived from in-pond PON leaching. Research is needed to determine the contribution of PON to DON in urban stormwater retention ponds. Biodegradability of DON in Stormwater Runoff and Pond Outflow Water. After original samples were inoculated and bioassayed, both the stormwater runoff and the pond outflow water contained a pool of reactive DON, as shown by DON uptake during the bioassay experiment (Table 1). In the stormwater runoff, 10.3 ± 0.85% of DON was biodegradable, whereas 39.3 ± 0.65% of DON in the pond outflow water was biodegradable. The less biodegradability of DON in our stormwater runoff than pond outflow water implies that DON is dominated by lignin-like or otherwise refractory and semilabile compounds even though it originates from a humandominated landscape. Reports of DON bioavailability in urban stormwater and stormwater ponds are scarce in the literature. Seitzinger et al.2 reported that 42−57% of summer urban stormwater DON in New Jersey, United States was utilized by estuarine plankton over 12+ days of bioassay experiment. In our study, DON in the pond outflow was nearly four times more biodegradable than stormwater runoff, indicating that DON in water exported from the pond was more bioavailable to bacteria and phytoplankton than DON in the stormwater runoff. The greater reactivity of the pond DON supports our previous assertion that internally produced DON is an important player in the N signature of water exported from the pond: Autochthonous DON would be expected to be reactive because it tends to be mostly aliphatic and comprised of the highly labile carbohydrates, amino acids, and proteins.46−48 Molecular Composition of DON in Stormwater Runoff and Pond Outflow Water. The mass spectra of the samples included peaks ranging from 150 to 600 m/z (Figure 1A,B). The spectra contained 5939 and 5788 ions for which formulas could be assigned in the stormwater runoff and pond outflow water, respectively. The smaller insets in Figures 1A,B represent a 0.3 m/z unit spectral window isolated from stormwater runoff and pond outflow water and illustrate how complex the samples



RESULTS AND DISCUSSION Nitrogen Forms in Original Stormwater Runoff and Pond Outflow Water. The concentrations of TDN in the stormwater runoff and pond outflow water (before inoculation) were 1.60 and 2.14 mg/L, respectively. In these samples, DON was the dominant form, with lower concentrations in the stormwater runoff (1.11 mg/L) than the pond outflow (1.88 mg/L). It is not unusual for stormwater retention ponds to export water with greater DON than the stormwater runoff,27,41,42 as was the case for our samples. Recently, Li and Davis43 observed a net increase in dissolved N export from 0.70 to 1.31 mg/L in a retention cell receiving urban runoff and attributed the increase to DON derived from in situ particulate organic N (PON) leaching. The greater DON concentrations in the pond outflow water compared to those in the stormwater runoff could be an indication that DON is leaching from PON derived from neighborhood plant materials (e.g., leaves and grass clippings) that settle in the pond. Furthermore, in-pond processing of DIN (i.e., NOx-N and NH4-N) may add DON to the pond water. A meta-analysis of reports of Florida retention ponds in the International Stormwater BMP Database found that NH4-N and organic N were the N forms most likely to display net export whereas NOx-N showed net removal in the ponds.42 Increases in NH4-N and/or organic N concentrations as stormwater moved through the ponds overshadowed NOx-N removal, such that total N concentrations in pond outflow exceeded those in stormwater for 8 out of 17 Florida retention ponds identified in the database.42 These increases are likely due to the PON leaching, mineralization, and microbial uptake of DIN that then leads to increased release of DON in 3393

DOI: 10.1021/acs.est.5b05714 Environ. Sci. Technol. 2016, 50, 3391−3398

Article

Environmental Science & Technology

5000 N-bearing formulas in spectra taken from various types of samples, such as groundwater, wetlands, streams, or glaciers.19,49−51 Because our interest is in the DON component of the bulk DOM, we focused our analysis only on N-bearing formulas and found that the stormwater runoff and pond outflow water had 1229 and 1227 N-bearing formulas (CHON), respectively (Table 2). It should be pointed out that these are likely not all DON formulas in the samples because our ability to recover all DON formulas is determined by the extraction efficiency of the SPE and by the mode of FT-ICR-MS ionization. In particular, our use of +APPI may mean that some high-molecular-weight compounds were not recovered because +APPI has been shown to be most effective for compounds smaller than 700 Da.32 A comparison of the runoff and outflow water showed 87% overlap in identified molecular formulas, which could be due to the presence of refractory material that persists as the water evolves from runoff to pond outflow and/or that even though assigned formulas are the same there chemical structures are different.49 As FT-ICR-MS is not a quantitative technique and that the observed overlap means that only 87% of the identified formulas were present in some quantity in both the stormwater runoff and the pond outflow water. Therefore, the absolute quantity of any given compound cannot be compared between stormwater runoff and pond outflow water. We can, however, make semiquantitative conclusions from the data by using the combined relative abundances for spectral peaks in each heteroatom class (CHO, CHOS, CHON, etc.) to compare the relative abundance of N-bearing compounds to all other compound types (Figure 2). This resulted in combined

Figure 1. Mass spectra of (A) stormwater runoff and (B) pond outflow water. The small insets show the complexity of the spectra in a 0.40 Da window.

are and why there is a need for ultrahigh resolution data. It is important to note that these peaks represent the bulk DOM and not just N-bearing compounds, which we place in context with the bulk DOM later in this discussion. Even in the very narrow window, there are more than 40 peaks with identifiable formulas. The complexity of the mass spectra reflects the ultrahigh resolving power of FT-ICR-MS to characterize the thousands of components in a single DOM sample and illustrates the potential of FT-ICR-MS to compare the compositions and expected ecological significance of DON or bulk DOM in environmental samples. Other recent studies have also noted the power of FT-ICR-MS to identify upward of

Figure 2. Relative spectral abundances of CHON classes in stormwater runoff and pond outflow water.

CHON relative abundances of 10.34 and 9.87 in the stormwater runoff and pond outflow water, respectively. In

Table 2. Number of Peaks and Relative Spectral Abundance for Compound Classes in Stormwater Runoff and Pond Outflow Water stormwater runoff

pond outflow water

compound class

no. peaks

relative spectral abundance

no. peaks

relative spectral abundance

CHO CHOS CHON CHS HC total

3793 626 1229 (156 unique) 27 264 5939

64.72 2.75 10.34 0.10 2.66

3707 568 1227 (159 unique) 25 261 5788

65.38 2.27 9.87 0.095 2.81

3394

DOI: 10.1021/acs.est.5b05714 Environ. Sci. Technol. 2016, 50, 3391−3398

Article

Environmental Science & Technology

Figure 3. van Krevelen diagrams showing all N-bearing formulas in the mass spectra of (A) stormwater runoff and (B) pond outflow water. m/z = mass to charge ratio; DBE = double bond equivalency.

Figure 4. van Krevelen diagrams of (A) all N-bearing formulas found in both stormwater runoff and pond outflow water and (B) N-bearing formulas unique to stormwater runoff (orange) and unique to pond outflow water (green).

The 87% overlap between N-bearing molecular formulas in stormwater runoff and pond outflow water means that there was 13% difference in formulas between the runoff and pond outflow water samples, and this difference is part of the story of

both waters, CHON compounds were the second most abundant compound class after CHO compounds (Table 2), demonstrating that N-bearing compounds account for approximately 10% of all identified compounds. 3395

DOI: 10.1021/acs.est.5b05714 Environ. Sci. Technol. 2016, 50, 3391−3398

Article

Environmental Science & Technology

pond.59−61 In this way, pond microbial populations are simultaneously consumers and producers of DON.62,63 Although determining the exact mechanism of labile DON generation in pond outflow water is beyond the scope of this work, we can state that both the stormwater runoff and the pond outflow water contain DON displaying a wide range of expected lability, including many compounds that are highly labile or bioreactive. This finding is in line with Hosen et al.64 who reported enriched levels of protein-like DOM in waters that had been anthropogenically influenced by runoff from urban impervious surfaces. Likewise, McEnroe et al.25 observed that urban stormwater ponds in Ontario, Canada were characterized by high levels of internally produced labile DOM. This data also supports our earlier discussion on the bioassay results, which showed that both stormwater runoff and pond outflow water contain a biodegradable DON pool.

how the stormwater runoff potentially differs from the pond outflow. To understand these differences better, we used van Krevelen diagrams (Figure 3A,B), which plot molecular O/C ratios (x axis) against molecular H/C ratios (y axis) and are described in detail by Kim et al.52 The mean m/z for all N-bearing formulas was higher in the stormwater runoff (341.7) than in pond outflow water (299.7). The double bond equivalency (DBE), which is the number of rings plus double bonds expected for a given molecular formula and is an indication of aromaticity,53 was also higher for stormwater runoff (8.23) than for pond outflow (7.25). These values (m/z, DBE) indicate that the stormwater runoff contained slightly larger and more aromatic compounds than did the pond outflow water. Studies have shown that DON that is more aromatic in character is commonly less bioavailable.54 Our data suggest that DON in the stormwater runoff is less bioavailable because of its greater aromaticity than that the pond outflow water corresponds to the findings of our bioassay study, which showed a 4-fold increase in DON bioavailability in the pond outflow over that of the stormwater runoff (Table 1). Figure 4A shows points for all N-bearing formulas found in both stormwater runoff and pond outflow water, and Figure 4B shows formulas unique to runoff and pond outflow. We overlaid diagrams with rectangles to point out where important classes of biomolecules are known to fall on a van Krevelen diagram.52 These formulas fall under the full range of biomolecular classes including lipid-, protein-, amino sugar-, and lignin-like formulas as well as a wide swath of uncharacterized hydrocarbons with O/C ratios