Harmful Algae 10 (2011) 310–318

Contents lists available at ScienceDirect

Harmful Algae

journal homepage: www.elsevier.com/locate/hal

Stormwater nutrient inputs favor growth of non-native macroalgae (Rhodophyta) on O’ahu, Hawaiian Islands

Brian E. Lapointe *, Bradley J. Bedford

Center for Marine Ecosystem Health, Harbor Branch Oceanographic Institute at Florida Atlantic University, 5600 US 1 North, Ft. Pierce, FL 34946, United States

ARTICLE INFO ABSTRACT

Article history: In Hawaii, blooms of native and non-native macroalgae (limu) have become increasingly problematic in Received 27 August 2010 recent decades. Although the role of human vectors in introducing non-native macroalgae is well Received in revised form 23 November 2010 documented, the ecological role of nutrient pollution in facilitating blooms of these species is not. This Accepted 24 November 2010 study assessed the effects of stormwater discharges on the diversity, abundance, and nutrient content (C, Available online 1 December 2010 N, P and d15N) of native and non-native limu at three sites in the intertidal zone at Ewa Beach, O’ahu. The results showed that native limu species diversity and abundance decreased with proximity to a Keywords: stormwater outfall (ASWO), whereas non-native species abundance increased. Limu tissue d15N values Hawaii at all three sites were within the range reported for sewage N. 15N, %N, and N:P ratios all increased with Limu d Native proximity to the ASWO, supporting the hypothesis that stormwater was a primary source of N Nitrogen enrichment in the study area. In contrast to N, limu %P showed little change among the sites, suggesting Non-native that the generally high N:P ratios, indicative of P-limitation, resulted from high N:P ratios from the Macroalgae upland watershed. Abundance and tissue %N of the non-native rhodophyte Acanthophora spicifera Phosphorus increased with proximity to the ASWO and were strongly correlated (r2 = 0.94) compared to native Stormwater rhodophytes, indicating that stormwater N enrichment provided this invader a competitive advantage (lower C:N ratio) over native limu. These results indicate that the spread of non-native macroalgae in oligotrophic coral reef regions can be facilitated by anthropogenic nutrients in stormwater runoff, thereby threatening native species and ecosystem services. ß 2010 Elsevier B.V. All rights reserved.

1. Introduction aquaculture, has become extremely abundant, especially on Maui where large accumulations accumulate on beaches, interfering Non-native macroalgal invasions are a major driver of coastal with tourists’ use of beaches as a result of malodorous ecosystem change worldwide (UNEP, 2006; Williams and Smith, decomposition (Huisman et al., 2007). 2007). The geographically isolated Hawaiian Islands are especially Throughout the Hawaiian archipelago, there is growing concern vulnerable to biological invasions, which have resulted in about the displacement of native seaweeds, known as limu in the significant impacts on biodiversity (Staples and Cowie, 2001). Hawaiian language, by non-native species. Non-native invasive Biological invasions in the marine environment include introduc- limu compete with and displace native limu species important to tions of non-native seaweeds that date back to the 1950s when the Hawaiians for food, medicine, and religious purposes (Abbott, non-native rhodophyte Acanthophora spicifera was accidentally 1984). Russell (1992) documented how non-native A. spicifera and introduced to Pearl Harbor, O’ahu (Doty, 1961). Today, A. spicifera is H. musciformis displaced native populations of Laurencia nidifica considered the most pervasive non-native algal species in Hawaii and Hypnea cervicornis. In his seminal work on biological invasions, (Huisman et al., 2007). Since the 1950s, over 20 species of non- Elton (1958) emphasized the importance of human-mediated native macroalgae have been introduced to the Hawaiian Islands, vectors, especially physical transport. Humans have subsequently but only about five species have become established and form been recognized as the primary vector in the global epidemic of extensive blooms that alter coastal ecosystems (Russell, 1992; biotic invasions in aquatic ecosystems (Carlton and Geller, 1993). Rodgers and Cox, 1999; Smith et al., 2002). Recently, the non- Indeed, increasing evidence shows that human activities facilitate native rhodophyte Hypnea musciformis, imported from Florida for the physical spread of non-native species in Hawaiian coastal waters, despite recent progress in the prevention of non-native limu introductions (Staples and Cowie, 2001). * Corresponding author. Tel.: +1 772 242 2276; fax: +1 772 468 0757. Increased urbanization of upland watersheds is a major E-mail address: [email protected] (B.E. Lapointe). mechanism increasing nutrient pollution of coastal waters and

1568-9883/$ – see front matter ß 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.hal.2010.11.004 B.E. Lapointe, B.J. Bedford / Harmful Algae 10 (2011) 310–318 311 may facilitate invasions of non-native limu in Hawaiian coastal nitrogen sources on coral reefs, particularly that from human waters. Anthropogenic nutrient pollution of coastal waters has sewage, which can also be a significant source of nitrogen been widely recognized as a common factor linking an array of enrichment in urban stormwater runoff (Wanielista and Yousef, problems, including harmful algal blooms, dead zones, seagrass 1993; Dillon and Chanton, 2008). and coral reef die-offs, declining fisheries, and marine mammal and seabird deaths (ECOHAB, 1997; NRC, 2000; Howarth et al., 2. Materials and methods 2000; MEA, 2005; HARRNESS, 2005; UNEP, 2006). In Hawaiian coastal waters, Soegiarto (1972) and Johannes (1975) first 2.1. Selection of sampling sites suggested a linkage between nutrient pollution from sewage and the expansion of the non-native rhodophyte A. spicifera in To test the hypothesis that stormwater nutrient pollution Kaneohe Bay, O’ahu. Blooms of the native invasive chlorophyte affects the relative abundance of non-native versus native limu in Dictyosphaeria cavernosa, which overgrew and killed corals in intertidal communities, three study sites were chosen along a Kaneohe Bay, O’ahu, were also linked to nutrient enrichment from gradient of exposure to stormwater discharge in the Ewa Beach sewage (Banner, 1974; Smith et al., 1981). Following sewage area on O’ahu (Fig. 1). The urbanized Ewa Beach area had a diversion from Kaneohe Bay in the late 1970s, nutrient concentra- stormwater drainage system constructed when this neighborhood tions and D. cavernosa biomass both decreased (Hunter and Evans, was developed in the 1970s. Currently, several stormwater outfalls 1995), demonstrating the ecological importance of nutrient discharge into the intertidal zone where longshore currents in the enrichment to Kaneohe Bay. Since then, however, blooms of nearshore area generally flow from the urbanized Ewa Beach area non-native limu have expanded throughout the Hawaiian Islands westwardly towards One’Ula State Beach Park. The easternmost (Russell, 1992; Rodgers and Cox, 1999; Smith et al., 2002), sampling site (Amio) in our study, located near the Amio Street especially in coastal waters adjacent to urbanized watersheds. stormwater outfall (ASWO), was chosen to be representative of Although sewage has been identified as a significant nitrogen direct stormwater impacts (Fig. 2a). A second site near Papipi Road source supporting blooms of native and non-native limu (Dailer (Papipi), 625 m west of Amio, was chosen as a site less impacted et al., 2010), studies have not addressed the importance of nutrient by the ASWO discharges to the east (Fig. 2b). The third site at Kaloi enrichment (nitrogen, N and phosphorus, P) from urban storm- Gulch (Kaloi) in One’Ula State Beach Park, west of Papipi and water runoff to the relative abundance of native and non-native 975 m from Amio, was selected as a reference site least impacted limu, such as A. spicifera, in Hawaiian coastal waters. Urban by stormwater discharges to the east (Fig. 2c). The three intertidal stormwater runoff can contain relatively high concentrations of sites were sampled during low (minus) tides, March 3–7, 2008. ammonium, nitrate, total N, soluble reactive P, and total P, which combined with the high volumes of stormwater following rain 2.2. Sampling for taxonomic composition of limu communities events, can account for considerable nutrient loads to coastal waters (Wanielista and Yousef, 1993). Four separate 30 m survey transects were established end-to- We posed the hypothesis that nutrients from stormwater runoff end in the low intertidal zone at each site using Keson fiberglass would affect the nutrition and relative abundance of native and survey tapes. The four transects were independent, had no spatial non-native limu. To test this hypothesis, we studied intertidal limu overlap, and resulted in a total surveyed length of 120 m at each communities at three locations in Ewa Beach, O’ahu. The study was site. Qualitative collections of conspicuous limu species were multi-faceted, and involved measuring the following variables in sampled along the four transects at each site; specimens were intertidal communities at the three sites: macroalgal species identified according to Abbott (1999), Abbott and Huisman (2004), presence, percent cover of abundant taxa, tissue C:N:P contents to and Huisman et al. (2007). gauge the degree of N versus P limitation (Atkinson and Smith, Limu communities along the four transects at each site were 1983; Lapointe et al., 1992), and stable nitrogen isotope ratios quantitatively sampled by photogrammetric techniques, using a (d15N) to discriminate between anthropogenic and natural Nikon Coolpix 5000 camera to obtain high resolution digital color nitrogen sources such as sewage, fertilizers, upwelling, or nitrogen images along the transects. This non-destructive method yields fixation (Lapointe, 1997; Lapointe et al., 2005; Derse et al., 2007; parallax-free sampling that generates highly reproducible quanti- Parsons et al., 2008; Dailer et al., 2010). Risk et al. (2009) concluded tative data, and is one of the most widespread and sophisticated that the measurement of stable nitrogen isotopes in macroalgae techniques for permanently recording marine algal standing stocks 2 [()TD$FIG]provides a cost-effective and objective means of quantifying (Littler and Littler, 1985). Ten digital images (0.1 m quadrat size)

Fig. 1. Satellite image showing the Ewa Beach study sites. 312[()TD$FIG] B.E. Lapointe, B.J. Bedford / Harmful Algae 10 (2011) 310–318

Fig. 2. Photographs showing: (a) Amio stormwater outfall pipe (ASWO); (b) intertidal zone at Papipi; (c) transect sampling tape at Kaloi Gulch; (d) red limu at Amio showing dark pigmentation indicative of nitrogen enrichment; (e) Sargassum echinocarpum at Papipi; and (f) mixed native limu community at Kaloi Gulch showing reduced pigmentation indicative of nitrogen limitation. were recorded, at 3 m intervals along each of the four 30 m transport to the lab. At the ASWO pipe, replicate (n = 2 per species) transects, yielding 40 photo-quadrats per site. At some locations samples of two limu species, S. echinocarpum and L. majuscula, along the transects, lack of suitable substrate prevented sampling. were also collected, to establish a stormwater N isotope source At these locations, the closest suitable substrate was sampled, signal. The freshly collected limu were processed in Dr. Robert typically within 1 m of the original location. Images were Richmond’s laboratory at the Kewalo Marine Lab, Honolulu, where analyzed using the Randomized Point Count method (Littler and the specimens were cleaned of epibionts, rinsed in deionized water Littler, 1985). Ten random points were overlayed on each quadrat and dried in a lab oven at 65 8C for 48 h. Dried limu samples were image (displayed on a high resolution LCD monitor) and the limu or ground to powder, using a mortar and pestle, and stored in plastic other biota/substrate beneath each point identified. In the case of screw-cap vials for later analysis (Lapointe et al., 2005). some limu that were difficult to identify to species (such as Limu d15N ratios were measured to aid in identification of the N Laurencia spp.) identification was made only to genus. Small source(s) in stormwater sustaining the limu communities (Heaton, microfilamentous turfs <2 cm high were scored as either ‘‘algal 1986; Dawson et al., 2002). Previous studies of d15N ratios in turf’’ (greens) or ‘‘red turf’’ depending on pigmentation. macroalgae experiencing variable inputs of stormwater runoff in Sarasota Bay, FL, reported values ranging from +1.3% to +13.3% 2.3. Sampling limu for nutrient contents and stable nitrogen (Dillon and Chanton, 2008). Analyses of replicate limu tissue isotope ratios samples (n = 4 per sample) were performed at the University of Maryland Center for Environmental Science, Horn Point Laborato- To quantitatively compare nutrient availability among the ry, Cambridge, MD. Samples were packed in tin capsules and three sites, four limu species common to all three sites were analyzed using a Sercon isotope ratio mass spectrometer. The collected from each site for tissue C:N:P contents and d15N standard used for stable nitrogen isotope analysis was d15N in air. 15 analysis. These included two phaeophytes, Padina sanctae-crucis d N values (%) were calculated as [(Rsample/Rstandard) À 1] Â 100; and Sargassum echinocarpum, and two rhodophytes, Laurencia where R = 15N/14N. majuscula and A. spicifera. Replicate (n = 2 per species), composite Limu samples were also analyzed for percent C, N, and P, at samples (10–20 thalli per replicate) were collected at each site, Nutrient Analytical Services, Chesapeake Biological Laboratory, placed in Whirl-Pak baggies, and held over ice in a cooler for University of Maryland System, Solomons, MD. C and N were B.E. Lapointe, B.J. Bedford / Harmful Algae 10 (2011) 310–318 313 measured on an Exeter Analytical, Inc., CE-440 Elemental Analyzer; Table 1 P was measured following the methodology of Asplia et al. (1976) Taxonomic inventory at each site, including site totals. on a Technicon Autoanalyzer II (Keefe et al., 2004). Kaloi Papipi Amio

Phaeophyta 2.4. Statistical analyses Asteronema breviarticulatum X Colpomenia sinuosu XX X Macroalgal benthic cover (%), C:N:P ratios, and d15N(%) value Dictyota sandvicensis XX X Dictyota acutiloba XX comparisons among the three study sites (df = 2) and between Dictyopteris plagiogramma X Phyla (df = 1) were performed in SPSS 11 for Mac (www.spss.com) Hincksia indica X using the Generalized Linear Model procedure (GLM; Type III sum- Hydroclathrus clathratus X of-squares). When values among sites were found by GLM to differ Padina sanctae-crucis XX X significantly (p  0.05), Tukey’s HSD (THSD) post hoc testing was Sargassum echinocarpum XX X Sargassum obustifolium X performed to identify the source(s) of difference(s). Correlative Sphaceleria novae-hollandiae XX comparison of A. spicifera tissue N (%) with A. spicifera benthic cover Stypopodium flabelliforme X (%) was performed using least-squares linear regression in Chlorophyta Microsoft Excel. Avrainvillea amadelpha X Bornetella sphaerica X Caulerpa microphysa XX 3. Results Caulerpa antonensis X Caulerpa taxifolia X 3.1. Taxonomic composition of intertidal limu communities Chaetomorpha antennina XX Cladophora catenata X Codium edule XX X A total of 52 limu species were collected and identified from the Codium arabicum XX three study sites. The Kaloi site, farthest from stormwater Codium reediae X discharges, had 39 spp., the highest species richness among sites; Dictyosphaeria versluysii XX X Papipi, intermediate in distance from the stormwater discharges, Halimeda opuntia XX X had 30 species; Amio, closest to the stormwater inputs, had 25 Halimeda discoidea XX Microdictyon setchellianum X species (Table 1 and Fig. 3). Neomeris vanbosseae XX Benthic cover at Kaloi was dominated by native Laurencia spp. Ulva fasciata XX (34.5 Æ 5.4%) and P. sanctae-crucis (33.8 Æ 7.1%) with smaller Rhodophyta amounts of algal turf (11.3 Æ 2.9%), S. echinocarpum (7.8 Æ 6.6%), Acanthophora specifera XX X Asparagopsis taxiformis XX X and Dictyota spp. (5.8 Æ 3.5%), and trace amounts of the non-native Botryocladia skottsbergii XX invasive rhodophyte A. spicifera (0.25 Æ 0.5%) and other taxa (Figs. 2f Gracilaria coronopifolia XX X and 4). At Papipi, S. echinocarpum (see Fig. 2e; 20.3 Æ 17.7%), Laurencia Grateloupia phuquocensis XX spp. (13 Æ 9.8%) and Pterocladiella spp. (12.5 Æ 13.9%) were the most Griffithsia heteromorpha X abundant taxa. However, the native invasive chlorophyte Ulva fasciata Hypnea spinella X Hypnea chordaceae XX (8.8 Æ 16.2%) and the non-native invasive rhodophytes A. spicifera Hypnea musciformis XX (7.5 Æ 5.1%) and H. musciformis (2.5 Æ 2.6%) also contributed signifi- Jania micrarthrodia X cantly to overall cover. At Amio, A. spicifera was the single most Jania micrarthrodia XX abundant limu species (29.8 Æ 11.5%), followed by Laurencia spp. Liagora sp. X Laurencia spp. X (25 Æ 12%), S. echinocarpum (18.8 Æ 16%), H. musciformis (8.3 Æ 1.0%) Laurencia dotyi X and other taxa (Fig. 2d). Laurencia majuscula XX X Total benthic cover did not vary significantly among sites, Laurencia mcdermidiae XX X averaging 95.9 Æ 2.4% overall (Fig. 3). However, cover of the non- Plocamium sandvicense XX native rhodophytes A. spicifera and H. musciformis was significantly Portieria hornemanni X Pterocladiella capillaceae XX higher at Amio than at Kaloi and Papipi (p  0.005, THSD; Fig. 4). Per- Pterocladiella caerlescens XX species average non-native cover (Fig. 3) also was significantly higher Spyridia filamentosa XX X at Amio (19.0 Æ 13.8%) than at Kaloi (0.1 Æ 0.4%) and Papipi Tricholgloea requienii XX (5.0 Æ 4.6%; p  0.008, THSD). In contrast, average native species Wrangelia requienni X Total taxa 39 30 25 cover was significantly lower at Amio (3.4 Æ 8.3%) than at Kaloi (8.5 Æ 13.6%; p = 0.030, THSD). These decreases in native species cover indicate a loss of native species abundance, with a correspond- spicifera) and Phaeophyta (brown limu: P. sanctae-crucis and S. ing increase in non-native invasive species abundance with proximity echinocarpum) for all variables except C:P ratios. Therefore, to the stormwater discharges (Fig. 3). This pattern was maintained in comparisons among sites were made separately for the two Phyla the summed benthic cover data (Fig. 3), which showed that total non- – Rhodophyta and Phaeophyta. Overall and individual species native benthic cover was significantly higher at Amio (36.3 Æ 13.2%) means (Æ1 SD) for all measured variables are shown in Table 2. In one than at Kaloi (0.3 Æ 0.5%) and Papipi (10.0 Æ 5.0%; p  0.004, THSD); or both Phyla, significant trends in d15N values, %N, N:P ratios, %P conversely, total native cover was significantly lower at Amio (rhodophytes only) and C:N ratios (rhodophytes only) between Kaloi (59.8 Æ 12.5%) than at Kaloi (96.3 Æ 1.5%) and Papipi (85.3 Æ 5.6%; and the AWSO were observed (Fig. 5). However, no significant p  0.004, THSD). differences among sites were observed for %C or C:P ratios, either overall or by Phylum (Fig. 5). d15N values in both rhodophytes and 3.2. Nutrient composition of intertidal limu communities phaeophytes increased significantly from Kaloi to the ASWO (p  0.007, THSD; as indicated in Fig. 5). Rhodophyte d15N averaged Comparison of pooled species among sites can become +11.5 Æ 2.0%, and ranged from +10.4 Æ 1.2% at Kaloi and +10.0 Æ confounded when tissue nutrient composition differs significantly 1.3% at Papipi to +12.7 Æ 1.6% at Amio and +14.2 Æ 0.6% at the among taxa. In this study, tissue nutrient composition differed ASWO. Phaeophyte d15N averaged +11.0 Æ 1.3% and ranged from significantly between Rhodophyta (red limu: L. majuscula and A. +9.8 Æ 1.1% at Kaloi to +11.4 Æ 1.3% at the ASWO. 314[()TD$FIG] B.E. Lapointe, B.J. Bedford / Harmful Algae 10 (2011) 310–318

Tissue C (% organic) in both rhodophytes and phaeophytes was similar among sites, as were tissue C:P ratios (Fig. 5). Rhodophyte C averaged 28 Æ 3%, ranging from 27 Æ 2% at Amio to 29 Æ 2% at Papipi; C:P averaged 1046 Æ 306% and ranged from 1218 Æ 506% at Kaloi to 717 Æ 3% at the ASWO. Phaeophyte C averaged 25 Æ 4% and ranged from 23 Æ 5% at Amio to 29 Æ 1% at the ASWO; C:P averaged 1199 Æ 337% and ranged from 1079 Æ 273% at Kaloi to 1641 Æ 18% at the AWSO. Tissue N (%) of both rhodophytes and phaeophytes increased from Kaloi to the ASWO (p  0.026; THSD), but was significantly higher (p = 0.012, GLM) and had a wider range in the rhodophytes than the phaeophytes (Fig. 5). Rhodophyte N averaged 2.0 Æ 0.6% and ranged from 1.5 Æ 0.2% at Kaloi to 3.5 Æ 0.1% at the AWSO. This increase in tissue N was most evident in the invasive, non-native A. spicifera, which increased from 1.5 Æ 0.3% at Kaloi Gulch to 3.5 Æ 0.1% at the ASWO (Table 2) and correlated strongly with benthic cover (r2 = 0.94; Fig. 6). Phaeophyte N averaged 1.2 Æ 0.4% and ranged from 0.8 Æ 0.1% at Kaloi to 1.7 Æ 0.2% at the ASWO. Tissue C:N ratios decreased from Kaloi to the ASWO, but the decrease was significant only for the rhodophytes (p  0.013), which averaged 16 Æ 5 and ranged from 22 Æ 5 at Kaloi to 10 Æ 0.2 at the AWSO (Fig. 5). Phaeophyte C:N averaged 27 Æ 10 and ranged from 35 Æ 13 at Kaloi to 20 Æ 2 at the AWSO. Tissue P (%) showed no consistent trends among sites, and only rhodophyte P differed significantly among sites (Fig. 5) with those at the ASWO having the highest P (0.10 Æ 0.00%; p  0.040, THSD). Rhodophyte P averaged 0.07 Æ 0.02% and was the lowest at Kaloi (0.06 Æ 0.01%), but not significantly lower than at Amio or Papipi. Phaeophyte P averaged 0.05 Æ 0.01% and ranged from 0.04 Æ 0.00% at the ASWO to 0.06 Æ 0.01% at Papipi. Fig. 3. (a) Mean (per species) native and non-native benthic limu cover (%) and Tissue N:P ratios also increased, in both rhodophytes (p = 0.037) numbers of taxa; (b) total benthic cover (%) of native and non-native limu at the and phaeophytes (p = 0.038), between Kaloi and the ASWO (Fig. 5). three study sites. Values are means Æ SE for percent cover (n = 4). Values with matching symbols differ significantly. Rhodophyte N:P averaged 70 Æ 19 and ranged from 54 Æ 12 at Kaloi to 88 Æ 22 at Amio. This increase in N:P ratio was again most evident in A. spicifera, which ranged from 45 Æ 6 at Kaloi to 107 Æ 1 at Amio (Table 2). Phaeophyte N:P ratios averaged 49 Æ 17 and ranged from [()TD$FIG] 32 Æ 4 at Kaloi to 82 Æ 8 at the ASWO.

4. Discussion

Multiple lines of evidence from our study support the hypothesis that nutrients from urban stormwater runoff support non-native limu invasions in the Ewa Beach area. First, as the non- native benthic cover increased with proximity to the stormwater outfall, native species diversity and abundance decreased. The two non-native rhodophytes, A. spicifera and H. musciformis, both increased in abundance with proximity to stormwater discharges at AWSO. Second, the abundance of A. spicifera, the most pervasive alien invader in Hawaii’s coastal waters (Russell, 1992; Huisman et al., 2007), was strongly correlated with tissue %N (r2 = 0.94), suggesting stormwater N enhances growth and abundance of this species in N-limited coastal waters of O’ahu (cf. Larned, 1998). Third, mean N:P ratios increased significantly in both red and brown limu from Kaloi to Amio, which further indicated increased N availability from ASWO discharges. Fourth, d15N values increased significantly between Kaloi and AWSO, pointing to stormwater discharges as the source of 15N enrichment. The highest tissue %N, %P, N:P ratio, and d15N values all occurred in limu collected from the ASWO, which provides compelling evidence that stormwater discharges were the primary source of both N and P enrichment in our study area. The significant increase in N:P ratio between the Kaloi and Amio sites indicates that the stormwater discharges have a relatively high N:P ratio. N:P ratios reported for stormwater on O’ahu are Fig. 4. Benthic cover (%) of the 10 most common taxa at each of the three study sites. typically low, ranging from 7.3 (Presley and Jamison, 2009)to16 Values are means Æ SE (n = 4). (USGS, 2010), and reflect relatively high P inputs from human and/ B.E. Lapointe, B.J. Bedford / Harmful Algae 10 (2011) 310–318 315

Table 2 Tissue C, N, and P (%); C:N, C:P, and N:P molar ratios; and d15N values (%) of limu collected at the study sites, including pooled values. Values are means Æ 1 SD.

n %C %N %P C:N C:P N:P d15N(%)

Kaloi Acanthophora spicifera 2 23.1 Æ 1.3 1.5 Æ 0.3 0.07 Æ 0.01 18 Æ 3 811 Æ 24 45 Æ 6 9.8 Æ 1.3 Laurencia majuscula 2 31.4 Æ 3.8 1.4 Æ 0.1 0.05 Æ 0.00 26 Æ 1 1625 Æ 326 63 Æ 9 10.9 Æ 0.8 Sargassum echinocarpum 2 29.0 Æ 0.7 0.8 Æ 0.1 0.06 Æ 0.02 46 Æ 8 1306 Æ 127 29 Æ 2 10.4 Æ 1.2 Padina sanctae-crucis 2 18.9 Æ 1.8 0.9 Æ 0.1 0.06 Æ 0.00 24 Æ 0 853 Æ 53 35 Æ 2 9.1 Æ 0.5 Kaloi mean 8 25.6 Æ 5.5 1.1 Æ 0.4 0.06 Æ 0.00 24 Æ 0 853 Æ 53 43 Æ 14 10.1 Æ 1.1 Papipi Acanthophora spicifera 2270Æ 0.6 2.1 Æ 0.2 0.08 Æ 0.00 15 Æ 1 906 Æ 50 62 Æ 8 10.2 Æ 1.1 Laurencia majuscula 2 31.1 Æ 0.1 2.2 Æ 0.2 0.08 Æ 0.01 17 Æ 2 1068 Æ 161 65 Æ 16 9.9 Æ 1.7 Sargassum echinocarpum 2 25.1 Æ 0.9 1.1 Æ 0.1 0.05 Æ 0.00 27 Æ 0 1218 Æ 30 45 Æ 2 12.6 Æ 1.0 Padina sanctae-crucis 2 25.0 Æ 1.8 1.7 Æ 0.1 0.07 Æ 0.01 17 Æ 0 956 Æ 41 57 Æ 2 11.2 Æ 1.1 Papipi mean 8 27.0 Æ 2.8 1.8 Æ 0.5 0.07 Æ 0.01 19 Æ 5 1037 Æ 145 57 Æ 11 11.0 Æ 1.6 Amio Acanthophora spicifera 2 25.1 Æ 1.0 2.9 Æ 0.5 0.06 Æ 0.01 10 Æ 2 1110 Æ 171 107 Æ 1 11.7 Æ 1.2 Laurencia majuscula 2 28.2 Æ 0.2 2.1 Æ 0.2 0.07 Æ 0.00 16 Æ 1 1087 Æ 58 69 Æ 1 13.8 Æ 1.1 Sargassum echinocarpum 2 27.6 Æ 0.4 1.1 Æ 0.1 0.05 Æ 0.01 31 Æ 3 1601 Æ 256 52 Æ 4 11.3 Æ 0.4 Padina sanctae-crucis 2 18.4 Æ 1.4 1.1 Æ 0.1 0.06 Æ 0.01 20 Æ 4 821 Æ 110 41 Æ 2 11.6 Æ 0.6 Amio mean 8 24.8 Æ 4.2 1.8 Æ 0.8 0.06 Æ 0.01 19 Æ 8 115 Æ 326 67 Æ 27 12.1 Æ 1.3 ASWO Laurencia majuscula 2 28.4 Æ 0.6 3.5 Æ 0.0 0.10 Æ 0.00 10 Æ 0 717 Æ 3 76 Æ 2 14.2 Æ 0.6 Sargassum echinocarpum 2 28.6 Æ 0.6 1.7 Æ 0.2 0.05 Æ 0.00 20 Æ 2 1641 Æ 18 82 Æ 8 15.1 Æ 0.4 ASWO mean 4 28.8 Æ 0.5 2.6 Æ 1.1 0.07 Æ 0.03 15 Æ 6 1179 Æ 534 79 Æ 6 14.7 Æ 0.7 Grand mean 28 26.2 Æ 4.0 1.7 Æ 0.8 0.06 Æ 0.02 21 Æ 9 1123 Æ 326 59 Æ 21 11.7 Æ 1.9 or animal waste, fertilizers, or other sources. However, high N:P Hawaiian Islands in recent decades (Smith et al. 2002; Huisman ratios ranging from 39 to 247 have been reported for human- et al., 2007). These blooms are generally located in bays (Kaneohe impacted streams in Waimana, O’ahu; the relatively low P Bay) and coastlines (Waikiki, Lahaina) impacted by urbanization concentrations in these streams probably reflected the high iron and increased stormwater and/or wastewater nutrient loads from content of Hawaiian soils, which effectively immobilizes P in human activities (Dailer et al., 2010). In southwest Florida, massive groundwater (Laws and Ferentinos, 2003). Similarly high limu N:P blooms of red drift macroalgae, including Gracilaria spp. and ratios were associated with stormwater discharges in our study, as Hypnea spp., have emerged in coastal waters that are enriched by N:P ratios averaged 43 at Kaloi, and increased to 79 at the ASWO, pulsed discharges of both N and P from the Caloosahatchee and indicating a shift towards stronger P-limitation along this gradient Peace rivers, respectively (Lapointe and Bedford, 2007). of stormwater enrichment (Atkinson and Smith, 1983; Lapointe Among all the native and non-native limu species assessed in et al., 1992). Elsewhere on O’ahu, in Kaneohe Bay, the mean N:P this study, A. spicifera responded most to increasing nutrient ratio of 30 limu species was 44 (Atkinson and Smith, 1983; Smith, enrichment from the stormwater inputs. Between the Kaloi and 1994), a value remarkably similar to the mean value of 43 at Kaloi Amio sites, tissue %N doubled (1.5–2.9%) while C:N ratio decreased in this study. Elevated limu N:P ratios at Amio (67) and the ASWO (18–10) to non-limiting levels in A. spicifera; in comparison, the (79) were consistent with those in other human-impacted, P- native L. majuscula had relatively small increases in %N (1.4–2.1%) limited, carbonate-rich marine environments. For example, high and its C:N ratio was relatively high at both sites (26–16), N:P ratios (>100) occur in groundwaters in the Florida Keys suggesting nitrogen limitation (Lapointe and Bedford, 2010). On impacted by septic tanks (Lapointe et al., 1990). the Belizean Barrier Reef, experimental N pulses significantly Nutrient availability is a major factor affecting competition increased %N and photosynthesis of A. spicifera, a native species in among limu in tropical oligotrophic settings but has been poorly this location (Lapointe et al., 1987). The %N of the Belizean A. documented for introduced invasive macroalgae (Williams and spicifera ranged from 0.48% to 0.69% dry weight, 3-fold lower Smith, 2007). Although Soegiarto (1972) and Johannes (1975) than the 1.5–2.9% range found in A. spicifera from Ewa Beach (Table noted an apparent link between sewage pollution and the 2). These N-enriched conditions in the Ewa Beach study area, expansion of A. spicifera in Kaneohe Bay four decades ago, studies especially at Amio, support increased productivity and cover of the examining the relationship(s) among nutrient source(s), tissue non-native A. spicifera at the expense of native limu (Russell, 1992). C:N:P ratios and abundance of this pervasive invader have not been The superior ability of A. spicifera to assimilate nitrogen allows this conducted. Rhodophytes, such as A. spicifera and Hypnea spp., are invader to compete favorably with native species in eutrophic favored in nutrient-enriched tropical coastal waters where they environments and explains why this species is such a pervasive can outcompete other species that dominate under more invader in Hawaiian waters experiencing land-based nutrient oligotrophic conditions (Lapointe et al., 2004). The significantly enrichment. Similarly, on coral reefs off highly populated higher %N, %P, and N:P ratios in the rhodophytes compared to the southeast Florida, the non-native Caulerpa brachypus had a phaeophytes in this study, along a gradient of nutrient enrichment, significantly lower C:N ratio than native species, enabling this illustrate the competitive advantage of fast-growing rhodophytes alga to overgrow and displace native species in these nutrient- that sequester growth-limiting nutrients when they become enriched habitats (Lapointe and Bedford, 2010). available during episodic stormwater discharge events. Our In urban areas that rely on cesspools and septic tanks for human observations are consistent with early aquaculture studies that sewage disposal, stormwater runoff can be enriched with human found highly productive rhodophytes are capable of rapid uptake waste through the process of infiltration and inflow of contami- and storage of N (D’Elia and DeBoer, 1978). This phenomenon is not nated groundwater into the stormwater collection system restricted to rhodophytes in the Ewa Beach area, but is also (Wanielista and Yousef, 1993). Concentrations of total nitrogen evidenced by blooms of non-native rhodophytes – A. spicifera, H. in stormwater on O’ahu have ranged from 40 mM(USGS, 2010)to musciformis, Gracilaria salicornia and Kappaphycus alvarezii – that 70 mM(Presley and Jamison, 2009). Although these concentrations have become common around urbanized coastlines of the are lower than that typical of sewage effluent, the high volume 316[()TD$FIG] B.E. Lapointe, B.J. Bedford / Harmful[()TD$FIG] Algae 10 (2011) 310–318

Fig. 6. Linear regression showing the relationship between Acanthophora spicifera tissue N contents (%; n = 2) and benthic cover (%; n = 4). Values are means Æ SE.

similar to wetland plant species on O’ahu, which average +8.2 Æ 1.2% (n = 11; Table 3); d15N values decrease in wetland species on other less populated Hawaiian Islands where lower levels of urbanization and d15N values occur (Bruland and MacKenzie, 2010). Laws et al. (1999) reported values of +6% for particulate nitrogen in downstream coastal waters off Ewa Beach, which they attributed to wastewater (cesspools) enrichment of groundwaters on the Ewa plain. Other studies in the Hawaiian Islands have reported variable d15N values in coastal macroalgae, depending on the type and degree of anthropogenic influence: highly elevated values of +17.8 to 50.1% off Maui (Lahaina, Kihei, Kahului) were indicative of wastewater effluent (Dailer et al., 2010) compared to lower values of À0.5% off Kauai that reflect fertilizer N as well as relatively low inputs of sewage N (Derse et al., 2007). Similar patterns have been reported globally. On coral reefs in the vicinity of sewage outfalls off highly populated southeast Florida, invasive blooms of native and non-native chlorophytes had elevated d15N values ranging from +8 to +10% (Lapointe et al., 2005). In the Buccoo Reef Complex in Tobago, West Indies, macroalgal blooms adjacent to a sewage outfall had d15N values ranging from +10 to +12% (Lapointe et al., 2010). All these studies support the recent conclusions of Risk et al. (2009) that measurement of stable nitrogen isotopes in macroalgae provides a cost-effective and objective means of quantifying sewage stress on coral reefs.

5. Summary and conclusions

Fig. 5. Tissue chemistry (d15N, %C, %N, %P, C:N, C:P, N:P) of red (Rhodophyta) and Our results showed that native limu species diversity and brown (Phaeophyta) limu at three study sites and the Amio stormwater outfall pipe abundance decreased with proximity to stormwater discharges, (ASWO). Values are means Æ SE (n = 4 for Kaloi, Papipi, and Amio; n = 2 for ASWO). whereas non-native limu increased. The highest tissue %N, %P, N:P Dumbbells (dots connected by a line) connect statistically similar values; unconnected ratio and d15N values all occurred in limu collected from the ASWO, columns differ significantly. which provides compelling evidence that cumulative impacts from episodic stormwater discharges were the primary source of associated with stormwater runoff during episodic events can nutrient enrichment in the study area. The abundance of the result in high overall N loadings to coastal waters. This mechanism non-native invasive A. spicifera was strongly correlated with tissue could explain why the d15N values that we observed in exposed %N (r2 = 0.94), suggesting that stormwater N conferred a competi- macroalgae were elevated to values typical of sewage N (Heaton,

1986). In general, limu that rely on natural N from nitrogen fixation Table 3 15 have d N values close to atmospheric N (0%; France et al., 1998) Tissue d15N values (%) of coastal wetland plants in the Hawaiian and become progressively enriched as a result of assimilation of Islands from Bruland and MacKenzie (2010). d15N values are fertilizer (0 to +3%) and wastewater (+3 to +15%) N sources means Æ S.E.; ‘‘n’’ values are the number of sites sampled per island. (Heaton, 1986; Lapointe, 1997; Costanzo et al., 2001; Lapointe Island d15N(%) n et al., 2004, 2005; Lapointe and Bedford, 2007). The significant 15 Oahu 8.2 Æ 1.2 11 increase in d N values from +10 to +15% between Kaloi and Maui 7.0 Æ 1.0 6 Amio supports the hypothesis that stormwater discharges from Molokai 4.4 Æ 0.1 2 ASWO were a primary source of d15N enrichment to limu in the Kauai 3.4 Æ 0.1 8 study area. The high background d15N value of +10% in this study Hawaii 3.0 Æ 0.2 7 Mean 5.2 Æ 0.2 15 is typical for limu growing on wastewater N (Risk et al., 2009) and B.E. Lapointe, B.J. Bedford / Harmful Algae 10 (2011) 310–318 317 tive advantage on this species compared with native species, Costanzo, S.D., O’Donohue, M.J., Dennison, W.C., Loneragan, N.R., Thomas, M., 2001. A new approach for detecting and mapping sewage impacts. Mar. Poll. Bull. 42, particularly phaeophytes. N:P ratios in limu were relatively high in 149–156. the study area, indicating P-limitation, and increased in proximity D’Elia, C.F., DeBoer, J.A., 1978. Nutritional studies of two red algae. 2. Kinetics of to ASWO discharges. Finally, d15N values increased with increasing ammonium and nitrate uptake. J. Phycol. 14, 266–272. 15 proximity to the ASWO and limu at all sites were in the range of Dailer, M.L., Knox, R.S., Smith, J.E., Napier, M., Smith, C.M., 2010. Using d N values in algal tissue to map locations and potential sources of anthropogenic sewage N, suggesting that human and/or animal waste from nutrient inputs on the island of Maui, Hawaii, USA. Mar. Poll. Bull. 60, 655– stormwater runoff influenced the entire study area. 671. Blooms of both native and non-native limu have been Dawson, T.E., Mambelli, S., Plamboeck, A.H., Templer, P.H., Tu, K.P., 2002. Stable isotopes in plant ecology. Annu. Rev. Ecol. Ecol. Syst. 33, 507–559. increasingly problematic in the Hawaiian Islands over the past Derse, E., Knee, K.L., Wankel, S.D., Kendall, C., Berg, C.J., Paytan, A., 2007. Identifying five decades (Doty, 1961; Johannes, 1975; Smith et al., 1981, 2002; sources of nitrogen to Hanalei Bay, Kauai, using the nitrogen isotope signature Russell, 1987, 1992; Rodgers and Cox, 1999; Huisman et al., 2007). of macroalgae. Environ. Sci. Technol. 41, 5217–5223. Dillon, K.S., Chanton, J.P., 2008. Nitrogen stable isotopes of macrophytes assess The results of this study indicate that the spread of non-native stormwater nitrogen inputs to an urbanized estuary. Estuaries Coasts 31, 360– rhodophytes in Ewa Beach is facilitated by nutrient pollution, in 370. this case from stormwater discharges. However, nutrient enrich- Doty, M.S., 1961. Acanthophora, a possible invader of the marine flora of Hawaii. Pac. Sci. 15, 547–552. ment via submarine groundwater discharges can also deliver ECOHAB, 1997. The ecology and oceanography of harmful algae blooms. In: anthropogenic nutrients that support invasive blooms of limu, Anderson, D.M. (Ed.), A National Research Agenda. Woods Hole Oceanographic such as on Maui where injection wells are used to dispose of Institution, Woods Hole. treated sewage (Dailer et al., 2010). Sea turtles forage on the Elton, C.S., 1958. The Ecology of Invasions by Animals and Plants. Methuen and Co., London. expanding blooms of non-native invasive limu and records of France, R., Holmquist, J., Chandler, M., Cattaneo, A., 1998. Evidence for nitrogen fibropapilloma indicate a cluster of the disease in Hawaiian coastal fixation associated with macroalgae from a seagrass–mangrove–coral reef waters impacted by human nitrogen enrichment (Van Houtan system. Mar. Ecol. Prog. Ser. 167, 297–299. HARRNESS, 2005. Harmful Algal Research and Response: A National Environmental et al., 2010). With an increase in non-native invasive limu, the Science Strategy 2005–2015. Woods Hole Oceanographic Institution, Woods corresponding decrease in native limu abundance and diversity Hole. will also diminish reef ecosystem services (cf. Ruddle and Heaton, T.H.E., 1986. Isotopic studies of nitrogen pollution in the hydrosphere and atmosphere: a review. Chem. Geol. 59, 87–102. Johannes, 1985) and threaten traditional cultural uses of native Howarth, R., Anderson, D., Cloern, J., Elfring, C., Hopkinson, C., Lapointe, B., Malone, limu species. T., Marcus, N., McGlathery, K., Sharpley, A., Walker, D., 2000. Nutrient pollution Globally, non-native macroalgal invasions are expected to of coastal rivers, bays, and seas. Ecology 7, 1–15 (issues). Huisman, J.M., Abbott, I.A., Smith, C.M., 2007. Hawaiian Reef Plants. Report # UNIHI- increase in coming years due to the anticipated increase in SEAGRANT-BA-03-02. University of Hawaii Sea Grant College Program, Hono- unintended introductions of non-native organisms as a side effect lulu. of globalization (UNEP, 2006). Land-based nutrient pollution has Hunter, C.L., Evans, C.W., 1995. Coral reefs in Kaneohe Bay, Hawaii: two centuries been identified as an important factor facilitating invasions of non- of western influence and two decades of data. Bull. Mar. Sci. 57 (1), 501– 515. native Caulerpa in southeast Florida (Lapointe et al., 2005, 2010) Jaubert, J.M., Chisholm, J.R.M., Minghelli-Roman, A., Marchioretti, M., Morrow, J.H., and the Mediterranean Sea (Chisholm et al., 1997; Jaubert et al., Ripley, H.T., 2003. Re-evaluation of the extent of Caulerpa taxifolia development 2003). Hence, the role of land-based nutrient pollution in Hawaii, in the northern Mediterranean using airborne spectrographic sensing. Mar. Ecol. Prog. Ser. 263, 75–82. and elsewhere, must be recognized if coastal ecosystems are to be Johannes, R.E., 1975. Pollution and degradation of coral reef communities. In: Wood, managed effectively in the future. E.J.F., Johannes, R.E. (Eds.), Tropical Marine Pollution. Elsevier Scientific Pub- lishing Company, Oxford, pp. 13–51. Keefe, C.W., Blodnikar, K.L., Boynton, W.R., Clark, C.A., Frank, J.M., Kaumeyer, N.L., Acknowledgements Weir, M.M., Wood, K.V., Zimmermann, C.F., 2004. Nutrient Analytical Services Laboratory Standard Operating Procedures. Special Publication Series SS-80-04- We would like to acknowledge the assistance of Dr. Phil CBL. Chesapeake Biological Laboratory, Center for Environmental and Estuarine Studies, University of Maryland System, Solomons. McGillivary, Mike Lee, Henry Chang Wo, David Kimo Frankel, the Lapointe, B.E., 1997. Nutrient thresholds for bottom-up control of macroalgal Office of Hawaiian Affairs, Dr. Robert Richmond of the University of blooms on coral reefs in Jamaica and southeast Florida. Limnol. Oceanogr. Hawaii, Manoa, and residents of Ewa Beach that provided access to 42, 1119–1131. Lapointe, B.E., Bedford, B.J., 2007. Drift rhodophyte blooms emerge in Lee County, their properties. The paper was improved by the constructive FL, USA: evidence of escalating coastal eutrophication. Harmful Algae 6, 421– comments of several anonymous reviewers. This is contribution # 437. 1817 from the Harbor Branch Oceanographic Institute at Florida Lapointe, B.E., Bedford, B.J., 2010. Ecology and nutrition of invasive Caulerpa Atlantic University, Ft. Pierce, FL.[SS] brachypus f. parvifolia blooms on coral reefs off southeast Florida, USA. Harmful Algae 9, 1–12. Lapointe, B.E., Littler, M.M., Littler, D.S., 1987. A comparison of nutrient-limited References productivity in macroalgae from a Caribbean Barrier Reef and from a mangrove ecosystem. Aquat. Bot. 28, 243–255. Abbott, I.A., 1984. Limu: An Ethnobotanical Study of Some Hawaiian Seaweeds, 4th Lapointe, B.E., O’Connell, J.D., Garrett, G.S., 1990. Nutrient couplings between on- edition. National Tropical Botanical Garden, Kauai. site sewage disposal systems, groundwaters, and nearshore surface waters of Abbott, I.A., 1999. Marine Red Algae of the Hawaiian Islands. Bishop Museum Press, the Florida Keys. Biogeochemistry 10, 289–307. Honolulu. Lapointe, B.E., Littler, M.M., Littler, D.S., 1992. Nutrient availability to marine Abbott, I.A., Huisman, J.M., 2004. Marine Green and Brown Algae of the Hawaiian macroalgae in siliciclastic versus carbonate-rich coastal waters. Estuaries 15, Islands. Bishop Museum Press, Honolulu. 75–82. Asplia, I., Agemian, H., Chau, A.S.Y., 1976. A semi-automated method for the Lapointe, B.E., Barile, P.J., Yentsch, C.S., Littler, M.M., Littler, D.S., Kakuk, B., 2004. The determination of inorganic, organic, and total phosphate in sediments. Analyst relative importance of nutrient enrichment and herbivory on macroalgal com- 101, 187–197. munities near Norman’s Pond Cay, Exumas Cays, Bahamas: a ‘‘natural’’ enrich- Atkinson, M., Smith, S.V., 1983. C:N:P ratios of benthic marine plants. Limnol. ment experiment. J. Exp. Mar. Biol. Ecol. 298 (2), 275–301. Oceanogr. 28 (3), 568–774. Lapointe, B.E., Barile, P.J., Littler, M.M., Littler, D.S., 2005. Macroalgal blooms on Banner, A.H., 1974. Kaneohe Bay, Hawaii: urban pollution and a coral reef ecosys- southeast Florida coral reefs II. Cross-shelf discrimination of nitrogen sources tem. In: Proc. 2nd Int. Symp. Coral Reefs, vol. 2. Great Barrier Reef Committee, indicates widespread assimilation of sewage nitrogen. Harmful Algae 4, 1106– Brisbane. 1122. Bruland, G.L., MacKenzie, R.A., 2010. Nitrogen source tracking with d15N content of Lapointe, B.E., Langton, R., Bedford, B.J., Potts, A.C., Day, O., Hu, C., 2010. Land-based coastal wetland plants in Hawaii. J. Environ. Qual. 39, 409–419. nutrient enrichment of the Buccoo Reef Complex and fringing coral reefs of Carlton, J.T., Geller, J.B., 1993. Ecological roulette: the global transport of non- Tobago, West Indies. Mar. Poll. Bull. 60, 334–343. indigenous marine organisms. Science 261, 78–82. Larned, S.T., 1998. Nitrogen versus phosphorus limited growth and sources of Chisholm, J.R.M., Fernex, F.E., Mathieu, D., Jaubert, J.M., 1997. Wastewater dis- nutrients for coral reef macroalgae. Mar. Biol. 132, 409–421. charge, seagrass decline and algal proliferation on the Cote d’ Azur. Mar. Poll. Laws, E.A., Ferentinos, L., 2003. Human impacts on fluxes of nutrients and sediment Bull. 34, 78–84. in Waimanalo stream, O’ahu, Hawaiian Islands. Pac. Sci. 57, 119–140. 318 B.E. Lapointe, B.J. Bedford / Harmful Algae 10 (2011) 310–318

Laws, E., Ziemann, D., Schulman, D., 1999. Coastal water quality in Hawaii: the Russell, D.J., 1992. The ecological invasion of Hawaiian reefs by two marine red importance of buffer zones and dilution. Mar. Environ. Res. 48, 1–21. algae Acanthophora spicifera (Vahl) Boerg. and Hypnea musciformis (Wulfen) J. Littler, M.M., Littler, D.S., 1985. Nondestructive sampling. In: Littler, M.M., Littler, Ag., and their association with two native species, Laurencia nidifica J. Ag. and D.S. (Eds.), Handbook of Phycological Methods, Ecological Field Methods: Hypnea cervicornis. J. Agric. ICES Mar. Sci. Symp. 194, 110–125. Macroalgae. Cambridge University Press, Cambridge, pp. 161–175. Smith, A.A.C., 1994. The effect of nutrient loading on C:N:P ratios of marine (MEA) Millenium Ecosystem Assessment, 2005. Ecosystems and Human Well- macroalgae in Kanehoe Bay, Hawaii. MS Thesis, University of Hawaii, Honolulu. Being: Biodiversity Synthesis. World Resources Institute, Washington, DC. Smith, S.V., Kimmerer, W.J., Laws, E.A., Brocknand, R.E., Walsh, T.W., 1981. Kaneohe (NRC) National Research Council, 2000. Clean Coastal Waters: Understanding and bay sewage diversion experiment: perspectives on ecosystem responses to Reducing the Effects of Nutrient Pollution. Ocean Studies Board, Water Science nutritional pertubation. Pac. Sci. 35 (4), 279–402. and Technology Board, National Academy Press, Washington, DC. Smith, J.E., Hunter, C.L., Smith, C.M., 2002. Distribution and reproductive character- Parsons, M.L., Walsh, W.J., Settlemier, C.J., White, D.J., Ballauer, J.M., Ayotte, P.M., istics of nonindigenous and invasive marine algae in the Hawaiian Islands. Pac. Osada, K.M., Carman, B., 2008. A multivariate assessment of the coral ecosystem Sci. 56, 299–315. health of two embayments on the lee side of the island of Hawaii. Mar. Poll. Bull. Soegiarto, A., 1972. The role of benthic algae in the carbonate budget of the modern 56, 1138–1149. reef complex, Kaneohe Bay. Thesis, University of Hawaii, Honolulu. Presley, T.K., Jamison, M.T.J., 2009. Rainfall, discharge, and water quality data during Staples, G.W., Cowie, R.H., 2001. Hawaii’s Invasive Species. Bishop Museum Press, stormwater monitoring, July 1, 2008, to June 30, 2009, Halawa stream drainage Honolulu. basin and the H-1 storm drain, Oahu, Hawaii. United States Geological Survey, (UNEP) United Nations Environmental Programme, 2006. Marine and Coastal Open-File Report, 2009-1162. . Ecosystems and Human Well-being: A Synthesis Report Based on the Findings Risk, M.J., Lapointe, B.E., Sherwood, O.A., Bedford, B.J., 2009. The use of d15Nin of the Millenium Ecosystem Assessment. . assessing sewage stress on coral reefs. Mar. Poll. Bull. 58, 793–802. United States Geological Survey, 2010. http://nwis.waterdata.usgs.gov/hi/nwis/ Rodgers, S.K., Cox, E.F., 1999. Rate of spread of introduced rhodophytes Kappa- qwdata/?site_no=212353157533001&agency_cd=USGS. phycus ahwarrzii, Kappaphycus striatum,andGracilaria salicornia and their Van Houtan, K., Hargrove, S.K., Balzas, G.H., 2010. Land use, macroalgae, and tumor- current distributions in Kaneohe Bay, Oahu, Hawaii. Pac. Sci. 53, 232–241. forming disease in marine turtles. PLOS One 5 (9), e12900. Ruddle, K., Johannes, R.E., 1985. The traditional knowledge and management of Wanielista, M.P., Yousef, Y.A., 1993. Stormwater Management. John Wiley and Sons, coastal systems in Asia and the Pacific. UNESCO, Jakarta. New York. Russell, D.J., 1987. Introductions and establishment of alien marine algae. Bull. Mar. Williams, S.L., Smith, J.E., 2007. A global review of the distribution, taxonomy, and Sci. 42, 641–642. impacts of introduced seaweeds. Annu. Rev. Ecol. Evol. Syst. 38, 327–359. See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/313480082

Sources and dispersal of land-based runoff from small Hawaiian drainages to a coral reef: Insights from geochemical signatures

Article · February 2017 DOI: 10.1016/j.ecss.2017.02.013

CITATIONS READS 0 60

2 authors, including:

Curt D Storlazzi United States Geological Survey 126 PUBLICATIONS 2,046 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Coral Reefs View project

Quantifying the long-term anthropogenic and natural effects of sediment resuspension along the coast, shelf and slope View project

All content following this page was uploaded by Curt D Storlazzi on 01 December 2017.

The user has requested enhancement of the downloaded file. Estuarine, Coastal and Shelf Science 188 (2017) 69e80

Contents lists available at ScienceDirect

Estuarine, Coastal and Shelf Science

journal homepage: www.elsevier.com/locate/ecss

Sources and dispersal of land-based runoff from small Hawaiian drainages to a coral reef: Insights from geochemical signatures

* Renee K. Takesue , Curt D. Storlazzi

U.S. Geological Survey, Pacific Coastal and Marine Science Center, 2885 Mission Street, Santa Cruz, CA 95060, USA article info abstract

Article history: Land-based sediment and contaminant runoff is a major threat to coral reefs, and runoff reduction efforts Received 12 September 2016 would benefit from knowledge of specific runoff sources. Geochemical signatures of small drainage Received in revised form basins were determined in the fine fraction of soil and sediment, then used in the nearshore region of a 3 February 2017 coral reef-fringed urban embayment on southeast Oahu, Hawaii, to describe sources and dispersal of Accepted 6 February 2017 land-based runoff. The sedimentary rare earth element ratio (La/Yb) showed a clear distinction between Available online 8 February 2017 N the two main rock types in the overall contributing area, tholeiitic and alkalic olivine basalt. Based on this geochemical signature it was apparent that the majority of terrigenous sediment on the reef flat origi- Keywords: fi Coral reef nated from geologically old tholeiitic drainages. Sediment from one of ve tholeiitic drainages had a fl Runoff distinct geochemical signature, and sediment with this signature was dispersed on the reef at 2 km Sediment provenance west and 150 m offshore of the contributing basin. Sediment and the anthropogenic metals Cd, Pb, and Rare earth elements Zn were entrained in runoff from the most heavily urbanized region of the watershed. Although Anthropogenic metals anthropogenic Cd and Zn had localized distributions close to shore, anthropogenic Pb was found asso- USA ciated with fine sediment on the westernmost part of the reef flat and 400 m offshore, illustrating how Hawaii trade-wind-driven sediment transport can increase the scale of runoff impacts to nearshore commu- Oahu nities. Our findings show that sediment geochemical signatures can provide insights about the source Maunalua Bay and dispersal of land-based runoff in shallow coastal environments. The application of such knowledge to watershed management and habitat remediation efforts can aid in the protection and restoration of runoff-impacted coastal ecosystems worldwide. Published by Elsevier Ltd.

1. Introduction smothers live corals and exposes them to contaminants, and it is associated with lower live coral cover, lower species diversity, and Coastal areas have long been desirable locations for human degraded fisheries (e.g., Fabricius, 2005; Knowlton, 2001; Rogers, settlement and economic activity (Mee, 2012). Human use of 1990). Runoff prevention and control measures in watersheds up- coastal areas and watersheds can, however, exert a heavy toll on stream of coral reefs are recognized globally as means to prevent ecosystems by altering natural processes and habitats and over- further degradation and facilitate recovery of runoff-impacted coral using resources (Jackson, 2008; Lotze et al., 2006). For example, reefs (e.g., Bartley et al., 2014; Hughes et al., 2010; Richmond et al., increased erosion and runoff of sediment, nutrients, and contami- 2007). In addition, the coupling of land-based runoff management nants can degrade coastal water quality and lead to the loss of vital with an understanding of local hydrodynamic controls on near- ecosystems including coral reefs (Fabricius, 2005; Pandolfi et al., shore sediment and contaminant transport is an important 2005), seagrasses (Short and Wyllie-Echeverria, 1996; Waycott component of comprehensive and effective remediation strategies et al., 2009), and kelp forests (Jackson et al., 2001). Land-based for runoff-impacted reefs (Done, 1995; Hunter and Evans, 1995; sediment and contaminant runoff is harmful to coral reefs in Restrepo et al., 2016). Recent studies have demonstrated the use many ways: it inhibits photosynthesis and larval recruitment, it of sediment-geochemical tracers in identifying sources of land- derived sediment to the coastal zone (Araújo et al., 2002; Prego et al., 2009, 2012; Roussiez et al., 2013; Smith et al., 2008). The goals of this study were to identify geochemical signatures of * Corresponding author. E-mail addresses: [email protected] (R.K. Takesue), [email protected] terrigenous sediment and trace metal runoff to a coral reef-fringed (C.D. Storlazzi). http://dx.doi.org/10.1016/j.ecss.2017.02.013 0272-7714/Published by Elsevier Ltd. 70 R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80 urbanized embayment from several short, steep drainages; to use Hawaii Kai Marina complex (Coles et al., 2002). Urban stream these signatures to identify sediment sources and infer nearshore sediment and roadside soil in southeast Oahu have been found to transport; and to describe the distribution of anthropogenic metals contain elevated anthropogenic trace metals (De Carlo et al., 2005; in urbanized basins and the reef flat. The trace metals cadmium Sutherland and Tolosa, 2000). (Cd), copper (Cu), lead (Pb), and zinc (Zn) are enriched by anthro- Maunalua Bay lies in the lee of Koolau Ridge relative to the di- pogenic activities, particularly the operation of motor vehicles rection of the northeast trade winds. Rainfall in the watershed is (Alloway, 1995). Insights from geochemical signatures about sour- higher in winter than in summer due to a higher frequency of ces of land-based sediment and contaminants and their nearshore southerly (Kona) storms and other low pressure disturbances dispersal can help runoff management efforts target priority (Giambelluca et al., 2013; Oki and Brasher, 2003). At high elevations contributing areas and guide nearshore habitat remediation of on Koolau Ridge where many streams have their headwaters, mean sediment-impacted coastal ecosystems. annual rainfall is approximately 1500e2000 mm, whereas rainfall on the urbanized coastal plain is about half that amount (Giambelluca et al., 2013; Oki and Brasher, 2003). Sediment 2. Site description retention structures were built in valleys above residential areas to control runoff and stream channels have been straightened and 2.1. Environmental setting hardened to varying extents. Storm runoff is flashy in nature (Tomlinson and De Carlo, 2003) and exacerbated by the high degree Maunalua Bay is an urbanized embayment on the southeast of impervious surface in urban areas (Wolanski et al., 2009). shore of the Island of Oahu, Hawaii, U.S.A. (Fig. 1). Hawaiian wa- The Maunalua Bay reef flat ranges from 0.2 to 1.0 km wide and is tersheds are generally composed of several small, steep valleys approximately 1 m deep and 10 km long. It is subject to water with streams that enter a larger body of water. Maunalua Bay is one quality impairments due to elevated nutrients and chlorophyll, such body, receiving runoff from 10 small drainage basins with a non-native algae, low live coral cover, and diminished fish and total contributing area of 57 km2. A major highway parallels about seagrass communities (Coles et al., 2002). half of the shoreline (Fig. 1) and dense residential development Currents in Maunalua Bay are driven by winds and tides (Presto occupies the coastal plain, valley floors, and some ridges. Kuapa et al., 2012; Storlazzi et al., 2010). At the surface to a depth of 1 m, Pond, a 2 km2 shallow lagoon in the east part of the watershed, was the prevailing trade winds drive westward transport (Presto et al., breached permanently, its marshland filled, and developed into the

Fig. 1. Shaded relief map of southeast Oahu showing Maunalua Bay (white line marks the shoreline) and its watershed (dashed line). Drainage basins are identified with numbers and abbreviations described in Section 3.1. Streams in basins 1e5 are shown by name. The sub-basin 3-WPE-W (Wiliwilinui) is denoted by a ‘w’ inside a gray triangle. Black triangles show USGS gaging stations (GS) on Wailupe and Waiakeakua Streams. ‘terr’, terrestrial; ‘ns’, nearshore. Black lines on the reef flat show reef transects. The 5 m isobath is shown for reference. Inset shows the location of Maunalua Bay in the Main Hawaiian Islands. R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80 71

2012), and the corresponding transport of suspended sediment from Hawaii Loa Ridge to address the local perception that con- over the reef flat is to the west (Storlazzi et al., 2010). Wind-driven struction activities made it a source of runoff to the reef. The lower currents are weakest in the eastern part of the bay in the lee of Koko reaches of Wailupe, Niu, and Kuli'ou'ou Streams were sampled from Head and strongest in the middle of the bay offshore of Wailupe highway bridges, and marina bottom sediment from a small boat, ® Stream (Storlazzi et al., 2010). using a hand-deployed Petite Ponar benthic sampler. The upper 0e5 cm of seabed sediment on the reef flat was collected by hand at 2.2. Geology low tide in acid-cleaned polypropylene jars along shore-normal transects offshore of Black Point, Wai'alae Nui Stream, Wiliwilinui The uplands around Maunalua Bay consist of layered tholeiitic Stream, Wailupe Stream, Hawaii Loa Ridge, Kuli'ou'ou Stream, and basalt lava flows from the Makapuu stage of Koolau Volcano Portlock at distances of 0, 50, 100, 250, and 400 m, where possible. (Haskins and Garcia, 2004), which erupted 2e3Ma(Jackson et al., Sampling locations and the numbers of samples are shown in Fig. 1 1999; Stearns and Vaksvik, 1935). Lava beds dip 4e10 to the south and Table 1. toward Maunalua Bay (Wentworth and Winchell, 1947). After about The 10 small basins that drain to Maunalua Bay and Hawaii Loa a million years of quiescence, subsidence, and erosion of Koolau Ridge are for convenience denoted with numbers increasing from Volcano (Gramlich et al., 1971), alkalic basalts erupted from rift west to east and an abbreviation: Wai'alae Nui (1-WAN), Wai'alae zones and vents, forming the Honolulu Volcanics. Around Maun- Iki (2-WAI), Wailupe (3-WPE), which includes Wiliwilinui (3-WPE- alua Bay, rocks of the Honolulu Volcanics form Koko Head, Koko W), Hawaii Loa Ridge (3.5-HLO), Niu (4-NIU), Kuli'ou'ou (5-KUL), Crater, and Black Point (Clague and Frey, 1982; Stearns and Vaksvik, Ka'alakei (6-KAL), Haha'ione (7-HAH), Kamilo Nui (8-KAN), Kamilo 1935; Winchell, 1947)(Fig. 1). Mantle-incompatible trace element Iki (9-KAI), and Portlock (10-KOK) (Fig. 1, Table 1). Runoff from 6- contents of Makapuu stage lavas can vary widely (Huang and Frey, KAL, 7-HAH, 8-KAN, and 9-KAI enters the Hawaii Kai Marina 2005) but do not vary systematically with age (Jackson et al., 1999). (MAR) and can become trapped there until removal from the sys- Compared to Koolau basalt, rocks of the Honolulu Volcanics are tem by dredging. Because the linkage between runoff into the enriched in alkali and alkali earth elements, light rare earth ele- marina and discharge into Maunalua Bay may be weak, sediment ments, and incompatible trace elements (Clague and Frey, 1982; geochemistry from these basins and the marina will not be dis- Roden et al., 1984). cussed in detail.

3. Approach 3.2. Sediment geochemical analyses

The element aluminum (Al) is a major component of terrestrial Geochemical analyses were performed on the sediment fine sediment (Windom et al., 1989) and a trace component of marine fraction (particle diameter <63 mm) to reduce grain size bias among carbonates (Milliman and Syvitski, 1992). This orders-of-magnitude samples. Bulk sediment was dried at 60 C, disaggregated gently to difference makes Al a sensitive indicator of terrigenous sediment in preserve original grains in an acid-cleaned agate mortar and pestle, environments where carbonate sediment predominates. In other and dry-sieved using stainless steel sieves to obtain the <63 mm coastal environments, Al contents of coastal sediment have been fraction. Fine sediment was decomposed according to EPA Method used to indicate riverine sediment inputs in the Gulf of Papua 3052, a near-total microwave-assisted digestion of siliceous (Brunskill et al., 1995) and the Portuguese shelf (Araújo et al., 2002). matrices using hydrochloric and hydrofluoric acids (USEPA, 1996). Furthermore, Al is not generally enriched by anthropogenic activ- Contents of major, minor, and trace elements including Al, Ba, Ca, ities (Windom et al., 1989) nor is it altered by oxidation-reduction Cd, Co, Cu, Fe, K, Mg, Mn, Na, Pb, Sr, Ti, V, and Zn were determined processes as is the element iron (Fe), another major component on a ThermoFinnigan Element I High Resolution inductively- of basalt. coupled plasma mass spectrometer (ICP-MS) at the Marine To characterize geochemical signatures of drainages contrib- Analytical Laboratories of the University of California at Santa Cruz. uting sediment to Maunalua Bay, immobile and relatively immobile Internal standardization was with germanium (72Ge); external trace element contents were determined in the fine fraction of standardization was with sediment reference materials (SRMs): stream sediment, where possible, to exploit the integrative nature National Institute of Standards and Technology 1646a and 2702 and over space and time of sediment stored in streams (Frissell et al., Canadian Certified Reference Materials Stream Sediment 2 and 3. 1986). Rare earth elements (REE), scandium (Sc), and thorium The reproducibility, expressed by the relative standard deviation (Th) are the most effective sediment provenance indicators (RSD) of a consistency standard analyzed five times, was 6% or (McLennan, 1989; McLennan et al., 1993). Barium (Ba), chromium better for all target elements except strontium (Sr) which had a RSD (Cr), cobalt (Co), nickel (Ni), niobium (Nb), rubidium (Rb), and zir- of 10%. The reproducibility of Fe measurements was better than 3%. conium (Zr) are can also be informative but could be affected by Contents of target elements in 86 soil and sediment samples sorting and weathering (McLennan et al., 1993). The anthropogenic analyzed on the Element I were several orders of magnitude higher trace metals cadmium (Cd), copper (Cu), lead (Pb), and zinc (Zn) than analytical detection limits, which were defined as three times were explored as urban overprints on sediment geochemistry in the standard deviation of blanks. Contents of rare earth elements drainage basins surrounding Maunalua Bay and on the reef flat. (REE), elements in resistant minerals (Cr, Hf, Nb, Y, Zr); elements with inadequate external standards (La, Rb, Th); and nickel (Ni) and 3.1. Sediment collection scandium (Sc), which may have had molecular interferences, were obtained by total digestion of 70 samples that had sufficient fine A low degree of fine-grained sediment storage in the short, material for quantification by SGS Inc., a nationally-recognized hardened stream channels around Maunalua Bay in many cases testing laboratory. SGS used a sodium peroxide sinter for total necessitated sediment collection from retention basins and from digestion and quantified major and minor elements by ICP-AES features such as culverts on the urbanized coastal plain. The upper (atomic emission) and trace elements by ICP-MS. The reproduc- 1e3 cm of sediment, where possible, composed of redeposited soil ibility of three replicate samples analyzed by SGS was better than and organic matter were collected in catchments 7e10 June 2010 10% for all target elements except the rare earth element thulium using acid-cleaned polypropylene sampling tools. Culverts gener- (Tm), which had a reproducibility of 12%. Target element contents ally contained only a veneer of sediment. In situ soil was collected determined by SGS were five or more times higher than analytical 72 R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80

Table 1 Drainage basin codes, names, area estimates, numbers of samples, and numbers of rare earth element (REE) analyses in the Maunalua Bay watershed. ‘TERR’, terrestrial; ‘NS’, nearshore.

Basin code Basin name Drainage area Number of samples Number of REE analyses

(km2)a TERR NSb TERR NS

1-WAN Wai'alae Nui 11.2 5 7 5 3 2-WAI Wai'alae Iki 4.6 2 2 2 1 3-WPE-W Wiliwilinuic 1.0 2 7 2 6 3-WPE Wailupe 9.5 4 6 3 2 3.5-HLO Hawaii Loa Ridge 0.7 4 4 4 3 4-NIU Niu 7.4 5 1 5 1 5-KUL Kuli'ou'ou 4.7 3 7 2 6 MAR Marinad 16.0 2 17 2 17 10-KOK Portlock 1.8 4 4 4 2

a Estimated from ridge to shoreline from Google Earth Pro. b Nearshore includes estuarine sites. c Wiliwilinui area included in Wailupe. d Marina drainages include Ka'alakei, Ha'ha'ione, Kamilo Nui, Kamilo Iki.

detection limits. Major element contents are reported as weight 3.4. Environmental data percent (wt %) and minor and trace element contents as micro- grams per gram (mg/g). Geochemical data are tabulated in the Time series of daily stream discharge and total suspended supplementary material. sediment concentration were obtained from USGS gaging station 16247550 (Wailupe Gulch at E. Hind Drive Bridge) for the two years 3.3. Normalization and criteria for geochemical signatures preceding the study. Annual stream discharge at station 16240500 (Waiakeakua Stream at the head of nearby Manoa Valley) over the Elemental contents of fine sediment were normalized to iron past three decades provided a long-term context for southeast (Fe) to account for basaltic parent rock compositions, a convention Oahu. Wind data were obtained from NOAA National Ocean Service used in Hawaiian soil and sediment (e.g., De Carlo and Spencer, (NOS) Station 1612340 (Honolulu, Hawaii), approximately 15 km 1995) because it can contain primary volcanic minerals (Nelson west of Maunalua Bay. et al., 2013). REE contents were normalized to a North American shale composite (NASC), which represents the composition of average sedimentary rock (McLennan, 1989). Normalized values are 4. Results denoted with subscripts MFe or REEN. Statistics were calculated with StatPlus:mac Pro software. REE ratios were normally distrib- 4.1. Environmental conditions uted in basaltic and in alkalic fine sediment (Shapiro-Wilk). Anal- ysis of variance (ANOVA) was performed on log-transformed data In the 8 months preceding this study, northeast trade winds for non-normally distributed parameters. were prevalent 55% of the time and there were only 16 d when the Geochemical tracers used for sediment-source attribution mean daily wind direction was from the west (Fig. 2). Water year should vary little within each source region so that end members (WY) 2010, which began 1 October 2009, was a dry year. Honolulu are well-constrained, and considerably among source regions to received about half the amount of precipitation as in the preceding allow discrimination (Collins and Walling, 2002). Ratios of 29 major year (Presley and Jamison, 2010). Discharge in Wailupe Stream was and trace elements relative to Al, Fe, Nb, Sc, and Th, and seven REE four times lower than in WY 2009, and discharge in Waiakekua ratios were examined in the initial data review. A high degree of Stream was four times lower than the decadal average (Oki and chemical weathering has been observed in Hawaiian soil and Brasher, 2003). suggests that only immobile element ratios are representative of source compositions (Kurtz et al., 2000; Vitousek et al., 1997). Accordingly, further data exploration focused on immobile ele- 4.2. Terrigenous sediment on the reef flat ments (REE ratios, ScFe,ThFe) and relatively immobile elements (BaFe,CrFe,CoFe,NiFe,NbFe,RbFe,ZrFe) as potential geochemical Aluminum contents of fine-grained sediment on the reef flat signatures in sediment derived from the two types of basalt in the ranged from 0.6 to 6.6 wt % with a mean of 2.7 ± 1.5 wt % (1s) Maunalua Bay watershed, henceforth denoted by THO (sediment compared to a mean of 9.8 ± 1.2 wt % (1s)infine-grained upland derived from tholeiitic basalt, n ¼ 25) and AOB (sediment derived sediment. Assuming fine sediment on the reef flat was a mixture from alkalic olivine basalt, n ¼ 4). One criterion was that the primarily of terrigenous and carbonate material, and that the Al geochemical property should have low coefficients of variation (CV, content of terrigenous sediment on the reef flat was similar to that defined as ratio of the standard deviation to the mean) in AOB and of upland sediment, then 6e66% (mean of 27%) of fine sediment on THO. The other was that the geochemical property should have a the reef flat was land-derived. Fine-sediment Al contents generally large enrichment factor (EF) between AOB and THO. EFs were decreased with distance from shore, except offshore of Wiliwilinui calculated as the ratios of median values (EF ¼ medianAOB/ and Portlock, where Al maxima occurred at 50 m. The three near- medianTHO). A conservative value of 2 was used for the minimum EF shore sites with the largest fractions of terrigenous sediment based criterion, which yielded geochemical signatures that were on on their Al contents were, in decreasing order: the mouth of Wai- average two or more times higher in AOB than THO. EFs of lupe Stream (66%), 50 m offshore of Wiliwilinui Stream (58%), and immobile and relatively immobile elements in AOB and THO were the mouth of Kuli'ou'ou Stream (44%). At two reef sites 400 m from plotted relative to the corresponding CVs in order to identify the the shore, the fine fraction of bed sediment contained only 1.2% and most effective geochemical tracers. 1.4% terrigenous material. R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80 73

Fig. 2. Time series plots of wind direction (top) and wind speed (middle) at NOS Meteorological Station 1612340 in Honolulu. Bottom panel shows stream discharge and total suspended sediment concentration (TSS) in Wailupe Stream at USGS gaging station 16247550 from 1 October 2009 through 30 June 2010.

4.3. Geochemical signatures were evaluated. (La/Yb)N ratios in THO and AOB fell along distinctly different Fine-sediment contents of ScFe,CrFe,CoFe,NiFe, and ZrFe were trends (Fig. 4). Fine sediment at 31 of 36 reef sites had (La/Yb)N relatively similar in AOB and THO (0.8 < EF < 1.3, Fig. 3) and so were not investigated further as geochemical signatures. RbFe had EF ¼ 10.2, however its high CVs precluded its ability to constrain end member compositions (Fig. 3). The median BaFe value was almost five times higher in AOB relative to THO and its CVs were intermediate (Fig. 3). EF values for (La/Yb)N,NbFe, and ThFe were 2.4, 2.5, and 2.4, respectively (Fig. 3). Because (La/Yb)N had the lowest CVs and EF > 2(Fig. 3), it was selected as the geochemical signature by which sources of land-based fine sediment to Maunalua Bay

Fig. 4. Plot of fine-sediment lanthanum (LaN) relative to ytterium (YbN) contents in terrestrial (large symbols) and marine environments (small symbols). The average composition and standard deviation (error bars) of tholeiitic basalt (THO) are shown Fig. 3. Comparisons of enrichment factors (EF) of Fe-normalized immobile and rela- for reference, data from Frey et al. (1994). The dashed line shows the least-squares tively immobile elements in the fine fraction of terrestrial soil and sediment derived regression through HLO (r ¼ 0.9980,n¼ 4) extrapolated from YbN ¼ 0.5 to low YbN from tholeiitic (THO) and alkalic olivine basalt (AOB) relative to their coefficients of values. The compositions of alkalic olivine basalts (AOB) are shown for reference, data variation (CV). Circles show groupings for THO and AOB corresponding to the labeled from Clague and Frey (1982) with a least-squares regression line (r ¼ 0.9949,n¼ 3, element. Dashed line shows EF ¼ 2. Symbols below the line are for ScFe,CrFe,CoFe,NiFe, dotted line). The star shows the composition of Asian dust in North Pacific pelagic and ZrFe. sediment near Hawaii (Nakai et al., 1993). 74 R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80 compositions that more closely followed the trend of tholeiitic orographic rainfall and thus varies spatially and with elevation on basalt than alkalic olivine basalt (Fig. 4). An estuarine sample Oahu (Jackson et al., 1971). In the contributing area of Maunalua collected in Niu Stream (4-NIU) had a (La/Yb)N ratio indicative of an Bay, dust deposition should be greater at high elevations than on AOB source whereas Niu basin consisted of tholeiitic basalt. Stream the coastal plain and in western drainages compared to eastern sediment in Wiliwilinui basin (3-WPE-W) had elevated LaN con- ones (Giambelluca et al., 2013). Following the approach of Kurtz tents relative to YbN values (LaN ¼ 0.5e0.6), representing an et al. (2001) who used soil Th/Nb ratios to estimate the Asian enrichment of 0.1e0.2 units relative to 3-WPE sediment (Fig. 4); dust contribution to soil on the Island of Hawaii, the fine fraction of this difference was significant in a one-way ANOVA (p ¼ 0.02). This soil and sediment collected for this study was estimated to contain enrichment is also apparent at five sites on the reef flat: 0 m and 0e15% Asian dust, assuming Th/Nb ratios of 0.52 for Asian dust 50 m offshore of the mouth of Wiliwilinui Stream (3-WPE-W), at (Kurtz et al., 2001), 0.05 for Koolau basalt (Frey et al., 1994), and the shoreline of 2-WAI, and at the shoreline and 150 m offshore of 0.08 for alkali olivine basalt and basanite (Clague and Frey, 1982). 1-WAN (Fig. 4). One sample from 5-KUL had an elevated Th content and Th/Nb ratio Stream sediment from tholeiitic basins 1e5 and soil from Hawaii that yielded an estimated dust contribution of 36%, but its (La/Yb)N Loa Ridge could not be distinguished individually based on their ratio was not indicative of a large dust fraction so it was considered (La/Yb)N ratios, except for Wiliwilinui. Instead, BaFe,NbFe, and ThFe an outlier and disregarded. The mean ± 1s estimated dust content were examined as potential basin-specific geochemical signatures was lowest at 10-KOK (4 ± 2%), the driest and easternmost basin, because these immobile or relatively immobile elements were and highest on Hawaii Loa Ridge (12 ± 4%), the highest elevation more variable among tholeiitic fine sediment. There were 2e5 site. There were no other longitudinal patterns in the mean esti- samples from individual source areas (Table 1). The fine fraction of mated dust content of fine sediment in drainage basins surround- soil and sediment from basins 1e5 and Hawaii Loa Ridge had ing Maunalua Bay: 1-WAN (7 ± 2%), 2-WAI (7 ± 1%), 3-WPE, generally similar median BaFe and NbFe values with the exception of including 3-WPE-W (7 ± 4%), 4-NIU (8 ± 5%), 5-KUL (7%). NbFe at Wiliwilinui (Fig. 5). Median ThFe contents of fine sediment Wiliwilinui (3-WPE-W) was the only basin whose sediment had appeared to be slightly lower in basins 1 and 2 compared to Hawaii (La/Yb)N ratios that were intermediate between those of AOB and Loa Ridge and basins 4, 5 (Fig. 5). However, log-transformed BaFe, THO (Fig. 5). Asian dust [(La/Yb)N ¼ 1.22 (Nakai et al., 1993),] would NbFe, and ThFe contents of fine sediment were statistically indis- increase the (La/Yb)N ratio of soil developed on THO [0.67 (Frey tinguishable among individual tholeiitic basins except for NbFe at et al., 1994),] and decrease the (La/Yb)N ratio of soil developed on Wiliwilinui (one-way ANOVA with Fisher LSD post-hoc analysis for AOB [1.50 (Clague and Frey, 1982),]. Based on the percentages of p < 0.01), as were other commonly used sediment provenance in- Asian dust estimated from the Th/Nb ratios of two samples from dicators such as log-transformed La/Th, Th/Sc, and Sm/Nd and Eu/ Wiliwilinui Stream, 7% and 14%, Asian dust inputs could explain * Eu . Thus, except in the case of Wiliwilinui, it was not possible to approximately half of the increase of (La/Yb)N ratios relative to THO geochemically discriminate sediment from individual tholeiitic in 3-WPE-W. basins and Hawaii Loa Ridge based on this set of samples and these In the nearshore region, fine sediment with elevated (La/Yb)N geochemical parameters. This result was not wholly unexpected ratios relative to THO were observed at the highway bridge over Niu because immobile elements were found to covary in Makapuu Stream and to the west of 3-WPE-W. No sediment in Niu basin had stage lavas (Huang and Frey, 2005). similarly elevated values, making it likely that the higher ratio re- flected the presence of AOB rather than Asian dust. Estimated dust contents of nearshore fine sediments were insufficient to account 4.4. Geochemical inputs to soil from Asian dust for elevated (La/Yb)N ratios offshore of 3-WPE-W, but could possibly account for those offshore of 1-KAN and 2-KAI. Dust from Asia is transported in the atmosphere across the North Pacific Ocean and deposited on the land surface of Hawaii (Dymond et al., 1974; Rex et al., 1969) where it can be incorporated 4.5. Trace metals in soil (Kurtz et al., 2001; Porder et al., 2007). Because the chemical composition of Asian dust can differ substantially from that of Chromium and Ni contents of Maunalua Bay terrestrial, estua- Hawaiian soil (e.g., Ferrat et al., 2011; Nakai et al., 1993), it can alter rine, and reef fine sediment and soil were correlated with those of geochemical mass balances in the soil column (Kurtz et al., 2001; Fe (Fig. 6). Nickel contents relative to Fe fell along a linear trend for Vitousek et al., 1997). Dust deposition is associated with all basins and depositional environments, whereas Cr contents fell

Fig. 5. Box and whisker plots of Fe-normalized barium (Ba), niobium (Nb), and thorium (Th) contents of fine-sediment from five tholeiitic basins (1e5) and Hawaii Loa Ridge (3.5). The height of the box shows the interquartile range, the whiskers show the maximum and minimum values, and the line crossing the whiskers shows the median value. The composition of sediment derived from alkalic olivine basalt (AOB) in 10-KOK is shown for comparison. R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80 75

Fig. 6. Plots of chromium and nickel content relative to iron content for all samples. The dashed lines show the effects range median values (ERM) for reference. into two populations, one with Cr contents that increased with Wiliwilinui streams. All nearshore fine-sediment contents of Cd, increasing Fe, and one with low Cr and high Fe contents (Fig. 6). All Cu, Pb, and Zn were below marine and estuarine ERM levels. Pb was terrestrial soil and sediment had Cr contents that exceeded the the only anthropogenic trace metal that was consistently elevated probable effects concentration (PEC) for freshwater ecosystems, above background levels in estuarine and reef flat fine sediment. 111 mg/g (MacDonald et al., 2000), as did 24 for Ni [PEC ¼ 48.6 mg/g Total Pb contents ranged from 3 to 114 mg/g (Fig. 7) compared to an (MacDonald et al., 2000),]. At 25 marine and estuarine sites sedi- estuarine background of 14 ± 1 mg/g (Table 2). When normalized to ment had Cr contents that exceeded the level at which adverse Fe to account for Pb in the geologic fraction, patterns of Pb biological effects are probable in marine and estuarine environ- enrichment were less extreme closer to shore (Fig. 7), and the ments [ERM, 370 mg/g (Long et al., 1995)], as did 42 with respect to highest PbFe value was found in carbonate-dominated sediment Ni (51.6 mg/g, Fig. 6). The sites with the highest Cr contents in fine- 400 m offshore of Wiliwilinui (3-WPE-W). The PbFe ratio of that grained sediment were nearshore sites: Wiliwilinui (3-WPE-W) sample, 52, was similar to the ratio in estuarine sediment before the stream mouth (0 m) and 50 m offshore of the stream mouth, and a phase-out of leaded gasoline (De Carlo and Spencer, 1995). In storm drain 125 m to the west of Wiliwilinui stream that carried comparison, the mean Pb content of five other carbonate- runoff from Wailupe basin. dominated sediment samples on the Maunalua Bay reef flat was Among terrestrial soil and sediment samples collected in June 6 ± 1 mg/g, lower than the pre-1927 value from the Ala Wai canal 2010, approximately twice as many had Cu and Pb contents (Table 2). exceeding background levels determined in forested conservation e lands during the 1998 2000 National Water Quality Assessment on 5. Discussion Oahu than for Cd and Zn (Table 2). Trace metal enrichments were up to 6 (Cd), 1.8 (Cu), 65 (Pb), and 6 (Zn) times background levels in High-standing and young islands in the tropics and subtropics fi ne soil and sediment in the contributing area of Maunalua Bay; can undergo high soil erosion rates and contribute large material fi however, median ne-sediment contents of Cd, Cu, Pb, and Zn were fluxes to the coastal ocean (Hilton et al., 2008; Kao and Milliman, at or below background levels (Table 2). Soil and sediment with 2008; Lyons et al., 2002; Milliman and Syvitski, 1992). Such land- elevated Cd, Cu, and Zn generally occurred on land, whereas more derived runoff of sediment, nutrients, carbon, and contaminants marine sites had above-background Pb contents than terrestrial can have large impacts on global biogeochemical cycles (Milliman ones (Table 2). et al., 1999; Nittrouer et al., 1995; Sholkovitz et al., 1999) and ma- Two of 31 terrestrial sites had Pb (1 site) and Zn (2 sites) con- rine and coastal ecosystems (Restrepo et al., 2006; Waycott et al., fi tents of ne sediment that exceeded the PEC (Table 2). These were 2009). Many Hawaiian watersheds are small, steep, urbanized, culverts adjacent to a major highway on the south shore of Hawaii and coupled to downstream ecologic communities, such as coral Kai Marina and Koko District Park. Runoff near the highway had the reefs and seagrasses, and socio-economic activities such as tourism m highest Zn and Pb contents measured in this study (1253 g/g and and recreation, that rely on clean and clear water in coastal areas to m 325 g/g, respectively). The Cu background level in Maunalua soil flourish (GPA, 2006; Nurse et al., 2014). Sediment-geochemical was three times higher than the PEC (Table 2), and all but three signatures that identify areas contributing runoff could aid in the fi terrestrial sites had soil or sediment Cu contents of the ne fraction protection and restoration of coastal waters and ecosystems by m that exceeded the PEC. The highest overall Cu value (391 g/g) was identifying priority areas for management and remediation measured in forested parkland on Hawaii Loa Ridge at 340 m (Bartley et al., 2014). elevation. At no sites did soil or sediment Cd contents of the fine fraction exceed the PEC, which was more than an order of magni- 5.1. Watershed sources and nearshore dispersal of terrigenous tude higher than the Cd background in Maunalua soil (Table 2). sediment In the nearshore region, Cd, Cu, and Zn were elevated above estuarine background levels in fine sediment deposited in estua- Geochemical signatures of tholeiitic and alkalic olivine basalt rine reaches of streams, in marina bottom sediment, at the land-sea were able to distinguish fine sediment from basins in Koolau Range interface, and up to 75 m from shore offshore of Wai'alae Nui and versus a contributing area on the west slope of Koko Head. Based on 76 R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80

Table 2 Levels of anthropogenic trace metals in unimpacted areas, sediment quality criteria, and in Maunalua Bay terrestrial and marine (reef and estuarine) sediment.

Cd (ppm) Cu (ppm) Pb (ppm) Zn (ppm)

Background levels in unimpacted areas Upland forested soila 0.20 ± 0.08 223 ± 17 5 ± 3 210 ± 27 Estuarine sediment, Ala Wai Canalb 0.29 ± 0.03 141 ± 614± 1 122 ± 11

Freshwater sediment quality criteriac

Threshold effects concentration (TEC) 0.99 31.6 35.8 121 Probable effects concentration (PEC) 4.98 149 128 459

Marine and estuarine sediment quality criteriad

Effects range low (ERL) 1.2 34 46.7 150 Effects range median (ERM) 9.6 270 218 410

Terrestrial sediment, June 2010 (n ¼ 31)

Minimum 0.10 126 3 113 Maximum 1.16 391 325 1253 Median 0.18 218 5 167 Mean 0.24 215 23 219 Standard deviation 0.19 50 61 208 # samples exceeding upland background 6 13 14 5

Reef and estuarine sediment, June 2010 (n ¼ 55)

Minimum 0.04 17 3 18 Maximum 2.36 262 114 292 Median 0.09 78 15 86 Mean 0.14 90 21 104 Standard deviation 0.31 55 21 56 # samples exceeding estuarine background 1 11 29 17

a De Carlo et al. (2005), leeward sites (n ¼ 4). b De Carlo and Spencer (1995), core G8B (107e110 cm, n ¼ 3). c Consensus-based values from MacDonald et al. (2000). d Long et al. (1995). its geochemical signature, the majority of fine terrigenous sediment basin Wiliwilinui. Fine sediment from Wiliwilinui had a distinct on the reef flat originated from basins in Koolau Range. Basin- (La/Yb)N and NbFe signature and its dispersal up to 2 km west and specific sources could not be distinguished individually using the 150 m offshore of its source was consistent with the direction of geochemical signatures explored here, except for the small sub- trade-wind-driven sediment transport on south-facing fringing Hawaiian reefs (Ogston et al., 2004; Presto et al., 2006; Storlazzi et al., 2004). The size of Wiliwilinui was small compared to other basins, but the scale of the dispersal of its runoff signature on the reef flat was not. Therefore, runoff mitigation in this small basin could result in a relatively large improvement in land-based runoff impacts on the nearshore ecologic community. The alkalic (La/Yb)N signature in fine sediment near the mouth of Niu Stream was another example of westward terrigenous sediment transport over several km. The predominance of trade- wind-driven westward sediment transport has ecological implica- tions for the reef community in Maunalua Bay in two ways. First, sediment and contaminants entering the bay will be entrained in westward flow and undergo repeated cycles of resuspension, deposition, and interaction with organisms (Ogston et al., 2004) before exiting the bay near Black Point (Presto et al., 2012; Storlazzi et al., 2010). Runoff plumes arising from winter storms, on the other hand, have short residence times in Maunalua Bay, exiting the reef rapidly in the offshore direction (Storlazzi et al., 2010; Wolanski et al., 2009). Second, runoff from the east part of the watershed near Hawaii Kai Marina, which consists of almost 20 km of shore- line with high-density residential and commercial development, had the highest levels of anthropogenic trace metals. Runoff from this region likely contains other compounds associated with ur- banization (Brasher and Wolff, 2004) such as PAHs, pesticides, flame retardants, pharmaceuticals, and personal care products that are growing concerns in urban stormwater (Daughton and Ternes, 1999; Schwarzenbach et al., 2007) and groundwater (Barnes Fig. 7. Plots of total lead (Pb) and iron (Fe)-normalized Pb contents of estuarine and et al., 2008). Urban contaminants that are transported to the reef fine sediment with distance offshore. Estuarine values are shown to the left of 0 m. coastal ocean in surface runoff or groundwater can impact ecologic Dotted lines show marine background levels (Table 2). R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80 77 communities downstream of discharge sites. part of the watershed and a six-fold or higher enrichment over There were few rainfall events during the 2008e2009 winter geologic background levels. This is consistent with previous studies and spring preceding field sampling for this study, and flow in local showing that Cd and Zn in Hawaiian soils are anthropogenically streams was below normal, so runoff to the nearshore was also influenced (e.g., De Carlo and Spencer, 1995; De Carlo et al., 2005; likely below normal. In this context, the relatively large amount of Sutherland, 2000). Elevated fine-sediment contents of Cd and Zn fine-grained terrigenous sediment on the Maunalua Bay reef flat, in the estuarine reaches of streams and near storm drain outfalls which averaged more than 25%, was somewhat surprising, since indicate that these trace metals were entrained in runoff from the the same component averaged less than 20% of the sediment urbanized watershed and deposited at the land-sea interface, as is transported off the reef flat during winter 2008e2009 (Storlazzi typical in estuaries (Li et al., 1984; Turekian, 1977). Although et al., 2010). From the absence of strong Kona storms during nearshore fine sediment with anthropogenic Cd and Zn was winter 2009e2010, it can be inferred that there were fewer large confined to the inner 75 m of the reef flat under the dry conditions wave events that resuspended and transported sediment, resulting preceding this study, sediment and contaminant runoff and its in greater storage of terrigenous material from previous flood dispersal could be higher in wetter years. Furthermore, because events (Draut et al., 2009). Thus calm winter conditions appear to motor vehicle traffic is the primary urban source of Cd and Zn (De have contributed to the storage of land-derived sediment, and by Carlo et al., 2005; Sutherland, 2000), loading of these metals to the association sediment-bound contaminants, on the Maunalua Bay coastal ocean is expected to increase as population increases in the reef flat. state of Hawaii (HOP, 2013). Increasing runoff of sediment and ur- The naturally high and variable contents of Cr and Ni in Ha- ban contaminants is a global concern as coastal regions become waiian basalts are related to the minerals spinel and olivine, more populated and urbanized (Newton et al., 2012). respectively (Frey et al., 1994; Haskins and Garcia, 2004; Jackson Pb contents of urban soils on Oahu have been decreasing since et al., 1999). The strong correlation of Ni and Fe in terrestrial fine the phase-out of leaded gasoline in the 1980s (De Carlo and sediment was indicative of the presence of olivine (De Carlo et al., Anthony, 2002). Fine-grained sediment with high PbFe values 2005; Frey et al., 1994). When olivine is subaerially exposed it is relative to background levels indicate that legacy Pb-contaminated susceptible to alteration; however, Ni was tightly coupled to Fe in soil and sediment are still present in urban watersheds of southeast fine sediment on the reef flat, indicating it was immobile Oahu, though the Pb levels reported here are lower than during the (Marsaglia, 1993; Moberly et al., 1965) over the timescales of USGS National Water Quality Assessment a decade earlier. The Pb erosion and transport in the small drainages surrounding Maunalua content of nearshore carbonate sediment can reflect marine as well Bay and thus not likely to be bioavailable (Sutherland, 2000). The as anthropogenic processes. Carbonate has a strong affinity for Pb high Cr content of upland sediment was also due to a volcanic in seawater (Talbot and Chegwidden, 1983), and Pb enrichment of source (De Carlo et al., 2005; Frey et al., 1994), and its biomodal marine carbonate can occur by passive adsorption (Sturesson, distribution relative to Fe was indicative of two Cr-bearing min- 1976), also called scavenging, particularly when sediment is erals. Cr-spinel is ubiquitous in Hawaiian soil (Frey et al., 1994; Oze resuspended, which likely occurs almost daily on the Maunalua Bay et al., 2004) and its composition would account for sediment with a reef flat for more than 6 months of the year when trade winds high Cr to Fe ratio (Oze et al., 2004), whereas sediment with a lower prevail (Presto et al., 2006). Scavenging can account for nearshore Cr to Fe ratio could have contained Cr-bearing pyroxene (Nelson Pb contents of fine-grained carbonate sediment that were two to et al., 2013; Oze et al., 2004). Spinel and pyroxene are heavy min- three times higher than the marine background, values that were erals with densities approximately two times higher than terrige- comparable to Pb levels in calcareous sediment in other shallow nous alumiosilicates and marine carbonates, so stronger hydraulic water environments in Australia (Esslemont, 2000; Talbot and forcing is required for their suspension and transport than for Chegwidden, 1983) and Central America (Guzman and Jimenez, similarly sized particles of lower density. As a result, heavy mineral 1992). Lead enrichments in nearshore fine sediment in excess of transport can be decoupled from that of other particles in reef Pb-scavenging were attributed to anthropogenic legacy Pb from the environments (Marsaglia, 1993; Moberly et al., 1965). Cr contents of dispersal of land-based runoff across the Maunalua Bay reef flat. about 1000 mg/g found in carbonate-dominated sediment (<3% Legacy Pb was found along three of five transects on the reef flat; terrigenous) offshore of basins Wai'alae Nui and Portlock shows however, it did not occur at levels where biological impacts would that high-Cr mineral grains remained on the Maunalua Bay reef flat be expected. after other terrigenous sediment was winnowed away. Therefore the distribution of Cr in fine-grained sediment on the reef flat is not 5.3. Implications representative of overall fine-sediment transport and deposition patterns. Urbanization and development of coastal zones are increasing worldwide, and concomitant increases in material fluxes from land 5.2. Anthropogenic trace metals in the Maunalua Bay system to sea could irreparably disrupt coastal processes and ecosystems if corrective measures are not undertaken (Hughes et al., 2010; The elevated Cu content of soil and sediment in the Maunalua Jackson, 2008; Steffen et al., 2011). Sediment geochemical signa- Bay watershed, although affected by anthropogenic activities (De tures represent one means to improve the effectiveness of runoff Carlo et al., 2004), is due primarily to a large geologic component reduction and control efforts by identifying runoff-contributing (De Carlo et al., 2005), almost 60% of the total (Sutherland and areas, information that can inform watershed management prac- Tolosa, 2000), which is not biologically available. The occurrence tices (Bartley et al., 2014). In the coastal zone, geochemical signa- of some of the highest soil Cu contents in forested parkland where tures can provide insights about land-based sediment and motor vehicle operation is minimal underscores the geologic con- contaminant transport in relation to ecological communities, in- trol of this trace metal. The three-fold variation of terrestrial soil formation that is important for restoration efforts. and sediment CuFe contents reflects the natural variability of Cu in tholeiitic and alkalic olivine basalt lavas. 6. Conclusions A stronger anthropogenic influence was found for fine sediment Cd and Zn contents in the Maunalua watershed, as evidenced by the The health and resilience of coral reefs has been shown to occurrence of maximum values near the urban center in the east improve when land-based sediment and contaminant runoff is 78 R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80 reduced. Insights about runoff sources gained from sediment 160e162. http://dx.doi.org/10.1007/BF01204458. geochemical signatures would make runoff management more Clague, D.A., Frey, F.A., 1982. Petrology and trace element geochemistry of the Honolulu Volcanics, Oahu: implications for the oceanic mantle below Hawaii. effective by targeting priority contributing areas. A sediment J. Petrol 23, 447e504. http://dx.doi.org/10.1093/petrology/23.3.447. geochemical signature based on (La/Yb)N showed a clear distinction Coles, S.L., DeFelice, R.C., Eldredge, L.G., 2002. Nonindigenous Marine Species at between the two main rock types, tholeiitic and alkalic olivine Waikiki and Hawaii Kai, Oahu, Hawaii. Bishop Musuem Technical Report No. 25, Honolulu, Hawaii. 255 pp.. basalt, in the contributing area of a coral-reef fringed embayment in Collins, A.L., Walling, D.E., 2002. Selecting fingerprint properties for discriminating southeast Oahu, Hawaii. Asian dust was only a minor component of potential suspended sediment sources in river basins. J. Hydrol. 261, 218e244. soil and sediment in southeast Oahu. One of five tholeiitic basins in http://dx.doi.org/10.1016/S0022-1694(02)00011-2. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the overall contributing area could be individually distinguished the environment: agents of subtle change? Environ. Health Perspect. 107 (6), using (La/Yb)N and NbFe. Fine sediment from this small sub-basin 907e938. http://www.justor.org/stable/3434573. was dispersed over a large section of the reef flat, illustrating De Carlo, E.H., Anthony, S.S., 2002. Spatial and temporal variability of trace element concentrations in an urban subtropical watershed, Honolulu. Hawaii. Appl. how hydrodynamic sediment transport can increase the scale of Geochem 17, 475e492. http://dx.doi.org/10.1016/S0883-2927(01)00114-7. runoff impacts in coastal areas, and how geochemical signatures De Carlo, E.H., Beltran, V.L., Tomlinson, M.S., 2004. Composition of water and sus- can identify potential target areas for runoff control. Anthropogenic pended sediment in streams of urbanized subtropical watersheds in Hawaii. e metals Cd, Pb, and Zn were associated with fine sediment from the Appl. Geochem 19, 1011 1037. http://dx.doi.org/10.1016/0025-326X(95)98338- W. most heavily urbanized portion of the watershed, and were De Carlo, E.H., Spencer, K.J., 1995. Sedimentary records of anthropogenic inputs of entrained in runoff to the reef flat. Other trace metals Cr, Cu, and Ni heavy metals to the Ala Wai a small man-made estuary in Honolulu. Hawaii. e were associated with volcanic minerals in Hawaiian soil and sedi- pac. Sci. 49, 471 491. De Carlo, E.H., Tomlinson, M.S., Anthony, S.S., 2005. Trace elements in streambed ment, so were unlikely to pose ecological risks. Sediment-source sediments of small subtropical streams on O'ahu, Hawai'i: results from the attribution with geochemical signatures can be a valuable tool for USGS NAWQA program. Appl. Geochem 20, 2157e2188. http://dx.doi.org/ the implementation of effective runoff control measures, a strategy 10.1016/0025-326X(95)98338-W. Done, T., 1995. Remediation of degraded coral reefs: the need for a broad focus. Mar. that has become critical for the protection and restoration of coastal Pollut. Bull. 30, 686e688. http://dx.doi.org/10.1016/0025-326X(95)98338-W. ecosystems as urbanization and exploitation of coastal areas in- Draut, A.E., Bothner, M.H., Field, M.E., Reynolds, R.L., Cochran, S.A., Logan, J.B., crease worldwide. Storlazzi, C.D., Berg, C.J., 2009. Supply and dispersal of flood sediment from a steep, tropical watershed: Hanalei Bay, Kaua'i, Hawai'i, USA. Geo. Soc. Am. Bull. 121, 574e585. http://dx.doi.org/10.1130/B26367.1. Acknowledgments Dymond, J., Biscaye, P.E., Rex, R.W., 1974. Eolian origin of mica in Hawaiian soils. Geo. Soc. Am. Bull. 85, 37e40. http://dx.doi.org/10.1130/0016-7606(1974) 85<37:EOOMIH>2.0.CO;2. This work was funded by and carried out as part of the USGS Esslemont, G., 2000. Heavy metals in seawater, marine sediments and corals from Pacific Coral Reef Project, an effort in the U.S. and its trust territories the townsville section, great barrier reef marine park. Qld. Mar. Chem. 71, e to better understand the effect of geologic processes on coral reef 215 231. http://dx.doi.org/10.1016/S0304-4203(00)00050-5. Fabricius, K.E., 2005. Effects of terrestrial runoff on the ecology of corals and coral systems. Funding was provided by the USGS Coastal and Marine reefs: review and synthesis. Mar. Pollut. Bull. 50, 125e146. http://dx.doi.org/ Geology Program. The authors thank Cheryl Hapke (USGS) for 10.1016/j.marpolbul.2004.11.028. laying the groundwork for the field study; Alyssa Miller and Jolie Ferrat, M., Weiss, D.J., Strekopytov, S., Dong, S., Chen, H., Najorka, J., Sun, Y., Gupta, S., Tada, R., Sinha, R., 2011. Improved provenance tracing of Asian dust sources Wagner, on staff at the time of this study with Malama Maunalua, using rare earth elements and selected trace elements for paleomonsoon for guidance and logistical support during upland and marina studies on the eastern Tibetian Plateau. Geochim. Cosmochim. Acta 75, sediment collection; Leticia Hallas for assistance processing sam- 6374e6399. http://dx.doi.org/10.1016/j.gca.2011.08.025. Frey, F.A., Garcia, M.O., Roden, M.F., 1994. Geochemical characteristics of Koolau ples; and Rob Franks of the University of California at Santa Cruz for Volcano: implications of intershield geochemical differences among Hawaiian assistance with ICP-MS analyses. John Crusius (USGS) and anony- volcanoes. Geochim. Cosmochim. Acta 58, 1441e1462. http://dx.doi.org/ mous reviewers provided excellent suggestions than improved the 10.1016/0016-7037(94)90548-7. fi Frissell, C.A., Liss, W.J., Warren, C.E., Hurley, M.D., 1986. A heirarchial framework for manuscript. Any use of trade, product, or rm names is for stream habitat classification: viewing streams in a watershed context. Environ. descriptive purposes only and does not imply endorsement by the Manage 10, 199e214. http://dx.doi.org/10.1007/BF01867358. U.S. Government. Giambelluca, T.W., Chen, Q., Frazier, A.G., Price, J.P., Chen, Y.-L., Chu, P.-S., Eischeid, J.K., Delparte, D.M., 2013. Online rainfall atlas of Hawai'i. B. Am. Meteorol. Soc. 94, 313e316. http://dx.doi.org/10.1175/BAMS-D-11-00228.1. Appendix A. Supplementary data GPA, 2006. Protecting Coastal and Marine Environments from Impacts of Land- based Activities: a Guide for National Action. United Nations Environment Programme (UNEP) Global Programme of Action for the Protection of the Ma- Supplementary data related to this article can be found at http:// rine Environment from Land-Based Activities (GPA), The Hague, p. 107. dx.doi.org/10.1016/j.ecss.2017.02.013. Gramlich, J.W., Lewis, V.A., Naughton, J.J., 1971. Potassium-argon dating of Holocene basalts of the Honolulu volcanic series. Geo. Soc. Am. Bull. 82 http://dx.doi.org/ 10.1130/0016-7606(1971)82[1399:PDOHBO]2.0.CO;2, 1,399-391,404. References Guzman, H.M., Jimenez, C.E., 1992. Contamination of coral reefs by heavy metals along the Caribbean coast of Central America. Mar. Pollut. Bull. 24, 554e561. Alloway, B.J., 1995. Heavy Metals in Soils. Blackie Academic & Professional, London. http://dx.doi.org/10.1016/0025-326X(92)90708-E. Araújo, M.F., Jouanneau, J.-M., Valerio, P., Barbosa, T., Gouveia, A., Weber, O., Haskins, E.H., Garcia, M.O., 2004. Scientific drilling reveals geochemical heteroge- Oliveira, A., Rodrigues, A., Dias, J.M.A., 2002. Geochemical tracers of northern neity within the Ko'olau shield, Hawai'i. Contrib. Mineral. Petrol 147, 162e188. Portuguese estuarine sediments on the shelf. Prog. Ocean. 52, 277e297. http:// http://dx.doi.org/10.1007/s00410-003-0546-y. dx.doi.org/10.1016/j.csr.2004.02.010. Hilton, R.G., Galy, A., Hovius, N., Chen, M.-C., Horng, M.-J., Chen, H., 2008. Tropical- Barnes, K.K., Kolpin, D.W., Furlong, E.t., Zaugg, S.D., Meyer, M.T., Barber, L.B., 2008. cyclone-driven erosion of the terrestrial biosphere from mountains. Nat. Geosci. A national reconaissance of pharmaceuticals and other organic wastewater 1, 759e762. http://dx.doi.org/10.1038/ngeo333. compounds in the United States - I). Groundw. Sci. Total Environ. 402, 192e200. HOP, 2013. Hawaii Ocean Resources Management Plan. Hawaii Office of Planning. http://dx.doi.org/10.1016/j.scitotenv.2008.04.028. Coastal Zone Management Program, Honolulu, p. 152. Bartley, R., Bainbridge, Z.T., Lewis, S.E., Kroon, F.J., Wilkinson, S.N., Brodie, J.E., Huang, S., Frey, F.A., 2005. Recycled oceanic crust in the Hawaiian Plume: evidence Silburn, D.M., 2014. Relating sediment impacts on coral reefs to watershed from temporal geochemical variations within the Koolau Shield. Contrib. Min- sources, processes, and management: a review. Sci. Total Environ. 468e469, eral. Petrol 149, 556e575. http://dx.doi.org/10.1007/s00410-005-0664-9. 1138e1153. http://dx.doi.org/10.1016/j.scitotenv.2013.09.030. Hughes, T.P., Graham, N.A.J., Jackson, J.B.C., Mumby, P.J., Steneck, R.S., 2010. Rising to Brasher, A.M.D., Wolff, R.H., 2004. Relations between land use and organochlorine the challenge of sustaining coral reef resilience. Trends Ecol. Evol. 25, 633e642. pesticides, PCBs, and semi-volatile organic compounds in streambed sediment http://dx.doi.org/10.1016/j.tree.2010.07.011. and fish on the Island of Oahu, Hawaii. Archive Environ. Contam. Toxicol. 46, Hunter, C.L., Evans, C.W., 1995. Coral reefs in Kaneohe Bay, Hawaii: two centuries of 385e398. http://dx.doi.org/10.1007/s00244-003-3019-4. western influence and two decades of data. B. Mar. Sci. 57, 501e515. Brunskill, G.J., Woolfe, K.J., Zagorskis, I., 1995. Distribution of riverine sediment Jackson, J.B.C., 2008. Ecological extinction and evolution in the brave new ocean. chemistry on the shelf, slope and rise of the Gulf of Papua. Geo-Mar. Lett. 15, P. Natl. Acad. Sci. U. S. A. 105, 11458e11465. http://dx.doi.org/10.1073/ R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80 79

pnas.0802812105. 559e569. http://dx.doi.org/10.1007/s00338-004-0415-9. Jackson, J.B.C., Kirby, M.X., Berger, W.H., Bjorndal, K.A., Botsford, L.W., Bourque, B.J., Oki, D.S., Brasher, A.M.D., 2003. Environmental Setting and the Effects of Natural Bradbury, R.H., Cooke, R., Erlandson, J., Estes, J.A., Hughes, T.P., Kidwell, S., and Human-related Factors on Water Quality and Aquatic Biota, Oahu, Hawaii. Lange, C.B., Lenihan, H.S., Pandolfi, J.M., Peterson, C.H., Steneck, R.S., Tegner, M.J., U.S. Geological Survey Water-Resources Investigations Report 03-4156, p. 98. Warner, R.R., 2001. Historical overfishing and the recent collapse of coastal Oze, C., Fendorf, S., Bird, D.K., Coleman, R.G., 2004. Chromium geochemistry of ecosystems. Science 293, 629e638. http://dx.doi.org/10.1126/science.1059199. serpentine soils. Int. Geol. Rev. 46, 97e126. http://dx.doi.org/10.2747/0020- Jackson, M.C., Frey, F.A., Garcia, M.O., Wilmoth, R.A., 1999. Geology and geochem- 6814.46.2.97. istry of basaltic lava flows and dikes from the Trans-Koolau tunnel, Oahu. Pandolfi, J.M., Jackson, J.B.C., Baron, N., Bradbury, R.H., Guzman, H.M., Hughes, T.P., Hawaii. b. Volcanol. 60, 381e401. Kappel, C.V., Micheli, F., Ogden, J.C., Possingham, H.P., Sala, E., 2005. Are U.S. Jackson, M.L., Levelt, T.W.M., Syers, J.K., Rex, R.W., Clayton, R.N., Sherman, G.D., coral reefs on the slippery slope to slime? Science 307, 1725e1726. http:// Uehara, G., 1971. Geomorphological relationships of tropospherically derived dx.doi.org/10.1126/science.1104258. quartz in the soils of the Hawaiian Islands. Soil Sci. Soc. Am. J. 35, 515e525. Porder, S., Hilley, G.E., Chadwick, O.A., 2007. Chemical weathering, mass loss, and http://dx.doi.org/10.2136/sssaj1971.03615995003500040015x. dust inputs across a climate by time matrix in the Hawaiian Islands. Earth. Kao, S.J., Milliman, J.D., 2008. Water and sediment discharges from small moun- Planet. Sci. Lett. 258, 414e427. http://dx.doi.org/10.1016/j.epsl.2007.03.047. tainous rivers, Taiwan: the roles of lithology, episodic events, and human ac- Prego, R., Caetano, M., Bernardez, P., Brito, P., Ospina-Alvarez, N., Vale, C., 2012. Rare tivities. J. Geol. 116, 431e448. http://dx.doi.org/10.1086/590921. earth elements in coastal sediment of the northern Galician shelf: influence of Knowlton, N., 2001. The future of coral reefs. P. Natl. Acad. Sci. U.S.A. 98, 5419e5425. geologic features. Cont. Shelf Res. 35, 75e85. http://dx.doi.org/10.1016/ http://dx.doi.org/10.1073/onas.091092998. j.csr.2011.12.010. Kurtz, A.C., Derry, L.A., Chadwick, O.A., 2001. Accretion of Asian dust to Hawaiian Prego, R., Caetano, M., Vale, C., Marmolejo-RodrIguez, J., 2009. Rare earth elements soils: isotopic, elemental, and mineral mass balance. Geochim. Cosmochim. in sediments of the vigo ria, NW iberian peninsula. Cont. Shelf Res. 29, Acta 65, 1971e1983. http://dx.doi.org/10.1016/S0016-7037(01)00575-0. 896e902. http://dx.doi.org/10.1016/j.csr.2009.01.009. Kurtz, A.C., Derry, L.A., Chadwick, O.A., Alfano, M.J., 2000. Refractory element Presley, T.K., Jamison, T.J., 2010. Rainfall, Discharge, and Water-quality Data during mobility in volcanic soils. Geology 28, 683e686. http://dx.doi.org/10.1130/ Stormwater Monitoring, H-1 Storm Drain, Oahu, Hawaii, July 1, 2009 to June 30, 0091-7613(2000)28<683:REMIVS>2.0.CO;2. 2010. U.S. Geological Survey Open-File Report 2010-1161, Reston, Virginia, p. 30. Li, Y.-H., Burkhardt, L., Teraoka, H., 1984. Desorption and coagulation of trace ele- Presto, M.K., Ogston, A.S., Storlazzi, C.D., Field, M.E., 2006. Temporal and spatial ments during estuarine mixing. Geochim. Cosmochim. Acta 48, 1879e1884. variability in the flow and dispersal of suspended-sediment on a fringing reef http://dx.doi.org/10.1016/0016-7037(84)90371-5. flat, Molokai. Hawaii. Estuar. Coast. shelf. Sci. 67, 67e81. http://dx.doi.org/ Long, E.R., MacDonald, D.D., Smith, S.L., Calder, F.D., 1995. Incidence of adverse 10.1016/j.ecss.2005.10.015. biological effects within ranges of chemical concentrations in marine and Presto, M.K., Storlazzi, C.D., Logan, J.B., Reiss, T.E., Rosenberger, K.J., 2012. Coastal estuarine sediments. Environ. Manage 19, 81e97. http://dx.doi.org/10.1007/ Circulation and Potential Coral Larval Dispersal in Maunalua Bay, Oahu, Hawaii. BF02472006. U.S. Geological Survey Open-File report 2012-1040, p. 67. Lotze, H.K., Lenihan, H.S., Bourque, B.J., Bradbury, R.H., Cooke, R.G., Kay, M.C., Restrepo, J.D., Park, E., Aquino, S., Latrubesse, E.M., 2016. Coral reefs chronically Kidwell, S.M., Kirby, M.X., Peterson, C.H., Jackson, J.B.C., 2006. Depletion, exposed to river sediment plumes in the southwest Caribbean: rosario Islands, degradation, and recovery potential of estuaries and coastal seas. Science 312, Colomiba. Sci. Total Environ. 553, 316e329. http://dx.doi.org/10.1016/ 1806e1808. http://dx.doi.org/10.1126/science.1128035. j.scitotenv.2016.02.140. Lyons, W.B., Nezat, C.A., Carey, A.E., Hicks, D.M., 2002. Organic carbon fluxes to the Restrepo, J.D., Zapata, P., Diaz, J.M., Garzon-Ferreira, J., Garcia, C.B., 2006. Fluvial ocean from high-standing islands. Geology 30, 443e446. http://dx.doi.org/ fluxes into the Caribbean Sea and their impact on coastal ecosystems: the 10.1130/0091-7613(2002)030<0443:OCFTTO>2.0.CO;2. Magdalena River, Colombia. Glob. Planet. Change 50, 33e49. http://dx.doi.org/ MacDonald, D.D., Ingersoll, C.G., Berger, T.A., 2000. Development and evaluation of 10.1016/j.gloplacha.2005.09.002. consensus-based sediment quality guidelines for freshwater ecosystems. Arch. Rex, R.W., Syers, J.K., Jackson, M.L., Clayton, R.N., 1969. Eolian origin of quartz in soils Environ. Contam. Toxicol. 39, 20e31. http://dx.doi.org/10.1007/s002440010075. of Hawaiian Islands and in Pacific pelagic sediments. Science 163, 277e279. Marsaglia, K.M., 1993. Basaltic island sand provenance. Geological Society of http://dx.doi.org/10.1126/science.163.3864.277. America Special Papers 284, Boulder, CO. In: Johnsson, M.J., Basu, A. (Eds.), Richmond, R.H., Rongo, T., Golbuu, Y., Victor, S., Idechong, N., Davis, G., Kosta, W., Processes Controlling the Composition of Clastic Sediments, pp. 41e65. Neth, L., Hamnett, M., Wolanski, E., 2007. Watersheds and coral reefs: conser- McLennan, S.M., 1989. Rare earth elements in sedimentary rocks: influence of vation science, policy, and implementation. BioScience 57, 598e607. http:// provenance and sedimentary processes. In: Lipin, B.R., McKay, G.A. (Eds.), Rare dx.doi.org/10.1641/B570710. Earth Elements. Reviews in Mineralogy, Washington D.C, pp. 169e200. Roden, M.F., Frey, F.A., Clague, D.A., 1984. Geochemistry of tholeiitic and alkalic lavas McLennan, S.M., Hemming, S., McDaniel, D.K., Hanson, G.N., 1993. Geochemical from the Koolau Range, Oahu, Hawaii: implications for Hawaiian volcanism. approaches to sedimentation, provenance, and tectonics. Geological Society of Earth. Planet. Sci. Lett. 69, 141e158. http://dx.doi.org/10.1016/0012-821X(84) America Special Paper 284, Boulder. In: Johnsson, M.J., Basu, A. (Eds.), Processes 90079-7. Controlling the Composition of Clastic Sediment, pp. 21e40. Rogers, C.S., 1990. Responses of coral reefs and reef organisms to sedimentation. Mee, L., 2012. Between the devil and the deep blue sea: the coastal zone in an era of Mar. Ecol-Prog. Ser. 62, 185e202. http://dx.doi.org/10.3354/meps062185. globalization. Estuarine. Coast. Shelf Sci. 96, 1e8. http://dx.doi.org/10.1016/ Roussiez, V., Aubert, D., Heussner, S., 2013. Continental sources of particles escaping j.ecss.2010.02.013. the Gulf of Lion evidenced by rare earth elements: Floods vs. normal conditions. Milliman, J.D., Farnsworth, K.L., Albertin, C.S., 1999. Flux and fate of fluvial sedi- Mar. Chem. 153, 31e38. http://dx.doi.org/10.1016/j.marchem.2013.04.010. ments leaving large islands in the East Indies. J. Sea Res. 41, 97e107. http:// Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., von dx.doi.org/10.1016/S1385-1101(98)00040-9. Gunten, U., Wehrli, B., 2007. The challenge of micropollutants in aquatic sys- Milliman, J.D., Syvitski, J.P.M., 1992. Geomorphic/tectonic control of sediment tems. Science 313, 1072e1077. http://dx.doi.org/10.1126/science.1127291. discharge to the ocean: the importance of small mountainous rivers. J. Geol. Sholkovitz, E.R., Elderfield, H., Szymcak, R., Casey, K., 1999. Island weathering: river 100, 525e544. http://www.jstor.org/stable/30068527. sources of rare earth elements tot he Western Pacific Ocean. Mar. Chem. 68, Moberly, R., Baver, L.D., Morrison, A., 1965. Source and variation of Hawaiian littoral 39e57. http://dx.doi.org/10.1016/S0304-4203(99)00064-X. sand. J. Sediment. Petrol 35, 589e598. Short, F.T., Wyllie-Echeverria, S., 1996. Natural and human-induced disturbance of Nakai, S., Halliday, A.N., Rea, D.K., 1993. Provenance of dust in the Pacific Ocean. seagrasses. Environ. Conserv. 23, 17e27. http://dx.doi.org/10.1017/ Earth. Planet. Sci. Lett. 119, 143e157. http://dx.doi.org/10.1016/0012-821X(93) S0376892900038212. 90012-X. Smith, J., Douglas, G.B., Radke, L.C., Palmer, M., Brooke, B.P., 2008. Fitzroy River Nelson, S.T., Tingey, D.G., Selck, B., 2013. The denudation of ocean islands by ground Basin, Queensland, Australia. III. Identification of sediment sources in the and surface waters: the effects of climate, soil thickness, and water contact coastal zone. Environ. Chem. 5, 231e242. http://dx.doi.org/10.1071/EN07094. times on Oahu, Hawaii. Geochim. Cosmochim. Acta 103, 276e294. http:// Stearns, H.T., Vaksvik, K.N., 1935. Geology and Ground-water Resources of the Island dx.doi.org/10.1016/j.gca.2012.09.046. of Oahu, Hawaii. Hawaii Division of Hydrography. Bulletin 1, Honolulu, p. 479. Newton, A., Carruthers, T.J.B., Icely, J., 2012. The coastal syndromes and hotspots on Steffen, W., Persson, A., Deutsch, L., Zalasiewicz, J., Williams, M., Richardson, K., the coast. Estuar. Coast. Shelf. Sci. 96, 39e47. http://dx.doi.org/10.1016/ Crumley, C., Crutzen, P., Folke, C., Gordon, L., Molina, M., Ramanathan, V., j.ecss.2011.07.012. Rockstrom,€ J., Scheffer, M., Schellnhuber, H.J., Svedin, U., 2011. The Anthro- Nittrouer, C.A., Brunskill, G.J., Figueiredo, A.G., 1995. Importance of tropical coastal pocene: from global change to planetary stewardship. Ambio 40, 739e761. environments. Geo-Mar. Lett. 15, 121e126. http://dx.doi.org/10.1007/ http://dx.doi.org/10.1007/s13280-011-0185-x. BF01204452. Storlazzi, C.D., Ogston, A.S., Bothner, M.H., Field, M.E., Presto, M.K., 2004. Wave- and Nurse, L.A., McLean, R.F., Agard, J., Briguglio, L.P., Duvat-Magnan, V., Pelesikoti, N., tidally-driven flow and sediment flux across a fringing coral reef: southern Tompkins, E., Webb, A., 2014. Small islands. In: Barros, V.R., Field, C.B., Molokai. Hawaii. Cont. Shelf Res. 24, 1397e1419. http://dx.doi.org/10.1016/ Dokken, D.J., Mastrandrea, M.D., Mach, K.J., Bilir, T.E., Chatterjee, M., Ebi, K.L., j.csr.2004.02.010. Estrada, Y.O., Genova, R.C., Girma, B., Kissel, E.S., Levy, A.N., MacCracken, S., Storlazzi, C.D., Presto, M.K., Logan, J.B., Field, M.E., 2010. Coastal Circulation and Mastrandrea, P.R., White, L.L. (Eds.), Climate Change 2014: Impacts, Adaptation, Sediment Dynamics in Maunalua Bay, Oahu, Hawaii. U.S. Geological Survey and Vulnerability. Part B: Regional Aspects. Contribution of Working Group II to Open-File Report 2010-1217, p. 59. the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Sturesson, U., 1976. Lead enrichment in shells of Mytilus edulis. Ambio 5, 253e256. Cambridge University Press, Cambridge, UK, pp. 1613e1654. Sutherland, R.A., 2000. Bed sediment-associated trace metals in an urban stream, Ogston, A.S., Storlazzi, C.D., Field, M.E., Presto, M.K., 2004. Sediment resuspension Oahu. Hawaii. Environ. Geol. 39, 611e627. http://dx.doi.org/10.1007/ and transport patterns on a fringing reef flat, Molokai. Hawaii. Coral Reefs 23, s002540050473. 80 R.K. Takesue, C.D. Storlazzi / Estuarine, Coastal and Shelf Science 188 (2017) 69e80

Sutherland, R.A., Tolosa, C.A., 2000. Multi-element analysis of road-deposited GSA Today 7, 1e8. sediment in an urban drainage basin, Honolulu. Hawaii. Environ. Pollut. 110, Waycott, M., Duarte, C.M., Carruthers, T.J.B., Orth, R.J., Dennison, W.C., Olyarnik, S., 483e495. http://dx.doi.org/10.1016/S0269-7491(99)00311-5. Calladine, A., Fourqurean, J.W., Heck Jr., K.L., Hughes, A.R., Kendrick, G.A., Talbot, V., Chegwidden, A., 1983. Heavy metals in the sediments of Cockburn Sound, Kenworthy, W.J., Short, F.T., Williams, S.L., 2009. Accelerating loss of seagrasses Western Australia, and its surrounding areas. Environ. Pollut. B 5, 187e205. across the globe threatens coastal ecosystems. P. Natl. Acad. Sci. U. S. A. 106, http://dx.doi.org/10.1016/0143-148X(83)90003-4. 12377e12381. http://dx.doi.org/10.1073/pnas.0905620106. Tomlinson, M.S., De Carlo, E.H., 2003. The need for high resolution time series data Wentworth, C.K., Winchell, H., 1947. Koolau basalt series, Oahu, Hawaii. Geo. Soc. to characterize Hawaiian streams. J. Am. Water Resour. As 39, 113e123. http:// Am. Bull. 58, 49e78. http://dx.doi.org/10.1130/0016-7606(1947)58[49:KBSOH] dx.doi.org/10.1111/j.1752-1688.2003.tb01565.x. 2.0.CO;2. Turekian, K.K., 1977. The fate of metals in the coeans. Geochim. Cosmochim. Acta 41, Winchell, H., 1947. Honolulu series, Oahu, Hawaii. Geo. Soc. Am. Bull. 58, 1e48. 1139e1144. http://dx.doi.org/10.1016/0016-7037(77)90109-0. http://dx.doi.org/10.1130/0016-7606(1947)58[1:HSOH]2.0.CO;2. USEPA, 1996. Method 3052: Microwave Assisted Acid Digestion of Siliceous and Windom, H.L., Schropp, S.J., Calder, F.D., Ryan, J.D., Smith, R.G., Burnery, L.C., Organically Based Matrices, Part of the Test Methods for Evaluating Solid Waste, Lewis, F.G., Rawlinson, C.H., 1989. Natural trace metal concentrations in estu- Physical/chemical Methods. U.S. Environmental Protection Agency, p. 20. arine and coastal marine sediments of the southeastern United States. Environ. https://www.epa.gov/hw-sw846/sw-846-test-method-3052-microwave-assis- Sci. Technol. 23, 314e320. http://dx.doi.org/10.1021/es00180a008. ted-acid-digestion-siliceous-and-organically-based. Wolanski, E., Martinez, J.A., Richmond, R.H., 2009. Quantifying the impact of Vitousek, P.M., Chadwick, O.A., Crews, T.E., Fownes, J.H., Hendricks, D.M., watershed urbanization on a coral reef: Maunalua Bay. Hawaii. Estuar. Coast. Herbert, D., 1997. Soil and ecosystem development across the Hawaiian islands. shelf. Sci. 84, 259e268. http://dx.doi.org/10.1021/es00180a008.

View publication stats Marine Pollution Bulletin 71 (2013) 92–100

Contents lists available at SciVerse ScienceDirect

Marine Pollution Bulletin

journal homepage: www.elsevier.com/locate/marpolbul

Can stormwater be detected by algae in an urban reef in Hawai‘i? ⇑ T. Erin Cox a, , Celia M. Smith a, Brian N. Popp b, Michael S. Foster c, Isabella A. Abbott a a University of Hawai‘i, Department of Botany, 3190 Maile Way, Room 101, Honolulu, HI 96822, USA b University of Hawai‘i, Department of Geology and Geophysics, 1680 East–West Road, Honolulu, HI 96822, USA c Moss Landing Marine Laboratories, 8272 Moss Landing Road, Moss Landing, CA 95039, USA article info abstract

Keywords: Nitrogen (N) enrichment of tropical reefs can result in the dominance of invasive algae. The invasive alga ‘Ewa Beach Acanthophora spicifera and the native alga Laurencia nidifica are part of a diverse reef assemblage in ‘Ewa O‘ahu Beach, O‘ahu. Their N contents and d15N values were investigated to determine if N was enriched and to Acanthophora spicifera evaluate potential nitrogenous sources near and removed from storm-drain outlets. d15N values of algae Laurencia nidifica (3.8–17.7‰) were within and above the range for algae around the island (1.9–11.9‰). Elevated algae N Nitrogen isotope isotope values (d15N>+7‰, [N] > 1.6%) and seawater nitrate + nitrite levels (0.59–7.93 M) indicated a Effluent l 15 Nutrient inputs mixed, high nutrient environment. The overlap in d N values with multiple nitrogenous sources pre- cluded identification. However, spatial and temporal patterns did not support stormwater as the domi- nant, nitrogenous source. Patterns were congruent with algal incorporation of terrestrial derived N, subjected to a high degree of biogeochemical cycling. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction on algal communities and near shore water quality (Bernardo, 2008). The land in the ‘Ewa Beach area, once used for extensive su- Storm-drains have the potential to act as conduits during rain gar agriculture, is currently being developed into neighborhood events, collecting and focusing nitrogen (N) enriched runoff into subdivisions (Schaefers, 2006). Although climate in ‘Ewa Beach is coastal systems. In contrast, regions with histories of intensive tropical and dry, episodic rainfall events in winter months often re- agricultural production and/or use of septic tanks in high density sults in coastal low-land flooding. The average precipitation be- housing areas could lead to elevated background nutrient levels tween 1949 and 2001 was 508 mm but, 381 mm fell during in groundwater that obscure episodic rain events. Very little is October–March (Otkin and Martin, 2004). To prevent flooding known about the fate of these effluents into oligotrophic waters. and continue neighborhood development, new storm drains have Increased N input into aquatic systems can cause increased macro- been proposed (Schaefers, 2006; Bernardo, 2008) that would dis- algal production, changes in assemblages of organisms, alteration charge into nearby intertidal habitat. The shallow coastal area of food webs, and disruption of nutrient biogeochemical cycling along southwest O‘ahu has historically been an area of high algal (Valiela et al., 1992; Walsh, 2000; Cole, 2003; Choi et al., 2007). abundance and is culturally important to Hawaiians for collection In tropical pristine waters, where nutrient levels are low, addition of edible limu (macroalgae) (Abbott, 1996; Leone, 2004; Ohira, of nutrients can potentially be detrimental. There are several 2005). Longtime residents familiar with ‘Ewa Beach coastline recall examples of blooms of non-indigenous, nuisance seaweeds that the ability to collect burlap bags full of edible macroalgae (Abbott, out-compete native algae and corals in areas adjacent to sewage 1996). The recent macroalgal community appears to be different inputs (Banner, 1974; Russell, 1992; Stimson et al., 2001, 2002, from past descriptions. The local perception is the edible, native 2004, 2005). Also, increase in nutrient supply to oligotrophic marine plants are in decline (Leone, 2004). Terrestrially-derived waters is often cited as a catalyst for community phase shifts from nutrients potentially from past agricultural and waste manage- coral to algal dominated reefs (McCook, 1999). ment may explain the wide availability of limu in the ‘Ewa Beach In ‘Ewa Beach, a growing coastal community on the island of area in the past and the coincident reef algal decline with recent O‘ahu, a primary concern is the impact(s) of storm-drain effluent changes in land use. An investigation (Lapointe and Bedford, 2011) along ‘Ewa Beach coastline used N in algae to identify nitrog- ⇑ Corresponding author. Address: Univ. Pierre et Marie Curie-Paris, Laboratoire enous contributions from a storm-drain. They concluded that d’Oceanographie de Villefranche (UMR7093), 06234 Villefranche-sur-Mer Cedex, storm-drain effluent favors the growth of non-indigenous algal France. Tel.: +33 (0) 4 93 76 38 33; fax: +33 (0) 4 93 76 38 34. species along this shore but did not thoroughly consider all poten- E-mail addresses: [email protected], [email protected] (T. Erin Cox), celia@ tial sources of N or the effects of variation in habitat on algal abun- hawaii.edu (C.M. Smith), [email protected] (B.N. Popp), [email protected] dance and species composition. (M.S. Foster).

0025-326X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.marpolbul.2013.03.030 T. Erin Cox et al. / Marine Pollution Bulletin 71 (2013) 92–100 93

Identifying nutrient sources and fates can help predict potential hypothesis of Lapointe and Bedford (2011) that N in storm-drain impacts from storm-drain discharge on water quality and reef algal effluent supports the physiology of introduced and native algae. physiology and abundance. Reef algae are generally N-limited so if We sampled the N content and the d15N values of algae along a gra- a particular N source is dominant, reef algae are well suited for dient leading offshore, in the dry and rainy season, at four sites tracing N in aquatic environments. Many algal species have simple with and six sites without storm-drains that discharge into the morphologies that allow uptake of nutrients from the water col- intertidal zone. Isotopic values in algae are compared to the con- umn and integration into growth or storage (Larned, 1998; centrations and d15N values (nitrate + nitrite) of potential N McCook, 1999; Schaffelke, 1999; Fong et al., 2001, 2003; Lapointe sources. Support for a ‘‘drain effect’’ would be indicated by a spatial et al., 2004; Lin and Fong, 2008; Teichberg et al., 2008; Dailer et al., effect; reef algae with enriched N concentration and elevated d15N 2010, Dailer et al., 2012b). values in near-shore macroalgae at sites with a storm-drain and The N isotopic composition of macroalgae can also be used to after large rain events. Finally, to gain a wider perspective on water identify and track sources of N. The d15N values of macroalgae vary quality and reef physiology, we compare d15N values of algae col- as a function of N sources, the biogeochemical history of the N uti- lected at ‘Ewa Beach to values of algae collected from several other lized, species of N used, and the extent of dissolved N use (Peterson nearshore sites around O‘ahu island. and Fry, 1987; Kendall, 1998; Robinson, 2001). Atmospheric nitro- gen (N ) can be fixed and reduced to ammonium (NHþ, ammonifi- 2 4 2. Methods cation) and converted under low oxygen conditions to nitrite and nitrate (NO and NO, nitrification) by bacteria and archaea. Dur- 2 3 2.1. Description of sites along ‘Ewa Beach ing these transformations the N isotopic composition is altered via kinetic isotope fractionation, or the differential reaction of Ten sites (numbered 1–10 from west to east) were selected along 15N relative to 14N. N fixed in the atmosphere typically has d15N the shore of O‘ahu, from One‘ula Beach Park (21°18036.5900 values of 4‰ to +4.0‰ (Owens, 1987; Macko and Ostrom, N–158°0027.5200W) to ‘Ewa Beach proper (21°18042.1700N–158°00 1994) while dissolved N in groundwater including that derived 15.6600W), to examine the N sources of reef algae and to determine from manure and waste from sewage treatment with increased the extent of N contribution from storm-drain discharge (Fig. 1). bacterial processing has d15N values that can range from about Four sites (sites 4, 6, 8, 9) each contain an existing large storm-drain +6‰ to +22‰ (Macko and Ostrom, 1994; Kendall, 1998). For that directly discharges accumulated urban runoff into the intertidal macroalgae with simple morphologies that completely use the zone. These sites are referred to as ‘‘Drain sites’’. ‘‘Control sites’’ surrounding available pool of nitrate and/or ammonium for growth (sites: 1, 2, 3, 5, 7, 10), were selected for comparison and are inter- the incorporation of 15N occurs with little fractionation (Peterson spersed between sites with large storm-drains’ in areas with similar and Fry, 1987; Gartner et al., 2002; Cohen and Fong, 2005; topography, limestone and sand substrate, and shore direction as Umezawa et al., 2007) and quickly, within 7 days (Gartner et al., the adjacent Drain site. 2002). Hence when a single N source is dominant in the ecosystem, The large existing storm-drains have been in use for over the d15N values of macroalgae can reflect the d15N values of the 30 years (C. Morgan, Planning Solutions, personal communication). source of N. The storm-drains at Drain 4, 6, and 8 drain a watershed of approx- Previous investigations of the d15N values and N content in reef imately 13–31 acres (Hiyakumoto, 2012). Drain 9 is an ocean out- algae have had limited replication of sites of interest, limited sea- let for a larger watershed that extends more than 420 acres (C. sonal data, limited tissue replication, and no investigation of d15N Morgan, Planning Solutions, personal communication). values of nitrogenous sources (Derse et al., 2007; Lapointe and The ‘Ewa Plain consists of a Pleistocene age highly porous lime- Bedford, 2011). As such these earlier studies fail to take into ac- stone caprock, a wedge of sediment overlying volcanic rock (Mink, count the complexity of nutrient dynamics and species specific as- 1989). Highly rugose carbonate platforms occur from the intertidal pects of plant metabolism. Identifying sources of N can be zone into the shallow subtidal zones and are dominated by algae. challenging because the isotopic compositions of nitrogenous The urbanized area surrounding these sites is served by sewer sys- nutrients in land and marine environments is dynamic. N not only tems, but some households use on-site disposal of wastewater de- varies in form but increased N inputs result from upwelling of off- fined as either soil treatment, seepage pits, or cesspools (Whittier shore seawater (nitrates), discharge of wastewater (nitrates and and El-Kadi, 2009). ammonium), fertilizer runoff (nitrates and ammonium) and natu- ral sources of N in runoff. Mixing of these multiple sources occurs under field conditions. Mixing and inputs can also vary temporally 2.2. Species selection for fine scale examination with seasonal changes in the environment and with rainfall. For in- stance, groundwater discharge can vary with rainfall amounts and Two closely related reef algae, Acanthophora spicifera and Lau- deliver N into coastal waters (Lapointe et al., 2004). Temporal as- rencia nidifica (Order Ceramiales, Family Rhodomelaceae) were pects could be particularly important when assessing impact from sampled because they are abundant and have a simple morphology 15 a pulsed N source like a storm-drain. Furthermore, biogeochemical that should allow d N values of these reef algae to quickly reach reactions can occur naturally in groundwater and can be associated steady state with dissolved nitrogenous nutrients. A. spicifera,an with septic systems. Thus, a higher d15N value in algae does not introduced species, in Hawai‘i (Doty, 1961; Russell, 1992), is necessarily mean they have been exposed to elevated N inputs known to grow faster in N enriched waters and has been previ- from drainage effluent. Despite these challenges several studies ously used to identify anthropogenic and natural nitrogenous 2 have successfully used bulk d15N values of macroalgae to identify sources (Lin and Fong, 2008). For all sampling, small (1–2 cm ) dis- land-based N inputs into pristine coastal zones where the differ- crete clumps consisting of several whole plants were collected. ences between end points of nutrient gradients are well resolved Laurencia nidifica was rare at sites 8–10 and few or no individuals (Sammarco et al., 1999; Schaffelke, 1999; Umezawa et al., 2002; were collected. Cohen and Fong, 2005; Garrison et al., 2007; Lin et al., 2007; Teichberg et al., 2008; Dailer et al., 2010, 2012a). 2.3. Sampling methodology We used the N content and the d15N values of common reef al- gae to identify the potential source and fate of N along the coastal Sampling occurred from the upper edge of the intertidal zone or region of ‘Ewa Beach, O‘ahu. Specifically, we rigorously test the storm-drains to a distance of approximately 20 m (1–3 m depth) 94 T. Erin Cox et al. / Marine Pollution Bulletin 71 (2013) 92–100

The HawaiianThe Hawaiian Archipelago Archipelago

N O‘aO‘ahu h u

‘EwaBeach‘Ewa‘E w aBeach B e a c h

Fig. 1. Map of the Hawaiian Islands (top) showing southwest O‘ahu and an aerial image of ‘Ewa Beach (Google EarthÓ, bottom) with selected sites 1–10 (circles are control sites with no drain present, squares are drain sites, i.e. drain present), nearby parks (picnic tables), and golf course (golfer). The and enclosed area marked with are the approximate locations of nearby on-site sewage disposal systems (OSDSs) as identified by Whittier and El-Kadi, 2009. More inland OSDS identified in 2009 within the Hasekeko development (northwest cleared portion of image) have been excluded from this representation as these OSDS are temporary structures and possibly not present when this study was conducted. offshore. One algal sample of each species was collected, when occurred, from 12 shallow water sites around the island of O‘ahu encountered, at set intervals (1 m) along a transect line. If both for determination of d15N values in reef algae. Results of analyses species did not occur, the available species was collected. of these specimens provide perspective on whether d15N values Replicate algal samples were collected in months that are typi- from algae within the ‘Ewa Beach area differ from those at other cally dry (July–August) and rainy (November–December) to ac- intertidal shores. Nine sites are located in a different watershed count for seasonality and discharge events. Because of permit from sites 1–10. These watersheds have variable rainfall amounts, requirements, sites 1–6 were sampled in 2007 while sites 7–10 drainage areas, land use, onshore wave action, and nearshore were sampled in 2008. The accumulated rainfall 7 days prior to this depths, and are thus predicted to have different N concentrations collection ranged from 1.09 to 8.66 cm in the rainy season and and varying isotopic compositions (Van Houtan et al., 2010). On trace amounts to 1.04 cm in the dry season (Ewa Kalaeloa Airport the 13th of March 2008 samples (n = 5–7) were collected along Station Id:GHCND:USW00022551). the Wai‘anae (West) coast. On the 9th of October 2011 algae were To determine the nutrient inputs delivered by storm drain flow collected from the six other sites and from Control 2 and Drain 4 and water column nutrients, at time of plant collection in the dry (for a total of 9 sites). Control 2 and Drain 4 were sampled again season, 80 ml water samples (n = 1/site) were collected from on this day in 2011 to account for any temporal variation that near-shore or at the mouth of drains. On a large rain event in may have occurred from 2007. One to two replicate samples of 2007, samples of flowing water were collected from street storm each species were collected when encountered at a site. Samples grates that empty directly into Drain 4, 6, and 8 and were used were from separate, growing plants found within a 100 m dis- as a proxy for storm-drain nutrient source. For comparison, on tance from each other in intertidal habitats. the same rain event, water was collected from the ocean at Control 3 and from a shallow water habitat 100 m to the west of Site 1. In 2.5. Nitrogen content and d15N value determination the dry season in 2008, an 80 ml water sample was collected from standing water in Drain 9. One 80 ml water sample was collected Algal samples were immediately cleaned of epiphytes, rinsed in from a nearby non-potable irrigation well that taps into groundwa- de-ionized water, and placed into an oven at 60 °C until dried. ter as a water source. Dried samples were ground into a fine powder, stored in glass scin- tillation vials until further analyses. Water samples were filtered 2.4. Methodology for island wide d15N value comparison (0.22 lm Millipore filter) and stored frozen (20 °C) until analyzed. Reef algae used in similar studies examining N inputs in Hawai‘i Carbon and N isotope compositions of reef algae were deter- (Dailer et al., 2010: A. spicifera, Laurencia mcdermidiae, Astronema mined using an on-line carbon–nitrogen analyzer coupled with breviarticulatum, or Ulva lactuca) were collected, when they an isotope ratio mass spectrometer (Finnigan ConFlo II/Delta-Plus). T. Erin Cox et al. / Marine Pollution Bulletin 71 (2013) 92–100 95

Isotope values are reported in standard d-notation relative to an Lastly, to determine if N inputs in the ‘Ewa Beach area were 15 international standard (V-PDB and atmospheric N2 for carbon similar to other areas in O‘ahu, we spatially mapped the d N and N, respectively). Glycine reference compounds with well-char- values of reef algae collected from around the island, following acterized isotopic compositions were used to ensure accuracy of all methods used by Dailer et al. (2010). isotope measurements. Several samples were measured in dupli- cate or triplicate and the reproducibility associated with these 3. Results measurements was typically 0.2‰ for both carbon and N isotopic measurements. 3.1. Fine scale examination Analyses of water samples for dissolved inorganic nutrients (NHþ,NO,NO,PO, SiO) were performed by the Analytical 4 2 3 4 4 The non-linear relationship (n =7, p = 0.05, r-squared = 0.58) Lab at the Marine Science Institute, University of California, Santa d15 Barbara using a continuous flow technique with a Quick Chem between the ln [nitrate] concentration and N (nitrate + nitrite) values in seawater suggested the d15N values of reef algae in the 800 Flow Injection Analyzer manufactured by Lachat Instruments, Inc. The precision associated with these measurements was typi- ‘Ewa Beach area were the result of a mixing of multiple sources of N with different isotopic compositions (Fig. 2). cally 0.05%. 15 A comparison of nutrient concentrations within the water col- Analyses of water samples for d N (nitrate + nitrite) values were conducted at the University of Washington, Isolab using the umn at sites 1–10 with the concentrations measured in potential sources indicated mixing and dilution of N (Table 1). Silicate con- bacterial denitrifier method (following Sigman et al., 2001) along with an autosampler, PreCon GasBench II assembly coupled to a centrations in groundwater were well above those measured in the seawater at any site. Similarly, the standing water in Drain 9 Finnigan Delta Plus. The isotopic compositions of the international reference materials USGS34 and IAEA-NO-3 were used to ensure was particularly high in silicate. The nutrient concentrations were variable among replicate storm-drain samples and tended towards accuracy. Several samples were measured in duplicate and had a standard deviation from 0.02‰ to 0.18‰. relatively higher phosphate, nitrate, and ammonium concentra- tions than those measured in seawater at nearby sites and from concentrations measured in ocean water collected simultaneously, after the same rain event. 2.6. Analyses The average d15N values of reef algae at sites 1–10 ranged from 4.8‰ to 14.7‰ (Table 2); in comparison the d15N values of nitrate A linear regression was used to examine the relationship be- 15 ‰ ‰ tween the ln [nitrate] and d N (nitrate + nitrite) values of seawa- ranged from 1.5 to 2.4 for open ocean water from 150 m and 7.0–7.1‰ from 500 m (as an estimate of upwelled nitrate) at Sta- ter at sites 1–10. Linearity would indicate a closed system with one source of N while a non-linear relationship indicates an open tion ALOHA located 100 km north of O‘ahu (Casciotti et al., 2008) to 27.9‰ for ‘Ewa Plain groundwater. Reef algae collected during system with mixing of N sources (Robinson, 2001). In addition, the d15N values of reef algae, nearby water sources and the concen- both the dry and rainy season at Drains 4 and 6 had higher mean d15N values (4: 12.0–13.3‰, 6: 10.0–11.9‰) than the d15N values tration of nutrients in the water column and source samples were compared in an attempt to identify N sources for algae. of nitrate+nitrite measured from runoff draining into the associ- 15 ated storm-drains (7.7–7.9‰) but algal isotopic values were much Statistical analyses were performed on total N content and d N 15 values (separately) for each species. Specifically for each species, a lower than the d N values measured for nitrate + nitrite in groundwater (27.9‰). Similarly, the d15N values of reef algae at two-way analysis of variance (ANOVA) was used to compare reef ‰ algal values between drain and control sites in the dry and rainy Drain 9 (5.3–7.9 ) were lower than the isotopic values of nitra- te + nitrite measured for standing water in Drain 9 (19.7‰) and season. Data met the requirements for normality and homogeneity groundwater (27.9‰). However, the reef algae at Drain 8 had sim- of variance. Multiple linear regressions were used to examine the 15 15 ilar mean d N values (8.8–10.4‰) as the storm-water collected relationship between d N values of L. nidifica and A. spicifera and ‰ distance from shore. Each site was examined separately because near Drain 8 (10.0 ). In contrast to reef algae, the two ocean water samples collected after the rain event had d15N values of nitrate + of unique site characteristics. Season and an interaction of Dis- nitrite (22.3‰, 20.2‰) that were similar to the d15N values for tance Season were included in the models as seasonal rainfall groundwater (27.9‰) and the values of water in Drain 9 (19.7‰). could alter delivery of N to the shore. Regressions were not per- formed on %N in algae because of the correlation with d15N values. Because nitrogenous nutrients can be derived from a wide 30 range of sources (e.g., natural runoff, groundwater seeps, submar- y = 13.96 - (3.39 + x) 28 r-squared = 0.58, p > 0.05 ine groundwater discharge, agricultural development, etc.) and 26 have the potential to deliver N to coastal environments on a local scale and large scale, two analyses were conducted. First, to deter- 24 mine if inputs were related to fine-scale location, we statistically 22 examined the geographic affinity of total N content and d15N val- 20

ues for the two reef algae collected at sites 1–10. In these fine-scale N (‰) 18 15 analyses, separate statistical analyses were performed for total N δ 16 15 content and d N values for each species in both seasons. Using 14 Euclidean distance, a similarity matrix was constructed between 12 each value at every site. Second a separate, similarity matrix of geographical distance between sites also was constructed. RELATE 10 in PRIMER-E (Clarke and Warwick, 2001), a Mantel-like test, was 8 -101234 used to examine the statistical relationship between the two types ln NO3- of similarity matrices for both species in both dry and rainy season. RELATE performs Spearman Rank correlation to determine Rho Fig. 2. The non-linear relationship between seawater concentration and d15N with 999 permutations and this is used to determine statistical values (nitrate + nitrite) suggesting an open system with mixing of multiple sources significance. with different d15N values. 96 T. Erin Cox et al. / Marine Pollution Bulletin 71 (2013) 92–100

Table 1 Phosphate, Silicate, Nitrite, Nitrate, and Ammonia as ppm values and d15N values from water samples collected inshore at Sites 1–10, storm-drain grates, and well. n = 1 unless otherwise specified, n > 1 values are expressed as mean ± SE. Values with an are below detectable values, and are samples collected on the same rain event.

Site name Dry season (lM) Rainy season (lM) d15N(‰) dry season 3 SiO NO NO þ NO NH 3 SiO NO NO þ NO NH PO4 2 2 2 3 3 PO4 2 2 2 3 3 Control 1 – – – – – 0.10 7.9 0.20 0.96 1.18 – Control/proposed Drain 2 – – – – – 0.09 12.7 0.38 2.53 2.02 – Control 3 0.15 10.0 0.22 1.03 0.81 0.10 7.0 0.29 1.53 0.94 10.7 Drain 4 0.23 10.9 0.18 2.88 2.71 0.12 6.6 0.29 2.57 1.86 15.9 Control 5 0.15 3.3 0.18 0.72 0.54 0.07 8.2 0.26 1.29 0.34 13.2 Drain 6 0.07 16.5 0.14 7.93 7.79 0.07 13.5 0.37 1.71 0.87 15.2 Control 7 – – – – – – – – – – 9.0 Drain 8 0.08 3.7 0.20 1.02 0.82 0.08 3.7 0.20 0.82 3.10 – Drain 9 0.09 2.6 0.13 0.59 0.46 – – – – - 14.0 Control 10 – – – – – 0.09 12.7 0.38 2.53 2.02 – n 3 SiO NO NO +NO NH n d15N PO4 2 2 2 3 3 Drain water near 4 1 36.2 159.0 6.8 37.4 32.9 1 7.9 Drain water near 6 1 17.5 47.2 5.2 128.0 26.5 1 7.7 Drain water near 8 – – – – – – 1 10.0 Standing water in Drain 9 1 1.4 515.6 3.1 99.2 15.6 1 19.7 Ocean water at 3 1 0.2 4.4 0.4 2.4 3.3 1 22.3 Ocean water at One‘ula 1 – – – – – 1 20.2 Well water, proxy for groundwater 1 2.5 ± 0.1 574.2 ± 20.0 0.8 ± 0.0 23.5 ± 0.4 10.1 ± 0.0 1 27.9 Deep ocean (Casciotti et al., 2008) 7.0–7.1 Open ocean (Casciotti et al., 2008) 2.4–3.0

Table 2 Thalli mean ± SE Total Nitrogen as % dry weight and d15N values for A. spicifera and L. nidifica in dry and rainy seasons at sites 1–10.

Site name Species Dry season Rainy season n Total N (%) d15N n Total N (%) d15N(‰) Control 1 A. spicifera 5 2.6 ± 0.1 14.0 ± 0.5 3 1.6 ± 0.1 8.2 ± 0.7 L. nidifica 2 2.6 ± 0.3 14.7 ± 0.9 3 1.6 ± 0.1 9.5 ± 1.1 Control/proposed drain 2 A. spicifera 6 2.1 ± 0.1 11.6 ± 0.2 1 2.3 10.8 L. nidifica 4 2.1 ± 0.1 12.3 ± 0.1 4 2.1 ± 0.4 11.4 ± 0.3 Control 3 A. spicifera 5 1.8 ± 0.1 6.8 ± 0.1 3 1.8 ± 0.1 7.0 ± 0.0 L. nidifica 4 2.1 ± 0.1 6.9 ± 0.1 4 2.0 ± 0.2 7.3 ± 0.1 Drain 4 A. spicifera 5 2.5 ± 0.2 12.2 ± 0.4 5 2.5 ± 0.1 12.0 ± 0.4 L. nidifica 5 2.6 ± 0.2 13.3 ± 0.7 4 2.1 ± 0.3 12.0 ± 0.7 Control 5 A. spicifera 5 2.1 ± 0.1 9.2 ± 0.2 7 1.9 ± 0.2 8.5 ± 0.1 L. nidifica 7 2.1 ± 0.2 9.5 ± 0.2 6 2.0 ± 0.2 8.7 ± 0.1 Drain 6 A. spicifera 8 2.7 ± 0.1 11.3 ± 0.4 8 2.4 ± 0.1 10.5 ± 0.2 L. nidifica 6 2.5 ± 0.2 11.9 ± 0.2 6 2.5 ± 0.1 11.3 ± 0.3 Control 7 A. spicifera 6 1.8 ± 0.1 11.4 ± 0.4 5 1.7 ± 0.1 12.5 ± 0.7 L. nidifica 4 1.5 ± 0.3 11.0 ± 0.5 – – – Drain 8 A. spicifera 6 1.7 ± 0.2 12.4 ± 1.5 5 2.3 ± 0.1 10.4 ± 1.5 L. nidifica 1 1.3 8.8 – – – Drain 9 A. spicifera 6 1.1 ± 0.1 7.5 ± 0.5 5 1.2 ± 0.0 5.3 ± 0.5 L. nidifica –– – –– – Control 10 A. spicifera 6 1.1 ± 0.0 4.8 ± 0.3 5 1.4 ± 0.1 5.3 ± 0.3 L. nidifica –– – –– –

Although the d15N values and N contents of A. spicifera differed controls, p-value = 0.40, drain x season p-value = 0.53). In addition, among sites, differences were not significant between drain and N contents of L. nidifica did not vary between dry and rainy seasons control sites and did not differ between season (Table 2)(d15N: (Table 2)(p-value = 0.98). Two-Way ANOVA, drain vs control p-value = 0.45, season A. spicifera and L. nidifica at several drain and control sites had p-value = 0.41, drain season p-value = 0.89% N: Two-way ANOVA, higher d15N values nearshore with lower values further from shore. drain vs control p = 0.42, season p = 0.92, drain season p = 0.63). This offshore gradient was seen in both seasons (Fig. 3). For nine of Average values of d15N and% N in L. nidifica from sites 1–8 were 10 sites, distance from shore was a significant factor in a multiple similar to those from A. spicifera at sites 1–10 (Table 2). The d15N regression model used to determine d15N values of A. spicifera values of L. nidifica varied among sites but differences were not sig- (Table 3). Values of d15N tended to be lower in the rainy season nificant (two-way ANOVA) between drain and control sites and for most sites did not alter the slope of the relationship (p-value = 0.21), rainy and dry season (p-value = 0.45), or for the between d15N values and distance from shore. However, an inter- interaction of site type (drain or control) and season action of distance and season was observed in samples of A. spicif- (p-value = 0.33). Although N content of L. nidifica at Drains 4 and era at three sites: Control 1, Control 3, and Control 5 (Table 3). 6 was higher in both seasons, the values at these sites were statis- The d15N values of L. nidifica differed at some sites with distance tically similar to Controls 1, 2, 3, 5, 7 and similar to the sample col- from shore and varied for some sites with season (Fig. 3). Three out lected in the dry season at Drain 8 (Two-way ANOVA, drain vs. of six sites sampled showed significant variation in d15N values of T. Erin Cox et al. / Marine Pollution Bulletin 71 (2013) 92–100 97

20 C1 C/PD2 C3 D4 C5 15

10

5 ) 00 / 0 0

N ( 20

15 D6 C7 D8 D9 C10 δ 15

10

5

0 0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25 Distance from shore (meters)

Fig. 3. d15N(‰) values of A. spicifera N and L. nidifica d in the dry (closed) and rainy (open) season as distance from shore (meters) for sites.

Table 3 Multiple linear regressions show the relationship between d15NofA. spicifera and L. nidifica with distance from shore, season, and the interaction of terms at drain (D) and control (C) sites (rel = direction of relationship, NS = not significant at a = 0.05, NA = not applicable).

Site Acanthophora spicifera Laurencia nidifica

Terms rel df Fp-Value R2 (100) Terms rel df F p-Value R2 (100) C1 Distance, p = 0.010 7 27.8 0.004 95.4 Distance, NS 4 11.3 0.043 79.4 Season, NS – Season, p = 0.043 Distance Season, p = 0.011 + Distance Season, NS C2 Distance, NS 5 0.6 0.488 12.7 Distance, NS 7 1.1 0.433 39.6 Season, NA Season, NS Distance Season, NA Distance Season, NS C3 Distance, p = 0.032 8 20.6 0.085 70.6 Distance, NS 8 9.3 0.021 61.6 Season, NS Season, p = 0.021 + Distance Season, p = 0.041 + Distance Season, NS D4 Distance, p = 0.001 9 22.4 0.001 73.6 Distance, p = 0.002 8 37.1 <0.001 95.7 Season, NS Season, NS Distance Season, NS + Distance Season, p = 0.019 + C5 Distance, p = 0.010 11 6.2 <0.001 85.9 Distance, p = 0.003 12 21.4 <0.001 81.1 Season, p = <0.001 Season, p = 0.045 Distance Season, p = 0.016 + Distance Season, NS D6 Distance, p = 0.017 15 14.3 <0.001 68.8 Distance, p = 0.021 11 9.4 0.021 48.4 Season, p = <0.001 Season, NS Distance Season, NS Distance Season, NS C7 Distance, p = 0.046 10 5.30 0.046 37.4 NA Season, NS Distance Season, NS D8 Distance, p = <0.001 10 18.0 0.001 81.8 NA Season, p = 0.009 Distance Season, NS D9 Distance, p = < 0.040 10 10.8 0.005 73.0 NA Season, p = 0.002 Distance Season, NS C10 Distance, p = 0.044 10 2.7 0.128 53.4 NA Season, NS Distance Season, NS

L. nidifica with distance from the shoreline. Distance was a signifi- related to geographic location for the dry season (RELATE test:, cant predictor of d15N values at Drain 4, Control 5, and Drain 6. For Rho = 18.9%, p-value = 0.02), but not in the rainy season these sites, the d15N values were lower with distance from the (Rho = 3.5%, p-value = 0.20). shoreline. Season was a significant predictor of L. nidifica d15N val- Total N content of A. spicifera was related to fine-scale geo- ues for Control 1, Control 3, and Control 5. At two of these sites, graphic location where collection occurred but for L. nidifica, sam- values were lower during the rainy season while Control 3 values pled only at sites 1–8, the relationship was not evident. There was of d15N were higher in the rainy season. Drain 4 was the only a significant relationship between the similarity matrices of geo- regression model that had a significant interaction between dis- graphic location and total N content in dry (Rho = 36.4%, tance from shore and season (Table 3). p = 0.001) and rainy (Rho = 36.3%, p-value = 0.001) seasons for A. Significant relationships were found between geographic loca- spicifera. However there was no relationship between geographic tion and d15N values of A. spicifera for both seasons (RELATE: dry location and N content for either season for L. nidifica (RELATE: season, Rho = 80.9%, p-value = 0.001; rainy season, Rho = 65.4%, dry season, Rho = 4.0%, p-value = 0.69; wet season, Rho = 3.0%, p-value = .001). The d15N values of L. nidifica were significantly p-value = 0.29). 98 T. Erin Cox et al. / Marine Pollution Bulletin 71 (2013) 92–100

δ15N (‰) Laniakea 1.0 - 4.9 10.1 5.0 - 8.9 6.1

9.0 - 12.9 Kaena Point

>13.0 1.9 Species A. spicifera A. breviarticulatum L. mcdermidiae U. lactuca Kane‘ohe Bay Wai‘anae Coast 9.8 Maili-Kahe

Makapu‘u 11.9

9.6 6.1 8.4 3.8 2.7

N 16.8 9.7 km 7.3 7.7

One‘ula, Control 2, Drain 4 3.1 2.6 Diamond Head

Fig. 4. The island of O‘ahu showing mean d15N values (‰, n = 1–7) of reef algae (A. spicifera, A. breviarticulatum, Laurencia mcdermidae, Laurencia nidifica, and Ulva lactuca) collected along the coast in shallow waters in March 2008 (gray box) and October 2011.

3.2. Island wide d15N value comparison Storm-drain effluent alone does not dominate the input of nutrients supporting nearshore algal growth in the ‘Ewa Plain re- The mean d15N values of reef algae sampled at sites around gion studied. This conclusion is based on (1) the N content and O‘ahu ranged from 1.9‰ to 16.8‰; reef algae collected in 2011 d15N values in algae, which did not vary in a predictable manner from the ‘Ewa Beach area fell within the upper end of this range with drain presence (2) the nutrient concentrations in drain-water (7.3–16.8‰)(Fig. 4). Reef algae with the lowest mean d15N value that differed from concentrations measured in seawater (3) the of 1.9‰ were collected at Ka‘ena Point and the highest mean distinct differences in d15N values between algae collected from d15N value was determined from algal samples collected along drain sites and the d15N values of nitrate + nitrite measured in the ‘Ewa Beach area. The d15N values of A. spicifera collected at Con- storm-drain effluent and (4) the relatively consistent d15N values trol 2 and Drain 4 differed between 2011 and 2007. In 2007, A. spi- regardless of season and rainfall amounts (see online materials). cifera from Control 2 during the dry and rainy season had mean The latter is particularly revealing for a fast-growing species which values of 11.6‰ and 10.8‰ (respectively) but in 2011 had a lower likely uses all available substrate quickly and depletes internal N mean value of 7.7‰. The mean d15N values of A. spicifera at Drain 4 stores. Furthermore, algae located at Drain 9 which services the increased from 2007 to 2011 from 12.0‰ to 12.2‰ to 16.8‰. largest drainage area within the study region had the lowest values of d15N (4.8‰), a value that in other systems is more indicative of 4. Discussion non-anthropogenic sources (Garrison et al., 2007) or anthropo- genic sources that derived N from atmospheric sources (Kendall, The results of this study do not support the hypothesis that ni- 1998). trate in storm-drain effluent dominated the input of nutrients for Our findings are contrary to the conclusions drawn in previous nearshore algal growth over a several year evaluation with two investigations (Derse et al., 2007; Lapointe and Bedford, 2011). The species and end member water sources included in the analyses. discrepancy can be explained by differences in sampling design 15 Instead results suggested that reef algae at ‘Ewa Beach incorporate and interpretation of high d N values in algae. Assuming all avail- 15 N from a mixture of sources that cannot be readily identified, able nitrogenous nutrients are used by algae, the d N values of al- requiring more thorough investigation that goes beyond the typi- gae can provide information about nutrient sources particularly 15 cal approaches used for the measurements of d15N values in algae. when compared to d N values (nitrate + nitrite) of potential N The N content and d15N values in algae were often higher nearest sources (Peterson and Fry, 1987; Kendall, 1998; Robinson, 2001). to shore, tied to geographic location, and were similar or above val- However, elevated N microbial processing in an aquifer can lead 15 ues for algae collected from other nearshore environments. These to high d N values of nitrate (e.g., Lindsey et al., 2003) similar to findings support the conclusion that the nitrogenous nutrients values that often indicate anthropogenic enrichment (Dailer used by these plants are derived predominantly from terrestrial et al., 2012a). For this reason, a suite of additional indicators or sources that may more closely model usual coastal development. incorporation of spatial arrangement and temporal changes of al- 15 In areas with coastal development nutrients are predicted to be gae d N values can aid in interpretations (see Gartner et al., anthropogenic in origin, and undergo N cycling. These results are 2002; Lapointe et al., 2005; Cohen and Fong, 2005; Lin and Fong, in marked contrast to Dailer et al. (2010) but those Maui sites ran- 2008; Dailer et al., 2012a). 15 ged from near pristine to sites with focused delivery of up to Despite the differences between studies, elevated d N values 5 M gal d1 of wastewater, nutrient loading not observed on O‘ahu. (>7‰) and N contents (>2%) in both species of macroalgae in the T. Erin Cox et al. / Marine Pollution Bulletin 71 (2013) 92–100 99

‘Ewa Beach area fell above the range commonly cited for animal Peterson et al., 2007, 2009; Johnson et al., 2008; Street et al., waste (Kendall, 1998; Macko and Ostrom, 1994) but within the 2008; Peterson et al. 2009). Further investigations are needed to range from studies in other locations that investigated anthropo- identify if d15N values that occur along ‘Ewa Beach are associated genic enrichment in tropical and subtropical waters. Published val- with seep locations and studies including quantification of excess 15 ues of d N of various warm water macroalgae range from +1.0‰ N2 gas in groundwater (Lindsey et al., 2003) are needed to distin- to +5.5‰ in the Florida Keys (interpreted as being indicative of guish whether high d15N values in reef algae result from the incor- agricultural runoff, Lapointe et al., 2004), from +5.7‰ to +12.0‰ poration of a land based anthropogenic or natural nitrogenous in waste water influenced locations of South Florida (Lapointe sources that have been modified significantly by microbial reaction et al., 2005), from +7.7‰ to +11.4‰ in southeastern Gulf of Califor- in aquifers. nia (plausible sources included sewage, agriculture, and shrimp In conclusion, our study revealed a complex pattern of nutrient farms, Piñón-Gimate et al., 2009), a mean of 0.5‰ in Hanalei sources that is most congruent with localized delivery of terrestrial Bay, Kaua‘i island, Hawai‘i (concluded to be indicative of synthetic N source (s) for much of the reef algae along the coast that pro- fertilizers from nearby Resorts, Derse et al., 2007), from +2‰ to duces an onshore–offshore trend. No drain effect was evident +8‰ in Ishigaki, Japan (Umezawa et al., 2002), and +1.0‰ to but, we caution that results are not necessarily applicable to other +4.8‰ in Ofu, Samoa (concluded to be indicative of non-anthropo- areas. For instance, the effects of storm-drain effluent on bloom- genic sources, Garrison et al., 2007). Nonetheless, we cannot distin- forming species could have a strong impact in more closed, oligo- guish unequivocally whether the elevated d15N values (>7‰) trophic waters. The results of this study should be considered in observed for Ewa Beach macroalgae result from microbial cycling the design of future assessment schemes using similar techniques, of natural or anthropogenic N sources in soils or groundwater in decisions about how to manage water flow, and can be used as aquifers. baseline to measure future alterations to nutrient inputs in ‘Ewa The general enrichment in 15N contents of algae collected from Beach. The study also highlights the value of considering physio- ‘Ewa Beach shorelines relative to many other sites around O‘ahu logical differences among the algae sampled, the spatial-temporal suggests that the N source in this area could be anthropogenic in patterns of their d15N values and the biogeochemical environment origin. Recent studies by Dailer et al. (2010) on Maui have shown where they occur. remarkably high macroalgal d15N values (>40‰) for field collected materials, well above those recorded in this study for algal material Acknowledgments collected from O‘ahu. In that study, there was a single clear anthro- pogenic N source that could be linked to a suite of sewage injection We would like to acknowledge K. Boyle, P. Buxton, M. Lurie, E. wells using multiple tracers including elevated N and phosphorous Donham, S. V. Gent, M. nGiriou, and M. Kawachi for their assistance levels and high concentrations of pharmaceuticals (Hunt and Rosa, in collecting or processing samples for this study. Also we thank 1 2009). These wells deliver up to 5 M gal d of partially denitrified the staff of University of Hawai‘is Stable Isotope Biogeochemistry primary treated sewage effluent into groundwater (Dailer et al., Laboratory for guidance in stable isotope analyses. The study was 2010, 2012a). In this study, no single signal was detected. Further, supported by Hasekeko Development, Inc. We also acknowledge there is no obvious relationship between elevated marine macroal- the State of Hawai‘i, Division Aquatic Resources for permitting us 15 gal d N values and proximity to known on-site sewage disposal to collect inside the ‘Ewa Beach Limu Management Area. This is SO- systems (Whittier and El-Kadi, 2009). EST contribution #8904. Our d15N values of reef algae were often related to the fine- scale, geographic location where collection occurred. Average d15N values of reef algae varied among sites from 4.8‰ to 14.7‰, Appendix A. Supplementary material a difference of nearly 10‰ that was greater than the largest varia- tion between seasons (5.8‰ difference), the two species at a site Supplementary data associated with this article can be found, in within a season (1.3‰ difference), and individuals (3.8‰ differ- the online version, at http://dx.doi.org/10.1016/j.marpolbul.2013. ence) at a site within a season. Furthermore, this variation occurred 03.030. over a small geographic area and was also observed along shore- lines in other areas of O‘ahu (e.g., Wai‘anae Coast). The geographic References affinity of d15N values and %N of reef algae reveal that the N sources in these areas act on a local scale. However, obvious point Abbott, I.A., 1996. Limu, An Ethnobotanical Study of Some Hawaiian Seaweeds, sources of N were difficult to identify. fourth ed. National Tropical Botanical Garden, Lawai, Kauai, Hawai‘i. Banner, A.H., 1974. Kane‘ohe Bay, Hawaii: Urban pollution and a coral reef Groundwater seeps are thought to occur along the coast in ‘Ewa ecosystem. In: Proc. 2nd Int. Coral Reef Symp. vol. 2, pp. 685-702. Beach (Spengler et al., 1998; Laws et al., 1999). Groundwater can Bernardo, R., 2008. Limu Delays Project to Ease Ewa Flooding. Star Bulletin, deliver substantial amounts of nutrients into coastal systems Honolulu. . 15 Casciotti, K.L., Trull, T.W., Glover, D.M., Davies, D., 2008. Constraints on nitrogen (Dulaiova et al., 2008; Taniguchi et al., 2008). The very high d N cycling at the subtropical North Pacific Station ALOHA from isotopic value for nitrate + nitrite in our single groundwater sample sug- measurements of nitrate and particulate nitrogen. Deep-Sea Res. II 55, 1661– gests that the surrounding aquifer is likely to have an active com- 1672. Choi, W.-J., Han, G.-H., Lee, S.-M., Lee, G.-T., Yoon, K.-S., Choi, S.-M., Ro, H.-M., 2007. munity of denitryfing bacteria resulting in partial denitrification of Impact of land-use types on nitrate concentration and d15N in unconfined the available nitrate prior to being discharged into nearshore groundwater in rural areas of Korea. Agric. Ecosyst. Environ. 120, 259–268. waters. The delivery of terrestrial N could possibly occur through Clarke, K.R., Warwick, R.M., 2001. Changes in Marine Communities: An Approach to localized groundwater seeps mixing with other sources with lower Statistical Analysis and Interpretation. Primer-E Ltd., Plymouth. Cohen, R.A., Fong, P., 2005. Experimental evidence supports the use of d15N content N isotopic values. The locations of these seeps are not well docu- of the opportunistic green macroalga Enteromorpha intestinalis (Chlorophyta) to mented for ‘Ewa Beach but, are plausible given there are similar determine nitrogen sources to estuaries. J. Phycol. 41, 287–293. seeps documented elsewhere in the state of Hawai‘i. For instance, Cole, M.L., 2003. Detection of Eutrophication in Aquatic Ecosystems: Nitrogen Stable Isotopes in Macrophytes and Groundwater. Boston University, Boston, seeps occur along Kahana Bay located along the north shore of Ph.D Dissertation. O‘ahu (Garrison et al., 2003) and Tomlinson et al. (2011) docu- Dailer, M.L., Knox, R.S., Smith, J.E., Napier, M., Smith, C.M., 2010. Using d15N values in mented persistent negative salinity anomalies after rainfall events algal tissue to map locations and potential sources of anthropogenic nutrient inputs on the island of Maui, Hawai‘i, USA. Mar. Pollut. Bull. 60, 655–671. offshore from urban Honolulu. In addition, there are many well Dailer, M.L., Ramey, H.L., Saephan, S., Smith, C.M., 2012a. Algal d15N values detect a known groundwater seeps off the island of Hawaii (Hunt, 2006; wastewater effluent plume in nearshore and offshore surface waters and three- 100 T. Erin Cox et al. / Marine Pollution Bulletin 71 (2013) 92–100

dimensionally model the plume across a coral reef on Maui, Hawai‘i, USA. Mar. Owens, N.P.J., 1987. Natural variations in d15N in the marine environment. Adv. Mar. Pollut. Bull. 64, 207–213. Biol. 24, 389–451. Dailer, M.L., Smith, J.E., Smith, C.M., 2012b. Responses of bloom forming and non- Peterson, B.J., Fry, B., 1987. Stable isotopes in ecosystem studies. Ann. Rev. Ecol. bloom forming macroalgae to nutrient enrichment in Hawai‘i, USA. Harmful Syst. 18, 293–320. Algae 17, 111–125. Peterson, R.N., Burnett, W.C., Glenn, C.R., Johnson, A.J., 2007. A box model to Derse, E., Knee, K.L., Wankel, S.D., Kendall, C., Berg, C., Paytan, A., 2007. Identifying quantify groundwater discharge along the kona coast of hawaii using natural sources of nitrogen to Hanalei Bay, Kauai utilizing the nitrogen isotope tracers. In: Sanford, W., Langevin, C., Polemio, M., Povinec, P. (Eds.), A New Focus signature of macroalgae. Environ. Sci. Technol. 41, 5217–5223. on Groundwater–Seawater Interactions. IAHS Press, pp. 142–149. Doty, M.S., 1961. Acanthophora, a possible invader of the marine flora of Hawaii. Pac. Peterson, R.N., Burnett, W.C., Glenn, C.R., Johnson, A.G., 2009. Quantification of Sci. 15, 547–552. point-source groundwater discharges to the ocean from the shoreline of the Big Dulaiova, H., Gonneea, M.E., Henderson, P.B., Charette, M.A., 2008. Geochemical and Island, Hawaii. Limnol. Oceanogr. 54, 890–904. physical sources of radon variation in a subterranean estuary – implications for Piñón-Gimate, A., Soto-Jiménez, M., Ochoa-Izaguirre, M.J., García-Pagés, E., Páez- radon groundwater end-member activities in submarine groundwater Osuna, F., 2009. Macroalgae blooms and d15N in subtropical coastal lagoons discharge studies. Mar. Chem. 110, 120–127. from the Southeastern Gulf of California: Discrimination among agricultural, Fong, P., Kamer, K., Boyer, K.E., Boyle, K.A., 2001. Nutrient content of macroalgae shrimp farm and sewage effiuents. Mar. Pollut. Bull. 58, 1144–1151. with differing morphologies may indicate sources of nutrients for tropical Robinson, D., 2001. D15N as an integrator of the nitrogen cycle. Trends Ecol. Evol. 16, marine systems. Mar. Ecol. Prog. Ser. 220, 137–152. 153–162. Fong, P., Boyer, K.E., Kamer, K., Boyle, K.A., 2003. Influence of initial tissue nutrient Russell, D.J., 1992. The ecological invasion of Hawaiian reefs by two marine red status of tropical marine algae on response to nitrogen and phosphorus algae, Acanthophora spicifera (Vahl) Boerg. and Hypnea musciformis (Wulfen) J. additions. Mar. Ecol. Prog. Ser. 262, 111–123. Ag., and their association with two native species, Laurencia nidifica J. Ag and Garrison, G.H., Glenn, C.R., McMurtry, G.M., 2003. Measurement of submarine Hypnea cervicornis. J. Agric. ICES Mar. Sci. Symp. 194, 110–125. groundwater discharge in Kahana Bay, O‘ahu, Hawai‘i. Limnol. Oceanogr. 48, Sammarco, P.W., Risk, M.J., Schwarcz, H.P., Heikoop, J.M., 1999. Cross-continental 920–928. shelf trends in coral d15N on the Great Barrier Reef: further consideration of the Garrison, V., Kroeger, K., Fenner, D., Craig, P., 2007. Identifying nutrient sources to reef nutrient paradox. Mar. Ecol. Prog. Ser. 180, 131–138. three lagoons at Ofu and Olosega, American Samoa using d15N of benthic Schaefers, A., 2006. Ocean Pointe Drainage Plan Under Attack. Star Bulletin, macroalgae. Mar. Pollut. Bull. 54, 1813–1838. Honolulu. . forms of macroalgae and filter-feeders to reveal temporal and spatial patterns Schaffelke, B., 1999. Short-term nutrient pulses as tools to assess responses of coral in sewage dispersal. Mar. Ecol. Prog. Ser. 235, 63–73. reef macroalgae to enhanced nutrient availability. Mar. Ecol. Prog. Ser. 182, Hiyakumoto, G. (R.M. Towill Corporation), 2012. Estimate Rainfall Depth for Area 305–310. Makai of Papipi Road. Prepared for Haseko (Ewa) Inc. Sigman, D.M., Casciotti, K.L., Andreani, M., Barford, C., Galanter, M., Bohlke, J.K., Hunt Jr., C.D., 2006. Ground-Water Nutrient Flux to Coastal Waters and Numerical 2001. A bacterial method for the nitrogen isotopic analysis of nitrate in Simulation of Wastewater Injection at Kihei, Maui, Hawaii. Sci. Invest. Rpt., seawater and freshwater. Anal. Chem. 73, 4145–4153. 2006–5283. Smith, J.E., Hunter, C.L., Smith, C.M., 2002. Distribution and reproductive Hunt Jr., C.D., Rosa, S.N., 2009. A multi-tracer approach to detecting wastewater characteristics of nonindigenous and invasive marine algae in the Hawaiian plumes from municipal injection wells in nearshore marine waters at Kihei and Islands. Pac. Sci. 56, 299–315. Lahaina, Maui, Hawaii. USGS. Sci. Invest. Rep. 2009–5253, 1–166. Smith, J.E., Hunter, C.L., Conklin, E.J., Most, R., Sauvage, T., Squair, C., Smith, C.M., Johnson, A.G., Glenn, C.R., Burnett, W.C., Peterson, R.N., Lucey, P.G., 2008. Aerial 2004. Ecology of the invasive red alga Gracilaria salicornia (Rhodophyta) on infrared imaging reveals large nutrient-rich groundwater inputs to the ocean. Oahu, Hawaii. Pac. Sci. 58, 325–343. Geophys. Res. Lett. 35, 1944–8007. Smith, J.E., Runcie, J.W., Smith, C.M., 2005. Characterization of a large-scale Kendall, C., 1998. Tracing nitrogen sources and cycling in catchments. In: Isotope ephemeral bloom of the green alga Cladophora sericea on the coral reefs of Tracers in Catchment Hydrology. Elsevier, Amesterdam, pp. 519–576. West Maui, Hawaii. Mar. Ecol. Prog. Ser. 302, 77–91. Lapointe, B.E., Bedford, B.J., 2011. Stormwater nutrient inputs favor growth of non- Spengler, S.R., Yamauchi, R.S.W., Pabingwit, B.M.B. 1998. Evaluating the native macroalgae (Rhodophyta) on O‘ahu, Hawaiian Islands. Harmful Algae 10, Environmental Impact from Injection of Treated Wastewater in a Coastal 310–318. Aquifer, City and County of Honolulu, Department of Environmental Services. Lapointe, B.E., Barile, P.J., Matzie, W.R., 2004. Anthropogenic nutrient enrichment of Stimson, J., Larned, S.T., Conklin, E.J., 2001. Effects of herbivory, nutrient levels, and seagrass and coral reef communities in the lower florida keys: discrimination of introduced algae on the distribution and abundance of the invasive macroalga local versus regional nitrogen sources. J. Exp. Mar. Biol. Ecol. 308, 23–58. Dictyosphaeria cavernosa in Kaneohe Bay, Hawaii. Coral Reefs 19, 343–357. Lapointe, B.E., Barile, P.J., Littler, M.M., Littler, D.S., 2005. Macroalgal blooms on Street, J.H., Knee, K.L., Grossman, E.E., Paytan, A., 2008. Submarine groundwater southeast Florida coral reefs: II. Cross-shelf discrimination of nitrogen sources discharge and nutrient addition to the coastal zone and coral reefs of leeward indicates widespread assimilation of sewage nitrogen. Harmful Algae 4, 1106– Hawai‘i. Mar. Chem. 109, 355–376. 1122. Taniguchi, M., Burnett, W.C., Dulaiova, H., Siringan, F., Foronda, J., Wattayakorn, G., Larned, S.T., 1998. Nitrogen-versus phosphorus-limited growth and sources of Rungsupa, S., Kontar, E.A., Ishitobi, T., 2008. Groundwater discharge as an nutrients for coral reef macroalgae. Mar. Biol. 132, 409–421. important land–sea pathway into Manila Bay, Philippines. J. Coastal Res. 24, 15– Laws, E., Ziemann, D., Schuman, D., 1999. Coastal water quality in Hawaii: the 24. importance of buffer zones and dilution. Mar. Environ. Res. 48, 1–21. Teichberg, M., Fox, S.E., Aguila, C., Olsen, Y.S., Valiela, I., 2008. Macroalgal responses Leone, D., 2004. Ewa Ban on Limu Picking Proposed. Honolulu Star Bulletin, to experimental nutrient enrichment in shallow coastal waters: growth, Honolulu. . internal nutrient pools, and isotopic signatures. Mar. Ecol. Prog. Ser. 368, Lin, D.T., Fong, P., 2008. Macroalgal bioindicators (growth, tissue N, d15N) detect 117–126. nutrient enrichment from shrimp farm effluent entering Opunohu Bay, Moorea, Tomlinson, M.S., De Carlo, E.H., McManus, M.A., Pawlak, G., Steward, G.F., Sansone, French Polynesia. Mar. Pollut. Bull. 56, 245–249. F.J., Nigro, O.D., Timmerman, R.E., Patterson, J., Jaramillo, S., Ostrander, C.E., Lin, H.-J., Wu, C.-Y., Kao, S.-J., Kao, W.-Y., Meng, P.-J., 2007. Mapping anthropogenic 2011. Characterizing the effects of two storms on the coastal waters of O‘ahu, nitrogen through point sources in coral reefs using d15N in macroalgae. Mar. Hawai‘i, using data from the Pacific Islands Ocean Observing System. Ecol. Prog. Ser. 335, 95–109. Oceanographic 24, 182–199. Lindsey, B.D., Phillips, S.W., Donnelly, C.A., Speiran, G.K., Plaummer, L.N., Bohkle, J.- Umezawa, Y., Miyajima, T., Yamamuro, M., Kayanne, H., Koike, I., 2002. Fine-scale K., Focazio, M.J., Burton, W.C., Busenberg, E., 2003. Residence Times and Nitrate mapping of land-derived nitrogen in coral reefs by d15N in macroalgae. Limnol. Transport in Ground Water Discharging to Streams in the Chesapeake Bay Oceanogr. 47, 1405–1416. Watershed. USGS Water-Resources Investigations, Report 03-4035. Umezawa, Y., Miyajima, T., Tanaka, Y., Hayashibara, T., Koike, I., 2007. Variation of Macko, S.A., Ostrom, N.E., 1994. Stable Isotopes in Ecology and Environmental internal d15N and d13C distribution and their bulk values in a brown macroalgae. Science. Blackwell Scientific Publications, Oxford. J. Phycol. 43, 437–448. McCook, L.J., 1999. Macroalgae, nutrients and phase shifts on coral reefs: scientific Valiela, I., Foreman, K., LaMontagne, K., Hersh, D., Costa, J., Peckol, P., Demeo- issues and management consequences for the Great Barrier Reef. Coral Reefs 18, Andreson, B., D’Avanzo, C., Babione, M., Sham, C., Brawley, J., Lajtha, K., 1992. 357–367. Couplings of watersheds and coastal waters. Sources and consequences of Mink, J., 1989. Groundwater Resources and Sustainable Yield Ewa Plain Caprock nutrient enrichment in Waquoit Bay, Massachusetts. Estuaries 15, 443–457. Aquifer, Oahu, Hawaii. Prepared by George Yuen and Associates, Commission on Van Houtan, K.S., Hargrove, S.K., Balaz, G.H., 2010. Land use, macroalgae, and a Water Resource Management, State of Hawaii, Report 1-79. tumor-forming disease in marine turtles. PLoS ONE 5, 1–9. Ohira, R., 2005. Hau Bush Reunion Evokes Memories of Ewa Beach’s Jewel. Honolulu Walsh, C.J., 2000. Urban impacts on the ecology of receiving waters: a framework Advertiser, Honolulu. . Whittier R.B., El-Kadi, A., 2009. Human and Environmental Risk Ranking of Onsite Otkin, J.A., Martin, J.E., 2004. A Synoptic Climatology of the Subtropical Kona Storm. Sewage Disposal Systems, State of Hawaii, Department of Health Safe Drinking Monthly Weather Review 132, pp. 1502–1517. Water Branch, Honolulu, pp. 1–7. Marine Environmental Research 48 (1999) 1±21

Coastal water quality in Hawaii: the importance of bu€er zones and dilution E.A. Laws a,*, D. Ziemann b, D. Schulman a aUniversity of Hawaii, Department of Oceanography, 1000 Pope Road, Honolulu, HI 96822-2285, USA bOceanic Institute, 41-202 Kalanianaole Highway, Waimanalo, HI 96795-1898, USA

Received 20 June 1998; received in revised form 10 December 1998; accepted 27 December 1998

Abstract

A study of the relationship between point and nonpoint source freshwater discharges and marine water quality were studied during a period of 1 year in Mamala Bay, a coastal inden- tation on the south shore of the island of Oahu, Hawaiian Islands. Despite the fact that 100± 300Â106 m3 year1 of land runo€/groundwater seepage and 150Â106 m3 year1 of treated sewage e‚uent enter Mamala Bay and its tributaries, coastal water quality as judged by stand- ard chemical and physical parameters is high at virtually all locations in the bay. The expla- nation for the high water quality re¯ects several important factors. First, much of the nonpoint source discharge enters either estuaries or harbors, which function as bu€er zones by trapping some of the sediment and nutrients that would otherwise enter the coastal ocean. Second, the principal point source discharges are located in water suciently deep that their wastewater plumes are trapped below the surface most of the time. When the plumes surface they are suciently diluted that their impact on parameters, such as nutrient concentrations, is unde- tectable. Third, the coastal current system is greatly diluted by exchange with the o€shore ocean. Based on a simple box model, the degree of mixing with the o€shore ocean is roughly 40 times the rate of input of fresh water from point and nonpoint sources. The o€shore waste- water outfalls have no discernible e€ect on water quality at any recreational beach along the shoreline. The principal impact on water quality at the recreational beaches comes from non- point source discharges, and with the exception of one beach located directly adjacent to a stream mouth, that impact is on the composition rather than the concentration of the plank- ton. There is a systematic shift from a chlorophyte- to a diatom-dominated phytoplankton community due to the high silicate concentration in groundwater and land runo€, and there is a systematic increase in the d15N of suspended particles due to the high d15N of the biologically available nitrogen in groundwater seepage. # 1999 Elsevier Science Ltd. All rights reserved. Keywords: Bu€er zone; Coastal zone; Dilution; Estuaries; Mixing processes; Ocean disposal; Outfalls; Recreational waters; Sewage; Water quality

* Corresponding author. Tel.: +1-808-956-7633; fax: +1-808-956-9225.

0141-1136/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0141-1136(99)00029-X 2 E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21

1. Introduction

Mamala Bay is a coastal indentation that extends a distance of about 30 km along the southern shoreline of the island of Oahu in the Hawaiian Islands, from Diamond Head in the east to Barbers Point in the west (Fig. 1). The city of Honolulu borders much of the shoreline of Mamala Bay, and along roughly the eastern third of the bay's shoreline are located the hotels and beaches of Waikiki, one of the major tourist destinations of the world. The attraction of the Hawaiian Islands to tourists depends very much on the quality of the environment, including in particular near- shore water quality. Because of the importance of the tourism industry to the state of Hawaii, the state has been much concerned with maintaining high water quality standards in nearshore recreational waters. Maintenance of high water quality is a particularly sensitive issue in an area such as Mamala Bay, which currently receives wastewater in the form of primary treated sewage from a population of roughly 750,000 persons. During 1993 and 1994 a comprehensive study of water quality in Mamala Bay was carried out under the direction of the Mamala Bay Commission (1996). One of the concerns of that study was the impact on water quality of nutrients introduced into the bay from both point and nonpoint sources. Stream water runo€ and ground- water seepage to the bay have been estimated to be somewhere between 100Â106 m3 year1 (Freeman, 1993) and 300Â106 m3 year1 (Stevenson, O'Connor, & Aldrich,

Fig. 1. Mamala Bay showing locations of principal wastewater treatment plant outfalls. Beaches moni- tored were Diamond Head (DH), Queens Surf (QS), Fort DeRussy (FD), Ala Moana (AM), Sand Island (SI), Keehi Lagoon (KL), Fort Kamehameha (FK), Ewa Beach (EB), Oneula Beach (OB), and Barbers Point (BP). Dashed line indicates approximate path of o€shore transects and locations of Stations 1±17, where vertical pro®ling work was done. GPS=Global positioning system. E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 3

1996). The associated inputs to Mamala Bay of nutrients such as nitrogen (N) and phosphorus (P) are known with less accuracy, since much of the fresh water enters estuaries and harbors, which can be rather e€ective nutrient traps (Laws, Hiraoka, Mura, Punu, & Yamamura, 1994). Roughly 65% of the stream runo€ and ground- water seepage enters Pearl Harbor, about 20% enters Keehi Lagoon/Honolulu Harbor, and 10% enters the Ala Wai Canal (Fig. 1). The nutrient delivery from stream runo€ and groundwater seepage to these estuaries/harbors has been esti- mated to be 30±70 tonnes year1 of total phosphorus (TP) and 300±700 tonnes year1 of total nitrogen (TN) (Freeman, 1993; Stevenson et al., 1996). How much of this N and P reaches Mamala Bay is problematic. Estuaries can be ecient sediment and nutrient traps, but their trapping eciency is a function of many factors, including loading rate, estuarine morphology, and water residence time. They may function as sources of inorganic nutrients due to the remineralization of organic matter, particularly on the bottom (Nixon, 1981; Correll, Jordan, & Weller, 1991), but in general they are sinks rather than sources of TN and TP. The Ala Wai Canal is a type B estuary as de®ned by Pritchard (1967). In such estuaries, ``Very little of the suspended sediment introduced from upland sources escapes through the estuary to the sea'' (Biggs & Cronin, 1981, p. 10). Laws et al. (1994) have estimated that sedimentation removes about 40% of the allochthonous inputs of organic carbon to the Ala Wai Canal. It seems likely that a comparable percentage of the organic N and P inputs are also removed via sedimentation. In the case of N, an additional mechanism of removal is denitri®cation (Nixon, 1981). Pearl Harbor is a type C estuary, with an area of 20.1 km2 and a mean tidal range of 0.37 m (Buske, 1974); its trapping eciency is unknown. Point source inputs are known with more accuracy and are summarized in Table 1. The principal inputs come from sewage treatment plants operated by the City and County of Honolulu, namely the Sand Island wastewater treatment plant (WWTP) and the Honouliuli WWTP. Both plants discharge primary treated e‚uent directly into Mamala Bay (Fig. 1). An additional source of sewage is the Fort Kamehameha WWTP, which discharges secondary treated sewage into the mouth of Pearl Harbor (Fig. 1). The discharge of TN and TP from all point sources is estimated to be 2800

Table 1 Principal point source inputs of total nitrogen (TN) and total phosphorus (TP) to Mamala Bay (Stevenson et al., 1996)

Source Nature of Freshwater discharge TN discharge TP discharge discharge (106 m3 year1) (tonnes year1) (tonnes year1)

Sand Island WWTP Primary treated sewage 100 1890 274 Honouliuli WWTP Primary treated sewage 33 800 123 Fort Kamehameha Secondary treated sewage 10 110 16 WWTP Other 7 Total 150 2800 413

WWTP, wastewater treatment plant. 4 E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 and 413 tonnes year1, respectively (Stevenson et al., 1996). While it seems clear from this analysis that point sources dominate the nutrient loading to Mamala Bay, it is important to bear in mind that the Sand Island and Honouliuli WWTPs dis- charge their e‚uent 2 km from shore at depths of 73 and 61 m, respectively. The discharges from harbors/estuaries occur directly along the shoreline and impact the uppermost portion of the water column. Furthermore, stream runo€ during storms can easily be 10±100 times dry weather ¯ow (Laws, 1993). Thus, from the standpoint of a€ecting water quality in recreational areas, the nonpoint source discharges are potentially of comparable or even greater importance than the sewer outfalls. The study reported here was aimed at investigating patterns in nutrient-related water quality parameters on recreational beaches and along the approximate isobath of the two major sewer outfalls in Mamala Bay, with the expectation that some insights could be gained about the mechanisms that control water quality in the bay and about the relative importance of point and nonpoint nutrient sources at locations where impacts from fresh water were discernible.

2. Materials and methods

2.1. Beach sampling

Water samples were collected at the 10 beaches shown in Fig. 1 on a monthly basis from August 1993 through July 1994. The samples were collected from just below the surface in acid-cleaned 20-l plastic carboys at a distance from the shoreline where the water column was 1-m deep. Salinity was measured with an Extech Oyster con- ductivity meter calibrated with Copenhagen seawater. Upon return to the laboratory, aliquots of the water were ®ltered through glass ®ber Whatman GF/F ®lters for fur- ther analysis. The ®ltrates were used to determine inorganic nutrient concentrations. Silicate, molybdate-reactive phosphate (MRP), nitrate+nitrite, and ammonium +ammonia concentrations were determined by colorimetric methods on a Technicon AutoAnalyzer. Concentrations of particulate carbon (PC) and particulate nitrogen (PN) were determined from aliquots ®ltered onto precombusted GF/F ®lters and analyzed using a Perkin-Elmer model 2400 elemental analyzer. In addition, the iso- topic composition of the PN (d15N) was determined using either a Finnigan MAT 252 or Delta-S mass spectrometer. A modi®ed Dumas sealed tube method was used to convert the PN to N2 as described by Minawaga, Winter, and Kaplan (1985). Dried ®lters containing the PN were placed in 9-mm outer diameter Vycor brand quartz tubes along with 3 g of pre-combusted cupric oxide and 1 g of copper wire. The tubes were evacuated, sealed, and combusted at 650C overnight. The tempera- ture was then lowered to 500C for 1 h, after which time they were allowed to cool slowly to room temperature. Slow cooling ensured that the oxides of N created by combustion were all converted to N2.N2 was isolated by cryogenic distillation, using silica gel to adsorb the N2 (Mariotti, 1983). Samples for chlorophyll a (chl a) and diagnostic carotenoid and chlorophyll pigment analyses were collected on GF/F

®lters and placed in liquid N2 prior to extraction. The pigments were extracted by E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 5 placing the ®lters in 5 ml of acetone for 12 h. The ®lters were then ground using a Wheaton tissue grinder and the ®lter residue removed by centrifugation. The extract- ed pigment concentrations were measured on a Varian 5000 high-performance liquid chromatograph (HPLC) as described by Laws, DiTullio, Betzer, and Hawes (1990).

2.2. O€shore ®eld work

Six cruises were carried out between December 1993, and December 1994, at approximately 2-month intervals. The cruises consisted of horizontal transects along the 60-m isobath between Diamond Head and Barbers Point and vertical pro®les at selected stations located at depths of 20, 60, and 120 m. Typical horizontal transect tracks and the locations of the vertical pro®le stations are shown in Fig. 1. Hydro- graphic pro®les of temperature, salinity, and beam transmissometry to a maximum depth of 100 m were conducted with an Applied Microsystems CTD12 system ®tted with a 25-cm beam transmissometer (CTTD). The CTTD was ®tted with internal RAM and operated in a stand-alone mode. The CTTD was programmed to record data at 2-s intervals for all vertical pro®les, and data were downloaded to a portable computer at the completion of each cast. Water samples for analysis of chemical and biological parameters were collected by Niskin bottles during the December 1993 cruise and with a submersible pump system on subsequent cruises. The system con- sisted of a 110-volt submersible pool pump, portable generator, 60 m of 2-cm inside- diameter opaque plastic hose, and assorted ®ttings. Samples were transferred to polyethylene bottles and chilled in ice. In the laboratory the samples were split, with one fraction ®ltered onto 0.2-mm Nuclepore membrane ®lters for analysis of extract- ed chl a. The other fraction was analyzed (un®ltered) for dissolved nutrients and turbidity. Turbidity was measured using a Turner Designs nephelometer. Inorganic nutrients were analyzed using colorimetric methods with a Technicon Auto- Analyzer, with the exception of nitrate+nitrite, which was analyzed using the chemiluminescent procedure of Garside (1982) on an Antek model 703C nitrogen oxides analyzer.

2.3. Sewage bioassay experiments

Bioassays were carried out using e‚uent from either the Sand Island or Honouliuli WWTPs. The e‚uent was, in each case, a 24-h composite sample obtained from the City and County of Honolulu's Sand Island wastewater laboratory. The sewage was ®ltered through precombusted GF/F ®lters. The particulate matter on the ®lters was analyzed for PC and PN and the d15N of the PN as previously described. The ®ltrate was analyzed for inorganic nutrient concentrations as previously described using colorimetric methods on a Technicon AutoAnalyzer. Concentrations of total dissolved nitrogen (TDN) and total dissolved phosphorus (TDP) were determined by ultraviolet oxidation of the ®ltrates followed by colorimetric analysis of the inorganic N and MRP. Salinity was measured as previously described with an Extech Oyster conductivity meter. 1-week incubations were conducted to determine the d15Nof the biologically available dissolved N in the sewage e‚uent and to determine the 6 E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 impact of sewage enrichment on the composition of the phytoplankton community. For these experiments, 100 ml of sewage ®ltrate were added to each of two clear poly- carbonate bottles containing 900 ml of surface seawater collected 1 km o€ Diamond Head. The bottles were incubated at a temperature of 25C in front of a bank of day- light ¯uorescent lamps at a continuous irradiance of 150 mmol quanta m2 s1 (PAR). One bottle was incubated without further nutrient additions and the cells harvested by ®ltration after 1 week for pigment analysis. The second bottle was enriched with the mixture of trace metals, vitamins, and phosphate recommended for IMR (Institute of Marine Resources) medium (Eppley, Holmes, & Strickland, 1967). The phosphate concentration produced by this enrichment was about 50 mM, and the resultant ratio of dissolved inorganic N (DIN) to phosphate was about 2 on a molar basis. The enrich- ment thus assured that the phytoplankton would strip the water of available N before signi®cantly depleting the phosphate concentration, since the ratio of N to P in phyto- plankton is about 16 and virtually never falls outside the range 3±30 on a molar basis (Ryther & Dunstan, 1971). After 1 week virtually all DIN had been stripped from the water. The phytoplankton cells were harvested by ®ltration onto GF/F ®lters and the particulate material analyzed for d15N as previously described.

2.4. Bioassays with stream water and groundwater

In addition to the 1-week incubations conducted with sewage e‚uent, similar bioassays were conducted with stream runo€ and groundwater to determine the d15N of the biologically available dissolved N. Seven streams were sampled, ®ve of which ¯ow into Pearl Harbor (Aiea, Kalauao, Waiawa, Waikele, Waimalu streams), one of which ¯ows into Keehi Lagoon/Honolulu Harbor (Manaiki stream), and one of which ¯ows into the Ala Wai Canal (Manoa-Palolo stream). The groundwater samples were taken from two wells in the Ewa Plain. For the stream water assays, several liters of water were collected and transferred to clear polycarbonate bottles, which were then enriched with IMR concentrations of trace metals, vitamins, and phosphate. The natural phytoplankton community in the streams was allowed to grow until virtually all DIN had been stripped from the water. In the case of the groundwater, 200 ml were added to 800 ml of Diamond Head surface seawater, and the mixture enriched with IMR concentrations of trace metals, vitamins, and phos- phate. The Diamond Head phytoplankton community was then allowed to grow until virtually all DIN had been stripped from the water. The phytoplankton cells were then harvested by ®ltration onto GF/F ®lters and the particulate material analyzed for d15N as previously described.

3. Results

3.1. Sewage, stream, and groundwater analyses

A summary of the analyses of the Sand Island and Honouliuli sewage e‚uent is given in Table 2. The silicate concentrations of 700±1100 mM re¯ect interactions E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 7

Table 2 Summary of results of analyses of e‚uent from Sand Island and Honouliuli wastewater treatment plant e‚uents

Sand Island Honouliuli

Phosphate (MRP) 54‹8 72‹3 Nitrate+nitrite 5.7‹8.1 7.7‹2.8 Ammonium+ammonia 857‹115 1302‹58 Silicate 698‹59 1081‹21 Total dissolved N 1094‹61 1684‹153 Total dissolved P 58‹9 74.3‹3.4 Dissolved organic N 231‹179 376‹95 Dissolved organic P 3.6‹1.2 2.7‹2.6 Particulate C 2670‹219 3162‹760 Particulate N 474‹334 379‹85 PC/PN 6.3‹2.5 7.1‹0.2 d15N 1.9‹3.1 1.8‹3.5 d15N of biologically available N 6.9‹5.8 10.1‹2.2 Salinity 4.4‹0.2 0.7‹0.1

Nutrient concentrations and concentrations of particulate carbon (PC) and particulate nitrogen (PN) are micromolar. d15N is expressed per mil. Salinity is given in practical salinity units. Error bars are ‹one standard deviation based on three analyses. MRP, molybdate-reactive phosphate; N, nitrogen; P, phos- porous; C, carbon; d15N, isotopic composition of PN. between groundwater and basaltic rock and are typical of groundwater in Hawaii. Virtually all the DIN in the sewage consisted of ammonium+ammonia. The d15Nof the biologically available dissolved N in the sewage averaged 7±10%, about 5±8% higher than the d15N of the PN in the sewage. The d15N of the biologically available dissolved N in stream water and groundwater averaged 2.4‹2.7 and 10.4‹0.7%, respectively.

3.2. Beach analyses

The results of the beach sampling are shown in Figs. 2±4. Salinities were relatively constant and close to the salinity of o€shore surface water (34.5%) at beaches from Diamond Head to Sand Island. Salinity was clearly depressed at Keehi Lagoon and recovered only gradually toward more typical seawater values at Oneula Beach and Barbers Point. The much higher variability of the salinity readings at Keehi Lagoon Beach Park re¯ects the highly variable nature of stream runo€ and the fact that the beach park lies directly adjacent to the mouths of the Moanalua and Kalihi streams. Silicate concentrations were approximately a mirror image of the salinity values due to the high concentration of silicate in Hawaiian streams. The con- centrations from Diamond Head to Sand Island were uniformly low and typical of o€shore ocean values (1.5 mM). The values at Keehi Lagoon again showed by far the greatest variability due to the intermittent nature of storm runo€ events. Mean PN and chl a values were highly correlated (r2=0.97). The highest values and greatest variability occurred at Keehi Lagoon. Chl a values were uniformly low at 8 E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21

Fig. 2. (A) Mean salinities, (B) concentrations of silicate, (C) chlorophyll a, and (D) particulate nitrogen (PN) at the 10 recreational beaches. Error bars are standard errors. BP, Barbers Point; OB, Onuela Beach; EB, Ewa Beach; FK, Fort Kamehameha; KL, Keehi Lagoon; SI, Sand Island; AM, Ala Moana; FD, Fort DeRussy; QS, Quenns Surf; DH, Diamond Head.

Diamond Head, Queens Surf, Sand Island, Fort Kamehameha, and Barbers Point. Slightly higher and more variable values occurred at Ala Moana, Fort DeRussy, Oneula Beach, and Ewa Beach. PN values followed a similar pattern with the exception of Queens Surf, where the results were slightly higher and more variable compared to the pattern of chl a concentrations. Based on a one-way analysis of variance (ANOVA) there was no signi®cant di€erence ( p>0.05) between the MRP concentrations at the 10 beaches. The results were uniformly variable, and all mean values fell within the range 0.15‹0.07 mM. There was a highly signi®cant di€erence in nitrate+nitrite concentrations (ANOVA on log-transformed concentrations, p=0.02), but the di€erence was due entirely to the high and variable results at Keehi Lagoon Beach Park. Mean concentrations at the remaining beaches were more- or-less uniform, and all were less than 0.4 mM. Concentrations of ammonium +ammonia were uniformly higher than nitrate+nitrite, with all mean values falling in the range 0.8‹0.4 mM. There was again a signi®cant di€erence in concentrations between beaches ( p=0.03), the di€erence in this case being due to the combina- tion of the high values at Ewa Beach and the low values at Fort Kamehameha. The d15N of the PN di€ered signi®cantly ( p=108) and dramatically between beaches. Values systematically increased from east to west, from a low of about 3.5% at Diamond Head to a high of almost 7% at Barbers Point. Analysis of the diagnostic carotenoid and chlorophyll pigments in the suspended particulate material revealed two general patterns. At Keehi Lagoon and at beaches E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 9

Fig. 3. Mean concentrations of (A) phosphate, (B) nitrate+nitrite, (C) ammonium+ammonia, and (D) the d15N of particulate nitrogen (PN) at the 10 recreational beaches. Error bars are standard errors. BP, Barbers Point; OB, Onuela Beach; EB, Ewa Beach; FK, Fort Kamehameha; KL, Keehi Lagoon; SI, Sand Island; AM, Ala Moana; FD, Fort DeRussy; QS, Quenns Surf; DH, Diamond Head. east of Sand Island the dominant diagnostic pigment was chl b, and the ratio of 190- hexanoyloxyfucoxanthin (190-hex or hex) to chl b was consistently about 0.5. At Sand Island and at beaches west of Keehi Lagoon the dominant diagnostic pigment was either fucoxanthin or (at Ewa Beach) 190-hex, and the 190-hex/chl b ratio was consistently greater than 0.5. In Diamond Head water enriched with sewage e‚uent, the dominant diagnostic pigment was fucoxanthin, but the 190-hex/chl b ratio was only about 0.1, much lower than at any of the beaches.

3.3. O€shore ®eld work

The results of the o€shore ®eld work are shown in Figs. 5±9. Plots of ammonium +ammonia versus silicate and nitrate+nitrite versus silicate reveal the presence of three water types. In most cases the water contained low concentrations of all three nutrients. However, some water samples contained relatively high concentrations of both ammonium+ammonia and silicate, and other water samples contained rela- tively high concentrations of both nitrate+nitrite and silicate. Examination of the depths from which these high-nutrient samples were collected reveals that the former group came from depths of 10 m or greater, and the latter group came from depths of 5 m or less. The samples with the highest concentrations of ammonium+ammo- nia and silicate came from depths of 60 m, and the samples with the highest con- centrations of nitrate+nitrite and silicate came from depths of either 1 or 2 m. 10 E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21

Fig. 4. Pro®les of normalized diagnostic pigment concentrations for the ®ve beaches at which (A) chlo- rophyll b was the dominant accessory pigment, (B) either fucoxanthin (fuc) or 190-hexanoyloxyfucox- anthin (hex) was the dominant accessory pigment, and (C) sewage-enriched Diamond Head water. Error bars are standard errors.

Further analysis of the relationship between nutrient concentrations and sample location reveals that the high ammonium+ammonia results came almost exclusively from stations in the vicinity of the Sand Island and Honouliuli WWTP outfalls, whereas the high nitrate concentrations came from stations o€ the mouths of the Ala Wai Canal and Pearl Harbor. Some additional insight into the nature of o€shore water quality can be gained by examining the frequency distribution of nutrient concentrations. For example, when results from samples collected from depths of 1 or 2 m are excluded, the nitrate+ nitrite concentrations give an excellent ®t to a log-normal distribution Fig. 7, with a median of 8 nM. This ®gure is well below the limit of detection by traditional col- orimetric methods (0.03 mM), but is comfortably above the detection limit (0.3 nM) of Garside's (1982) chemiluminescent assay. Determining the background marine distribution of phosphate, silicate, and ammonium+ammonia was not as straightforward since both land runo€/groundwater seepage and the e‚uent from the o€shore WWTP outfalls contain high concentrations of phosphate and silicate, and the e‚uent from the o€shore WWTP outfalls is greatly enriched in ammonium +ammonia. In order to estimate the concentrations associated with strictly marine waters, we excluded samples taken from depths of 1±2 m and from stations in the immediate vicinity of the WWTP outfalls (Stations 1±3 and 13±15). The median phosphate, silicate, and ammonium+ammonia concentrations at the other stations E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 11

Fig. 5. Scatter diagrams of (A) silicate versus nitrate+nitrite concentrations, and (B) silicate versus ammonium+ammonia concentrations. Numbers next to symbols indicate depths from which samples with high concentrations were taken. and depths were 0.06, 1.5, and 0.06 mM, respectively. The median phosphate and ammonium+ammonia concentrations approach the limit of detection (0.03 mM) by colorimetric methods. The turbidity of most water samples was less than 0.2 nepholometric turbidity units (NTUs), but exceptions to this pattern were clearly evident o€ the mouths of 12 E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21

Fig. 6. Concentrations of (A) nitrate+nitrite, (B) turbidity, and (C) ammonium+ammonia versus o€- shore station number. Numbers next to symbols indicate depths from which samples with high con- centrations were taken. NTUs, nepholometric turbidity units. the Ala Wai Canal and Pearl Harbor. Turbidity was closely correlated with the dis- tribution of chl a. By far the highest and most variable chl a and turbidity were found o€ the mouth of the Ala Wai Canal (Station 4) and Pearl Harbor (Station 9). Comparisons were made between the chl a and turbidity readings at depths of 20 and 60 m in the immediate vicinity of the Sand Island and Honouliuli WWTP out- falls and at other stations in the bay to determine whether the impact of the outfalls E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 13

Fig. 7. Frequency distribution of o€shore nitrate+nitrite concentrations, excluding results from samples taken at depths of 1±2 m. Asterisks indicate expected frequencies based on a log-normal distribution of values with the same mean and standard deviation. on these parameters was discernible in a statistical sense. Based on a t-test, there was no signi®cant di€erence between chl a concentrations at either 20 or 60 m near the Sand Island outfall and at other stations in the bay ( p>0.35). However, chl a at both 20 and 60 m was judged to be higher near the Honouliuli outfall than at other stations ( p<0.05). There was no di€erence in turbidity at 20 m ( p>0.55), but at 60 m turbidity was signi®cantly di€erent and higher at stations near both the Sand Island and Honouliuli outfalls compared to other stations in the bay ( p<0.04). We estimated the chl a concentration of strictly marine waters in the bay by excluding shallow water (<5 m) samples collected immediately o€ the mouth of the Ala Wai Canal (Station 4) and Pearl Harbor (Station 9) and samples from the immediate vicinity of the two WWTP outfalls. The median chl a concentration of the remaining samples was 0.12 mg/l. Vertical pro®les of temperature, salinity, and turbidity sometimes revealed very obvious gradients in the upper 60 m of the water column and at other times a rela- tively well-mixed system. The data from August 1994 are illustrative of the former condition. The temperature of the water dropped by about 2C between 40 and 55 m, and this gradient was sucient to trap the sewage plumes. The plumes were clearly evident as increases in turbidity at roughly 50 m depth (Fig. 9B,C). No such behav- ior was apparent in December 1993. 14 E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21

Fig. 8. Mean turbidity and chlorophyll (chl) a concentrations measured at Stations 4 (Pearl Harbor) and 9 (Ala Wai Canal) and at depths of 20 and 60 m over the Sand Island and Honouliuli outfalls and at depths of 20 and 60 m at other stations in the bay. Error bars are standard errors. NTUs, nepholometric turbidity units.

4. Discussion

4.1. Recreational beaches

Of all the recreational beaches included in this study, Keehi Lagoon Beach Park was clearly the most impacted by fresh water. This conclusion is evidenced by its low mean salinity and high concentrations of silicate, chl a, PN, and nitrate+nitrite. The highly variable nature of all these parameters at Keehi Lagoon re¯ects the fact that the impact comes from stream runo€, which is temporally highly variable (Laws, 1993). The temporal variability would be much less if the impact were primarily from groundwater seepage or wastewater discharges, which are relatively constant. Both the magnitude of the impact and the temporal variability would be less extreme if the impact were due to the out¯ow from an estuary as opposed to the direct dis- charge from several streams, since estuaries can be rather ecient traps of sediment and N (Biggs & Cronin, 1981; Nixon, 1981; Laws et al., 1994) and can e€ectively convolute the time series of fresh water and nutrient inputs on a time scale com- parable to the residence time of water in the estuary. The latter consideration is particularly relevant to Pearl Harbor, which has a narrow mouth and within which the ratio of volume to fresh water discharge is 1.0 year (Freeman, 1993). The impacts of freshwater discharges on water quality at beaches other than Keehi Lagoon appear to be relatively subtle. Of the remaining beaches, the one E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 15

Fig. 9. Vertical pro®les of temperature, salinity, and turbidity above the Sand Island outfall measured in August 1994 (A±C) and December 1993 (D±F). experiencing the greatest impact would appear to be Fort Kamehameha Beach, which lies directly adjacent to the mouth of Pearl Harbor and within close proximity to the Fort Kamehameha WWTP. The Fort Kamehameha WWTP discharges 26Â103 m3 of secondary treated sewage directly into the mouth of Pearl Harbor. The treated wastewater contains 360 mM nitrate+nitrite and 55 mM TP. The high phosphate concentration and anomalously low ammonium+ammonia concentra- tion at Fort Kamehameha Beach probably re¯ect microbial uptake of ammonium stimulated by the phosphate in the wastewater and the fact that the N/P ratio in the wastewater is less than half the Red®eld ratio. Hamilton, Singer, and Waddell (1996, p. 1) have noted that, ``Circulation patterns in Mamala Bay are complex, with substantial changes occurring over short distances and times.'' While agreeing with this assessment, Blumberg and Connolly (1996, p. 2) concluded that, ``The mean ¯ow is typically westward at all depths.'' This latter conclusion is not surprising, since the Hawaiian Islands lie within the North Equa- torial Current. Since roughly 65% of the stream ¯ow and groundwater seepage to Mamala Bay enters Pearl Harbor, one would expect Fort Kamehameha Beach and beaches immediately to the west of the mouth of the harbor to experience the greatest impact from that fresh water. These considerations probably explain the low salinities and high silicate concentrations at Fort Kamehameha and Ewa Beach. Following this line of reasoning, one would expect the ammonium+ammonia con- centration at Ewa Beach to be low, since the out¯ow from Pearl Harbor contains only about 0.4 mM ammonium+ammonia. Obviously this expectation is not 16 E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 supported by the experimental data. The ammonium+ammonia concentration at Ewa Beach was higher than at any of the other beaches. Relevant to this observation is the fact that during the time of this study a block of approximately 150 homes within about 200 m of the Ewa Beach shoreline were not connected to the Honolulu municipal sewer system but instead were serviced by cesspools. Seepage from those cesspools probably accounts for the high ammonium+ammonia concentrations at Ewa Beach. Of the 10 beaches studied, nutrient concentrations at one (Keehi Lagoon) seem clearly impacted by stream runo€, at another (Fort Kamehameha) by discharges of treated wastewater, and at a third (Ewa Beach) perhaps by cesspool seepage. The remaining beaches show very little impact from fresh water on the concentrations of either dissolved or particulate materials. However, there appear to be some very signi®cant changes in the composition of the particulate material. Most striking is the steady increase in the d15N of the PN from Diamond Head to Barbers Point. The approximate twofold increase in the d15N of the PN from Diamond Head to Barbers Point and the absence of any corresponding change in the concentration of PN can be readily understood with the simple model shown in Fig. 10. According to this model the coastal zone receives an input of fresh water from the land and exchanges water with the o€shore ocean. The net exchange with the o€shore ocean is assumed to exactly balance the input of fresh water from the land and, hence, the westward ¯ux of water in the coastal zone remains constant. The rate of change of the quantity of any substance in the coastal zone is then given by the equation:

d VX† ˆ F X X †‡F X ‡ F X F ‡ F †X; 1† dt P DH BP L L O O L O where X is the average concentration of the substance in the coastal zone box; V is the volume of the coastal zone box; XDH and XBP are the concentrations of X at Diamond Head and Barbers Point, respectively; XL and XO are the concentrations of X in land runo€/seepage and the o€shore ocean, respectively; and FP, FL, and FO

Fig. 10. Box model of water movement in the coastal zone used to derive Eq. (1). Arrows indicate direc- tion of water movement. FP, FL, and FO are the ¯uxes of water (volume/time) parallel to the coastline (Diamond Head to Barbers Point), from the land to the coastal zone box, and from the ocean into the coastal zone box, respectively. E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 17 are the ¯uxes of water (volume/time) parallel to the coastline (Diamond Head to Barbers Point), from the land to the coastal zone box, and from the ocean into the coastal zone box, respectively. Under steady state conditions Eq. (1) equals zero, and the right-hand side can be rearranged to give:

fP XDH XBP†‡ 1 fO†XL ‡ fOXO ˆ X; 2†

where fO and fP are dimensionless numbers equal to FO/(FO+FL) and FP/(FO+FL); clearly 0

1 fO†XL ‡ fOXO ˆ X: 3†

1 The mean PN concentration at the 10 beaches was 42.7 mgl =3.05 mM, and XO is 0.5 mM (Laws & Terry, 1983). Fresh water entering the coastal zone from the land comes from a variety of sources, including groundwater seepage, streams, and street runo€. The e€ective concentration of PN in those various inputs, including DIN which is converted to PN via microbial uptake, can be estimated from our data by plotting salinity versus the concentration of PN+DIN and extrapolating to a salinity of zero. The correlation coecient between salinity and PN+DIN is 0.86, and the intercept at zero salinity is 120 mM. This ®gure is comparable to an estimate of 80 mM based on calculations of N loading to Mamala Bay from freshwater sources other than the o€shore sewer outfalls (Stevenson et al., 1996). Substituting

XL=100 mM in Eq. (3) gives fO=0.9744. In other words, the ¯ux of o€shore ocean water to the coastal zone is about 40 times the ¯ux of fresh water from land into the same system. The conclusion that fO is large compared to fL is insensitive to the value assigned to XL as long as XL is large compared to X and XO. Based on the work of Chun (1994), we assumed the d15N of o€shore marine PN to be 3.5%. The fact that the d15N of the PN in the coastal zone rises to almost 7% at Barbers Point suggests that the d15N of a signi®cant portion of the freshwater N is at least 7%. Between Ewa Beach and Barbers Point the only freshwater input to the ocean from the land is groundwater seepage, for which d15N=10%. Assuming that the d15N of the ter- restrial N is 10%, Eq. (2) accounts for the rise of d15PN between Diamond Head and Barbers Point if fP=0.667. As a cross check on these calculations, we used Eq. (2) to estimate the salinity and silicate concentration at Barbers Point, with fP=0.667 and fO=0.9744. Assuming the salinity of the o€shore ocean to be 34.5% based on our o€shore sampling, the salinity at Barbers Point was predicted to be 34.2%, which is identical to the mean of the measured values. In the case of silicate, we assumed the concentration in the o€shore ocean to be 1.5 mM, again based on our o€shore sampling. The concentra- tion of silicate in the terrigenous freshwater input was estimated from the intercept of a regression line ®t to the mean silicate and salinity concentrations at the 10 18 E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 beaches. The intercept at zero salinity was 303 mM, which is well within the range of values reported for streams on Oahu (Laws & Atkinson, 1994). If we assumed that microorganisms took up none of the silicate, the silicate concentration predicted at Barbers Point was 4.2 mM. If we assumed that half the terrigenous DIN was taken up by diatoms with a molar N:Si ratio of 1:1 (Dugdale & Wilkerson, 1998), the silicate concentration predicted for Barbers Point was 3.35 mM. The mean silicate concentration at Barbers Point was 3.40 mM. The implication of this exercise is that the general characteristics of water quality at recreational beaches from Diamond Head to Barbers Point can be explained by a simple model of the coastal current system which assumes that o€shore ocean water and land runo€/groundwater seepage enter the coastal current system in a roughly 40:1 ratio and, in particular, if about half of the DIN associated with the fresh water is assumed to be taken up by diatoms. The latter conclusion is consistent with the beach pigment pro®les in the sense that fucoxanthin was the dominant diagnostic pigment at most beaches in the western half of the bay and chl b at most beaches in the eastern half of the bay. Fucoxanthin, 190-butanoyloxyfucoxanthin (190-but or but), and 190-hex are all found in chrysophytes, prymnesiophytes, and diatoms, but in dif- ferent ratios (Letelier et al., 1993). To a good approximation fucoxanthin may be taken as diagnostic of diatoms, 190-but of chrysophytes, and 190-hex of prymnesio- phytes. Prasinoxanthin (pra) and zeaxanthin (zea) are diagnostic for prasinophytes and cyanobacteria, respectively. These two algal groups apparently contributed little to the biomass of microalgae at any of the beaches. In the absence of prasinophytes, chl b is diagnostic for chlorophytes (Letelier et al., 1993). The pigment pro®les imply that the most important algal groups at the 10 beaches were chlorophytes, diatoms, chryso- phytes, and prymnesiophytes, with chlorophytes being more important in the eastern half of the bay and diatoms in the western half. Peridinin (per), which is diagnostic for dino¯agellates, was more abundant at beaches in the western half of the bay. This observation is consistent with the circulation of the coastal current system and the fact that dino¯agellates are very abundant in the Ala Wai Canal (Laws et al., 1994). Several aspects of the beach results suggest that the two o€shore sewer outfalls have little impact on beach water quality. First, none of the beach pigment pro®les closely resemble the results obtained by enriching ocean water with sewage e‚uent. Although diatoms were the dominant algal group at ®ve of the recreational beaches, prymnesiophytes were consistently more abundant at those recreational beaches than in sewage-enriched ocean water. Prymnesiophytes and chlorophytes seemed to be outcompeted by other algal groups in sewage-enriched seawater. Second, the roughly 10-fold higher concentration of DIN in sewage e‚uent versus stream run- o€/groundwater requires that fL=1fO=0.0026 in order to explain the absence of any increase in PN from Diamond Head to Barbers Point. Given this very small amount of freshwater input, Eq. (2) predicts no change in salinity between Diamond Head and Barbers Point. Furthermore, because silicate and DIN in the sewage e‚uent occur in a molar ratio of 0.8 as compared to the ratio of 3 in stream water/groundwater, Eq. (2) predicts very little change or even a decrease in silicate between Diamond Head and Barbers Point if sewage e‚uent is assumed to be the primary source of the N and silicate. E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 19

4.2. O€shore water quality

The results of the o€shore work leave little doubt about the mechanisms that a€ect water quality along the approximate isobath of the WWTP outfalls. In surface waters the out¯ows from Pearl Harbor and the Ala Wai Canal are the principal factors that cause water quality characteristics to deviate from open ocean values. This fact is clearly evident in plots of nitrate, chl a, and turbidity. When the mixed layer is suciently shallow, the plumes from the sewer outfalls are trapped within the thermocline, and the signature of the e‚uent is clearly apparent in pro®les of ammonium, silicate, turbidity, and salinity. The trapping typically occurs at depths of 50 m. At other times the mixed layer is suciently deep that no trapping occurs. Under such conditions at least portions of the plumes undoubtedly ®nd their way to the surface. Of the parameters we measured, ammonium+ammonia is the most sensitive chemical tracer of the sewage, and the concentrations of ammonium +ammonia measured at depth provide some measure of the dilution of the sewage by physical processes. Near the Sand Island outfall the highest ammonium +ammonia concentrations measured at 60 m were in the range 1.5±8.5 mM, imply- ing dilution of the sewage by a factor of 100±600. At Honouliuli similar calculations imply dilution at 60 m by a factor of 200±900. These results are consistent with the physical oceanographic studies of Roberts (1996) who concluded that dilution of the submerged plume at the Sand Island outfall was in the range 87±3700 with a mean of 627 and that dilution of the submerged Honouliuli plume was even higher due to its lower ¯ow rate per unit di€user length. At a depth of 5 m near the outfalls the highest measured silicate concentration was <2 mM and the average ammonium +ammonia concentration was 0.18‹0.10 mM. The latter ®gure is less than half the mean ammonium+ammonia concentration measured at any of the recreational bea- ches. The implication is that by the time the sewage plumes rise to within a few meters of the surface, dilution and/or uptake have reduced the concentrations of both ammonium+ammonia and silicate to almost background levels. The absence of any discernible impact on the recreational beaches is therefore not surprising. These results are again consistent with the physical studies of Roberts (1996), who con- cluded that when the outfall plumes surface they are diluted by an average factor of 1350 (range: 600±5400), which would be sucient to reduce the concentration of both silicate and ammonium+ammonia to less than 1.0 mM even in the absence of uptake. The two o€shore WWTP outfalls have some impact on water quality in the immedi- ate vicinity of the outfalls, but the impact is small compared to the washout from Pearl Harbor and the Ala Wai Canal. The fact that chl a concentrations are virtually identical near the Sand Island outfall and other stations in Mamala Bay is intriguing, but chl a concentrations are signi®cantly higher near the Honouliuli outfall, despite the fact that it discharges only about half as much DIN, phosphate, and silicate as the Sand Island outfall. The di€erence in impact may re¯ect the fact that the Honouliuli di€user is 12 m shallower (61 vs 73 m) than the Sand Island di€user. An alternative explanation may be that the Honouliuli outfall lies downcurrent from the Sand Island outfall and that at least some of the elevation in chl a evident at 20 and 60 m near the Honouliuli outfall re¯ects the impact of nutrients discharged by the Sand Island WWTP. 20 E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21

5. Conclusions

We conclude from this analysis that a variety of physical factors are responsible for the high water quality at recreational beaches along the south shore of Oahu, particu- larly in the Waikiki area, despite the fact that 100±300Â106 m3 year1 of land runo€/ groundwater seepage and almost 150Â106 m3 year1 of treated sewage e‚uent enter Mamala Bay or its tributaries. First, most of the land runo€ enters Pearl Harbor, Keehi Lagoon/Honolulu Harbor, or the Ala Wai Canal. The Ala Wai Canal traps about 40% of allochthonous organic matter. The trapping eciency of the other harbors/estuaries is unknown, but they almost certainly function as sinks for sediment, TN, and TP. Second, the beaches of Waikiki lie to the east of these three tributaries, and the mean ¯ow in the bay is westward at all depths. Thus materials exported from Pearl Harbor and Keehi Lagoon/Honolulu Harbor are transported away from the recreational bea- ches of Waikiki. Although a recirculation cell in the nearshore region o€ Waikiki may transport materials discharged from the Ala Wai Canal eastward (Blumberg & Con- nolly, 1996), the impact of such transport on water quality at Waikiki beaches appears to be minimal. Our studies along the 60-m isobath indicates that a substantial percen- tage of the out¯ow from both the Ala Wai Canal and Pearl Harbor moves o€shore, i.e. it does not hug the shoreline. The exception to this scenario is Keehi Lagoon Beach Park, which is located directly adjacent to the mouths of the Moanalua and Kalihi streams. Water quality at this beach is greatly impacted by stream runo€ because there is no bu€er zone to moderate the impact of sediment and nutrient loading from the streams. The impact of the o€shore sewer outfalls is greatly reduced by the fact that they are located below the mixed layer much of the time. Studies by Roberts (1996) indicate that the plumes from these outfalls surface only about 15±20% of the time, and that when they do surface they are diluted by a factor that averages 1350. Since the outfalls are located roughly 2 km o€shore and in the western two-thirds of Mamala Bay, their potential impact on beaches in the Waikiki area is nil. While there is some potential for the e‚uent from the Sand Island and Honouliuli WWTPs to reach bea- ches in the western portion of Mamala Bay, any such impacts on water quality are undetectable due to the high degree of dilution of the e‚uent by the time it reaches the shoreline. The principal impacts on the nearshore water column in the western portion of Mamala Bay seem to be a shift from a chlorophyte- to a diatom-dominated phyto- plankton community and an increase in the d15N of suspended particles. These e€ects appear to re¯ect the high silicate concentration of land runo€/groundwater seepage in Hawaii and the high d15N of groundwater seepage, particularly in the Ewa Plain area. Finally, exchange of water between the coastal current system and the o€shore ocean appears to be an important process by which impacts of freshwater discharges to Mamala Bay are minimized. That exchange is roughly 40 times greater than the input of fresh water to the coastal zone.

Acknowledgements School of Ocean and Earth Science and Technology contribution number 4753. This work was supported by a grant from the Mamala Bay Commission. E.A. Laws et al. / Marine Environmental Research 48 (1999) 1±21 21

References

Biggs, R. B., & Cronin, L. E. (1981). Special characteristics of estuaries. In B. J. Neilson, & L. E. Cronin (Eds.), Estuaries and nutrients (pp. 3±23). Clifton, NJ: Humana Press. Blumberg, A. F., & Connolly, J. P. (1996). Modeling transport and fate of pathogenic organisms in Mamala Bay. Mamala Bay Study (Final report, Vol. I). Honolulu, HI: Mamala Bay Study Commission. Buske, N. L. (1974). Tides, runo€ and currents. In E. C. Evans (Ed.), Pearl Harbor Biological surveyÐ ®nal report (pp. 3.3-1±3.3-79). San Diego: Naval Undersea Center. Chun, A. (1994). The e€ects of nutrient loading on C:N:P ratios of marine macroalgae in Kaneohe Bay, Hawaii. MS thesis, University of Hawaii, Hawaii. Correll, D. L., Jordan, T. E., & Weller, D. E. (1991). Nutrient ¯ux in a landscape: e€ects of coastal land use and terrestrial community mosaic on nutrient transport to coastal waters. Estuaries, 15, 431±442. Dugdale, R. C., & Wilkerson, F. P. (1998). Silicate regulation of new production in the equatorial Paci®c upwelling. Nature, 391, 270±273. Eppley, R. W., Holmes, R. W., & Strickland, J. D. H. (1967). Sinking rates of marine phytoplankton measured with a ¯uorometer. Journal of Experimental Marine Biology and Ecology, 1, 191±208. Freeman, W. (1993). Revised total maximum daily load estimates for six water quality limited segments. Island of Oahu, Hawaii. Honolulu, HI: Report prepared for the State of Hawaii Department of Health Environmental Planning Oce. Garside, C. (1982). A chemiluminescent technique for the determination of nanomolar concentrations of nitrate and nitrite in seawater. Marine Chemistry, 11, 159±167. Hamilton, P., Singer, J., & Waddell, E. (1996). Ocean current measurements. Mamala Bay Study (Final report, Vol. I). Honolulu, HI: Mamala Bay Study Commission. Laws, E. A. (1993). Aquatic pollution: an introductory text (2nd ed.). New York: Wiley-Interscience. Laws, E. A., & Terry, K. (1983). The impact of sewage discharges at ocean outfalls on phytoplankton populations in waters surrounding the Hawaiian Islands. Marine Environmental Research, 8, 101±117. Laws, E. A., & Atkinson, M. L. (1994). A study of nutrient inputs, nutrient limitation, and their implications for ®sh production in Kaneohe Bay (Final report). State of Hawaii Department of Land and Natural Resources Aquatic Resources Division, Honolulu, HI. Laws, E. A., DiTullio, G. R., Betzer, P. R., & Hawes, S. (1990). Primary production in the deep blue sea. Deep-Sea Research, 37, 715±730. Laws, E. A., Hiraoka, J., Mura, M., Punu, B., & Yamamura, C. (1994). Impact of land runo€ on water quality in an Hawaiian estuary. Marine Environmental Research, 38, 225±241. Letelier, R. M., Bidigare, R. R., Hebel, D. V., Ondrusek, M., Winn, C. D., & Karl, D. M. (1993). Temporal variability of phytoplankton community structure based on pigment analysis. Limnology and Oceanography, 38, 1420±1437. Mamala Bay Commission. (1996). Mamala Bay Study (Final report). Honolulu, HI: Mamala Bay Commission. Mariotti, A. (1983). Atmospheric nitrogen is a reliable standard for 15N abundance measurements. Nature, 303, 687±695. Minawaga, M., Winter, D. A., & Kaplan, I. R. (1985). Comparison of kjeldahl and combustion methods for measurement of nitrogen isotope ratios in organic matter. Analytical Chemistry, 56, 1859±1861. Nixon, S. W. (1981). Remineralization and nutrient cycling in coastal marine ecosystems. In B. J. Neilson, & L. E. Cronin (Eds.), Estuaries and nutrients (pp. 111±138). Clifton, NJ: Humana Press. Pritchard, D. W. (1967). Observations of circulation in coastal plain estuaries. In G. H. Lau€ (Ed.), Estuaries (pp. 37±44). Washington, DC: American Association for the Advancement of Science. Roberts, P. J. W. (1996). Plume modeling. Mamala Bay Study (Final report, Vol. I). Honolulu, HI: Mamala Bay Study Commission. Ryther, J. H., & Dunstan, W. M. (1971). Nitrogen, phosphorus, and eutrophication in the coastal marine environment. Science, 171, 1008±1013. Stevenson, M., O'Connor, J., & Aldrich, J. (1996). Pollutant source identi®cation. Mamala Bay Study (Final report, Vol. I). Honolulu, HI: Mamala Bay Study Commission. Science of the Total Environment 324 (2004) 173–182

Cu, Pb and Zn contamination in Nuuanu watershed, Oahu, Hawaii

Stephanie Andrews, Ross A.Sutherland*

Department of Geography, University of Hawaii, 2424 Maile Way, Honolulu, HI 96822, USA

Accepted 31 October 2003

Abstract

Trace metal contamination in urban aquatic ecosystems in Hawaii is a significant problem, especially in terms of Cu, Pb, and Zn.These trace metals are linked to automobile usage.An in-depth study was designed to determine the influence of road sediments and storm sewers on bioavailable (0.5 M HCl) trace metal concentrations in bed sediments of Nuuanu stream, Oahu.Lead was the most enriched trace metal in the watershed.Compared to baseline Pb concentrations of -3mgykg, road sediments averaged 186 mgykg, with a maximum value of 3140 mgykg. Stream bed sediments had average Pb values of 122 mgykg, with a maximum of 323 mgykg.Al-normalized enrichment ratios (ERs) for the -63 mm fraction indicated that the watershed was significantly polluted in the lower, urbanized reaches, with maximum ER values of 560 and 94 for Pb in road sediments and stream sediments, respectively.Median ER values for Cu, Pb, and Zn in stream sediments were 2, 36, and 5, respectively.Rainfall events prior to sediment sampling masked any influence that storm sewer outlets might have had on the localized spatial distribution of metals associated with bed sediments.However, there was a general pattern of increasing trace metal concentrations downstream as the fluvial network traversed residential areas and commercial, highly trafficked areas in the lower portions of the watershed. ᮊ 2003 Elsevier B.V. All rights reserved.

Keywords: Lead; Copper; Zinc; HCl leach; Road-deposited sediments; Stream sediments; Enrichment ratios; Storm outlets; Spatial distribution; Sediment quality guidelines

1. Introduction emission products from vehicles, and weathered inorganic particles from paved road surfaces and Organic and inorganic contaminants have been sidewalks (Dempsey et al., 1993).The accumulat- emitted from automobiles since the 1920s, and ed debris temporarily stored on roads is termed accumulate in local soil profiles.As these local road-deposited sediment (RDS), and there is soils are eroded, contaminants are transported potential for this material to be remobilized by through the modern environment.Sediment trans- wind, traffic, and runoff.Urban streets are ported from polluted roadside soils accumulates designed to funnel runoff into storm drains, and on road surfaces, and is combined with a variety this design facilitates the removal of RDSs.Partic- of materials, including organics, refuse, wear and ulate contaminants in runoff have the potential to be temporarily stored along the transport path, in *Corresponding author. runoff channels, at the storm sewer inlet, in the E-mail address: [email protected] (R.A. Sutherland). storm sewer system, at the storm sewer outlet, and

0048-9697/04/$ - see front matter ᮊ 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2003.10.032 174 S. Andrews, R.A. Sutherland / Science of the Total Environment 324 (2004) 173–182 near the stream outlet (Parker et al., 2000).Recent where the stream is fully channelized are Ewa studies have indicated a tentative association mollisols. between storm-sewer outlets and contaminated The Nuuanu drainage basin is approximately depositional zones (bed sediments) in river sys- 11.7 km2 , from the crest of the Koolau range to tems.For example, Pb levels in the river Seine in the confluence with Waolani stream (Fig.1 ). France were highest just downstream of storm Nuuanu stream has a length of ;10 km, with the drains (Estebe´ et al., 1997).In Manoa stream, lower 1.5 km completely channelized to the mouth Hawaii, Sutherland (2000) noted that in some of the stream in Honolulu Harbor. situations the bed sediments below storm drain Rainfall in the Nuuanu watershed is spatially outlets had higher concentrations of Cu, Pb, and variable.From 1989 to 1999, mean annual rainfall Zn than those upstream of the outlet.Rhoads and at Nuuanu Reservoir No.4 at an elevation of 320 Cahill (1999) also noted a similar association for m above sea level, varied from 189 to 322 cm, Kaskaskia river, Illinois.These authors observed a with an average of 256 cm (The State of Hawaii rapid decay in trace element concentrations down- Data Book, 2000). stream of storm-sewer outfalls, but there was Nuuanu Valley is one of the oldest developed substantial local variability within the contaminat- valleys on Oahu.The first development occurred ed reaches. in what is now the Puunui neighborhood between This study examines the distribution and con- 1900 and 1910.More recently, French SPOT centrations of Cu, Pb, and Zn in an urban stream, satellite images from 1985 and 1996 were com- particularly in relation to storm sewers.Lead pared and results indicated that there has been no contamination of the Nuuanu system will be com- significant change in land-use during this period pared to other streams in Honolulu as well as to (Andrews, 2002).Fifty-one percent of the surface the US Geological Survey’s national water-quality area in Nuuanu watershed is forested conservation assessment (NAWQA) study units across the con- lands, while 49% is ‘developed’.Developed lands terminous United States. in the lower watershed consist of: commercial (approx.9% ), manufacturingyindustry (6%), open ( ) ( ) 2. Materials and methods space 14% , public infrastructure 16% , residen- tial (46%), and social services (9%)(Steve Anthony, personal communication, 2002). 2.1. Study area characteristics The population density of Nuuanu Valley is 1287 persons per km2 (Brasher and Anthony, Situated on the southern side of the Koolau 2000).Daily traffic volumes on streets range from volcanic range on the island of Oahu, Nuuanu 3440 to 19 500 vehicles per 24 h, with highway Valley is a long, deep valley, formed through the volumes between 37 800 and 46 000 (Gail Oka- fluvial action of Nuuanu, Waolani, and Pauoa neku, personal communication, 2001). streams (Fig.1 ).Much of the valley is now covered with layers of alluvium and soil.The 2.2. Sample collection and processing underlying geology consists of basaltic lavas.A variety of soils have formed in the valley (Foote Bed sediment samples were collected between et al., 1972).The upper reaches of the watershed January 26 and February 27, 2002.January was are dominated by a silty clay loam (Lolekaa an especially wet month, with 53 cm of rainfall at ultisol).Downstream in the low-density residential Nuuanu Reservoir No.4 (Fig.1 ), 181% above area, the soils are silty, kaolinitic mollisols (Ewa), normal.Sampling was conducted at near-baseflow and clay loams (Kawaihapai mollisol).Further conditions, when water turbidity was negligible. downstream, in the high-density residential areas Baseline bed sediment samples were collected the soil is a stony montmorillonitic vertisol from a tributary to Nuuanu stream in the headwa- (Kaena).The soils of the lower watershed near ters of the watershed, above all residential, com- Honolulu Harbor and the commercial district mercial, and industrial areas (Fig.1 ).From a S. Andrews, R.A. Sutherland / Science of the Total Environment 324 (2004) 173–182 175

Fig.1.Map of study area and sample sites in Nuuanu watershed, Oahu, Hawaii. random starting point, samples were collected from was used to collect sediments from curbside areas 10 downstream depositional zones with 50 m (the mean sample mass was 390 g, with a range between sites.At each site six cores, to a bed from 180 to 610 g). sediment depth of 5 cm, were collected and com- For the stream samples, six storm-sewer outlets posited.Fifteen road sediment samples were col- were selected in Nuuanu stream based on access, lected from the watershed.Ten samples were the presence of sediment deposition zones, and a collected from roadside locations near storm-sewer direct and obvious connection from the road inlet inlets in the lower, heavy traffic areas.Five sam- to the storm sewer outlet.An additional outlet was ples came from locations in the upper watershed, added that emptied directly into Waolani stream which typically experience light traffic.For the just above the confluence with Nuuanu stream samples from heavy traffic areas, inlets were cho- (Fig.1 ).Sediment samples were collected at the sen based on proximity to storm sewer outlets that outlet and 20, 10, and 5 m above, and 5, 10, and were to be the focus of the stream-based bed 20 m below the outlet.This sampling methodology sediment sampling.An acid-washed plastic scoop was devised for two reasons: (1) to determine if 176 S. Andrews, R.A. Sutherland / Science of the Total Environment 324 (2004) 173–182

Table 1 Sediment quality guidelines as defined by the Canadian Council of Ministers of the Environment (2001) for Cu, Pb, and Zn

Trace metal Threshold effects level (mgykg) Probable effects level (mgykg) Cu 35.7 197 Pb 35.0 91.3 Zn 123 315 trace metal concentrations exhibit a downstream ment) certified by the national institute of stan- distance decay relationship; and (2) to allow for a dards and technology (NIST). paired statistical comparison of sample units above Organic matter was estimated using the loss-on- and below storm sewer outlets. ignition technique (LOIOM).Approximately 0.5g In the laboratory, samples were weighed, and was placed in a dried and weighed ceramic dish oven-dried for 48 h.Dried samples were gently and then in a muffle furnace, set at 450 8C, for a disaggregated and passed through a 2-mm stainless minimum of 16 h.Samples were cooled in a steel sieve.Samples were then sieved with new dessicator and re-weighed.The difference in mass stainless steel sieves for 10 min using a Ro-Tap was used as an index of organic matter. sieve shaker. To establish the general water chemistry of the system, six sites were sampled during baseflow 2.3. Analytical techniques conditions, from the headwaters (baseline area) to the furthest downstream storm-sewer outlet (see One hundred -63 mm samples were digested ‘Beretania’, Fig.1 ).Electrical conductivity (EC), using a partial extraction procedure.Samples of pH, and trace element concentrations by ICP-MS 0.50 g were agitated with 10 ml of 0.5 M HCl for were determined on unfiltered samples.Further 1hat208C (Sutherland, 2002).This procedure details are given in Andrews (2002). was used because it releases environmentally rel- 2.4. Sediment quality guidelines (SQGs) evant (labile) fractions of elements (Agemian and Chau, 1976; Campbell et al., 1988).A dilute HCl The SQGs for Cu, Pb, and Zn used in this study leach liberates adsorbed detrital and non-detrital were established by the Canadian Council of Min- carbonate-bound metals and much of the FeyMn isters of the Environment (2001) to protect aquatic oxide and organic-associated metals while mini- health (Table 1).These SQGs are empirically mizing the loss of residual silicate-bound metals based and rely on field sediment chemistry paired (Sutherland and Tolosa, 2000).Each sample solu- with field or laboratory biological effects data tion was filtered and analyzed for elements with (Burton, 2002).Two levels have been established, inductively coupled plasma-mass spectrometry a lower threshold effects level (TEL), and a higher (ICP-MS).The partial digestion data set included probable effects level (PEL).Trace element con- 10 baseline stream sediments, 15 road sediments, centrations, based on a total digestion, below the 49 outlet stream bed sediments, 14 duplicate sam- TEL rarely cause adverse effects in the biotic ples, and 12 replicates of a certified reference community, while values above the PEL are likely material (CRM).The CRM chosen was NCS to cause adverse effects.The SQGs are used as a DC73317 (GBW07307), a stream sediment certi- mechanism to compare Nuuanu watershed bed fied by the China National Analysis Center for sediments to two other systems sampled on Oahu, Iron with a Pb concentration of 350 mgykg (total and to 20 study units of the NAWQA program. digestion).Recovery of the partial digestion pro- 2.5. Trace metal normalization and enrichment cedure was determined relative to a four-acid total ratios digestion of 53 samples.Precision and accuracy of the total digestion procedure were established To evaluate the level of trace metal contamina- using eight replicates of RM 8704 (stream sedi- tion in the system, enrichment ratios (ERs) were S. Andrews, R.A. Sutherland / Science of the Total Environment 324 (2004) 173–182 177

Table 2 Quality assurance data

Element HCl precision Total digestionc Duplicates (%)a CRM replicates (%)b Precision (%) Accuracy (%) Al 8.4 9.1 3.8 q0.9 Cu 5.0 10.6 3.95 y9.1 Fe 13.1 14.0 5.2 q0.3 Mn 4.0 9.7 5.5 q10.8 Pb 6.0 5.8 4.6 q4.4 Zn 5.8 10.2 6.6 y3.3 a Median values for 14 Nuuanu samples with duplicate analyses digested using 0.5 M HCl. b Mean values for 12 replicates of CRM GBW07307 digested using 0.5 M HCl. c Mean values based on a total 4-acid digestion of eight samples of RM 8704. computed.An ER is the ratio between the trace five NAWQA study sites with median Pb concen- element concentration of the sample and that same trations above the TEL. element’s baseline concentration.Here we normal- The non-parametric (paired) Wilcoxon signed ize by grain size (-63 mm) and use Al as the rank test was used to quantitatively assess whether conservative element: statistical differences existed (as0.05) in Cu, Pb, and Zn between: (1) bed sediment samples at ( s C Sample equal distances from the storm sewer outlet n n 21, seven outlets paired at three locations, e.g. C Sample ERs Al pairing at distances of 5, 10, and 20 m up and BEn Baseline downstream of the outlets); and (2) road sediments

BEAl Baseline at storm-sewer inlet sites with bed sediments at the outlet of the same storm sewer (ns7). Spearman’s non-parametric rank order correla- where, BE is the best estimate of the element’s ( ) concentration (median) in the ten baseline sam- tion coefficient rs was used to make global assessments of the relationships between trace ples.The five-category pollution index of Suther- ( ( ) metals and potential controlling factors LOIOM , land 2000 is used to qualitatively assess degree ) of enrichment.Briefly, ER -2, is defined as no siltyclay, sand, Fe, and Mn for stream and road pollution; ER 2–5, moderate pollution; ER 5–20, sediment samples. significant pollution; ER 20–40, very strong pol- 3. Results and discussion lution; and ER)40, extreme pollution. Quality assurance data for the partial HCl and 2.6. Statistical analyses total digestions are shown in Table 2.A prelimi- nary study of the water chemistry of Nuuanu One way analysis of variance (ANOVA) of stream indicated no spatial pattern in pH, with a ( ) log10 transformed data was followed by the Bon- median value of 8.0 range 8.0–8.1 .Electrical ferroni–Dunn post-hoc test to characterize differ- conductivity increased from approximately 100 ences or similarities between trace metal mSycm in the headwaters to over 22 000 mSycm concentrations: (1) in baseline sediments (ns10), in the salt-water intrusion zone near storm-sewer road sediments (ns15), and stream outlet samples outlet no.7 (i.e. Beretania).The following ele- (ns49); (2) in bed sediment samples from the ments showed significant increases from headwater upper, residential sites (ns21) and the lower, areas to the stream outlet: Cu, 0.8–60 mgyl; Fe, commercial sites (ns28); and (3) between bed 44–230 mgyl; Mn, 4–43 mgyl; Pb, -0.1–1.1 mgy sediment samples from three Oahu streams and l; Se, -0.5–101 mgyl; and Zn, 1.4–6.3 mgyl. 178 S. Andrews, R.A. Sutherland / Science of the Total Environment 324 (2004) 173–182

Table 3 Descriptive statistics for trace metal concentrations (mgykg) in Nuuanu watershed samples (-63 mm), including baseline bed sediments (ns10), bed sediments associated with storm-sewer outlets (ns49), and road-deposited sediments (ns15)

Metalymedia Mean"S.D. Median"M.A.D. Minimum Maximum Cu-baseline 22.2"6.1a 21.3"3.2 16.5 37.2 Cu-outlets 68.6"25.3b 61.4"18.6 33.6 136 Cu-road sediment 235"207c 185"135 11.8 685 Pb-baseline 2.9"1.5a 2.4"0.5 1.8 6.8 Pb-outlets 133"54.2b 122"26.7 46.6 332 Pb-road sediment 445"787b 186"155 11.2 3140 Zn-baseline 20.1"3.6a 19.0"2.8 15.5 25.0 Zn-outlets 175"90.5b 175"77.0 61.0 380 Zn-road sediment 902"1058c 675"370 36.0 4537 Notes: For a given metal, sediment samples with the same letter are not significantly different at an a-level of 0.05 using one- way analysis of variance on 1og10 transformed data, followed by Bonferroni–Dunn post hoc testing. M.A.D. represents the median absolute deviation from the median.

Mean LOIOM values for baseline bed sediment had statistically similar concentrations to those in samples were 20"2% ("standard deviation), with road sediments. a range from 17 to 23%, compared to 22"7% for Trace metal concentrations were highly variable road sediments (range 14–35%), and 16"4% for in RDSs, with coefficients of variation for Cu of storm-sewer outlet samples (range 7–28%).All 88%, Pb, 177%, and Zn, 117%.High variability bed sediment and road sediment samples were was expected because the sampling design includ- dominated by sand-sized particles (63–2000 mm) ed sites that varied widely in traffic density, pop- and gravel-sized particles ()2000 mm).Silt and ulation density, and land use (i.e. low density clay contents were highly variable, but averaged residential to high density commercialymanufac- less than 2.5%. For example, baseline bed sedi- turing).The RDS sample with the highest Pb ments, 0.6"0.5%, range 0.3–22.6%; road sedi- concentration, 3140 mgykg, was in a moderately ments, 1.8"1.8% (range 0.2–35.3%); and outlet dense residential neighborhood in the upper valley samples, 2.3"4.7% (range 0.1–27.6%). (see ‘Henry’ Fig.1 ).This site may have been influenced by Pb wheel weights from vehicles that 3.1. Trace metal concentrations collect in curbside areas.According to Root (2000), Pb weights commonly fall off automobile A statistical summary of the trace metal data is wheels, they are soft and are rapidly abraded, and given in Table 3 for all baseline sediments, storm accumulate as fine particles along the curb.Two outlet sediments, and road sediments (concentra- sites in the lower portion of the watershed (King tions are reported for a HCl digestion unless and Beretania) had relatively low Pb concentra- otherwise specified).Concentrations of Cu, Pb and tions in their RDSs (Pbs105 mgykg and 164 Zn in baseline samples were the lowest encoun- mgykg, respectively).Both roads are very heavily tered in this study (P-0.0001).Median concen- traveled with a high density of commercialyman- trations in baseline bed sediments were, Cus21 ufacturing activity, and one would expect concen- mgykg, Pbs2.4 mgykg, and Zns19 mgykg.Bas- trations to be higher.As these streets are in high eline concentrations were 1y3 (Cu),1y50 (Pb), traffic areas, it is possible that they are swept more and 1y9 (Zn) of those for storm-sewer outlet bed often, thereby removing much of the contaminated sediments.Results of ANOVA and post hoc testing RDS.Another controlling factor is that there is (Table 3) indicated that road sediments were more limited exposure of soils in these roadside areas, highly concentrated in Cu and Zn than outlet bed therefore precluding significant contributions of sediments.However, Pb in outlet bed sediments contaminated sediment to the street surfaces. S. Andrews, R.A. Sutherland / Science of the Total Environment 324 (2004) 173–182 179

Non-parametric comparisons of metal concen- trations in road sediments, at seven inlets, with their concentrations in bed sediments at their respective outlet sites indicated significantly higher Cu (Ps0.018),Pb(Ps0.043), and Zn (Ps 0.018) in road sediments.Median concentrations of Cu were 3.2-fold higher (185 vs.58 mg ykg), Pb, 1.6-fold higher (186 vs.117 mg ykg), and Zn, 4.0-fold higher (755 vs.189 mg ykg) in road sediments.These data indicate that the more highly concentrated road sediments that directly enter the stream system are significantly diluted by less contaminated sediment from upstream.However, despite this dilution effect, bed sediments in this stream system are highly contaminated (see dis- cussion of SQGs and ERs). Bed sediments associated with inputs from res- idential areas (ns21, three sewer outlet sites with 7 samples per site) were compared to those influ- enced by high traffic volumes andyor commercialy Fig.2.Relative spatial distribution of Cu, Pb, and Zn down- ( ) manufacturing activity (ns28, four sewer outlet stream from baseline Base areas.SO represents storm-sewer outlets. sites with seven samples per site).Downstream translation of bed sediment from the upper resi- below (Ps0.835).Additionally, no statistically dentially dominated portions of the watershed ( s ) could have modified, either enriched or diluted, significant differences occurred for Al P 0.754 . the concentrations in the bed sediments classified This unexpected result reflects the frequent and as ‘commercial’.Greater concentrations of Cu, Pb, higher than normal precipitation inputs in the and Zn in the bed sediments were associated with month of January, prior to sampling.Increased the commercially influenced sites.Median concen- stormflow served to evenly distribute contaminants trations of Cu in the commercial sites were 1.8- downstream from storm outlets.During storm fold higher (86"14 vs.47 "8mgykg; P- events contaminants are carried further down- 0.0001); Pb, 1.4-fold higher (148"32 vs.108 "19 stream than normal under baseflow conditions, mgykg; P-0.0001); and Zn, 2.6-fold higher thereby affecting the ‘above’ sample sites in down- (233"32 vs.89 "15 mgykg; P-0.0001) than the stream locations.Additionally, sediments with low residential sites.Fig.2 displays the low baseline trace metal concentrations transported downstream concentrations in headwaters of the watershed and from the headwaters would dilute the sediments at the general progression in pollutant levels down- the lower outlet sites.Therefore it is probable that stream from ‘residential’ to ‘commercial’ areas. the hydrological conditions prior to the study Concentrations of Cu, Pb, and Zn below storm- period confounded the pattern of trace metal con- sewer outlets were not statistically different from centrations associated with storm sewer outlets. those above outlets (as0.05).Median Cu concen- trations above the sewer outlets were 61"19 mgy 3.2. Sediment quality guidelines kg ("median absolute deviation from the median, M.A.D.) compared to 60"16 mgykg (Ps0.520) Before our Cu, Pb and Zn concentrations were below; Pb values were 123"39 mgykg above compared to SQGs we corrected all data with the sewer outlets compared to 120"13 mgykg below following median recoveries (HClytotal): Cu, (Ps0.076); and Zn values were 175"77 mgykg 0.496; Pb, 0.749; and Zn, 0.640. This allowed above sewer outlets compared to 175"74 mgykg direct comparisons between sediments from 180 S. Andrews, R.A. Sutherland / Science of the Total Environment 324 (2004) 173–182

Nuuanu watershed to the SQGs (Table 1), based on a total digestion.Of the 64 road sediment and outlet bed sediment samples, 50 exceeded the Pb PEL (probable effects level), 91.3 mgykg.Of these 50 samples, 11 were road samples (73% of the road sediment samples) and the remaining 39 (80%) were bed sediment samples (not including baseline samples, because all were below TEL).It is likely that these data are lower than usual after sediment transport and dilution from headwater areas with the storm events encountered prior to sampling.Only seven road samples (47%) exceed- ed the Cu PEL (197 mgykg) and 11 road sedi- ments (73%), and four (8%) outlet samples exceeded the Zn threshold (315 mgykg). 3.3. Lead concentration intercomparison: US study units and Honolulu streams Lead concentration data from 20 NAWQA study ( ) Fig.3.Comparison of median Pb concentrations (-63 mm) units across the US were discussed by Rice 1999 , in bed sediments from 20 NAWQA study units and three and data published on the web (USGS, 2002). streams in Honolulu, Hawaii.TEL represents the threshold These raw data were analyzed and compared to effect level, and PEL the probable effects level.Error bars rep- data from this study, and two additional stream resent "median absolute deviations from the median.Sites systems in Honolulu.The first, Manoa stream, was with median concentrations greater than TEL were compared ( ) using ANOVA, and sites with the same letter are not signifi- examined by Sutherland 2000 , and the second, cantly different at the as0.05 level. CCPT, Central Columbia Palolo stream, by Bussen et al. (2000)(Fig.1 ). plateau unit; REDN, Red river of the North system; USNK, All median data for the 23 study units are shown Upper Snake river basin; WILL, Willamette basin; CNBR, in Fig.3.Lead concentrations from the Honolulu Central Nebraska basin; TRIN, Trinity river basin; SANJ, San watersheds are high, with median values all above Joaquin-Tulare basins; NVBR, Nevada basin and range unit; WMIC, Western Lake Michigan drainage basin; OZRK, Ozark the PEL threshold, the only NAWQA study unit plateaus; RIOG, Rio Grande valley; POTO, Potomac river with comparable values was the Connecticut, Hou- basin; WHIT, White river basin; SPLT, South Platte river basin; satonic, Thames river basins (CONN).Lead data ALBE, Albermarle-Pamlico drainage; GAFL, Georgia-Florida from NAWQA units (5) and Honolulu watersheds coastal plain; HDSN, Hudson river basin; ACFB, Apalachi- (3) were compared using ANOVA if their median cola-Chattahoochee-Flint river basin; LSUS, Lower Susque- hanna river basin; CONN, Connecticut, Housatonic, Thames concentrations exceeded the PEL threshold.The river basins; MANO, Manoa watershed; NUUA, Nuuanu post hoc results are shown on Fig.3; systems with watershed; and PALO, Palolo watershed. statistically equivalent Pb concentrations are indi- cated by similar letters.Of the NAWQA units, only CONN (110"58 mgykg) had Pb values pared to baseline concentrations (Table 4).Lead statistically similar to those from Manoa stream is the most enriched trace metal in the watershed, (97"23 mgykg), and both of these systems had with median ER-values of 67"47 for road sedi- statistically lower values than either Nuuanu ments, and 36"9 for outlet sediments.Eighty stream (163"36 mgykg) or Palolo stream percent of the road sediments and 98% of the (213"48 mgykg). outlet samples had Pb ER-values of 20 or more, indicating very strong to extreme pollution.Enrich- 3.4. Al-normalized enrichment ratios ment ratios were )5 for all road and outlet The ERs for Cu, Pb, and Zn, indicate that the samples, meaning that Pb is a significant contam- road and stream outlet samples are enriched com- inant throughout the urban Nuuanu system. S. Andrews, R.A. Sutherland / Science of the Total Environment 324 (2004) 173–182 181

Table 4 Percentage of road-deposited sediment (RDS) samples and bed sediment samples associated with storm-sewer outlets (BED) in five Al-normalized trace metal enrichment ratio (ER) categories in Nuuanu watershed; and ER summary statistics

Elementymedia ER-2 ER 2–5 ER 5–20 ER 20–40 ER)40 Median"M.A.D. Minimum Maximum Cu-RDS 20.0 (3)a 6.7 (1)a 66.7 (10)a 6.7 (1)a 07.9"3.0 1.3 25.2 Cu-BED 36.7 (18) 61.2 (30) 2.0 (1) 0 0 2.1"0.3 1.0 5.5 Pb-RDS 0 0 20.0 (3) 6.7 (1) 73.3 (11) 66.5"47.4 12.2 563 Pb-BED 0 0 2.0 (1) 65.3 (32) 32.7 (16) 35.9"9.0 17.4 94.3 Zn-RDS 0 6.7 (1) 20.0 (3) 46.7 (7) 26.7 (4) 28.0"16.4 4.9 127 Zn-BED 0 53.1 (26) 46.9 (23) 0 0 5.0"1.3 3.2 15.9 a Values in parentheses are the actual number of samples in each pollution category.Total number of bed sediment samples is 49; and total number of road-deposited sediments is 15.

The median Zn ER for road sediments was The associations between the physicochemical

28"16, and for outlet sediments was 5"1.Finally, variables LOIOM , siltyclay, and sand with the trace Cu exhibited the lowest ER.In road samples, 20% metals were not statistically significant at as0.05 of the samples have ER values of -2, indicating for either road or bed sediments.The only excep- ( s no pollution.The majority of road samples have tion to this was the significant rs-value 0.30; P ER values of 5–20, indicating that road samples 0.02) between siltyclay and Zn in bed sediments. are significantly enriched in Cu.For stream outlet The lack of a statistical association between Cu samples, approximately 60% of the samples were and ‘organic matter’ was surprising, but reflects moderately enriched with Cu. the black-box character of the LOIOM index.Fur- ther detailed work using sequential extractions is 3.5. Interelement and physicochemical associations necessary to investigate the phase associations of trace metals in this urban system.

The three trace metals were highly intercorrelat- 4. Conclusions ed, for both bed sediment and road sediment ( ) samples.Spearman correlation coefficients rs Based on concentrations, comparison to sedi- were greater than 0.77 (Ps0.004) for road sedi- ment quality guidelines, and computation of Al- ments, and greater than 0.79 (P-0.0001) for normalized enrichment ratios, trace metal contami- stream bed sediments.This suggests a common nation is a significant problem throughout the source or sources of contamination in the water- Nuuanu watershed, particularly Pb.In Hawaii, high shed.Iron and Mn were significantly associated population densities and significant traffic densi- with Cu and Pb (as0.05).Associations were ties, combined with dense commercialymanufac- typically stronger with Fe than Mn, especially in turing activities on such small watersheds probably the fluvial bed sediments.Overall, these positive contributes to the elevated contamination levels relationships suggest that the trace metals are likely compared to larger, continental watersheds.Road- associated with the reducible fractions of the sed- deposited sediments have very high contaminant iments, i.e. hydrous Fe and Mn oxides. Sutherland concentrations, and ultimately, these sediments are and Tack (2000) used a four-part sequential extrac- flushed into stream systems, where they have the tion procedure to identify the reducible fraction as potential to adversely affect aquatic health.Trace the dominant sink for Pb in roadside soils in metal pollution in Honolulu streams is not going Honolulu.Additionally, Sutherland (2000) found to improve as long as RDSs are allowed to flush Pb and Zn to be primarily associated with the directly into stream systems untreated.The domi- reducible fraction in bed sediments from Manoa nant source of material that forms road sediment stream, and Cu to be related to the oxidizable is erosion of roadside soils that have previously (organic) fraction. archived deposits from automobile use.Additional 182 S. Andrews, R.A. Sutherland / Science of the Total Environment 324 (2004) 173–182 sources of trace metals include contemporaneous entific Criteria for Environmental Quality, Publication No. wear of vehicles and contributions from the various NRCC 27694, Ottawa, Canada; 1988.p.298. Canadian Council of Ministers of the Environment.Canadian urban structures within a watershed.To reduce sediment quality guidelines for the protection of aquatic trace metal pollution, land use managers in Hawaii life: Summary tables, updated in: Canadian Environmental should consider an increased frequency of street Quality Guidelines, 1999, Canadian Council of Ministers of sweeping combined with filtering of stormwater the Environment, Winnipeg, 2001. runoff before allowing it to drain into stream Dempsey BA, Tai YL, Harrison SG.Mobilization and removal of contaminants associated with urban dust and dirt.Water channels. Sci Technol 1993;28(3-5):225 –230. Estebe´´ A, Boudries H, Mouchel J-M, Thevenot DR.Urban Acknowledgments runoff impacts on particulate metal and hydrocarbon con- centrations in river Seine: suspended solid and sediment ( ) The authors would like to thank Prof.M. transport.Water Sci Technol 1997;36 8-9 :185 –193. Foote DE, Hill EL, Nakamura S, Stephens F.Soil survey of McGranaghan and T.Vana for their help in field the Islands of Kauai, Oahu, Maui, Molokai, and Lanai, State sampling.The computer cartographic talents of D. of Hawaii.USDA, Soil Conservation Service, US Govern- Olsen and J.Silver are much appreciated.Financial ment Printing Office, Washington, DC, 1972, pp.232 q support for this work was provided to the Geo- Appendices and Maps. morphology Laboratory by the Department of Parker JTC, Fossum KD, Ingersoll TL.Chemical characteris- tics of urban stormwater sediments and implications for Geography, University of Hawaii.The constructive environmental management, Maricopa County, Arizona. comments of two anonymous reviewers have sig- Environ Manage 2000;26(1):99 –115. nificantly improved this manuscript and their con- Rhoads BL, Cahill RA.Geomorphological assessment of tributions are greatly appreciated. sediment contamination in an urban stream system.Appl Geochem 1999;14:459 –483. Rice KC.Trace-element concentrations in stream bed sediment References across the conterminous United States.Environ Sci Technol 1999;33:2499 –2504. Agemian H, Chau ASY.Evaluation of extraction techniques Root RA.Lead loading of urban streets by motor vehicle for the determination of metals in aquatic sediments.Analyst wheel weights.Environ Health Perspect 2000;108 (10):937 – 1976;101(1207):761 –767. 940. Andrews S.Heavy metal pollution in the Nuuanu watershed: The (2000) State of Hawaii Data Book.http: yywww.hawaii. Aquatic and roadside sediments.Masters of Arts Thesis, govydbedtydb00yindex.html (accessed November, 2002). Department of Geography, University of Hawaii, 2002. Sutherland RA.Bed sediment-associated trace metals in an Brasher AM, Anthony SS.Occurrence of organochlorine pes- urban stream, Oahu, Hawaii.Environ Geol 2000;39 (6):611 – ticides in stream bed sediments and fish from selected 627. streams on the island of Oahu, Hawaii, 1998.US Geological Sutherland RA.Multi-element removal from road-deposited Survey Fact Sheet 140-00.US Government Printing Office, sediments using weak hydrochloric acid.Environ Geol Washington D.C, 2000. 2002;42:937 –944. Burton GA Jr.Sediment quality criteria in use around the Sutherland RA, Tack FMG.Metal phase associations in soils world.Limnology 2002;3:65 –75. from an urban watershed, Honolulu, Hawaii.Sci Tot Environ Bussen JO, Sutherland RA, Tack FMG.Heavy metal pollution 2000;256:103 –113. in road deposited sediments, Palolo Valley, Honolulu, HI. Sutherland RA, Tolosa CA.Multi-element analysis of road In: Nriagu, J. (ed.), Eleventh Annual International Confer- deposited sediment in an urban drainage basin, Honolulu, ence on Heavy Metals in the Environment.August 6–10, Hawaii.Environ Pollut 2000;110:483 –495. 2000, University of Michigan, School of Public Health, Ann US Geological Survey.Trace-element data for 541 streambed- Arbor, Michigan (CD-ROM): 2000.pp.3. sediment samples across the conterminous US.National Campbell PGC, Lewis AG, Chapman PM, Crowder AA, Analysis of Trace Elements in ground water, streams, stream Fletcher WK, Imber B, Luoma SN, Stokes PM, Winfrey M. and reservoir sediment, and fish and clam tissue across the Biologically available metals in sediments.National United States.http: yywebserver.cr.usgs.govytraceypubsy Research Council of Canada Associate Committee on Sci- est_v34n2.html (accessed November, 2002). Applied Geochemistry

Applied Geochemistry 22 (2007) 1777–1797 www.elsevier.com/locate/apgeochem

Impact of storm runoff from tropical watersheds on coastal water quality and productivity

E. Heinen De Carlo *, Daniel J. Hoover, Charles W. Young, Rebecca S. Hoover, Fred T. Mackenzie

Department of Oceanography, School of Ocean and Earth Science and Technology, University of Hawaii at Manoa, Honolulu, HI 96822, United States

Available online 24 March 2007

Abstract

Storm runoff in the steep watersheds in Hawaii leads to sediment and freshwater pulses to coastal waters that quickly affect nearshore water quality. This is particularly true in semi-enclosed embayments, such as Kaneohe Bay, Oahu, where water has a relatively long residence time compared to more open coastal areas of the islands. In this paper the authors discuss water quality and productivity in Kaneohe Bay after back-to-back rain events in late November and early Decem- ber 2003, following a particularly dry summer. The short-term biogeochemical response of coastal waters and the ecosys- tem to runoff and physical forcing was evaluated through a combination of continuous in situ measurements and adaptive synoptic sampling carried out on a variety of temporal and spatial scales. Dissolved N:P ratios in Kaneohe Bay, which normally range from 2 to 4, consistent with a previously documented N- limited system, increase to as high as >25 during storm runoff. Order of magnitude increases in nutrients and chlorophyll in the bay shortly after the first storm and subsequent changes in the plankton community structure reflect an evolving bio- logical response stimulated by storm inputs to the bay. Phytoplankton blooms did not draw nutrients down to limiting levels, likely due to grazing pressure by zooplankton, yet phytoplankton were not grazed to limiting levels. As a result, a slow but steady increase of the phytoplankton standing stock was observed over time. Low phosphate levels (<0.2 lM) combined with very high N:P values are typical in Kaneohe Bay waters after most storms and P often becomes the ultimate limiting nutrient. Prior to and after the November–December 2003 storms, how- ever, dissolved P remained near or above 0.3 lM, implying that the system never became P limited and suggesting particle buffering of P concentrations. Furthermore, concentrations of NH3 became elevated (8–16 lM) following the initial storm, first in deep and subsequently in surface waters, and remained high for several months. Remineralization of organic matter transported into southern Kaneohe Bay during the storm possibly contributes nutrients that sustain phytoplankton pro- ductivity for extended periods. 2007 Elsevier Ltd. All rights reserved.

1. Introduction

Increased nutrient loading to coastal waters, * Corresponding author. E-mail address: [email protected] (E.H. De Carlo). principally in the form of N and P, and suspended

0883-2927/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.apgeochem.2007.03.034 1778 E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797 sediment, has been linked to numerous adverse the responses as well. The various processes in play effects on coastal water quality and ecosystems in tropical coastal systems, therefore, occur on time (Caraco, 1995; Tiessen, 1995; Howarth, 1996). His- scales of hours to days (e.g., Wu, 1969; Ringuet, torically, the most visible effects have been associ- 2003; Ringuet and Mackenzie, 2005; Tanaka and ated with point sources such as sewer outfalls, but Mackenzie, 2005). Unfortunately, there has been a legislation and management in the past few decades general lack of continuous measurements of water have dramatically reduced their impacts. Non-point quality, suspended sediment, and sedimentation in sources (NPS) of nutrients, however, are equally Hawaii. Traditional monitoring has largely relied problematic but generally present a less tractable on discrete water and sediment sampling methodol- problem. As a result, NPS pollution is now consid- ogies that are poorly suited to capturing variability ered to be the more pressing problem in environ- on time scales less than weeks or months. Due to mental water quality, and understanding the this mismatch in scales, much of the prior work linkages between land use, NPS nutrient and sus- on coastal ecosystems and reefs in Hawaii has not pended sediment fluxes to receiving waters, and eco- revealed a clear cause and effect relationship system function has become a key factor in between nutrient loading, suspended sediment load- managing high quality, productive coastal waters. ing, physical forcing, phytoplankton blooms, and Kaneohe Bay, in Hawaii, especially the southern overall ecosystem response (Tanaka and Mackenzie, sector of the bay, has a long history of nutrient and 2005). suspended sediment impacts from both point and This paper describes the first concerted effort in NPS. The effects of nutrients on the bay have been Hawaii to observe continuously, over space and studied in some fashion or another for at least a time, the evolution of a rainstorm-derived sediment quarter century (e.g., Smith et al., 1981; Smith and and nutrient laden freshwater plume in Kaneohe Atkinson, 1984; Laws and Allen, 1996; Kinzie Bay and its impact on water quality, productivity et al., 2001; Hoover, 2002; Ringuet, 2003; Ringuet and phytoplankton community structure, and how and Mackenzie, 2005; Tanaka and Mackenzie, physical forcing impacts the recovery of the system 2005). In spite of the abundant research on this from storm perturbations. The study extends the semi-enclosed coastal ecosystem, the nearshore work of Ringuet and Mackenzie (2005), who waters and tributary streams in southern Kaneohe described short-term impacts of freshwater dis- Bay continue to be listed by the Hawaii State charge on phytoplankton productivity in Kaneohe Department of Health as moderately to severely Bay after storms that occurred during several impaired. Point sources of nutrients to the bay have unusually dry seasons prior to the present study. been mostly eliminated, but bay ecosystems have not returned to pre-impact conditions. In particular, 2. Methods nuisance algae continue to out-compete corals in some areas of the bay (Abbot and Smith, unpub- 2.1. Study site lished data). Although a number of hypotheses con- tinue to be tested through investigations by various Kaneohe Bay is located on the eastern or ‘‘wind- researchers in the bay, comprehensive studies of the ward’’ side of the island of Oahu (Fig. 1). It is the potential for NPS nutrient subsidies to act as con- largest sheltered body of water in the Hawaiian trolling factors on the health of the Kaneohe Bay Islands, with a surface area of approximately ecosystem remain sparse. 52 km2, and hosts a large barrier reef and numerous A key problem with prior studies of water quality patch reefs. Multiple small streams that issue from in Hawaiian coastal waters is associated with the the surrounding steep Koolau Range feed into geography of the islands. The high islands of the Kaneohe Bay. The average long-term rainfall of Hawaiian archipelago are subject to intense short nearly 200 cm/a is, however, quite variable and bursts of rain, and the resulting runoff swells ranges from less than 100 cm/a to as high as streams in a few tens of minutes (Tomlinson and 365 cm/a (Giambelluca et al., 1986). Physical mix- De Carlo, 2003). Channelization of streambeds ing processes in the bay are driven by trade winds, associated with widespread urbanization, especially which dominate during most of the year, and mixed on the island of Oahu, accelerates the delivery of tides with an average amplitude of about 68 cm. [urban] runoff to coastal waters, leading to a reduc- The work focuses on the semi-enclosed southern tion in the time scales of inputs and potentially of part of Kaneohe Bay, a body of water with a rela- E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797 1779

Fig. 1a. Map of the Hawaiian Islands and detail of Oahu showing Kaneohe Bay on the East side of the island Modified after De Carlo et al. (2005). Annual rainfall isohyets (mm) are from Giambelluca et al. (1986).

tively long mean water residence time (13 days) and with other streams in the main Hawaiian Islands, an average depth of 9.5 m (Smith et al., 1981). is characterized by extended periods of low flow Southern Kaneohe Bay is generally well-mixed with sporadic periods of intense runoff associated except during periods of high rainfall, when stratifi- with either orographic or low-pressure rainstorms cation of the water column develops as a result of (Tomlinson and De Carlo, 2003). US Geological stream inputs (Smith et al., 1981; Ringuet and Mac- Survey (USGS) records at gauging station 1627200 kenzie, 2005). The extent of stratification is strongly in the Kaneohe watershed reveal an average stream dependent on wind forcing. Large, low-pressure flow from Kaneohe Stream of 0.3 m3/s. Discharge winter storms that originate to the SW of the from other streams the authors have studied on islands, locally known as ‘‘Kona weather’’, typically Oahu can occasionally reach near 500 m3/s during shut down the predominating trade winds and can Kona weather and is highly enriched in sediments contribute to maintenance of stratification for peri- (e.g., Tomlinson and De Carlo, 2003; De Carlo ods of up to a week. et al., 2004) and nutrients (Hoover, 2002). Storm Three perennial streams drain into southern discharge from these streams results in turbid Kaneohe Bay, with nearly 3/4 of the total runoff plumes, which can cover approximately 1/3 of the originating from Kaneohe Stream (Hoover, 2002 south sector of the bay and typically travel north- and Fig. 1b), the remainder of the runoff derives pri- ward along its western edge (Ringuet, 2003; Ringuet marily from Kawa Stream, Keaahala Stream, and, and Mackenzie, 2005). Kaneohe Bay has a well-doc- less importantly, fresh groundwater seeps along umented history of urban development, anthropo- the predominantly basaltic coastline (e.g., McGo- genic nutrient and sediment inputs, and reef wan, 2004). Stream discharge to Kaneohe Bay, as degradation (Cox et al., 1973; Smith et al., 1981; 1780 E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797

Fig. 1b. Site location map of Kaneohe Bay, showing CRIMP: Coral Reef Instrumented Monitoring Platform and other sampling sites (C-buoy, D, E, JD4-6 and KS) in the bay as well as USGS gage stations (Lululuku and Kamoalii) in the watershed.

Evans et al., 1986; Laws and Allen, 1996), which at the fixed stations. In the watershed, data were render the bay an ideal system to study the impact collected with the YSI sonde at 5-min intervals of runoff and NPS pollution on coastal tropical due to the potential for rapidly fluctuating condi- waters. tions in Hawaiian streams (Tomlinson and De Carlo, 2003). In the marine environment, data was 2.2. Field methods collected every 10 min to extend battery life without compromising data quality. Storm water samples Data collection followed a two-pronged were collected automatically from streams by ISCO approach of continuous multiparameter monitoring sequential samplers triggered by water rising above at the CRIMP station within Kaneohe Bay, comple- a preset level (stage). Samples were drawn through mented by two stations in Kaneohe Stream and its acid-cleaned silicone tubing into pre-cleaned 0.5 L upper watershed tributary, Luluku Stream HDPE bottles at 30-min intervals throughout storm (Fig. 1b). The continuous measurements in the events of sufficient magnitude to exceed the stage bay were supplemented during storms by synoptic required to trigger sampling. sampling at C-buoy and a network of other sites The CRIMP station is in Lilipuna channel, in in the bay (latter data not reported here). South Kaneohe Bay (Fig. 1b), through which pass Water quality monitoring was conducted with sediment laden freshwater plumes derived from the YSI multiparameter sondes (Model 6920 or 6600) three streams in the southernmost portion of the E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797 1781 bay (Ringuet, 2003). The Coral Reef Instrumented weighed Whatman GF/C (1.2 lm) glass-fiber filter, Measurement/Monitoring Platform (CRIMP) is an rinsing with deionized (DI) water to remove any approximately 1 m cube framework of galvanized remaining salts. Filters were then dried in an oven pipe to which is attached a series of instruments; at 60 C to a constant weight. Sediment trap sam- the platform rests on the bottom in approximately ples were processed in a manner analogous to 2–3 m of water depth, depending on tides. The water samples filtered for TSS. High precision CRIMP was originally based on a design used in salinity determinations were carried out using an the Southwest Florida Shelf Ecosystems Study AutoSal Model 8400B salinometer. Oxygen was (ESE, 1987). The components of CRIMP include determined by Winkler titration (Carpenter, a YSI Model 6600 multiparameter sonde, a Sontek 1965) using a computer-controlled potentiometric Triton ADV, a LISST-100 particle analyzer, and titration system. sediment traps. Individual probes on the YSI deter- Samples for dissolved inorganic nutrients þ 3 mine chlorophyll-a (chl-a), conductivity, depth, dis- ðDIN : NO3 þ NO2 ; NH4 ; DIP : PO4 ; SiO2Þ were solved O2 (DO), pH, temperature and turbidity. A filtered through acid- and DI water-rinsed What- Triton ADV was used to determine current speeds man GF/C glass-fiber filters and the filtrate frozen and direction in three dimensions. The sediment until analysis. Nutrient concentrations were deter- traps on CRIMP were serviced every three days. mined colorimetrically on a Technicon AutoAna- Automatic data collection devices were serviced lyzer (Parsons et al., 1984). Detection limits were and calibrated as per procedures and frequency rec- 0.1 lM for N and P species and 1 lM for Si. ommended by the respective manufacturers, gener- Total N (TN) and total P (TP) were determined ally on a bi-monthly basis. using the UV-oxidation method (Armstrong and Weather data (solar radiation, wind speed and Tibbitts, 1968) followed by analysis on a Techni- direction, rainfall) were obtained from the Hawaii con AutoAnalyzer. Dissolved organic N (DON) Institute of Marine Biology (HIMB) station on and dissolved organic P (DOP) were calculated Moku o Loe (Coconut Island), near the CRIMP by difference. site, and the USGS rain gage at Luluku. Runoff into Samples were processed for chl-a and pigment the bay was determined using USGS stream gage determinations by filtering a known aliquot (300– data, as well as from stream flow calculated from 1000 mL) of water onto Whatman GF/C filters. ISCO stage data and previously established USGS Samples for chl-a were extracted with 90% acetone rating curves. Tidal data were obtained from the and the solution analyzed on a Turner Designs 10- NOAA tidal gage on Coconut Island and from the AU fluorometer using the acidification method depth sensor on the YSI 6600 sonde on the CRIMP. (Strickland and Parsons, 1972). Filters for chloro- Adaptive synoptic water sampling was conducted phyll and carotenoid pigment extracts were frozen daily during storms and for the subsequent week, at 80 C until analysis. Immediately prior to anal- decreasing to weekly or bi-monthly during non- ysis, filters were ground in a glass tissue homoge- event periods. Surface water samples (10 cm nizer and extracted with 5 mL of acetone for 24 h depth) were collected manually in acid-washed at 0 C in the dark. Extracts were analyzed by high HDPE bottles for laboratory analysis of chl-a, phy- performance liquid chromatography (HPLC) using þ toplankton pigments, NO3 þ NO2 ; NH4 , soluble the method described by Wright et al. (1991).A 3 reactive PO4 (SRP), SiO2, total N (TN), total P Varian model 9012 HPLC equipped with a model (TP), salinity, total suspended solids (TSS), and 9300 autosampler, a Timberline column heater DO. In addition, water sampling was conducted (26 C) and Spherisorb 5 lm ODS2 analytical near the bay bottom at the CRIMP site. Manually (4.6 · 250 mm) column and guard cartridge were collected water samples were stored on ice and pro- used for the determinations. Pigments were detected cessed immediately upon return to the laboratory. with a UV detector (k = 436 nm) following the pro- Depth profiles of water quality parameters were cedures described by Bidigare et al. (2003). Relative obtained using the YSI 6600 sonde. abundances of large net phytoplankton (>64 lm) were determined from preserved net tows. Taxo- 2.3. Laboratory methods nomic composition was assessed by optical micros- copy cell counts; a detailed account of methods TSS were determined gravimetrically by filtering for plankton analysis is available in Scheinberg an aliquot of well-mixed sample onto a pre- et al. (2005). 1782 E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797

3. Results DO, temperature, pH, chl-a, turbidity, solar radia- tion and water depth (tides). The impact of the 3.1. CRIMP time-series data storm(s) is evident for DO, temperature, pH, chl- a, salinity, turbidity, and solar radiation, as well Selected time-series data at 10-min frequency col- as the LISST-100 particle volume data. lected by the CRIMP, wind and solar radiation for Daily water temperatures cover a range of only the period 18 November to 20 December 2003 are 1–2, with the lowest temperatures (23.5 C) presented in Fig. 2 and show diel fluctuations in observed during the period of strong northerly

Fig. 2. CRIMP 10-min frequency time-series data for dissolved O2, temperature, pH, chl-a, salinity, turbidity, particle volume and depth. Wind and solar radiation from the Hawaii Institute of Marine Biology weather station are also shown. E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797 1783 winds that preceded the 29 November 2003 storm. the water depth at the site decreased sufficiently to Temperatures then increase, with highs reaching allow the freshened surface water layer to reach about 25.5 C during several days of southerly the sensors on CRIMP. These decreases are also winds prior to the storm. Diel fluctuations are superimposed on a general trend of increasing peak strongly dampened during the storm, and tempera- salinities as the freshwater plume from the first tures remain near 25 C for several days after the storm begins to be mixed into the water column. storm. A gradual return to diel fluctuations coin- A less pronounced dip in salinity then follows as a cided with the re-establishment of the dominant result of the second storm on 7 December 2003. trade wind pattern about one week after the storm. Turbidity data from CRIMP are characterized DO values oscillate about saturation, as expected by relatively systematic variations within a narrow for a well-mixed shallow water body subject to pho- range (1–3 NTU) upon which are superimposed lar- tosynthesis and respiration, but are depressed dur- ger excursions (max 8 NTU). The first large tur- ing and immediately following the storm of 29 bidity spike in the record (20–21 November 2003) November 2003. The depression in DO likely results occurs during the period of strong northerly winds, from the initially low productivity associated with approximately one week prior to the first storm low light levels and highly turbid surface waters in event. Somewhat broader turbidity maxima occur the Bay immediately after the storm. A very similar during and following the two storms, although tur- trend is observed for fluctuations in pH, with oscil- bidity in bottom waters at CRIMP never reaches lations between pH 8 and 8.1 during non-storm peak values observed in surface waters (60 periods. The pH of the water is depressed immedi- NTU). Several of the turbidity spikes following ately after the first storm owing to the influx of the first storm, as well as other smaller amplitude low pH freshwater. diel fluctuations, correlate with lower salinity sug- The time-series data for chl-a from CRIMP gesting that the CRIMP was recording values in (Fig. 2), although noisy, also display what appear water nearer the surface where more turbid water to be daily oscillations. The noisy signals are most is typically observed. likely due to concentrations that are near the detec- Large increases in particle volume are observed tion limit of the fluorometric probe. An overall during and following the two major storms depression of the chl-a signal, however, is evident (LISST-100 particle data in Fig. 2). These are most during and immediately following the first storm. evident in the largest size class composite (110– It should be kept in mind that chl-a signals at 250 lm). The LISST-100 data also display effects CRIMP reflect activity at 2 m depth and not at of biofouling on the instrument windows (mani- the surface, and values recorded by CRIMP are gen- fested as steady upward drift in Fig. 2). When the erally lower than those measured in synoptic sam- instrument was retrieved for maintenance the need ples collected at the surface. Chl-a data from for window cleaning was clearly evident. Sharp synoptic water samples are described later in this drops in particle volumes were observed when accu- section. The diel fluctuations in chl-a measured by mulated material was physically dislodged or manu- CRIMP correlate positively with solar radiation, ally wiped from the instrument windows, thereby temperature and other chemical parameters that increasing light transmission. are affected by photosynthetic activity (i.e., DO, pH). 3.2. Climatic conditions The salinity of Kaneohe Bay water at the depth of the CRIMP remains close to 35 during non-storm The first storm of the study period, which fol- periods. A drop to a salinity of about 34 during the lowed three relatively dry years and an especially first rain event is followed by a more pronounced dry summer, delivered 26 cm of rain at the Luluku decrease to less than 30 several days later, upon pas- gage on 29 November 2003 (Fig. 3a), and led to sage of the freshwater plume over the CRIMP site. the release of 411,000 m3 of freshwater from Kane- Salinity at CRIMP remained near or below 34 for ohe Stream (Fig. 3b). This is close to what would approximately one week, although several sharp, normally be the highest flow day in an average year. but short-duration, decreases in salinity occur each Rainfall and runoff subsided over the next few days day for approximately five days following the storm but were followed by a second less intense storm, (Fig. 2). The sharp decreases in salinity coincide with 13 cm of rainfall on 7 December, and a release temporally with low tides and reflect times when of 295,000 m3. Cumulative freshwater release from 1784 E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797

Fig. 3a. Rainfall at Luluku and at Coconut Island (CI), and discharge at Kaneohe Stream during the period 22 November 2003–20 December 2003.

Fig. 3b. Discharge and cumulative flow (Cum F) from Kaneohe Stream during the period 22 November 2003–15 January 2004.

Kaneohe Stream over the period 29 November to 21 on 1 December 2003, but the upper 2 m were December 2003 exceeded 2 · 106 m3. strongly affected (Fig. 4a) for several days. As a Climatic conditions were notably different during result, the entire water column at the CRIMP sta- the two storm episodes. Normal predominating tion was influenced by lower salinity water during (NE) trade winds reversed (SW) in late November low tides (see data in Fig. 2). The plume gradually just prior to the first storm. Wind speeds dropped dissipated during a period of one week (Fig. 4a), considerably and were variable for several days; with the water column reverting to well-mixed con- trade wind conditions returned on December 3 but ditions by 6 December 2003. Although the second did not strengthen to normal levels (>30 km h1) storm was accompanied by about 75% of the stream until at least four days later (Fig. 2). Commensurate discharge of the first storm (Fig. 3), only a slight with the ‘‘Kona’’ nature of the first storm and the freshening of the bay occurred on 7 December attendant extensive cloud cover, solar radiation 2003. This fresh water was rapidly mixed into the over Kaneohe Bay decreased dramatically and higher salinity bay seawater within the next 24 h remained very low for three days; followed by vari- (Fig. 4b). able but relatively low insolation until the second storm. Climatic conditions during the latter storm 3.3. Nutrients were more typical, with persistent trade winds blow- ing throughout most of the following two weeks The large quantities of fresh water delivered into (Fig. 2). Kaneohe Bay by the storms strongly impacted Owing to the relatively calm winds and ensuing nutrient concentrations in surface waters (Fig. 5). low hydrodynamic turbulence in the bay following The storm on 29 November 2003 delivered large the first storm, a turbid plume with surface salinities amounts of freshwater with new nutrients to south below 15 (Table 1, Fig. 4a) developed across south- Kaneohe Bay and concentrations in surface waters ern Kaneohe Bay and extended to station SB (see spiked. Concentrations of nutrients did not, how- Fig. 1b). At the C-buoy location, salinity was most ever, increase as much in surface waters following depressed in the upper 20 cm of the water column the second storm (Table 1). Table 1 Water quality data in manually collected samples from southern Kaneohe Bay

Chl (lg/L) Salinity Si (lM) NH4 (lM) CRIMP sfc CRIMP C-buoy CRIMP sfc CRIMP C-buoy CRIMP sfc CRIMP C-buoy CRIMP sfc CRIMP C-buoy deep deep deep deep 11/22/2003 0.81 35.098 2.6 10.99 11/29/2003 1.52 26.569 12.8 0.37 11/30/2003 0.98 0.26 0.56 27.926 34.677 28.628 15.8 3.1 19.7 0.94 11.45 1.18 12/1/2003 0.61 0.78 14.866 13.758 57.2 106.2 0.34 0.59 12/2/2003 0.69 0.28 1.07 22.977 26.072 39.8 27.7 0.93 0.69 12/3/2003 2.13 0.5 4.47 26.944 32.936 23.815 27.1 8.5 41.3 0.86 7.97 0.065 ..D al ta./ApidGohmsr 2(07 1777–1797 (2007) 22 Geochemistry Applied / al. et Carlo De E.H. 12/4/2003 6.38 5.43 24.408 26.196 41.7 28.2 0.065 0.065 12/5/2003 2.77 3.79 14.345 29.518 122.7 23.8 0.065 0.53 12/6/2003 1.24 0.79 1.6 28.824 33.580 33.594 49.7 5.2 17.3 1.21 10.39 9.79 12/7/2003 0.83 0.87 30.461 31.621 15.3 11.5 3.08 4.3 12/8/2003 0.83 0.47 30.794 33.845 20.4 6.9 6.03 10.63 12/9/2003 1.37 0.75 0.99 28.041 33.806 34.032 55.5 11.4 6.5 0.81 9.78 12.8 12/10/2003 0.99 0.78 31.981 34.357 26.0 6.4 4.35 11.95 12/11/2003 0.73 1.34 31.402 34.460 31.4 9.1 4.16 13.11 12/12/2003 1.84 1.57 33.573 34.549 11.9 10.0 9.9 12.43 12/13/2003 1.66 1.38 1.29 31.983 34.429 34.529 28.0 8.8 5.2 8.76 9.15 13.45 12/14/2003 3.08 34.551 5.2 15.4 12/15/2003 2.19 2.88 33.113 34.261 14.5 5.6 9.46 14.4 12/16/2003 1.33 0.56 1.48 33.721 34.641 34.022 8.2 2.2 3.6 10.54 16.13 13.2 12/17/2003 1.47 1.75 34.168 34.375 1.7 1.2 12.94 12.71 12/18/2003 1.55 2.28 34.102 34.357 1.9 1.1 15.28 11.39 12/19/2003 2.8 2.15 33.584 34.404 2.5 1.4 11.4 11.63 12/20/2003 3.24 2.42 2.62 33.120 34.409 33.820 7.4 1.3 3.0 9.56 17.44 11.84 1/13/2004 0.55 0.72 34.095 33.863 3.7 3.8 16.77 16.35 1/27/2004 1.12 32.500 13.0 7.19 2/3/2004 1.81 0.82 1.19 2/7/2004 1.35 0.89 0.89 2/10/2004 0.67 0.55 1.24 33.816 33.453 6.6 10.5 15.49 12.22 2/14/2004 0.59 1.44 1.43 2/17/2004 0.9 1.15 1.72 2/21/2004 1.77 0.75 1.24 2/24/2004 0.8 1.1 1.18 34.232 1.2 15.74 2/28/2004 0.87 0.79 1.23 3/2/2004 1.37 1.16 3/5/2004 2.62 1.46 3/6/2004 1.74 0.44 1.07 3/9/2004 1.15 0.62 0.52 31.222 8.9 3.44 (continued on next page) 1785 1786

Table 1 (continued)

Chl (lg/L) Salinity Si (lM) NH4 (lM) CRIMP sfc CRIMP C-buoy CRIMP sfc CRIMP C-buoy CRIMP sfc CRIMP C-buoy CRIMP sfc CRIMP C-buoy deep deep deep deep

NO3 +NO2 (lM) SRP (lM) DON (lM) DOP (lM) 11/22/2003 0.12 0.51 5.8 0.19 11/29/2003 4.34 0.72 23.3 0.26 11/30/2003 3.39 0.78 3.86 0.58 0.53 0.67 15.3 5.7 24.8 0.14 0.17 0.16 1777–1797 (2007) 22 Geochemistry Applied / al. et Carlo De E.H. 12/1/2003 36.85 34.30 1.18 0.85 19.3 23.7 0.23 0.24 12/2/2003 4.97 4.83 0.53 0.54 17.1 14.2 0.22 0.22 12/3/2003 4.83 2.55 19.43 0.49 0.57 0.52 16.2 5.0 24.7 0.18 0.17 0.2 12/4/2003 7.10 5.81 0.37 0.41 19.4 21.4 0.19 0.16 12/5/2003 32.20 4.52 0.42 0.41 16.1 13.3 0.19 0.16 12/6/2003 6.60 1.65 1.00 0.48 0.52 0.48 11.1 8.8 4.8 0.18 0.15 0.15 12/7/2003 1.24 1.10 0.45 0.44 11.5 10.0 0.18 0.19 12/8/2003 2.74 1.31 0.49 0.48 11.4 7.3 0.14 0.12 12/9/2003 18.23 1.24 1.16 0.5 0.49 0.47 14.4 3.6 4.0 0.16 0.17 0.13 12/10/2003 0.79 0.90 0.53 0.49 14.9 3.6 0.16 0.18 12/11/2003 1.06 2.45 0.51 0.49 10.9 3.8 0.17 0.17 12/12/2003 0.94 0.86 0.45 0.48 4.0 3.9 0.22 0.15 12/13/2003 2.88 0.69 1.35 0.49 0.47 0.48 7.4 4.6 3.8 0.15 0.16 0.15 12/14/2003 0.35 0.46 2.5 0.19 12/15/2003 0.18 0.05 0.43 0.46 2.6 3.5 0.17 0.17 12/16/2003 0.05 0.05 0.27 0.42 0.46 0.31 1.7 3.3 4.1 0.17 0.18 0.33 12/17/2003 0.05 0.18 0.43 0.44 1.5 0.7 0.16 0.15 12/18/2003 0.05 0.05 0.45 0.43 2.2 2.5 0.14 0.15 12/19/2003 0.05 0.05 0.33 0.4 2.6 3.2 0.28 0.17 12/20/2003 0.05 0.05 0.05 0.39 0.37 0.32 13.4 4.7 2.5 0.01 0.06 0.24 1/13/2004 0.05 0.05 0.37 0.48 1/27/2004 0.05 0.29 2/3/2004 2/7/2004 2/10/2004 0.05 0.05 0.31 0.29 2/14/2004 2/17/2004 2/21/2004 2/24/2004 0.05 0.25 2/28/2004 3/2/2004 3/5/2004 3/6/2004 3/9/2004 0.05 0.27 Line missing DIN/DIP DON/DOP DIN (lM) TSS (mg/L) 11/22/2003 21.8 30.3 11.1 1.21 11/29/2003 6.5 89.5 4.7 19.07 011/30/ 7.5 23.1 7.5 109.2 33.8 154.9 4.3 12.23 5.0 3.78 2003 12/1/2003 31.5 41 84 98.9 37.2 34.9 16.12 12/2/2003 11.1 10.2 77.6 64.6 5.9 5.5 2.95 12/3/2003 11.6 18.5 37.5 90.1 29.4 123.3 5.7 10.52 19.5 2.68 12/4/2003 19.4 14.3 101.9 133.5 7.2 5.9 2.46 12/5/2003 76.8 12.3 84.9 82.8 32.3 5.1 2.85 12/6/2003 16.3 23.2 22.5 61.8 58.5 32.1 7.8 12.04 10.8 7.96 12/7/2003 9.6 12.3 63.7 52.6 4.3 5.4 0.98 1777–1797 (2007) 22 Geochemistry Applied / al. et Carlo De E.H. 12/8/2003 17.9 24.9 81.6 61 8.8 11.9 2.63 12/9/2003 38.1 22.5 29.7 90.3 21.4 30.4 19.0 11.02 14.0 0.88 12/10/2003 9.7 26.2 93.3 19.8 5.1 12.9 7.84 12/11/2003 10.2 31.8 64.2 22.3 5.2 15.6 9.57 12/12/2003 24.1 27.7 18.2 25.8 10.8 13.3 1.58 12/13/2003 23.8 20.9 30.8 49.4 28.5 25.1 11.6 9.84 14.8 2.12 12/14/2003 34.2 13.4 15.8 0.73 12/15/2003 22.4 31.4 15.1 20.7 9.6 14.5 2.22 12/16/2003 25.2 35.2 43.5 9.9 18.4 12.4 10.6 16.18 13.5 2.63 12/17/2003 30.2 29.3 9.4 4.5 13.0 12.9 1.39 12/18/2003 34.1 26.6 15.9 16.7 15.3 11.4 2.72 12/19/2003 34.7 29.2 9.3 18.8 11.5 11.7 8.06 12/20/2003 24.6 47.3 37.2 1340 78.5 10.4 9.6 17.49 11.9 2.22 1/13/2004 45.5 34.2 16.8 16.4 3.12 1/27/2004 25 7.2 1.99 2/3/2004 2/7/2004 2.89 2/10/2004 50.1 42.3 15.5 12.3 3.19 2/14/2004 2.59 2/17/2004 3.23 2/21/2004 3.08 2/24/2004 63.2 1.7 2/28/2004 15.8 1.29 3/2/2004 3.21 3/5/2004 3.14 3/6/2004 2.64 3/9/2004 12.9 3.5 3.06 1787 1788 E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797

Fig. 4. Salinity profiles recorded by the YSI 6600 multiparameter sonde at C-buoy during the period following the 11/29/2003 (a) and 12/7/2003 (b) storms.

Nitrogen was initially added as dissolved SRP declined slowly, but displayed no major vari- þ NO3 ðþNO2 Þ, although levels of NH4 increased sub- ability. By the end of the study period concentra- 3 stantially within a few days both in surface and deep tions remained in the range of 0.4–0.5 lMPO4 . waters (Fig. 5). Concentrations of ðNO3 þ NO2 Þ ISCO automatic samplers at the stream sites were and SiO2 peaked on 12 December 2003 at 35 and triggered by rising flow during the first storm, result- 100 lM, respectively, at a time when salinity was ing in the recovery of 24 samples from Luluku and lowest. Concentrations of NO3 and SiO2 subse- 17 samples from Kaneohe Stream (Table 2). Con- quently declined rapidly, especially at C-buoy. The centrations of dissolved nutrients in these samples þ concentration of NH4 remained below 1 lM in sur- were used to calculate the load of nutrients delivered face samples from the CRIMP station until 6 Decem- to Kaneohe Bay by the first storm and to evaluate if ber, at which time concentrations started to climb, stream delivery alone could account for the quickly reaching 10 lM, and remaining in a range observed nutrient concentrations in the freshened of 10–16 lM through the end of February 2004. surface water layer of the storm plume in southern Deep waters at the CRIMP site displayed concentra- Kaneohe Bay (Ha, 2005). In order to calculate the þ tions of NH4 in the range of 8–12 lM for approxi- predicted concentration of nutrients in the storm mately two weeks, and then increased to 16–18 lM plume, it was assumed that Kaneohe Stream dis- by 20 December 2003. charge accounts for 75% of the total freshwater dis- Concentrations of SRP (Fig. 6, Table 1) were ele- charge (Hoover, 2002) and that a freshened surface vated throughout the sampling period (0.5 lM) layer 0.2 m thick extended across 1/3 of southern although spikes of 0.85 and 1.2 lM were observed Kaneohe Bay, consistent with salinity profiles at C-buoy and the CRIMP site, respectively, coinci- (Fig. 4a) and the areal extent of the turbid plume. 3 dent with those for ðNO3 þ NO2 Þ and SiO2 Observed surface concentrations of PO4 at the described above. Subsequently, concentrations of CRIMP and C-buoy sites are quite similar to those E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797 1789

Fig. 5. Concentrations of inorganic nutrients in synoptic water samples collected manually during the study period. Samples were collected from 10 cm at CRIMP (surface) and C-buoy and from 2 m depth at CRIMP (deep). Note: SRP is a measure of dissolved 3 PO4 , the filterable (inorganic) fraction of P.

3 predicted, although the concentration of PO4 at December 2003 at the bay sites are also similar to the CRIMP site on 1 December 2003 is somewhat predicted values, except the observed concentrations higher than was predicted on the basis of the stream the day immediately after the storm are far below 3 input (Table 3). Concentrations of SiO2 on 1 the predicted values. In contrast to PO4 and 1790 E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797

Fig. 6. Absolute concentrations of chlorophyll (lg/L) and concentrations of pigments (ng/ng) relative to total chlorophyll determined by HPLC in manually collected synoptic surface water samples from CRIMP.

SiO2, however, observed concentrations of rapidly over the next two days. Concentrations of 1 NO3 þ NO2 are much lower than the predicted val- chl-a returned to <1 lgL by 7 December 2003 ues the day immediately after the storm and much but climbed at a more measured and steady rate higher than predicted two days after the storm. Pre- over the next three weeks, reaching between 2.4 þ 1 dicted concentrations of NH4 are considerably and 3.2 lgL throughout the entire water column higher than observed concentrations on either the at the CRIMP site and in surface waters at C-buoy second or the third day after the 1 December 2003 on 20 December 2003. Superimposed on this gener- storm. ally increasing trend was a 2-day secondary bloom on 14–15 December that reached 3.2 lgL1 at C- 3.4. Pigments/phytoplankton buoy. Sampling frequency was much lower through- out the remainder of the study period but a general Total chl-a concentrations (Figs. 5 and 6, Table decrease in concentrations of chl-a was observed 1) increased dramatically four days after the first through mid-January 2004 (Figs. 5 and 6). storm. Concentrations were <1 lgL1 before the Relative concentrations of various pigments storm and peaked at 6.4 and 5.4 lgL1 at the (normalized to total chl-a concentrations from the CRIMP surface and C-buoy sites, respectively, on same site) were used as diagnostic tools to evaluate 4 December 2003. Concentrations then declined short-term changes in the phytoplankton popula- E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797 1791

Table 2 Concentrations of nutrients in stream water samples collected during the 11/29/2003 storm 3 Sample ID Sample time Discharge (m /s) PO4 (lM) NO3 (lM) NH4 (lM) Si (lM) L1 12:25 0.600 3.15 6.84 2.46 328.98 L2 12:55 0.840 3.15 9.09 36.57 122.46 L3 13:25 1.639 2.46 9.75 23.46 85.92 L4 13:55 1.808 1.65 12.75 2.34 116.16 L5e 14:25 1.213 1.58 13.25 1.85 116.66 L6 14:55 1.293 1.50 13.74 1.35 117.15 L7e 15:25 2.230 1.44 13.30 1.77 128.28 L8 15:55 1.791 1.38 22.86 2.19 139.41 L9 16:25 1.334 1.17 21.57 1.26 185.85 L10e 16:55 0.796 1.07 23.24 1.11 218.90 L11 17:25 0.609 0.96 24.90 0.96 251.94 L12e 17:55 0.484 0.99 24.41 0.83 262.62 L13 18:25 0.433 1.02 23.91 0.69 273.30 L14 18:55 1.530 1.62 9.33 2.34 140.07 L15e 19:25 1.816 1.34 13.58 1.86 184.50 L16 19:55 1.040 1.05 17.82 1.38 228.93 L17e 20:25 0.867 1.04 15.33 1.22 128.54 L18 20:55 1.192 1.02 12.84 1.05 208.77 L19e 21:25 0.668 0.92 15.93 1.44 232.97 L20 21:55 0.509 0.81 19.02 1.83 257.16 L21e 22:25 0.424 0.86 22.38 1.50 282.81 L22 22:55 0.415 0.90 25.74 1.17 308.46 L23e 23:25 0.415 0.86 24.87 1.28 286.19 L24 23:55 0.401 0.87 24.00 1.14 263.91 K1 13:26 2.48 0.87 2.40 152.58 112.83 K2 13:55 3.87 1.89 15.09 4.02 66.12 K3 14:25 6.92 1.68 17.76 4.17 94.50 K4e 14:55 7.68 1.49 20.54 6.50 126.36 K5 15:25 9.53 1.29 23.31 8.82 158.22 K6e 15:55 10.98 1.05 18.63 6.17 149.66 K7 16:25 11.09 0.81 13.95 3.51 141.09 K8 16:55 11.15 1.05 17.88 3.08 151.68 K9 17:25 11.00 1.29 21.81 2.64 162.27 K10e 17:55 10.09 1.16 19.35 2.37 177.29 K11 18:25 8.06 1.02 16.89 2.10 192.30 K12 18:55 6.97 1.08 17.04 2.52 190.29 K13 19:25 8.77 1.05 21.96 27.24 140.43 K14 19:55 9.39 1.29 23.28 2.70 155.82 K15e 20:25 6.68 1.08 25.28 2.43 170.93 K16 20:55 5.63 0.87 27.12 2.16 186.03 K17 21:25 5.78 2.70 24.12 17.19 182.76 tion during the evolution of blooms (Fig. 6, and 2003 storm, but decreased gradually to a near zero Scheinberg et al., 2005; Hoover et al., 2006). An ini- proportion of chl-a at the end of the two-week study tial spike in the relative concentration of peridinin period. Concentrations of fucoxanthin increased occurred on 2 December 2003, but concentrations steadily towards the end of the pigment study per- of peridinin throughout the rest of the study period iod. Results of determinations of larger (>64 lm) were near or below the detection limit. Prasinoxan- plankton taxonomy and abundances are also pre- thin rapidly increased in proportion to total chl-a sented in Hoover et al. (2006). four days following the 29 November 2003 storm and peaked on the fifth day. Relative concentrations 4. Discussion of Zeaxanthin, a pigment indicative of cyanobacte- ria (among other phytoplankton), exhibited little The rain event on 29 November 2003 marked the variability immediately following the 29 November end of a particularly dry summer and a drought that 1792 E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797

Table 3 Estimated concentrations of nutrients in the southern Kaneohe Bay plume generated by the 11/29/2003 storm þ PO4 NO3 þ NO2 NH4 SiO2 Load (moles) 311 4930 2190 38,300 Estimated plume volume (L) 558 · 106 Estimated concentration based only on Kaneohe Stream 0.557 8.84 3.93 68.6 input Estimated concentrations of nutrients in the plume in 0.74 11.79 5.24 91.47 Kaneohe Bay Observed concentrations of nutrients at CRIMP (surface) 0.58 3.39 0.94 15.76 11/30/2003 Observed concentrations of nutrients at C-buoy 0.68 3.86 1.18 19.72 11/30/2003 Observed concentrations of nutrients at CRIMP (surface) 1.18 36.85 0.34 57.21 12/01/2003 Observed concentrations of nutrients at C-buoy 0.85 34.3 0.59 106.2 12/01/2003 All units in lM unless otherwise indicated. began in 2000 (Ringuet and Mackenzie, 2005). This 2003 storm plume and its subsequent dispersal are first storm of the season was followed approxi- also consistent with previously observed patterns mately one week later (i.e., 7 December 2003) by a (e.g., Ringuet, 2003; Ringuet and Mackenzie, second rain event of slightly smaller magnitude that 2005) and the expected general circulation pattern occurred under considerably different climatic con- that has been confirmed more recently using drifters ditions. The 28 cm of rainfall recorded at Luluku (Ostrander and McManus, 2005, unpublished data). gage (Fig. 3a) on 29 November 2003 resulted in The wind data (Fig. 2) and the salinity profiles the release of 411,000 m3 of freshwater into Kane- (Fig. 4) illustrate an important difference between ohe Bay. Approximately 15 cm of rain fell over the two storms. The first storm was what is locally the next two days thereby contributing to continued known as a ‘‘Kona storm’’. These storms are gener- strong freshwater influx into Kaneohe Bay ated by offshore low-pressure systems that move in a (Fig. 3b). The second event on 7 December 2003, generally northward direction and move across the however, was less intense. Rainfall was only 13 cm island from the SSW, typically several days after a at Luluku and the storm delivered 295,000 m3 of period dominated by strong northerly winds is freshwater to the bay. replaced by light to variable southerly winds. Kona The discharge on 29 November 2003 was close to storms occur primarily during the Hawaii winter that expected for the highest flow day in an average and are characterized by a shut down of the pre- year, resulting in commensurate direct impacts on dominating trade winds, which normally thor- the bay. A turbid plume initiating at the mouth of oughly mix the water column of Kaneohe Bay Kaneohe Stream (Fig. 1b) spread in a north–north- (Smith et al., 1981). The large volume of freshwater westerly direction and extended over about 1/3 of delivered to the bay during the November 2003 South Kaneohe Bay. The inner portion of the plume storm caused a strong stratification of the water col- along the shore extended through Lilipuna channel, umn. Maintenance of the stratification was which separates Moku o Loe (Coconut Island, enhanced by light/variable winds and the minimal Fig. 1b) from mainland Oahu, whereas its outer tidal range of Kaneohe Bay and lasted until 5 margin extended on the seaward side of Moku o December 2003, when the surface layer became Loe. To a first approximation, the C-buoy site much saltier and thinner. The water column was (Fig. 1b), at which synoptic water column sampling mixed completely by 6 December 2003, the day and profiling were carried out, can be considered prior to the second rain event, owing in large part representative of the middle of the plume generated to reestablishment of the predominating NE trade by this storm. The areal extent of the turbid plume wind pattern (Fig. 2). The very slight freshening of was similar to those described by Ringuet (2003) for the surface water (Fig. 4b) during and following considerably smaller rain events in December 2002 the second storm reflects the lack of development and February 2003. The track of the 29 November of a stratified water column. The difference in strat- E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797 1793 ification during the two storms is largely responsible slightly freshened water) was also observed in sur- for the different response of Kaneohe Bay to these face water at the CRIMP site the day after the sec- two storm inputs. ond storm. Because trade winds rapidly mixed in the The storm-derived pulses of freshwater from stream discharge entering the bay during the second Kaneohe Stream into the bay carried a large load event, it is also possible that this second pulse of of suspended particles and nutrients that propa- nutrients at CRIMP resulted from advection of rel- gated across the bay as a turbid low salinity surface atively nutrient-enriched shallow groundwater leak- (60.2 m; Fig. 4) plume. The evolution of the plume, ing into the bay from the coastline rather than from however, was relatively slow because of the calm cli- stream inputs. matic conditions following the first storm. Two days The second storm on 7 December 2003, although were required for concentrations of NO3 þ NO2 delivering only slightly less than 3/4 the freshwater and SiO2 to reach a maximum at the C-buoy and discharge to the bay observed during the first storm, surface CRIMP sites (Fig. 5). The subsequent rapid was not accompanied by a commensurate input of decrease in concentrations of nutrients at C-buoy is nutrients (Fig. 5), suggesting that the sources of attributed to a combination of dilution as the storm nutrients to the bay during the two storms may plume mixed into the water column, and uptake of have been different. It is proposed that an impor- nutrients by phytoplankton. Chlorophyll concentra- tant fraction of the nutrient load delivered to the tions did not peak until two days later, as light levels bay during the first storm was of anthropogenic ori- increased gradually, thereby allowing enhanced gin. For example, aerosol NOx derived from com- photosynthetic activity. The continuous records at bustion of automotive fuels in urbanized areas of CRIMP (e.g., DO, temperature, pH, chl-a, and Kaneohe could have accumulated in the watershed solar radiation; Fig. 2) are consistent with mainte- during the dry summer of 2003 and been mobilized nance of the turbid low salinity plume and low light by rain and runoff during the first storm of signifi- levels until approximately 2 December 2003, follow- cant magnitude. Additionally, accumulation in sur- ing which water quality records begin to once again face soils of N and P derived from fertilizer use in display diel cycles reflecting photosynthetic activity. the upper watershed would also provide an addi- Another pulse in NO3 þ NO2 and SiO2 at the tional source of readily mobilized nutrients. De surface CRIMP site on 5 December 2003 was not Carlo et al. (2004) reported that first flush samples observed at C-buoy. Because this peak occurred collected during rainstorms in the upper (and non- two days prior to the second storm, it is unlikely urbanized) reaches of a tributary to Kaneohe to derive from enhanced stream discharge. This Stream contain enhanced concentrations of trace spike in the surface waters at the CRIMP site was elements associated with fertilizers. These authors also accompanied by a large decrease in surface also noted that the elevated concentrations are not water salinity, suggesting another local freshwater sustained during the later stages of storms. It is thus source rather than propagation of a broadly defined plausible that the storm of 29 November 2003, plume northward from Kaneohe Stream. Because a because it followed a particularly dry summer, effec- somewhat freshened lens of water is often observed tively represented a first (albeit seasonal) flush of at the surface of the CRIMP site, with no corre- the watershed. The second storm, however, would sponding feature at C-buoy, only a short distance not likely mobilize as high a load of anthropogenic away but farther offshore, it is hypothesized that a nutrients because of the rather short accumulation separate pulse, possibly of groundwater, entered period between the two storms. the bay along the coastline and only affected the Lateritic soils in Hawaii contain high concentra- areas immediately adjoining the shore. McGowan tions of Fe and Al oxyhydroxides, which have a rel- (2004) has documented inputs of groundwater con- atively high pH of zero net surface charge (pHpzc, taining elevated concentrations of nutrients e.g., Parks, 1965, 1967; Sposito, 1984) in addition ðNO3 and SiO2Þ along the windward coast of to clay minerals. These soils consequently exhibit Oahu. Although coastal groundwater inputs are positive residual surface charges at the pH typical more pronounced in northern Kaneohe Bay than of soils in Hawaii (Uehara and Gillman, 1981;G. in the present study area, it is possible that ground- Uehara, personal communication 2000) and can water seeps occasionally contribute significant retain NO3 through electrostatic attractions (as well 3 amounts of both freshwater and nutrients. Interest- as PO4 through specific interactions). Over ingly, another smaller pulse of nutrients (and extended periods, some NO3 may penetrate the 1794 E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797

3 soils and become enriched in shallow groundwater, Concentrations of SRP ðPO4 Þ, the filterable especially if/when surface soils become saturated. inorganic fraction of P, displayed a behavior during Surface soils eroded and carried into streams by and after the event of 29 November 2003 that storm runoff subsequently enter saline waters of contrasted that observed in other runoff events. higher pH and ionic strength, where their surface Ringuet and Mackenzie (2005) reported that runoff charge reverses quickly. The charge reversal leads events in Kaneohe Bay are characterized by termi- 3 to a rapid release of NO3 , although PO4 release nation of phytoplankton blooms when DIN ðNO3 þ has been shown to be much slower, consistent with þNO2 ; and NH4 Þ and SRP become depleted. its specific binding to the soils (e.g., De Carlo and These authors also suggested that P may be the Dollar, 1997). primary limiting nutrient following runoff. This The lack of high dissolved concentrations of SiO2 hypothesis is supported by high DIN:DIP ratios and NO3 at C-buoy (Fig. 5) during the second (27) in stream runoff reported by Hoover storm, although not driven by an absence of anthro- (2002), and is consistent with the present observa- pogenic SiO2, is consistent with the anthropogenic tions of DIN:DIP ratios exceeding 40 in surface and soil sources discussed above. Freshwater runoff waters of Kaneohe Bay (Table 1). Ringuet and from the first storm likely flushed significant Mackenzie (2005) also reported that surface SRP amounts of groundwater from the wetted portion is depleted prior to DIN, further supporting the of the vadose zone and upper phreatic zone that P-limitation hypothesis. In the current study, how- had accumulated SiO2 during the antecedent dry ever, concentrations of dissolved SRP, which summer through water–rock interaction. Concen- peaked at 0.85 and 1.18 lM at C-buoy and surface trations of SiO2 were elevated in surface water sam- CRIMP, respectively, only declined slowly thereaf- ples from the CRIMP site after the second storm; ter and remained between 0.4 and 0.5 lM through- although this might initially seem inconsistent with out the study period (Fig. 6, Table 1). Furthermore, the arguments presented above, an input of dee- SRP was already elevated on 22 November 2003, a per-seated and older groundwater, as hypothesized week before the first storm. This suggests that bay earlier, could account for the high concentrations waters were not subject to P-limitation at any time of SiO2. Such deeper-seated groundwater, however, during the study period, possibly as a result of 3 would not be expected to display as elevated a con- PO4 being released slowly from soil particles sus- tent of NO3 , for the reasons described above. pended in seawater for several days following a Dissolved N in the first storm plume was pre- shorter period (hours) of somewhat enhanced dominantly present as NO3 ðþNO2 Þ, although some release (e.g., De Carlo and Dollar, 1997; Chun, water samples collected in the stream during the first unpublished results 2006). Recent experiments con- storm contained considerable concentrations of ducted with soils from Kaneohe and Kaneohe Bay þ þ NH4 (Table 2). Concentrations of NH4 in deep seawater resulted in final concentrations of SRP waters at CRIMP (Table 1, Fig. 5) were elevated near 1 lM (Chun, unpublished results, 2006). Final from the onset of sampling and remained relatively concentrations in release experiments varied with constant for several weeks, before increasing to near different particle types (i.e., stream bank soils vs. 18 lM approximately one week after the second stream sediments) but were remarkably close to þ storm event. Because concentrations of NH4 were peak concentrations of SRP in Kaneohe Bay below 1 lM in surface waters at the CRIMP and shortly after the storm. C-buoy sites until 6 December 2003, climbed to The highest concentrations of nutrients observed 10–16 lM over the next few days, and remained in during the current study were about one order of this range through the end of February 2004, a dis- magnitude greater than those reported at nearby tinct source of this readily utilized nutrient, other stations in Kaneohe Bay between 1998 and 2001 than freshwater from the stream, must be responsi- under background conditions (Kinzie et al., 2001). ble for the elevated concentrations. The two storms Although concentrations of NO3 þ NO2 and SiO2 delivered large amounts of particulate matter in the declined within three weeks after the storms to bay; hence, it is hypothesized that remineralization below typical levels observed during the 3-a CISNet þ of particulate organic N deposited by the storm rep- study (Kinzie et al., 2001), concentrations of NH4 þ resented a benthic source of NH4 to surface waters and SRP remained well above the baseline estab- and subsidized the nutrient budget of the bay for lished by these authors. It should also be noted that several months. although the storm on 7 December 2003 also deliv- E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797 1795 ered large quantities of freshwater and presumably reached those observed in the surface (Fig. 5). Fur- nutrients to south Kaneohe Bay, concentrations of thermore, absolute concentrations of chlorophyll in nutrients in bay waters did not spike significantly the YSI record (Fig. 2) should be viewed with cau- following this storm. tion owing to operation of the fluorometric probe A remarkable concurrence exists between the con- near its detection limit. centrations of SRP in Kaneohe Bay and dissolved Concentrations of chl-a shown in Fig. 5 (water 3 concentrations of PO4 predicted by Ha (2005). sample data) declined rapidly along what appears SRP measured in water samples from the two bay to be a linear dilution line as stratification of the sites are quite similar to those predicted, although water column broke down after the return of the SRP at the surface CRIMP site on 1 December trade winds. Concentrations dropped to <1 lgL1 2003 was somewhat higher than predicted. This sug- by 7 December 2003 but recovered steadily over gests that stream inputs can account, in some cases, the next three weeks. Concentrations of chl-a 3 1 for a significant fraction of the PO4 budget of at least reached between 2.4 and 3.2 lgL throughout proximal southern Kaneohe Bay and likely exert a the water column at the CRIMP site and in surface strong control on primary productivity. waters at C-buoy on 20 December 2003 as photo- synthetic activity was enhanced by strong insolation 4.1. Pigments/phytoplankton (Fig. 2) and continued nutrient subsidies likely pro- vided by remineralization of organic matter depos- The phytoplankton response in Kaneohe Bay fol- ited in the bay after the storm. Superimposed on lowing the 29 November 2003 event was influenced this generally increasing chl-a trend was a 2-day sec- by a combination of factors including turbidity of ondary bloom on 14–15 December 2003 that the water, nutrient concentrations and solar radia- reached 3.2 lgL1 at C-buoy. Although sampling tion. The relatively calm conditions after the storm frequency was much lower throughout the remain- (Fig. 2) prevented thorough mixing of the turbid der of the study period and trends are more difficult freshwater plume with bay water and resulted in to ascertain, a general decrease in concentrations of low light penetration through the water column. chl-a was observed through mid-January 2004. Most likely at this time with reduced wind-induced After this time, concentrations of chl-a (not shown) mixing of bay waters, much of the physical mixing returned to levels of 1–2 lgL1 with a minor peak was driven by tidal exchange that strongly influ- of 2.6 lgL1 observed at the CRIMP site on 5 ences the retention time of stream discharge waters March 2004. A substantially smaller peak was in southern Kaneohe Bay. Ringuet and Mackenzie observed at C-buoy on the same day. Overall, how- (2005) showed that the recovery time of Bay waters, ever, chl-a remained at concentrations near or to a storm perturbation, is related in part to the above the CISNet averages for south Kaneohe tidal range. Furthermore, reduced insolation Bay (Kinzie et al., 2001) through mid March 2004. (Fig. 2) owing to extensive cloud cover associated Smaller rain events that probably also delivered with the low-pressure system also limited light lev- nutrients to the bay occurred during the late winter els. This likely contributed to a delayed response and likely influenced water column conditions, but of the phytoplankton to the new input of nutrients. are not discussed here. Although diel cycles of chl-a shown in Fig. 2 are The increase in peridinin (Fig. 6), an accessory consistent with continued photosynthetic activity, pigment of dinoflagellates, indicated a rapid dinofla- the depressed concentrations of chl-a at CRIMP gellate response two days after the first storm as pre- correlate well with relatively constant water temper- viously observed by Ringuet and Mackenzie (2005). atures, lower salinities, lower pH and lower DO, The increase in prasinoxanthin, the main accessory and suggest an inhibition of photosynthetic activity pigment of prasinophytes that occurred several days even in the presence of enhanced nutrient concen- later, suggests that these small, autotrophic flagel- trations. Not surprisingly, the return of the trade lates bloomed immediately following dinoflagel- winds, attendant clearing of the cloud cover, and lates. Although zeathanthin concentrations were increased insolation were accompanied by a return low and showed little variability immediately fol- of enhanced diel cycles of biochemical constituents lowing the storm, the decreasing zeathanthin con- associated with photosynthesis and increased chl-a centrations at the end of the study period may even at the depth of the CRIMP. Concentrations reflect a gradual decrease in photosynthetic bacte- of chl-a at depth at the CRIMP site, however, never ria. It should be kept in mind, however, that the 1796 E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797 abundant, sub-micron sized Synechococcus spp. in participation this project could not have been car- Kaneohe Bay (Scheinberg et al., 2005) were likely ried out successfully. We are particularly grateful underrepresented in the samples because of the rel- for the able laboratory and field assistance of Steph- atively large pore size of the GF/C filters used in this anie Christensen, Mung Fa Chun, Fenny Cox, study. Hong Hyunh Ha, Dave Hashimoto, Franc¸ois Pa- Contrasting the pigments representative of the quay, Stephanie Ringuet, Mathieu Voluer and Yves smaller phytoplankton cells, a steady increase in Veillerobe. We are grateful for reviews and com- the relative concentrations of fucoxanthin suggests ments from Jason R. Price and two anonymous that diatoms dominated the phytoplankton commu- reviewers. This project was supported in part by a nity towards the end of the two-week study period. grant to EHDC and FTM by the NOAA/University This observation is consistent with results presented of Hawaii Sea Grant College Program (R/EL 33) by Ringuet and Mackenzie (2005) for storms during under institutional Grant Number NA16RG2254. the prior but much drier winter season of 2002– This is SOEST Contribution Number 6818 and 2003. An in-depth discussion of the pigment results UH Sea Grant Contribution Number JC-04-19. presented here, net (>64 lm) phytoplankton taxon- omy and abundances, and their relationship to the zooplankton community structure is available in References Scheinberg et al. (in press). Armstrong, F.A.J., Tibbitts, S., 1968. Photochemical combustion of organic matter in sea water for nitrogen, phosphorus and 5. Conclusions carbon determination. J. Marine Biol. Assoc. U.K. 48, 143– 152. This study demonstrated the utility of combining Bidigare, R.R., Benitez-Nelson, C., Leonard, C.L., Quay, P.D., real time observations of biogeochemically relevant Parsons, M.L., Foley, D.G., Seki, M.P., 2003. Influence of a parameters with periodic synoptic sampling over a cyclonic eddy on microheterotroph biomass and carbon export in the lee of Hawaii. Geophys. Res. Lett. 30 (6), broader geographic area to assess the response of 1318. doi:10.1029/2002GL016393. receiving waters to land-derived inputs of nutrients. Caraco, N.F., 1995. Influence of human populations on phos- The response of southern Kaneohe Bay to nutrient phorus transfers to aquatic systems: a regional scale study inputs associated with storm runoff is typically using large rivers. Phosphorus Global Environment. John rapid, generally occurring within a day of inputs, Wiley & Sons, Chichester, pp. 35–244. Carpenter, J.H., 1965. The accuracy of the Winkler method but is influenced by physical forcing mechanisms. for dissolved oxygen analysis. Limnol. Oceanogr. 10, 135– Because of the sluggish circulation of southern 140. Kaneohe Bay imposed by geographic constraints, Cox, D.C., Fan, P.F., Chave, K.E., Clutter, R.I. Gundersen, wind-driven, stress-induced mixing of the water col- K.R., Burbank Jr., N.C., Lau, L.S., Davidson, J.R. 1973. umn, in concert with tidal fluctuations, appears to Estuarine pollution in the State of Hawaii; Kaneohe Bay Study. University of Hawaii Water Resources Research play an important role in distributing nutrients Center WRRC Report #31. throughout bay waters. De Carlo, E.H., Dollar, S.J., 1997. Assessment of Suspended Storm inputs to the bay from streams and Solids and Particulate Nutrient Loading to Surface Runoff groundwater provide large quantities of dissolved and the Coastal Ocean in the Honokowai Drainage Basin, and particulate nutrients which quickly drive phyto- Lahaina District. Maui. Final Report to NOAA/Algal Blooms Project and Hawaii State DOH. plankton productivity, but changing N:P ratios De Carlo, E.H., Beltran, V.L., Tomlinson, M.S., 2004. Compo- potentially cause shifts in community structure that sition of water and suspended sediment in streams of ultimately affect the overall phytoplankton urbanized subtropical watersheds in Hawaii. Appl. Geochem. response. Remineralization of terrestrial particulate 19, 1011–1037. organic matter deposited in the bay during storms ESE, 1987. Environmental Science and Engineering, LGL Ecological Research Associates, and Continental Shelf Asso- appears to provide continued benthic nutrient ciates. Southwest Florida shelf ecosystems study, data inputs to the bay for periods extending weeks to synthesis report. Minerals Management Service [MMS] Con- months after the original storm delivery. tract nr 14-12-0001-30276. MMS, New Orleans, LA; MMS 87-0023. Acknowledgements Evans, C.W., Maragos, J.E., Holthus, P.F., 1986. Reef corals in Kaneohe Bay: six years before and after termination of sewage. In: Jokiel, P.L., Richmond, R.H., Rogers, R.A. The authors express their appreciation to a large (Eds.), Coral Reef Population Biology, Honolulu, Hawaii number of students and technicians, without whose Institute of Marine Biology, Tech. Report. 37, pp. 91–100. E.H. De Carlo et al. / Applied Geochemistry 22 (2007) 1777–1797 1797

Giambelluca, T.W., Nullet, M.A., Schroeder, T.A., 1986. Rain- Ringuet, S., Mackenzie, F.T., 2005. Controls on nutrient and fall atlas of Hawaii. Department of Land and Natural phytoplankton dynamics by storm runoff events southern Resources, State of Hawaii. Kaneohe Bay, Hawaii, Estuaries 28, 327–337. Ha, H.-H., 2005. Nutrient inputs to Kaneohe Bay during storms, Scheinberg, R.D., Calbet, A., Landry, M.R., 2005. Grazing of unpublished undergraduate thesis, Global Environmental two common appendicularians on the natural prey assem- Sciences, University of Hawaii. blage of a tropical coastal ecosystem. Mar. Ecol. Prog. Ser. Hoover, D.J., 2002. Fluvial nitrogen and phosphorus inputs to 294, 201–212. Hawaiian coastal waters: storm loading, particle-solution Smith, S.V., Atkinson, M.J., 1984. Phosphorus limitation of net transformations and ecosystem impacts, unpublished PhD production in a confined aquatic ecosystem. Nature 307, 626– discussion, Department of Oceanography, University of 627. Hawaii. Smith, S.V., Kimmerer, W.J., Laws, E.A., Brock, R.E., Walsh, Hoover, R.S., Hoover, D., Miller, M., Landry, M.R., De Carlo, T.W., 1981. Kaneohe Bay sewage experiment: perspectives on E.H., Mackenzie, F.T., 2006. Zooplankton response to storm ecosystem responses to nutritional perturbation. Pacific Sci. runoff in a tropical estuary: bottom-up and top-down 35, 379–395. controls. Mar. Ecol. Prog. Ser. 318, 187–201. Sposito, G., 1984. The Surface Chemistry of Soils. Oxford Howarth, R.W. (Ed.), 1996. Nitrogen Cycling in the North University Press, New York. Atlantic Ocean and its Watersheds. Kluwer Academic, Strickland, J.D.H., Parsons, T.R., 1972. A Practical Handbook Dordrecht. of Seawater Analysis, second ed. Fisheries Research Board of Kinzie III., R.A., Mackenzie, F.T., Smith, S.V., Stimson, J., Canada, Ottawa. 2001. CISNet: linkages between a tropical watershed and reef Tanaka, K., Mackenzie, F.T., 2005. Statistical and stability ecosystem. Final Project Report to NOAA, Honolulu, Uni- analysis of subtropical ecosystem dynamics in southern versity of Hawaii. Kaneohe Bay, Hawaii. Ecol. Model. 188, 296–326. Laws, E.A., Allen, C.B., 1996. Water quality in a subtropical Tiessen, H. (Ed.), 1995. Phosphorus in the Global Environment. embayment more than a decade after diversion of sewage John Wiley & Sons, Chichester. discharge. Pacific Sci. 50, 194–210. Tomlinson, M.S., De Carlo, E.H., 2003. The need for high- McGowan, M.P., 2004. Submarine Groundwater Discharge: resolution time series data to characterize Hawaiian Freshwater and Nutrient Input Into Hawaii’s Coastal Zone, streams. J. Am. Water Resour. Assoc. (JAWRA) 39, 113– University of Hawaii Masters Thesis. 123. Parks, G.A., 1965. The isoelectric points of solid oxides, solid Uehara, G., Gillman, G.P., 1981. The mineralogy, chemistry and hydroxides, and aqueous hydroxo complex systems. Chem. physics of tropical soils with variable charge clays. Westview Rev. 65, 177–198. Press, Boulder Colorado. Parks, G.A., 1967. Aqueous surface chemistry of oxides and Wright, S.W., Jeffrey, S.W., Mantoura, R.F.C., Llewellyn, C.A., complex oxide minerals. Adv. Chem. Ser. 67, 121–160. Bjoernland, T., Repeta, D., Welschmeyer, N., 1991. Improved Parsons, T.R., Maita, Y., Lalli, C.M., 1984. A Manual of HPLC method for the analysis of chlorophylls and carote- Chemical and Biological Methods for Seawater Analysis. noids from marine phytoplankton. Mar. Ecol. Prog. Ser. 77, Pergamon Press, New York. 183–196. Ringuet, S., 2003. Biogeochemical impacts of storm runoff on Wu, I.-P., 1969. Hydrograph study and peak discharge determi- water quality in southern Kaneohe Bay, Hawaii, M.S. Thesis, nation of Hawaiian small watersheds: Island of Oahu. Water Department of Oceanography, University of Hawaii. Resources Research Center, University of Hawaii.

Marine biological and water quality surveys for Hanapepe Loop drainage outfall improvements, Maunalua Bay, Honolulu, O‘ahu, Hawai‘i1

October 24, 2011 DRAFT AECOS No. 1262F

Stacey Kilarski AECOS, Inc. 45‐939 Kamehameha Hwy, Suite 104 Kāne‘ohe, Hawai‘i 96744 Phone: (808) 234‐7770 Fax: (808) 234‐7775 Email: [email protected]

Introduction

In October 2011, AECOS, Inc. biologists conducted water quality and marine surveys to assess the marine resources on the limestone bench and reef flat fronting a drainage outfall at 150 Hanapepe Loop, Portlock, O‘ahu. (Fig. 1). The storm drain line collects street runoff from Hanapepe Loop and the surrounding urbanized area and terminates at the drain outfall into Maunalua Bay. The reconstruction of Hanapepe Loop drainage outfall (the Project) proposed by the County and County of Honolulu includes replacement of the concrete drainage outfall and headwall structures along the shoreline. This involves removal and reconstruction of the existing concrete headwall structure (21 linear ft) with cast‐in‐place concrete and removal and replacement of a portion (7 linear ft) of the existing drain outlet. All in‐water work will be done with hand equipment. A water‐inflated dam will be used as a cofferdam to isolate the work area from the marine environment. Any water pumped from the construction site (and water from the dam itself) will be pumped to a GeotubeTM mobile dewatering system, located on Hanapepe loop within a 30 cu.yd roll‐off container. The purpose of this survey and report is to identify sensitive biological resources that may be impacted by the Project.

1 Report prepared for Bill’s Engineering for use in project permitting. This document will become part of the public record for the project.

AECOS, Inc. [FILE: 1238A.DOCX] Page | 1 Marine biology and water quality surveys MAUNALUA BAY

Figure 1. Project location on the Island of O‘ahu.

Site Description

The Project is located along the shoreline adjacent to a residential lot at 150 Hanapepe Loop, in Portlock on the southern shore of O‘ahu. The shoreline in the project vicinity faces nearly due west into Maunalua Bay. To the east is Koko Head, and to the north is Maunalua Bay Launch Ramp facility. A coastal access pathway near Kawaihoa Point provides public access to the western shoreline of Koko Head. People frequent the small beach just north of the project site (Koke`e Beach Park); surf at breaks “Pillars” or “China wall;” spearfish around Portlock Point; and fish from the breakwater, footbridge, and marine bench in Maunalua Bay Launch Ramp facility.

Fronting the drainage outfall is a wide (approximately 40 ft) intertidal limestone bench with depressions that are periodically filled with seawater (tidepools; Fig. 2). Other portions of the limestone bench remain inundated throughout the tidal cycle. A fringing reef extends about 915 m (3,000 ft) offshore and is primarily made up of an ancient limestone platform covered by algae and having very little coral cover, a characteristic typical of shallow reef areas off the south coast of O‘ahu.

AECOS Inc. [FILE: 1262F.docx] Page | 2 Marine biology and water quality surveys MAUNALUA BAY

Figure 2. Drainage outfall and headwall fronted by limestone bench and water filled depressions at Project site.

Methods

Marine Biota

On October 11, 2011, AECOS biologists conducted a biological reconnaissance survey of marine resources at the Project vicinity. Biologists walked along the intertidal limestone bench during an ebbing tide. The survey began at 9:30 am, 41 minutes before the 0.4‐ft low tide (higher low water or HLW). Biologists snorkeled the waters offshore from the Project area, approximately 21 m (70 ft) from the shoreline. Water visibility during the survey was about 2 m (6 ft) on the reef flat. Marine algae, fishes, and macroinvertebrates were identified in the field and verified with various texts (Hoover, 1999; Huisman, et al. 2007). A listing, including relative abundances, of species of macroalgae (limu) and marine animals observed in both areas is presented as Appendix A.

AECOS Inc. [FILE: 1262F.docx] Page | 3 Marine biology and water quality surveys MAUNALUA BAY

Water Quality

To characterize the water quality around the drainage outfall and headwall, and to contribute to establishing baseline water quality conditions for the Project area, three sampling stations were established (Fig. 3). Station “Impact” is located in the water‐filled limestone depression in front of the drainage outfall. Sta. “South” is located off the shoreline at the edge of the limestone bench, approximately 15 m (50 ft) south of the drainage outfall. Sta. “North” is located off the shoreline at the edge of the limestone bench, approximately 30 m (90 ft) north of the outfall.

Figure 3. Hanapepe Loop Drainage Outfall Improvement Project water quality sampling stations.

Field measurements for temperature, salinity, pH, and dissolved oxygen (DO) were taken in situ at each monitoring station. Water samples were collected from just below the surface at each station in appropriate containers, preserved on ice, and taken to AECOS laboratory in Kāne‘ohe, O‘ahu. Collected samples were analyzed for turbidity, total suspended solids (TSS), nitrate+nitrite, ammonia, total nitrogen, total phosphorus, and chlorophyll α. Table 1 lists the field instruments and analytical methods used to evaluate these samples.

AECOS Inc. [FILE: 1262F.docx] Page | 4 Marine biology and water quality surveys MAUNALUA BAY

Table 1. Analytical methods and instruments used for the October 11, 2011 water quality analyses to characterize nearshore waters off Hanapepe Loop drainage outfall, Maunalua Bay, O‘ahu.

Analysis Method Reference Instrument

Temperature EPA 170.1 USEPA (1983) YSI Model 85 DO meter Grasshoff et al. Salinity bench salinometer AGE Model 2100 salinometer (1999) pH EPA 150.1 USEPA (1983) Hannah pocket pH meter

Dissolved Oxygen EPA 360.1 USEPA (1983) YSI Model 85 DO meter Turbidity EPA 180.1, Rev. 2.0 USEPA (1993) Hach 2100N Turbidimeter Total Suspended SM 2540D SM (1998) Mettler H31 balance Solids Nitrate+Nitrite EPA 353.2 Rev. 2.0 USEPA (1993) Technicon AutoAnalyzer II nitrogen Grasshoff et al. Ammonia nitrogen SM 4500‐NH3 B/C Technicon AutoAnalyzer II (1999) persulfate digestion Grasshoff et al. Total Nitrogen Technicon AutoAnalyzer II EPA 353.2 (1999) Total Phosphorus EPA 365.1 Rev. 2.0 USEPA (1993) Technicon AutoAnalyzer II Chlorophyll α SM 10200 H SM (1998) Turner Model 112 fluorometer

Results

Water Quality

Water quality results are summarized in Table 2. Values for temperature, dissolved oxygen (DO), and salinity at Sta. Impact were elevated compared to the stations North and South. The water was supersaturated (saturation greater than 100%) with oxygen at all three stations. Salinity measured at Sta. Impact is indicative of some freshwater input, which is also reflected in the low pH. Chlorophyll α, a direct indicator of phytoplankton biomass, was slightly elevated at all three stations, as were turbidity and total suspended solids (TSS). Ammonia (a dissolved form of inorganic nitrogen) was elevated at Sta. Impact and North, and nitrate‐nitrite (another dissolved inorganic nitrogen moiety) was very high at Sta. Impact. Total nitrogen (TN), which includes inorganic, organic, and particulate nitrogen moieties, was high, especially at Sta. South. Total phosphorus (TP) was low at all three stations.

AECOS Inc. [FILE: 1262F.docx] Page | 5 Marine biology and water quality surveys MAUNALUA BAY

Table 2. Water quality characteristics of nearshore waters off Hanapepe Loop, Maunalua Bay, O‘ahu as determined at LHW on October 11, 2011.

Time Temp. Salinity DO DO sat. pH sampled (°C) (ppt) (mg/l) (%) Impact 11:09 27.4 28.9 9.61 143 7.99 North 11:20 26.4 33.4 7.43 112 8.37 South 11:15 26.0 33.4 6.93 104 8.30

Nitrate Total Total Turbidity TSS Ammonia + nitrite N P Chl α (ntu) (mg/l) (g N/l) (g N/l) (g N/l) (g P/l) (g/l) Impact 0.95 10 34 334 596 16 0.67 North 1.34 13 35 10 228 5 0.54 South 1.12 15 14 14 896 <4 0.67

Marine Biology

Drainage outfall and headwall ‐ The drainage outfall is sparsely covered with small numbers of barnacles (Chthamalus proteus). No other life was observed on the drainage outfall structure. The headwall structure was void of any growths. The limestone platform adjacent to the outfall and the water‐filled depression directly in front of the outfall and headwall does not host any macroalgae or marine animals (see Fig. 2, above).

Limestone bench ‐ The area of limestone bench closest to the drainage outfall and headwall is submerged only at high tide, and therefore hosts organisms adapted to conditions of the upper interidal. Most notable in this area are false ‘opihi (Siphonaria normalis or ‘opihi ‘awa) and littoral snails (dotted periwinkle; Littoraria pintado). Small numbers of nerite snails (Nerita picea and N. polita) and a‘ama crab (Grapsus tenuicrustatus) occur in the intertidal zone.

At the mid‐littoral zone, depressions in the limestone bench are regularly exposed and submerged by tides. The water‐filled depressions host a diverse assemblage of organisms including: goby (Bathygobius sp.), marbled blenny (Entomacrodus marmoratus), snakehead cowry (Cypraea caputserpentis), rock‐ boring urchin (Echinometra mathaei and E. oblonga), ashy sea cucumber (Holothuria cinerascens), zooanthids (Zoanthus sp.), and coralline algae nodules (Fig 4), One small black‐lipped pearl oyster (Pinctada margaritifera) was

AECOS Inc. [FILE: 1262F.docx] Page | 6 Marine biology and water quality surveys MAUNALUA BAY

observed. The pearl oyster is protected throughout the State of Hawai‘i and it is prohibited to “catch, take, kill, possess, remove, sell or offer for sale” a state protected species (HAR §13‐83‐1).

Figure 4. The mid intertidal zone of the Project vicinity, with a view of the south extent of limestone bench and water‐filled depressions.

A few small (<5 cm diameter) coral heads or fragments (Porites spp. and Pocillopora spp.) are present in the tide pools; these likely cast up during high sea conditions from parent colonies on the adjacent reef flat. Algae found in the tide pools include: Caulerpa taxifola, Cladophora catenata, Halimeda discoidea, Microdictyon setchellianum, Actinotrichia fragilis, Chamipa parvula, Galaxaura rugosa, Gelidiella acerosa, Laurencia mcdermidiae, Peyssonnelia rubra, Dictyota sandvicensis, Dictyoperis sp., Padina sanctae‐crucis, P. australis, Sargassum echinocarpum, S. polyphyllum, Turbinaria ornata, and Lyngba majuscule, with Padina spp. and Sargassum spp. being the most abundant. At the edge of the limestone bench, waves crest and with a rising tide, allow seawater to flood the

AECOS Inc. [FILE: 1262F.docx] Page | 7 Marine biology and water quality surveys MAUNALUA BAY

shoreline. Indicative of the lower littoral zone, algae is abundant, specifically the brown algal species, Sargassum polyphyllum and S. echinocarpum (Fig 5).

Figure 5. Lower littoral zone of limestone bench, where the brown algae, Sargassum echinocarpum and S. polyphyllum, are abundant.

Reef Flat ‐ The reef flat offshore of the limestone bench has roughly 1.5 to 2.5 m (5 to 7 ft) of water depth with a slightly undulating limestone bottom and widely scattered coral outcrops (Fig. 6). The limestone bottom is covered with fine sediment, low‐growing, turfy algae, and is deeply scoured by boring urchins (E. mathaei and E. oblonga), which are the most commonly seen macro‐ invertebrate on the reef flat. Other invertebrates are uncommon and include purse shells (Isognomon californicum and I. perna) and sea cucumbers (Actinopyga mauritiana, Holothuria atra, and H. cinerascens).

Common algae species observed in the area include Jania micrarthrodia. Lithophyllum kotschayanum, Galaxura rugosa, Padina australis, and Halimeda

AECOS Inc. [FILE: 1262F.docx] Page | 8 Marine biology and water quality surveys MAUNALUA BAY

discoidea. Other species less commonly seen include Caulerpa serrulata, C. sertularioides, Microdictyon setchellianum, Coelothrix irregularis, Dichotomeria marginata, Ganonema papenfussil, Peyssonnelia rubra, Portieria hornemannii, Tricleocarpa cylindrica, Padina sanctae‐crucis, Dictyota spp., Neomeris sp. and the invasive species, Acanthophora spicifera, is rarely seen.

Figure 6. Scattered coral outcrops and scoured limestone of the reef flat area offshore from the Project vicinity.

In the Project vicinity, corals are represented by at least 11 species. The most common coral genus is Porites with three species represented: P. lobata (lobe coral), P. lutea (mound coral), and P. evermanni (brown lobe coral). Next most common is Pocillopora, with two species: Poc. ligulata (thin cauliflower coral) and Poc. meandrina (cauliflower coral). Also present are Montipora capitata (rice coral), Montipora patula (sandpaper rice coral), Cyphastrea ocellina (ocellated coral), Leptastrea purpurea (crust coral), and L. bewickensis (bewick coral) all in low numbers and having low cover.

In the area directly seaward of the shoreline to approximately 12 m (40 ft) offshore, corals are generally small, ranging in size between 5 to 25 cm in

AECOS Inc. [FILE: 1262F.docx] Page | 9 Marine biology and water quality surveys MAUNALUA BAY

diameter. The diameter of one L. bewickensis colony was measured at 40 cm and a few Pocillopora colonies of diameter 30 cm were encountered. Remnant coral growth is evident by several large, partially dead, mound‐forming Porites colonies. In an area approximately 18 m (60 ft) offshore from the front of the bench, several large (>100 cm) Porites colonies are present. Most of these colonies appear quite healthy, with minimal mortality or damage.

Figure 7. Common fish species observed on reef flat offshore Project vicinity include convict tang (A. triostegus) and saddle wrasse (T. duperrey).

Thirty species of fishes were identified in the survey area (see Appendix A) The most common fishes on the reef flat are wrasses (Family Labridae), with numerous juvenile saddle wrasse (Thalassoma duperrey) and belted wrasse (Stethojoulis balteata) present. Various damselfish, including the brighteye damsel (Plectroglyphidodon imparipennis), Hawaiian sergeant (Abudefduf abdominalis), and Hawaiian Gregory (Stegastes marginatus) are also present. Small schools mullet (Mugil cephalus) are seen over the shallow reef flat. Convict tang (Acanthurus triostegus) and brown surgeonfish (A. nigrofuscus) feed on the sparse algae present (Fig. 7, above). Uncommonly seen species include palenose parrotfish (Scarus psittacus), square‐spot goatfish

AECOS Inc. [FILE: 1262F.docx] Page | 10 Marine biology and water quality surveys MAUNALUA BAY

(Mulloidichthys flavolineatus), and spotted boxfish (Ostracion meleagris). Reef triggerfish (Rhinecanthus rectangulus), Hawaiian whitespotted toby (Canthigaster jactator), Ambon toby (Canthigaster amboinensis), Raccoon butterflyfish (Chaetodon lunula), Hawaiian lizardfish (Synodus ulae), manybar goatfish (Parupeneus multifasciatus), bluespine unicornfish (Naso unicornis), ringtail surgeonfish (A. blochii), orangeband surgeonfish (A. olivaceus), and barred moray (Echnidna polyzona) are all rare fishes in the survey area.

Discussion

Water Quality

Water quality samples collected on October 11, 2011 represent low tide conditions on that date and results could vary depending upon tidal stage. Much of the environment immediately seaward of the Project site is intertidal. The upper tidal areas of the limestone bench is dry at low tide and nearly completely inundated at high tide, while the low intertidal area remains flooded throughout the tidal cycle. Waves crest over the limestone bench with the rising tide allowing seawater to flood the area.

The waters of Maunalua Bay between Paikō Peninsula and Koko Head are classified in the Hawai‘i Water Quality Standards (HDOH, 2009) as a Class A “embayment” and as a “Class II nearshore reef flat.” Maunalua Bay is listed on the Hawai‘i Department of Health (HDOH), 2006 list of impaired waters in Hawai‘i, prepared under Clean Water Act §303(d) (HDOH, 2008). This listing is based upon water quality data collected by HDOH in Maunalua Bay (Geocode ID HIW00016) and indicates Maunalua Bay may not meet Hawai‘i water quality standards for total nitrogen (TN), nitrate‐nitrite (NO2+NO3), ammonia (NH4), and chlorophyll α in the wet season (presumably meaning not meeting the wet criteria, applicable when freshwater inflow equals or exceeds 1% of the embayment volume per day).

The primary purpose of water quality measurements presented in this report is to characterize the existing aquatic environment, not to set baseline values or determine compliance with Hawaii’s water quality standards. In fact, state criteria for all nutrient measurements, turbidity, and chlorophyll α are based upon having a representative geometric mean value to compare with the standard; a minimum of three separate samples per sampling location would be required to generate this mean. Ideally, multiple samplings would encompass a range of conditions “typical” for this location, including but not limited to such events as rising, versus ebbing tide, wet versus dry weather periods, and even storm events. The criteria presented in Table 3 may be used, together with a

AECOS Inc. [FILE: 1262F.docx] Page | 11 Marine biology and water quality surveys MAUNALUA BAY

data collected from a series of preconstruction sampling events, to develop decision rules as part of the data quality objectives (DQO) process in an applicable monitoring and assessment program (AMAP) developed in accordance with the required Clean Water Act, Section 401, Water Quality Certification (AECOS, 2011).

Table 3. Selected State of Hawai‘i water quality criteria for embayments (HAR §11‐54‐5.2; HDOH, 2009).

Geometric mean Value not to be exceeded Value not to be not to exceed given more than 10% of the exceeded more than Parameter value time 2% of the time Turbidity 1.5* 3.00* 5.00* (NTU) 0.40** 1.00** 1.50** Total Nitrogen 200.00* 350.00* 500.00* (µgN/L) 150.00** 250.00** 350.00** Nitrate‐ Nitrite 8.00* 20.00* 35.00* (µgN/L) 5.00** 14.00** 25.00** Ammonia 6.00* 13.00* 20.00* (µgN/L) 3.50** 8.50** 15.00** Total 25.00* 50.00* 75.00* Phosphorus (µgP/L) 20.00** 40.00** 60.00** * Wet criteria apply when the average fresh water inflow from the land equals or exceeds one percent of the embayment volume per day. ** Dry criteria apply when the average fresh water inflow from the land is less than one percent of the embayment volume per day. The following non‐specific criteria are applicable to both “wet” and “dry” conditions.  pH shall not deviate more than 0.5 units from 8.1, except at coastal locations where and when freshwater may depress the pH to a minimum of 7.0.  Dissolved oxygen shall not be less than 75% saturation.  Temperature shall not vary more than 1 °C from ambient.  Salinity shall not vary more than 10 percent from natural or seasonal changes.

Water quality at the Project site is influenced by stormwater runoff and freshwater input. Project plans to use a water‐inflated dam will isolate the work area which will help to ensure that water quality of the adjacent reef flat is

AECOS Inc. [FILE: 1262F.docx] Page | 12 Marine biology and water quality surveys MAUNALUA BAY

protected from sedimentation and project‐related runoff. Any brief periods of impaired water quality associated with construction should have minimal impacts on the nearby reef flat as daily water exchange is high in this area.

ESA Listed Species

No listed (endangered or threatened; USFWS, 2009) species were encountered in the October 2011 surveys. Sea turtles, spinner dolphins, and humpback whales were not observed during the survey; however, they may occur in the Project vicinity (although well off the shore).

Green Sea Turtle — The most common sea turtle in the Hawaiian Islands is the honu or green sea turtle (Chelonia mydas), an inhabitant of the shallow waters of Maunalua Bay. In 1978, green sea turtle in Hawaiian waters became listed as threatened under the Endangered Species Act (USFWS, 1978, 2001). The National Marine Fisheries Service and Fish and Wildlife Service (NMFS‐FWS, 1998) developed a recovery plan for U.S. Pacific populations of the green sea turtle, a document that aids management decisions to protect the population towards recovery.

Threats to green sea turtles in Hawai‘i, in order of greatest to least, include: disease and parasites, accidental fishing take, and boat collisions. Lessor threats include: entanglement in marine debris, loss of foraging habitat to development, and ingestion of marine debris (NMFS‐USFWS, 1998). Turbidity (murky water) does not appear to deter green sea turtles from foraging and resting areas. Construction projects on the south shore of O‘ahu, at Hawaii Kai and off of Kapolei, have found sea turtles adaptable and tolerant of construction‐related disturbances (Brock, 1998a,b).

Traditionally, sea turtles rest in deeper water during the day where they use reef features to shelter themselves (Smith, 1999) and feed over the shallow reef flats at night (Balazs et al., 1987). Before acquiring a status of threatened in Hawaiian waters, green sea turtles would flee upon encountering human swimmers. In recent years, however, green sea turtles here have become exceedingly tolerant of human presence and now regularly come to shallows to feed during the day as well as night (Balazs, 1996).

The green sea turtle diet consists primarily of benthic macroalgae, which the shallow reefs of the main Hawaiian Islands provide in abundance. Red macroalgae make up 78% of the turtle diet and green macroalgae make up 12% (Arthur and Balazs, 2008). The single most consumed algal species is Acanthophora spicifera, which is an introduced species first recorded in Hawai‘i in 1950 (Huisman et al., 2007). A. spicifera was observed in the Project vicinity,

AECOS Inc. [FILE: 1262F.docx] Page | 13 Marine biology and water quality surveys MAUNALUA BAY

but in very small amounts and not likely to be of significance for the green sea turtle.

Humpback Whale — The humpback whale or kohola (Megaptera novaeangliae) was listed as endangered in 1970 under the Endangered Species Act. Prior to protection, the North Pacific humpback whale population was estimated at under 1,000 individuals, compared with an estimated original abundance of at least 15,000 (Rice, 1978; Johnson and Wolman, 1984). In 1993 it was estimated that there were 6,000 whales in the North Pacific Ocean, and that 4,000 of those regularly came to Hawai‘i. The population is estimated to be growing at between 4% and 7% per year. Today, as many as 10,000 humpback whales may visit Hawai‘i each year (HIHWNMS, 2008).

The waters of Maunalua Bay are within the Hawaiian Islands Humpback Whale National Marine Sanctuary (HIHWNMS). Humpback whales normally occur in Hawaiian waters annually from November to May with the peak between January and March (HIHWNMS, 2008). The Project will not directly affect humpback whales, and sounds generated from Project activities are not anticipated to be substantial enough to cause an acoustic disturbance to protected species in nearshore waters. The following in‐water acoustic impact thresholds are currently used by NMFS to assess potential impacts to marine mammals (NOAA, 2005; Don Hubner, Pers. Comm., 2011): Onset of Injury (also known as the Permanent Threshold Shift) is 180 dB for cetaceans (whales, porpoises) and 190 dB for pinnipeds (seals). The Onset of Behavioral Disturbance (also known as the Temporary Threshold Shift/Areal Avoidance) is 160 dB when an impulsive sound and 120 dB when a continuous, non‐impulsive sound.

Conclusions

Minimal direct impacts from the Project can be anticipated for the intertidal limestone bench. No sensitive biological resources occur in the immediate Project area. Because all Project work will be done with hand equipment transported through the house lot and no heavy equipment will be placed on the intertidal bench, impacts to the few small corals and one pearl oyster present in the tide pools will be avoided. The adjacent reef flat is expected to be only indirectly impacted by the Project.

Potential exists for short term impacts from construction activities on the water quality of the nearshore environment. Possible impacts from construction include introducing sediment into the bay and increasing pH from concrete

AECOS Inc. [FILE: 1262F.docx] Page | 14 Marine biology and water quality surveys MAUNALUA BAY

pours. Brief periods of impaired water quality associated with construction should have no long term impacts on the intertidal limestone bench and nearby reef flat as daily water exchange is high in these areas. Impacts can be mitigated by employing best management practices (BMPs). A requirement of the Nationwide Permit is to follow the standard BMPs (USACE, 2011)

An Applicable Monitoring and Assessment Program (AMAP) for the Project has been prepared for this Project (AECOS, 2011). The AMAP describes the monitoring requirements and the data quality objectives to be met during water quality monitoring efforts for the Clean Water Act, Section 401 Water Quality Certification that must be obtained from the Hawai‘i Department of Health for the Project. The intent of the AMAP is to conduct water quality sampling and analysis to monitor potential impacts caused by in‐water work. The AMAP includes baseline (preconstruction), during‐construction, and postconstruction monitoring. Data collected as part of the AMAP will be used to assess the adequacy of BMPs applied during construction and will facilitate assessing the impacts of the project on Maunalua Bay. If shown to be necessary by the monitoring data, BMPs will be modified during construction to protect water quality.

References

AECOS, Inc. (AECOS). 2002. An aquatic resources survey for shoreline improvements and nearshore dredging at 567 Portlock Road, O‘ahu, Hawai‘i. Prep. for: Wilson Okamoto & Associates. AECOS No. 933: 15 pp.

______. 2011. Applicable Monitoring and Assessment Program for Clean Water Act (CWA), Section 401 Water Quality Certification, Hanapepe Loop Drain Outfall Improvements, Maunalua Bay, O‘ahu, Hawai‘i. AECOS No. 1262H: 24 pp.

Arthur, K. E. and G. H. Balazs. 2008. A comparison of immature green turtles (Chelonia mydas) diets among seven sites in the main Hawaiian Islands. Pacific Science 62(2): 205–217.

Balazs, G. H, R. G. Forsyth, and A. K. H. Kam. 1987. Preliminary Assessment of Habitat Utilization by Hawaiian Green Turtles in their Residential Foraging Pastures. U.S. Department of Commerce, NOAANMFS‐SWFC 71, Honolulu. 115 pp.

Brock, R. E. 1988a. Green sea turtle population monitoring during blasting work at West Beach, Oahu. Final Report. Prep. for Alfred A. Yee & Assoc. 15 pp.

AECOS Inc. [FILE: 1262F.docx] Page | 15 Marine biology and water quality surveys MAUNALUA BAY

Brock, R. E. 1988b. Green turtles (Chelonia mydas) at Hawaii Kai, Hawaii: An analysis of the impacts with the development of a ferry system. Prep. for Sea Engineering, Inc. 26 pp

Hawai‘i Department of Health (HDOH). 2008. 2006 State of Hawai‘i Water Quality Monitoring and Assessment Report: Integrated Report to the U.S. Environmental Protection Agency and the U.S. Congress Pursuant To Sections §303(D) and §305(B), Clean Water Act (P.L. 97‐117). 279 pp.

______. 2009. Hawai‘i Administrative Rules, Title 11, Department of Health, Chapter 54, Water Quality Standards. State of Hawai‘i, Department of Health. 90 pp.

Hawaiian Islands Humpback Whale National Marine Sanctuary. 2008. Available online at URL: http://hawaiihumpbackwhale.noaa.gov/explore/whale_watching .html; last accessed October 18, 2011.

Hoover, J. P. 1999. Hawai‘i’s Sea Creatures: A Guide to Hawai‘i’s Marine Invertebrates. Mutual Publishing, Honolulu, Hawai‘i. 366 pp.

Hubner, D. 2011. USFWS, pers. communication (email).

Huisman, J. M., I. A. Abbott, C. M. Smith. 2007. Hawaiian Reef Plants. Hawai‘i Sea Grant College Program, Honolulu, Hawai‘i. 264 pp.

Johnson, J. H. and Wolman A. A. 1984. The humpback whale, Megaptera novaeangliae. Marine Fisheries Review, 46: 30‐37.

National Oceanic and Atmospheric Administration (NOAA). 2005. Department of Commerce, National Oceanic and Atmospheric Administration. Small Takes of Marine Mammals Incidental to Specified Activities; Low‐Energy Seismic Survey in the Southwest Pacific Ocean. Federal Register, 70 (35; February 10, 2005): 8768 ‐ 8783.

National Marine Fisheries Service and Fish and Wildlife Service (NMFS‐FWS) 1998. Recovery plan for U.S. Pacific populations of the green turtle (Chelonia mydas). National Marine Fisheries Service, Silver Spring, Maryland, USA.

Rice, D.W. 1978. The humpback whale in the North Pacific: Distribution, exploitation, and numbers. In: K. S. Norris and R. Reeves (Eds.), Report on a workshop on problems related to humpback whales (Megaptera

AECOS Inc. [FILE: 1262F.docx] Page | 16 Marine biology and water quality surveys MAUNALUA BAY

novaeangliae) in Hawai‘i, Report to the U.S. Marine Mammal Commission, Washington, D.C., pp. 29‐44)

U.S. Army Corps of Engineers (USACE). 2011. Letter to Collins Lam, City & County of Honolulu Department of Design & Construction, from USACE, dated September 8, 2011.

U.S. Fish and Wildlife Service (USFWS). 2001. 50 CFR 17. Endangered and Threatened Wildlife and Plants. Notice of Findings on Recycled Petitions. Federal Register, 66 (5; Monday, January 8, 2001): 1295‐1300.

______. 2005. Part II. Department of the Interior, Fish and Wildlife Service. 50 CFR 17. Endangered and Threatened Wildlife and Plants; Review of Species That Are Candidates or Proposed for Listing as Endangered or Threatened: Annual Notice of Findings on Resubmitted Petition: Annual Description of Progress on Listing Actions. Federal Register, 70 (90; Wednesday, May 11, 2005): 24870‐24934.

______. 2009. Endangered and Threatened Wildlife and Plants. 50CFR 17:11 and 17:12. Available online at URL: http://www.fws.gov/endangered/; last accessed on January 3, 2010.

______. 2010. USFWS Threatened and Endangered Species System (TESS). Available online URL: http://ecos.fws.gov/tess_public/StartTess.doc; last accessed on August 3, 2010.

AECOS Inc. [FILE: 1262F.docx] Page | 17

Appendix A. Inventory of aquatic biota observed in the Hanapepe Loop Project area, Maunalua Bay, O‘ahu, on October 11, 2011.

PHYLUM, CLASS, ORDER, Abundance by FAMILY location Genus species Tidal Reef flat Common name & Status Hawaiian name bench ALGAE CHLOROPHYTA GREEN ALGAE Caulerpa serrulata Ind. R Caulerpa sertularioides Ind. U O Caulerpa taxifolia Ind. U Cladophora catenata Ind. A Dictyosphaeria versluysii Ind. U Halimeda discoidea Ind. O C Microdictyon Ind. C O setchellianum Neomeris sp. Ind. U Ulva fasciata sea lettuceInd. U pālahalaha

RHODOPHYTA RED ALGAE Acanthophora spicifera Nat. U U Actinotrichia fragilis Ind. C Amphiroa sp. Ind. R Avrainvillea amadelpha Nat. R Champia parvula Ind. U Coelothrix irregularis Ind. U Dasya irridescens Ind. R Dichotomeria marginata Ind. U Galaxaura rugosa Ind. O C Ganonema papenfussil Ind. U U Gelidiella acerosa Ind. C O Hydrolithon onkodes Ind. U U Hydrolithon reinboldii Ind. U U Jania micrarthrodia Ind. C Laurencia mcdermidiae Ind. U Liagora sp. Ind. U Lithophyllum Ind. U C kotschayanum Peyssonnelia rubra Ind. C O Portieria hornemannii Ind. O Tricleocarpa cylindrica Ind. O

Appendix A [FILE: 1262F.docx] Page | 18

PHYLUM, CLASS, ORDER, Abundance by FAMILY location Genus species Tidal Reef flat Common name & Status Hawaiian name bench OCHROPHYTA BROWN ALGAE . Colpomenia sinuosa Ind. R R Dictyota acutiloba alani Ind. O O Dictyota sandvicensis alani End. C O Dictyota ceylanica Ind. A Dictyopteris sp. Ind. C R Hydroclathrus clathratus Ind. R Padina sanctae‐crucis Ind. A O Padina australis Ind. A C Ralfsia expansa Ind. U Sargassum kala Ind. A echinocarpum Sargassum polyphyllum kala Ind. A Turbinaria ornata Ind. C U

CYANOBACTERIA Lyngbya majuscule U Symploca hydnoides Ind. U CNIDARIA, ANTHOZOA, ZOANTHINARIA Palythoa caesia blue‐gray zoanthid Ind. O Zoanthus sp. O CNIDARIA, ANTHOZOA, SCELRACTINIA POCILLOPORIDAE Pocillopora ligulata thin cauliflower Ind. U coral Pocillopora meandrina cauliflower coral Ind. C PORITIDAE Porites lobata lobe coral, Ind. C pohaku puna Porites lutea mound coral Ind. O Porites evermanni brown lobe coral C Porites sp. R ACROPORIDAE Montipora capitata rice coral Ind. O Montipora patula sandpaper rice Ind. O coral FAVIIDAE Cyphastrea ocellina ocellated coral Ind. R Leptastrea purpurea crust coral Ind. R Leptastrea bewickensis bewick coral Ind. R

Appendix A [FILE: 1262F.docx] Page | 19

PHYLUM, CLASS, ORDER, Abundance by FAMILY location Genus species Tidal Reef flat Common name & Status Hawaiian name bench MOLLUSCA, GASTROPODA PATELLIDAE Siphonaria normalis false ‘opihi Nat. O ‘opihi‐‘awa NERITIDAE Nerita picea black nerite Nat. A pipipi Nerita polita polished nerite C kūpe‘e CYPRADIDAE Cyprae caputserpentis snake‐head cowry Ind. C U leho kupa Cyprea helvola THAIDADAE Morula uva grape drupe Ind. R LITTORINIDAE Littoraria pintado dotted periwinkle Ind. A pipipi kolea MOLLUSCA,BIVALVIA, PTERIIDAE Pinctada margaritifera black‐lipped pearl Ind. R oyster ISOGNOMONIDAE Isognomon black purse shell Ind. A californicum Isognomon perna brown purse shell Ind. U nahawele ARTHROPODA, CIRRIPEDIA, BALANIDAE Chthamalus proteus Proteus’ rock Ind. O† barnacle ARTHROPODA, MALACOSTRACA, DECAPODA, DIOGENIDAE Calcinus laevimanus. left‐handed hermit Ind. C crab GRAPSIDAE Grapsus tenuicrustatus thin shelled rock Ind. R crab;‘a‘ama MAJIDAE Schizophroida hilensis Hilo collector crab Ind. R pāpa‘limu

Appendix A [FILE: 1262F.docx] Page | 20

PHYLUM, CLASS, ORDER, Abundance by FAMILY location Genus species Tidal Reef flat Common name & Status Hawaiian name bench ECHINODERMATA, SEA URCHINS ECHNOIDEA, ECHINOMETRIDAE rock boring urchin Echinometra mathaei Ind. A A ‘ina kea oblong boring Echinometra oblonga Ind. C U urchin; ‘ina Heterocentrus red pencil urchin Ind. R mammillatus hā‘uke‘uke‘ula‘ula ECHINODERMATA, SEA CUCUMBERS

HOLOTHUROIDEA HOLOTHURIDAE Ind. white‐spotted sea Actinopyga mauritiana cucumber Ind. U loli black sea Holothuria atra cucumber loli okuhi Ind. U kuhi Holothuria cinerascens ashy sea cucumber Ind. A C VERTEGRATA, BONY FISHES ACTINOPTERYGII MURAENIDAE Echnidna polyzona barred moray Ind. R puhi leihala BLENNIIDAE Blenniella gibbifrons bullethead blenny Ind. U

pāo‘o Istiblennius zebra Hawaiian zebra End. C

blenny ACANTHURIDAE Acanthurus triostegus convict tang Ind. U A manini Acanthurus brown surgeonfish Ind. C nigrofuscus mā‘i‘i‘i Acanthurus blochii ringtail surgeonfish Ind. R pualu Acanthurus olivaceus orangeband Ind. surgeonfish R na‘ena‘e Naso unicornis bluespine Ind. unicornfish U kala MUGILIDAE Mugil cephalus striped mullet Ind. C C ‘ama‘ama

Appendix A [FILE: 1262F.docx] Page | 21

PHYLUM, CLASS, ORDER, Abundance by FAMILY location Genus species Tidal Reef flat Common name & Status Hawaiian name bench MULLIDAE Mulloidichthys square‐spot Ind. flavolineatus goatfish U weke‘ā Parupeneus manybar goatfish Ind. R multifasciatus moano POMACENTRIDAE Abudefduf abdominalis Hawaiian sergeant End. U mamo Abudefduf sordidus black spot sergeant Ind. C U kūpīpī Abudefduf vaigensis Indo‐Pacific Ind. R sergeant Plectroglyphidodon bright‐eye Ind. O imparipennis damselfish LABRIDAE Thalassoma duperrey saddle wrasse End. C hinalea lauwili Thalassoma Christmas wrasse Ind. C trilobatum ‘awela belted wrasse End. C Stethojulius balteata ‘omaka ZANCLIDAE Zanclus cornutus morrish idol Ind. R kihikihi CLUPEIDAE Spratelloides delicate Ind. A delicatulus roundherring SYNODONTIDAE Synodus ulae Hawaiian lizardfish Ind. R ulae SCARIDAE Calotomus sp parrotfish ‐‐ O Scarus psittacus palenose parrotfish Ind. U uhu CHAETODONIDAE Chaetodon lunula raccoon Ind. butterflyfish R kikakapu BALISTIDAE Rhinecanthus reef triggerfish, Ind. rectangulus humuhumu R nukunuku ahupua‘a

Appendix A [FILE: 1262F.docx] Page | 22

PHYLUM, CLASS, ORDER, Abundance by FAMILY location Genus species Tidal Reef flat Common name & Status Hawaiian name bench KUHLIIDAE Kuhlia xenura Hawaiian flagtail End. C āholehole OSTRACIIDAE Ostracion meleagris spotted boxfish: End. U moa GOBIIDAE Bathygobius sp. Tide pool goby Ind. C TETRAODONTIDAE Canthigaster jactator Hawaiian End. R whitespotted toby Canthigaster ambon toby Ind. R amboinensis

KEY TO SYMBOLS USED: Abundance categories: R – Rare – only one or two individuals observed. U – Uncommon – several to a dozen individuals observed. O – Occasional – seen irregularly in small numbers C – Common – observed everywhere, although generally not in large numbers. A – Abundant – observed in large numbers and widely distributed. Status categories: End – Endemic – species found only in Hawaii Ind. – Indigenous – species found in Hawaii and elsewhere Nat. – Naturalized – species were introduced to Hawaii intentionally, or accidentally. Other symbol used: † located on drainage outfall

Appendix A [FILE: 1262F.docx] Page | 23 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

Eutrophication and the dietary promotion of sea turtle tumors

! View article

Phosphoria @phosphoria31 231 days ago RT @thePeerJ: Eutrophication and the dietary promotion of sea turtle tumors https://t.co/6MicJ6503B #WorldOceansDay More oceans research… " reply # retweet ♥ like

PeerJ — Life and Environment Journal @thePeerJ 233 days ago Eutrophication and the dietary promotion of sea turtle tumors https://t.co/6MicJ6503B #WorldOceansDay More oceans research in our Aquatic Biology Section https://t.co/MWc5Ld7A7r https://t.co/BcW5Eg35rj " reply # retweet ♥ like

∠ PeerJ

Eutrophication and the dietary promotion of sea turtle tumors Kyle S. Van Houtan &​1,2, Celia M. Smith3, Meghan L. Dailer3, Migiwa Kawachi3

Published September 30, 2014 PubMed 25289187

The tumor-forming disease fibropapillomatosis (FP) has afflicted sea turtle populations for decades with no clear cause. A lineage of α-herpesviruses associated with these tumors has existed for millennia, suggesting environmental factors are responsible for its recent epidemiology. In previous work, we described how herpesviruses could cause FP tumors through a metabolic influx of arginine. We demonstrated the disease prevails in chronically eutrophied coastal waters, and that turtles foraging in these sites might consume arginine-enriched macroalgae. Here, we test the idea using High-Performance Liquid Chromatography (HPLC) to describe the amino acid profiles of green turtle (Chelonia mydas) tumors and five common forage species of macroalgae from a range of eutrophic states. Tumors were notably elevated in glycine, proline, alanine, arginine, and serine and depleted in lysine when compared to baseline samples. All macroalgae from eutrophic locations had elevated arginine, and all species preferentially stored environmental nitrogen as arginine even at oligotrophic sites. From these results, we estimate adult turtles foraging at eutrophied sites increase their arginine intake 17–26 g daily, up to 14 times the background level. Arginine nitrogen increased with total macroalgae nitrogen and watershed nitrogen, and the invasive rhodophyte Hypnea musciformis significantly outperformed all other species in this respect. Our results confirm that eutrophication substantially increases the arginine content of macroalgae, which may metabolically promote latent herpesviruses and cause FP tumors in green turtles.

Introduction

Fibropapillomatosis (FP) is a chronic and often lethal tumor-forming disease in sea turtles (Fig. 1A). It became a panzootic in green turtles in the 1980s, prompting concern that it was a serious threat to their global conservation (Chaloupka et al., 2008; Herbst, 1994). Though most green turtle population indices have increased steadily since (Seminoff et al., 2014), the disease remains prevalent and in several locations its incidence is still increasing (Van Houtan, Hargrove & Balazs, 2010). Advances in understanding the cause of FP

https://peerj.com/articles/602/ Page 1 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

have recently centered on environmental factors, with diverse lines of evidence from genomics to epidemiology supporting this hypothesis (Aguirre & Lutz, 2004; dos Santos et al., 2010; Ene et al., 2005; Herbst et al., 2004; Van Houtan, Hargrove & Balazs, 2010). The ecological promotion of the disease is made further interesting as FP tumors have a proposed viral origin.

Figure 1: (A) Juvenile green turtle (Chelonia mydas) severely afflicted with fibropapillomatosis, a tumor-forming disease associated with α-herpesviruses. Photo: August 2012 Makena, Maui (credit: Chris Stankis, Flickr/Bluewavechris). (B) Amino acid profiles from turtle tissues show fibropapilloma tumors are notably enriched in glycine, proline and arginine, and depleted in lysine. Glycine is a known tumor biomarker; proline aids herpesvirus infections; and arginine and lysine promote and inhibit herpesviruses, respectively. Bars represent the average difference between tumor and baseline tissue for 12 individual turtles, percent changes from baseline percent total protein listed in parentheses, error bars are SEM. Bar color indicates P values from two-tailed paired t-tests. (C) Underlying histograms for arginine content in baseline and tumor tissue samples, bars are raw values, curves are smoothed trend.

DOI: 10.7717/peerj.602/fig-1

Early studies discovered DNA from α-herpesviruses in FP tumors, but found adjacent tissues from diseased turtles, as well as samples from clinically healthy turtles, to be free of herpes DNA (Lackovich et al., 1999; Lu et al., 2000; Quackenbush et al., 1998). Though further progress has been limited by an inability to develop viral cultures, recent work with next generation genomic techniques have made important contributions. These studies (Alfaro-Núñez & Gilbert, 2014; Page-Karjian et al., 2012) found herpesvirus DNA to be rather ubiquitous—occurring in all hard-shelled sea turtles, in all populations tested, and even prevalent in clinically healthy turtles. If α-herpesviruses are the origin of FP, this represents a classic herpesvirus scenario where infections are pervasive, but latent or subclinical, in the host population (Stevens & Cook, 1971; Umbach et al., 2008) and revealing of its etymology from the Greek ερπης, meaning “to creep”. With this in mind, we recently described the epidemiological link between this disease and coastal eutrophication, detailing how green turtles could literally be eating themselves sick (Hall et al., 2007), activating latent herpes infections and promoting tumors by foraging on arginine-enriched macroalgae. A model built on this hypothesis (Van Houtan, Hargrove & Balazs, 2010) explained 72% of the spatial variability of the disease across the Hawaiian Islands while offering a detailed explanation of the disease that connects turtle ecology, plant physiology (e.g., Raven & Taylor, 2003), and herpes biology to known management problems of nutrient pollution and invasive species. At the forefront, this proposed pathway focuses on the role arginine might have in promoting FP tumors. A significant body of evidence supports this. In many chronic diseases, arginine is implicated in cell inflammation and immune dysfunction (Peranzoni et al., 2008) and in promoting viral tumors (Mannick et al., 1994). But arginine is specifically important for herpesviruses. Laboratory studies demonstrate that herpes infections require arginine, being stunted in its absence (Inglis, 1968; Mikami, Onuma & Hayashi, 1974; Olshevsky & Becker, 1970) and diminished when it is deprived (Mistry et al., 2001). Subsequent research revealed that arginine is a principal component of glycoproteins in the outer viral envelope of herpesviruses. These glycoproteins are conserved across a wide variety of herpesviruses (Alfaro-Núñez, 2014) and are critical to the herpes life cycle as they facilitate localization, fusion, and

https://peerj.com/articles/602/ Page 2 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

entrance to host cell nuclei (Hibbard & Sandri-Goldin, 1995; Klyachkin & Geraghty, 2008). Beyond its significance for herpesviruses, arginine is an emerging focus of human cancer treatments as well. Cancer tumors lacking enzymes that synthesize arginine must obtain arginine metabolically, and can therefore be regulated by arginine deprivation (Bowles et al., 2008; Feun et al., 2008; Kim et al., 2009). Perhaps coincidentally, arginine also plays an important role in how plants sequester environmental nitrogen. Nitrogen is a limiting factor for both plant and macroalgal growth (Raven & Taylor, 2003). As a result, in times of environmental availability plants acquire excess nitrogen through what is known as luxury consumption (Chapin III, 1980). Terrestrial plants, however, do not rely on a host of amino acids for luxury consumption; they preferentially store ambient nitrogen in arginine (Chapin III, 1980; Chapin III, Schulze & Mooney, 1990; Llàcer, Fita & Rubio, 2008). Little is known about how this functions in macroalgae, however. Previous studies in Hawaii suggest it might be relevant. Macroalgae from these limited surveys demonstrated that amino acids and stable isotope values for δ15N varied by species and by location (Dailer et al., 2010; McDermid, Stuercke & Balazs, 2007). Though the data were limited, arginine was specifically elevated at eutrophic sites for two invasive species of Ulva and Hypnea (McDermid, Stuercke & Balazs, 2007), prompting more systematic study. Eutrophication of coastal waters in Hawaii has spurred chronic nuisance algal blooms and dramatically altered the composition of reef ecosystems (Cox et al., 2013; Dailer et al., 2010; Lapointe & Bedford, 2011; Smith, Hunter & Smith, 2010). Non-native macroalgae introduced across the Main Hawaiian Islands after 1950 have been particularly influential (Abbot, 1999; Smith, Hunter & Smith, 2002), having displaced native algae and become the dominant forage for Hawaiian green turtles (Russell & Balazs, 2009). Despite the emergence of FP, and historical overharvesting (Kittinger et al., 2013; Van Houtan & Kittinger, 2014), numbers of nesting green turtles have grown steadily in Hawaii since their protection under state and federal regulations in the 1970s (Seminoff et al., 2014). Nonetheless, FP remains the greatest known mortality to Hawaiian green turtles (Chaloupka et al., 2008) and in some regions the incidence of FP tumors is still on the rise (Van Houtan, Hargrove & Balazs, 2010). Beyond its influence on green turtle populations, eutrophication is also associated with coral reef declines (Vega Thurber et al., 2014). Growth anomalies in Porites corals, for example, occur in the same eutrophied Hawaii reefs as diseased green turtles (Friedlander et al., 2008), and these coral tumors have Herpesviridae gene signatures (Vega Thurber et al., 2008). Understanding the promotion of FP tumors may therefore be broadly relevant for the conservation of coral reef ecosystems. Here we analyze tissues from tumored green turtles and dominant forage species of macroalgae from across Hawaii. We determine amino acid content using High-Performance Liquid Chromatography (HPLC) to establish tumor biomarkers (Jain et al., 2012) and examine nutrient changes in macroalgae and luxury consumption. We run a series of Generalized Linear Models (GLMs) to test for arginine levels, arginine storage, and to examine the role of eutrophication. Collectively, these analyses are an interdisciplinary test of a hypothesis we posed earlier (Van Houtan, Hargrove & Balazs, 2010; Van Houtan & Schwaab, 2013) that arginine would be elevated in invasive algae in eutrophic locations and this would in turn promote FP tumors in green turtles.

Materials and Methods Tissue collection and preparation We sampled turtle tissues during necropsies of stranded green turtles at the NOAA Pacific Islands Fisheries Science Center in Honolulu (US FWS permit # TE-72088A-0). For turtles with heavy tumor burdens, we collected both tumor and baseline tissue. Using a scalpel to make radial cross-sections, we obtained at minimum 0.5 cm3 for each tissue type, selecting subsurface material to avoid contamination. Tumors were sampled from the flipper or eye, and baseline tissue was subcutaneous muscle from the flipper or pectoral that appeared grossly subclinical. We collected samples from 12 turtles, representing males and females from a variety of life stages (Table S1 provides full details). At collection, we rinsed samples in water and stored in 90% alcohol in 1.5 mL cryovials (Thermo Scientific™ Nalgene™). After 24–48 h, we pressed the samples dry with forceps, and transferred to clean cryovials packed with SiO2 indicating gel desiccant (Fisher™, grade 48, 4–10 mesh). We replaced the spent silica beads every 24 h, repeating the process as needed until the samples were completely dried. We homogenized the resulting tissues with a porcelain mortar and pestle or by shaving samples with a #22 scalpel blade. We collected five species of macroalgae from coastal watersheds spanning a range of nutrient profiles (Van Houtan, Hargrove & Balazs, 2010) on Oahu, Maui, and Hawaii island. We focused on three invasive species—Hypnea musciformis (Wulfen) JV Lamouroux, Acanthophora spicifera(M Vahl) Borgesen, and Ulva lactuca Linnaeus—and two non-invasive native species —Pterocladiella capillacea (SG Gmelin) Santelices & Hommersand and Amansia glomerata C Agardh—that are representative turtle forage items (Russell & Balazs, 2009) and reasonably widespread. P. capillacea and A. glomerata were inconsistently found and combined into a single category. U. lactuca, the only chlorophyte, was recently reclassified (O’Kelly et al., 2010) from U. fasciata. For each species and location we collected three replicates at 0–5 m depth, where nearshore green turtles commonly forage (Van Houtan & Kittinger, 2014). We rinsed samples with deionized water and later dried in a 60 °C oven (Dailer et al., 2010). We homogenized dried samples with a mortar and pestle and stored in 5 mL cryovials (Thermo Scientific™ Nalgene™). When samples were difficult to obtain, we supplemented our samples with results from a published study in Hawaii (McDermid, Stuercke & Balazs, 2007). Online Supplemental Information provides full sample metadata. https://peerj.com/articles/602/ Page 3 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

Amino acid and statistical analysis We sent prepared samples to the Protein Chemistry Laboratory at Texas A&M University for amino acid determination. Samples were separated into three aliquots, weighed, and then placed in 200 µL of 6N HCl along with the Internal Standard and hydrolyzed at 110 °C for 22 h. The resulting amino acids were separated and quantified using an HPLC (Agilent™ 1260) with pre-column derivitization (Blankenship et al., 1989) by ortho-phthalaldehyde (OPA) and fluorenylmethyl chloroformate (FMOC). As both tryptophan and cysteine are destroyed during hydrolysis, the total protein measured is slightly underreported. As a result, this analysis reports the dry mass for 16 amino acids (Fig. 1B), which we calculate as the mass divided by the total sample mass, averaged between sample aliquot replicates. For turtle tissues, we calculate the change of amino acids from the baseline to tumor samples, expressed as the difference of the average percent total protein for each tissue type. For macroalgae samples, we first calculate the change in arginine levels in samples from eutrophic and oligotrophic sites. Collection sites were considered eutrophic if they had a nitrogen footprint statistic (Van Houtan, Hargrove & Balazs, 2010) above 0.50 and oligotrophic if not. (There was a clear separation here as no sites fell between 0.37 and 0.57.) Sites straddling two watersheds were also located on remote geographic peninsulas with little human impacts (Van Houtan, Hargrove & Balazs, 2010) and therefore given the lower of the two watershed footprint statistics. To describe how macroalgae store environmental nitrogen, we multiplied the amino acid dry mass (percent of total) for each sample by the proportional molecular weight of nitrogen. We formally test for statistical sample differences through a variety of generalized linear models (GLM). To assess amino acid changes between tumors and turtle baseline tissue, we run a paired t-test for sample means and plot the results to identify potential biomarkers (Jain et al., 2012). To examine arginine variability of individual algae across site types we use a one-tailed t-test. We follow this with a GLM that has site treatment and species as factors to predict arginine content. As a frame of reference, we combine these results with known energetic requirements of green turtles (Jones et al., 2005) and published energy content of alga in our study (McDermid, Stuercke & Balazs, 2007) to estimate the daily arginine intake. We calculate this for subadult turtles, the highest-diseased demographic in Hawaii (Van Houtan, Hargrove & Balazs, 2010), as well as for large adults, for different site-species comparisons. To test for luxury consumption, we fit normal distributions to the observed nitrogen amino acid dry mass values for each sample, and determine if arginine falls outside the distribution’s expected 95% interval. We then examine the cause of arginine nitrogen variability. We first build a GLM with total plant nitrogen and species as factors, and then a second with nitrogen footprint and species as factors.

Results

Figure 1B plots the amino acid profiles of tumor-baseline tissue sets for 12 green turtles. We observed significant differences in all 16 amino acids tested, highlighting the divergent metabolism of tumors. Methionine (tumor depleted) and arginine (tumor enriched) had the most statistically significant changes. Glycine, however, had the most dramatic shift. Tumor glycine increased on average 260% (range 93–382%, t = 11.4, P < 0.0001), meaning tumors had 2–5 times more glycine than baseline tissues. This is perhaps unsurprising as glycine is a building block for nucleic acids and is required in large amounts by rapidly proliferating cancer cells (Jain et al., 2012; Tomita & Kami, 2012). Proline also increased markedly in tumors (average 144%, range 40–269%, t = 10.1, P < 0.0001), the second largest change we observed. This may reflect the importance of proline for herpesviruses in counteracting host cell defenses. The herpesvirus protein Us11 has an arginine- and proline-rich binding domain that specifically inhibits PKR (protein kinase R), critical for cellular viral defense (Khoo, Perez & Mohr, 2002; Poppers et al., 2000). Proline synthesis was also important in recent analyses of cancer tumors (De Ingeniis et al., 2012; Nilsson et al., 2014). Arginine increased (average 25%, range 9–38%, t = 12.9, P < 0.0001) and lysine decreased (range: average 47%, range 23–67%, t = −12.1, P < 0.0001) in tumors, consistent with their respective demonstrated roles in herpes infections (Fatahzadeh & Schwartz, 2007; Griffith et al., 1987; Hibbard & Sandri-Goldin, 1995; Inglis, 1968; Mikami, Onuma & Hayashi, 1974; Olshevsky & Becker, 1970). Figure 1C plots the underlying raw histograms for arginine percent total protein in baseline samples and tumors, with smoothed distributions in the background. Aside from the above results, tumors were depleted in glutamine (53%), leucine (37%), isoleucine (44%), asparagine (18%), tyrosine (44%), valine (31%), threonine (24%), phenylalanine (22%), and histidine (35%)—listed in order of percent total protein change. Tumors were also enriched in alanine (37%) and serine (16%), the latter being essential in breast cancers (De Ingeniis et al., 2012; Possemato et al., 2011). These amino acid profiles serve as a first template for establishing FP biomarkers that may aid understanding this disease, and for herpesviruses and tumor formation more generally. Analyzing forage, Fig. 2 plots the arginine enrichment in common forage species of wild macroalgae between locations of low and high nutrient inputs. Arginine levels increased at eutrophic sites in all species (average 160%, range: 70–230%). Though A. spicifera had the highest increase (230%, t = 3.5, P = 0.01), H. musciformis had the highest arginine content at eutrophic (1.94% dry mass) and oligotrophic (0.79% dry mass) sites and the most statistically significant change (t = 4.1, P = 0.007). Of note, the H. musciformis arginine content at oligotrophic sites was higher than that of native rhodophytes sampled at eutrophic sites (0.73% dry mass). This may highlight the role of aggressively invasive macroalgae (Smith, Hunter & Smith, 2002) such as H. musciformis in this disease. A more complete GLM with site treatment and species as factors predicts arginine content (F(7,24) = 11.8, P < 0.0001, R = 0.91). https://peerj.com/articles/602/ Page 4 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

Figure 2: Common forage species for green turtles are arginine enriched at eutrophic coastal areas. Arginine content is 2–3 times higher at eutrophic sites, compared to the same alga sampled at oligotrophic sites. Increases are more pronounced, and arginine levels are higher, in the three nonnative invasive macroalgae species than for the two native Rhodophyta, Amansia glomerata and Pterocladiella capillacea. Error bars indicate SEM. Asterisks are one-tailed t-test results: ∗P < 0.05; ∗∗P = 0.01; ∗∗∗P < 0.01. A GLM with site treatment and species as factors to predict arg content is statistically significant (F(7,24) = 11.8, P < 0.0001, R = 0.91). Given energetic estimates, green turtles foraging on non-native algae at eutrophied sites can increase their daily arginine intake by 17–26 g.

DOI: 10.7717/peerj.602/fig-2

But this does not quite capture the nutrient intake of turtles foraging at different site treatments. From energetics we know a 45 kg subadult green turtle requires 2,435 kJ day-1 of dietary energy (Jones et al., 2005). If this turtle only foraged on H. musciformis—with an energy content of 4.3 kJ g-1 total dry mass (McDermid, Stuercke & Balazs, 2007)—it would require 567 g dry mass of H. musciformis daily to meet energetic demands. Based on our amino acid analysis (Fig. 2) this turtle would consume 11.1 g of arginine daily at eutrophic sites. If this same turtle only consumed the native species we tested (P. capillacea and A. glomerata, average energy 8.9 kJ g-1 total dry mass (McDermid, Stuercke & Balazs, 2007)) it would require 274 g dry mass of daily forage. At oligotrophic sites this turtle would consume 1.2 g arginine per day. In other words, foraging on invasive alga, H. musciformis, at eutrophic sites increases the average arginine intake by 9.9 g day-1 (range 7.8–11.8 g) by comparison to consuming native species at oligotrophic sites, which is 5–14 times the baseline arginine consumption. If we consider this for a 100 kg adult turtle requiring 5,364 kJ daily (Jones et al., 2005) the arginine boost is 21.8 g day-1 (range 17.2–26.0 g). Clearly, there is a substantial dietary influx of arginine for green turtles foraging in eutrophied watersheds. Figure 3 plots the nitrogen dry mass for each amino acid to examine how macroalgae sequester environmental nitrogen. In 13/13 samples from eutrophic sites and 9/12 (75%) samples from oligotrophic sites, nitrogen levels were anomalously high for arginine. That is, nitrogen dry mass was outside the expected 95% interval set by the fitted normal distribution parameters for that sample. Two samples from oligotrophic sites (Punaluu P. capillacea and Kaena A. spicifera) demonstrated no preferential nitrogen storage. One

https://peerj.com/articles/602/ Page 5 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

sample (Olowalu U. lactuca) had a positive anomaly for alanine, but its total nitrogen levels were the lowest measured, minimizing its significance. We generated Fig. 3 before receiving the results for a seventh Ulva sample (eutrophic Kanaha). This panel appears in the online supplement.

Figure 3: Like terrestrial plants, macroalgae preferentially sequester available environmental nitrogen in arginine. We quantify this luxury consumption by calculating the nitrogen dry mass in each amino acid across species and ecosystem treatments. Horizontal lines are the mean nitrogen dry mass for each sample, ! indicates value lies outside the expected 99% interval, and ∗ outside the expected 95% interval. All (13/13) of the eutrophic and 75% (9/12) of the oligotrophic site samples show preferential sequestration of nitrogen in arginine. Amino acids are arranged from left to right in average descending order of prevalence: arginine, R; aspartic acid, D; glutamic acid, E; alanine, A; lysine, K; glycine, G; leucine, L; serine, S; valine, V; histidine, H; threonine, T; proline, P; isoleucine, I; phenylalanine, F; tyrosine, Y; methionine, M.

DOI: 10.7717/peerj.602/fig-3

Though arginine was the clear preference for nitrogen storage (22/25 total samples, 88%), this was often extreme. In 15 samples (denoted by “!” in Fig. 3) arginine nitrogen storage is outside the 99% interval (above the expected 99.5% cumulative probability distribution) for that sample. Such extreme arginine preference occurred in all H. musciformis samples, followed by A. spicifera (4/6 samples, 67%), U. lactuca (3/7 samples, 43%), and the native rhodophytes (2/6 samples, 33%)—and was observed in 9/13 (69%) samples from eutrophic sites considering all species. Thus, the marine macroalgae we sampled demonstrated a clear tendency to sequester environmental nitrogen as arginine, which is an interesting convergence with terrestrial plants (Chapin III, 1980; Chapin III, Schulze & Mooney, 1990). Having documented elevated arginine at eutrophic sites and arginine luxury consumption, Fig. 4 assesses the relationship between arginine nitrogen storage, total tissue nitrogen, and watershed-level eutrophication metrics. Arginine nitrogen increased with total tissue nitrogen for all species (Fig. 4A), and a GLM with total tissue nitrogen and species as factors is highly significant (F(7,24) = 35.3, P < 0.0001, R = 0.97). Figure 4B shows that arginine nitrogen also increased with each site’s nitrogen footprint (Van Houtan, Hargrove & Balazs, 2010); an index of natural and anthropogenic factors that generate, deliver, and retain nitrogen in coastal watersheds, and a proxy for local nitrogen loading. Similar to the previous model, a GLM with nitrogen footprint and species as factors is significant (F(7,24) = 13.0, P < 0.0001, R = 0.92).

https://peerj.com/articles/602/ Page 6 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

Figure 4: Arginine sequestration of environmental nitrogen increases with (A) total plant N and (B) in proportion with watershed eutrophication. A GLM with total plant nitrogen and species as factors is statistically significant (F(7,24) = 35.3, P < 0.0001, R = 0.97), as is a model with nitrogen footprint (Van Houtan, Hargrove & Balazs, 2010) and species as factors (F(7,24) = 13.0, P < 0.0001, R = 0.92). The relationships are similar between species, except for H. musciformis, which has a steeper slope in both panels, indicating H. musciformis is more efficient at sequestering environmental nitrogen and in storing it in arginine.

DOI: 10.7717/peerj.602/fig-4

The relationships for these models are similar between species, except for H. musciformis, which has a steeper slope in both panels. In Fig. 4A, this indicates that given environmental levels, H. musciformis allocates proportionally more nitrogen to arginine than other amino acids, for the alga tested. Figure 4B suggests H. musciformis is more proficient at sequestering environmental nitrogen in arginine than the other species. Similar to Fig. 2, this perhaps underscores the importance of H. musciformis in promoting FP tumors. Though the arginine levels (Fig. 2) and the arginine nitrogen storage (Fig. 3) are not as high in the native macroalgae as for U. lactuca and A. spicifera, the statistical relationships between arginine nitrogen, tissue nitrogen, and ecosystem nitrogen are similar (Fig. 4).

Discussion

In this study we demonstrated how eutrophication increases the arginine in invasive marine macroalgae; that this significantly boosts the arginine intake by foraging green turtles, and that arginine is elevated in tumors from diseased green turtles. Based on energetic needs, we calculated adult turtles foraging in eutrophic habitats on invasive algae might boost their arginine intake 5–14 times, consuming a total of 22–28 g of arginine daily. We provide a first baseline set of amino acid biomarkers for FP tumors, and we documented arginine luxury consumption across a variety of macroalgae. We discuss the results and their implications for understanding the disease and its environmental promotion below. Though FP tumors were elevated in several amino acids, dietary shifts in arginine may be significant. Glycine, proline, alanine, arginine and serine all were elevated in tumors (Figs. 1B–1C). Of these amino acids, glycine (Jain et al., 2012), proline (De Ingeniis et al., 2012; Nilsson et al., 2014), and serine (Possemato et al., 2011) are known tumor biomarkers; where arginine and proline have https://peerj.com/articles/602/ Page 7 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

added significance for herpesviruses. Of the elevated amino acids, however, only arginine increased in macroalgae at eutrophied sites (where disease rates are elevated) and has a functional role in nitrogen luxury consumption (Fig. 3). Arginine nitrogen content, for example, was above the 95% expected interval in 88% of our macroalgae samples, and was extreme (above the 99% interval) in 60% of our samples (Fig. 3). This suggests arginine may be the critical ingredient linking nearshore eutrophication, luxury consumption, turtle diet, and FP tumors. The metabolic pathways here are uncertain, however. Metabolic reprogramming is a hallmark of rapidly proliferating cancer cells, and a growing body of literature has focused on carbon metabolism in tumors (De Ingeniis et al., 2012; Jain et al., 2012; Nilsson et al., 2014). Perhaps it is unsurprising that we detected significant tumor-baseline differences for all 16 amino acids. However, instead of focusing on carbon metabolism common to cancer studies, we profiled nitrogen due to its role in limiting macroalgae growth. Future progress in understanding FP may therefore come from a systematic characterization of the metabolic pathways in FP tumors, and in particular the recycling, salvaging, and biosynthesis of arginine. A dietary role for tumor promotion in human cancers may also benefit from a more comprehensive understanding of nitrogen metabolism. Our results help explain the epidemiology of this disease, and highlight the role of environmental factors in Hawaii and perhaps beyond. Though DNA from herpesviruses linked to FP tumors is found in all sea turtle species, this disease has only been widespread and a conservation concern for green turtles (Alfaro-Núñez, 2014). This is consistent with our proposed mechanism involving eutrophication and arginine intake. Green turtles are the only strictly herbivorous sea turtle and therefore would consume the most arginine-enriched algae in nearshore habitats (other omnivorous sea turtle species consume algae, though at lower rates). The spatial and demographic structure of green turtles may also be relevant. Juvenile green turtles have a pelagic phase until they recruit to nearshore habitats as young juveniles (Seminoff et al., 2014). Far away from human population centers, green turtles are disease-free during this pelagic phase (Ene et al., 2005; Van Houtan, Hargrove & Balazs, 2010). In the Main Hawaiian Islands (MHI), the incidence of FP tumors increases steadily as turtles mature, and then decreases when they begin migrating to the relatively pristine Northwestern Hawaiian Islands to breed. In other words, disease rates increase directly in proportion to their residency time in the Main Hawaiian Islands (Van Houtan, Hargrove & Balazs, 2010). In addition to this chronic exposure to eutrophic habitats, older turtles have greater energetic demands and therefore may additionally have higher disease rates due to increases in their consumption of MHI macroalgae and subsequent arginine intake. Though we are investigating environmental influences, it is possible that immunocompetence could factor in these patterns, but its influence is unknown. Aside from demographic patterns in turtles, invasive species of macroalgae also seem to be influential. Certain regions—such as the Kona coast of Hawaii island—have curiously low disease rates. While we previously demonstrated that this region has few nutrient inputs and invasive algae are uncommon (Van Houtan, Hargrove & Balazs, 2010), our results here help explain this pattern. In this study we showed foraging green turtles could be more easily satiated by native macroalgae, as they can have relatively higher energy contents. Combined with our amino acid results, the energy and arginine content of macroalgae may therefore act as a sort of one-two punch for promoting this disease. Native macroalgae have a fraction of the arginine content of invasive species (Fig. 2), but offer more calories per unit mass (see online Supplemental Information). Turtles foraging on invasive macroalgae in eutrophic areas would need twice the amount they would require of native algae, therefore multiplying the arginine enrichment effect. For so-called superweeds like H. musciformis, this low energy–high arginine combination is the most extreme we observed. Considering that H. musciformis energy content can vary inversely with growth rate (Guist Jr, Dawes & Castle, 1982), this may be a general result, and a topic for future research. Our estimates for turtle arginine consumption were often substantial, but the numbers could be even higher. For subadults we documented an average increase of 9.9 g arginine day-1 when shifting from native forage at oligotrophic sites to invasive forage at eutrophic sites. This number jumped to an average 21.8 g day-1 for adults. These numbers are based on our observed amino acid values and energetic demands. The metabolic rates we reference are baseline averages (Jones et al., 2005), which would underestimate the dietary needs of rapidly growing turtles, migrating animals, or adult females amassing resources for vitellogenesis. Our calculated dietary intake of arginine could therefore increase, making the already significant increase even more so. There are no daily nutritional guides for wild green turtles (Bjorndal, 1997). However, to put these numbers in context, they are well above the recommended dietary allowance for humans. Human adults (19–50 years) should consume 4.7 g of arginine and 51 g of total protein daily (Institute of Medicine, 2005). Our estimated arginine intake for adult turtles could reach 28 g day-1 (online Supplemental Information), which is half the recommended total daily protein for humans and 5 times the suggested arginine intake. Future studies can use our arginine intake estimates to guide treatments of turtles with FP tumors. Across green turtle populations, it is widely observed that FP occurs most frequently in eutrophied and otherwise impaired waterways (Herbst, 1994; Van Houtan, Hargrove & Balazs, 2010). Our efforts here have largely been to demonstrate why this might occur, and to detail the ecological mechanisms. A logical next step is to repeat this study comparing tumors and forage items for other green turtle populations and other species. Additional next steps could be developing a monitoring plan to assess ecosystem risk for the disease in Hawaii and other ecological regions. Stable isotope analyses of tissues from U. lactuca or H. musciformis are an effective method for monitoring water quality in an integrative manner. For example, for macroalgae that uptake nitrogen from the water column, δ15N values above 6.0 point to a significant wastewater presence (Dailer et al., 2010; Dailer, Smith & Smith, 2012). While these tests can reveal plant nitrogen sources, they do not comment on ecosystem nitrogen flux. Our test for nitrogen arginine sequestration (Fig. 3) may infer on nitrogen flux at a more informative level than total tissue nitrogen, however, more research here is https://peerj.com/articles/602/ Page 8 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

necessary. Combined stable isotope and amino acid analysis of macroalgae, therefore could be a powerful and reasonably inexpensive tool ($120 for both tests) to monitor and understand eutrophication in coastal ecosystems. The relevance of this tool extends beyond turtle diseases, but to ecosystem based management of coral reefs, estuaries, and seagrass systems.

Supplemental Information

Additional Information and Declarations Competing Interests The authors have no competing interests or ethical conflicts. Author Contributions Kyle S. Van Houtan conceived and designed the experiments, performed the experiments, analyzed the data, contributed reagents/materials/analysis tools, wrote the paper, prepared figures and/or tables, reviewed drafts of the paper. Celia M. Smith conceived and designed the experiments, performed the experiments, reviewed drafts of the paper. Meghan L. Dailer and Migiwa Kawachi performed the experiments, reviewed drafts of the paper. Animal Ethics The following information was supplied relating to ethical approvals (i.e., approving body and any reference numbers): We have an IACUC permit for our program but we do not need it for this study because all the samples involved dead, stranded sea turtles. For this, however, we require a USFWS ESA permit (#TE-72088A-0). Funding This study was supported by a grant from the Disney Worldwide Conservation Fund to CMS and KSVH, and a Presidential Early Career Award in Science and Engineering to KSVH. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. J Johnson conducted the amino acid analyses. E Cox, D Francke, T Jones, and N Sarto aided with tissue preparation, T Jones advised on the energetics calculations. C Henson and J Browning assisted with permits. C Stankis provided tumored green turtle photos. A Alfaro-Núñez, D Guo, A Page-Karjian, J Parr, S Pimm, J Polovina, A Rivero, and two anonymous reviewers made improvements to an earlier version of this manuscript.

References

Abbot IA. 1999. Marine red algae of the Hawaiian Islands. Honolulu: Bishop Museum Press. Aguirre AA, Lutz P. 2004. Marine turtles as sentinels of ecosystem health: is fibropapillomatosis an indicator? EcoHealth 1:275-283 Alfaro-Núñez A. 2014. Tracing the co-evolutionary history of the Chelonid fibropapilloma-associated herpesvirus and its host sea turtles. Ph.D. diss., University of Copenhagen, Denmark thesis Alfaro-Núñez A, Gilbert MTP. 2014. Validation of a sensitive PCR assay for the detection of Chelonid fibropapilloma-associated herpesvirus in latent turtle infections. Journal of Virological Methods 206:38-41 Bjorndal KA. 1997. Foraging ecology and nutrition of sea turtles. In: Lutz P, Musick J, eds. The biology of sea turtles, vol 1. Boca Raton, FL: CRC Press. 199-231 Blankenship DT, Krivanek MA, Ackermann BL, Cardin AD. 1989. High-sensitivity amino acid analysis by derivatization with o- phthalaldehyde and 9-fluorenylmethyl chloroformate using fluorescence detection: applications in protein structure determination. Analytical Biochemistry 178:227-232 Bowles TL, Kim R, Galante J, Parsons CM, Virudachalam S, Kung HÄ, Bold RJ. 2008. Pancreatic cancer cell lines deficient in argininosuccinate synthetase are sensitive to arginine deprivation by arginine deiminase. International Journal of Cancer 123:1950-1955 Chaloupka M, Work TM, Balazs GH, Murakawa SKK, Morris R. 2008. Cause-specific temporal and spatial trends in green sea turtle strandings in the Hawaiian Archipelago. Marine Biology 154:887-898 Chapin FS. 1980. The mineral nutrition of wild plants. Annual Review of Ecology and Systematics 11:233-260 Chapin FS, Schulze E-D, Mooney HA. 1990. The ecology and economics of storage in plants. Annual Review of Ecology and Systematics https://peerj.com/articles/602/ Page 9 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

21:423-447 Cox ET, Smith CM, Popp BN, Foster MS, Abbott IA. 2013. Can stormwater be detected by algae in an urban reef in Hawaii? Marine Pollution Bulletin 71:92-100 Dailer ML, Knox RS, Smith JE, Napier M, Smith CM. 2010. Using δ15 N values in algal tissue to map locations and potential sources of anthropogenic nutrient inputs on the island of Maui, Hawaii, USA. Marine Pollution Bulletin 60:655-671 Dailer ML, Smith JE, Smith CM. 2012. Responses of bloom forming and non-bloom forming macroalgae to nutrient enrichment in Hawaii, USA. Harmful Algae 17:111-125 De Ingeniis J, Ratnikov B, Richardson AD, Scott DA, Aza-Blanc P, De SK, Kazanov M, Pellecchia M, Ronai Z, Osterman AL. 2012. Functional specialization in proline biosynthesis of melanoma. PLoS ONE 7:e45190 dos Santos RG, Martins AS, Torezani E, Baptistotte C, Horta PA, Work TM, Balazs GH. 2010. Relationship between fibropapillomatosis and environmental quality: a case study with Chelonia mydas off Brazil. Diseases of Aquatic Organisms 89:87-95 Ene A, Su M, Lemaire S, Rose C, Schaff S, Moretti R, Lenz J, Herbst LH. 2005. Distribution of chelonid fibropapillomatosis-associated herpesvirus variants in Florida: molecular genetic evidence for infection of turtles following recruitment to neritic developmental habitats. Journal of Wildlife Diseases 41:489-497 Fatahzadeh M, Schwartz RA. 2007. Human herpes simplex virus infections: epidemiology, pathogenesis, symptomatology, diagnosis, and management. Journal of the American Academy of Dermatology 57:737-763 Feun L, You M, Wu C, Kuo M, Wangpaichitr M, Spector S, Savaraj N. 2008. Arginine deprivation as a targeted therapy for cancer. Current Pharmaceutical Design 14:1049-1057 Friedlander A, Aeby G, Brainard R, Brown E, Chaston K, Clark A, McGowan P, Montgomery T, Walsh W, Williams I+1 more. 2008. The state of coral reef ecosystems of the Main Hawaiian Islands. In: Waddell JE, Clarke AM, eds. The State of coral reef ecosystems of the United States and Pacific freely associated states. Silver Spring, MD: NOAA/NCCOS Center for Coastal Monitoring. 219-253 Griffith RS, Walsh DE, Myrmel KH, Thompson RW, Behforooz A. 1987. Success of L-lysine therapy in frequently recurrent herpes simplex infection. Dermatology 175:183-190 Guist GG, Dawes CJ, Castle JR. 1982. Mariculture of the red seaweed, Hypnea musciformis. Aquaculture 28:375-384 Hall SR, Sivars-Becker L, Becker C, Duffy MA, AJ Tessier, Caceres CE. 2007. Eating yourself sick: transmission of disease as a function of foraging ecology. Ecology Letters 10:207-218 Herbst LH. 1994. Fibropapillomatosis of marine turtles. Annual Review of Fish Diseases 4:389-425 Herbst L, Ene A, Su M, Desalle R, Lenz J. 2004. Tumor outbreaks in marine turtles are not due to recent herpesvirus mutations. Current Biology 14:R697-R699 Hibbard MK, Sandri-Goldin RM. 1995. Arginine-rich regions succeeding the nuclear localization region of the herpes simplex virus type 1 regulatory protein ICP27 are required for efficient nuclear localization and late gene expression. Journal of Virology 69:4656-4667 Inglis VB. 1968. Requirement of arginine for the replication of herpes virus. Journal of General Virology 3:9-17 Institute of Medicine. 2005. Dietary reference intakes for energy, carbohydrate, fiber, fat, fatty acids, cholesterol, protein and amino acids (macronutrients). Washington, DC: The National Academies Press. Jain M, Nilsson R, Sharma S, Madhusudhan N, Kitami T, Souza AL, Kafri R, Kirschner MW, Clish CB, Mootha VK. 2012. Metabolite profiling identifies a key role for glycine in rapid cancer cell proliferation. Science 336:1040-1044 Jones T, Seminoff J, Resendiz A, Lutz P. 2005. Energetics of the East Pacific green turtle at a Gulf of California foraging habitat. In: Coyne M, Clark RD, eds. Proceedings of the 21st annual symposium on sea turtle biology and conservation NOAA technical memorandum NMFS-SEFSC- 528. Miami, FL. NOAA Fisheries. 198-200 Khoo D, Perez C, Mohr I. 2002. Characterization of RNA determinants recognized by the arginine- and proline-rich region of Us11, a herpes simplex virus type 1-encoded double-stranded RNA binding protein that prevents PKR activation. Journal of Virology 76:11971-11981 Kim RH, Coates JM, Bowles TL, McNerney GP, Sutcliffe J, Jung JU, Gandour-Edwards R, Chuang FY, Bold RJ, Kung H-J. 2009. Arginine deiminase as a novel therapy for prostate cancer induces autophagy and caspase-independent apoptosis. Cancer Research 69:700-708 Kittinger JN, Van Houtan KS, McClenachan L, Lawrence AL. 2013. Using historical data to assess the biogeography of population recovery. Ecography 36:868-872 Klyachkin YM, Geraghty RJ. 2008. Mutagenic analysis of herpes simplex virus type 1 glycoprotein L reveals the importance of an arginine-rich region for function. Virology 374:23-32 Lackovich JK, Brown DR, Homer BL, Garber RL, Mader DR, Moretti RH, Patterson AD, Herbst LH, Oros J, Jacobson ER. 1999. Association of herpesvirus with fibropapillomatosis of the green turtle Chelonia mydas and the loggerhead turtle Caretta caretta in Florida. Diseases of Aquatic Organisms 37:89-97 Lapointe BE, Bedford BJ. 2011. Stormwater nutrient inputs favor growth of non-native macroalgae (Rhodophyta) on O’ahu, Hawaiian Islands. Harmful Algae 10:310-318

https://peerj.com/articles/602/ Page 10 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

Llàcer JL, Fita I, Rubio V. 2008. Arginine and nitrogen storage. Current Opinion in Structural Biology 18:673-681 Lu Y, Wang Y, Yu Q, Aguirre A, Balazs G, Nerurkar V, Yanagihara R. 2000. Detection of herpesviral sequences in tissues of green turtles with fibropapilloma by polymerase chain reaction. Archives of Virology 145:1885-1893 Mannick JB, Asano K, Izumi K, Kieff E, Stamler JS. 1994. Nitric oxide produced by human B lymphocytes inhibits apoptosis and Epstein- Barr virus reactivation. Cell 79:1137-1146 McDermid KJ, Stuercke B, Balazs GH. 2007. Nutritional composition of marine plants in the diet of the green sea turtle (Chelonia mydas) in the Hawaiian Islands. Bulletin of Marine Science 81:55-71 Mikami T, Onuma M, Hayashi T. 1974. Requirement of arginine for the replication of Marek’s disease herpes virus. Journal of General Virology 22:115-128 Mistry SK, Zheng M, Rouse BT, Morris SM. 2001. Induction of arginases I and II in cornea during herpes simplex virus infection. Virus Research 73:177-182 Nilsson R, Jain M, Madhusudhan N, Sheppard NG, Strittmatter L, Kampf C, Huang J, Asplund A, Mootha VK. 2014. Metabolic enzyme expression highlights a key role for MTHFD2 and the mitochondrial folate pathway in cancer. Nature Communications 5 O’Kelly CJ, Kurihara A, Shipley TC, Sherwood AR. 2010. Molecular assessment of the Ulva spp. (Ulvophyceae, Chlorophyta) in the Hawaiian Islands. Journal of Phycology 46:728-735 Olshevsky U, Becker Y. 1970. Synthesis of herpes simplex virus structural proteins in arginine deprived cells. Nature 226:851-853 Page-Karjian A, Torres F, Zhang J, Rivera S, Diez C, Moore P, Moore D, Brown C. 2012. Presence of chelonid fibropapilloma-associated herpesvirus in tumored and non-tumored green turtles, as detected by polymerase chain reaction, in endemic and non-endemic aggregations, Puerto Rico. SpringerPlus 1:1-8 Peranzoni E, Marigo I, Dolcetti L, Ugel S, Sonda N, Taschin E, Mantelli B, Bronte V, Zanovello P. 2008. Role of arginine metabolism in immunity and immunopathology. Immunobiology 212:795-812 Poppers J, Mulvey M, Khoo D, Mohr I. 2000. Inhibition of PKR activation by the proline-rich RNA binding domain of the herpes simplex virus type 1 Us11 protein. Journal of Virology 74:11215-11221 Possemato R, Marks KM, Shaul YD, Pacold ME, Kim D, Birsoy K, Sethumadhavan S, Woo H-K, Jang HG, Jha AK+17 more. 2011. Functional genomics reveal that the serine synthesis pathway is essential in breast cancer. Nature 476:346-350 Quackenbush SL, Work TM, Balazs GH, Casey RN, Rovnak J, Chaves A, duToit L, Baines JD, Parrish CR, Bowser PR. 1998. Three closely related herpesviruses are associated with fibropapillomatosis in marine turtles. Virology 246:392-399 Raven J, Taylor R. 2003. Macroalgal growth in nutrient-enriched estuaries: a biogeochemical and evolutionary perspective. Water, Air and Soil Pollution: Focus 3:7-26 Russell DJ, Balazs GH. 2009. Dietary shifts by green turtles (Chelonia mydas) in the Kaneohe Bay region of the Hawaiian islands: a 28 year study. Pacific Science 63:181-192 Seminoff JA, Balazs GH, Dutton PH, Eguchi T, Haas HL, Hargrove SA, Klemm DL, Lauritsen AM, MacPherson SL, Opay P+3 more. 2014. Green turtle (Chelonia mydas) Status Review under the U. S. Endangered Species Act. Silver Spring, MD: NOAA National Marine Fisheries Service. Smith JE, Hunter CL, Smith CM. 2002. Distribution and reproductive characteristics of nonindigenous and invasive marine algae in the Hawaiian Islands. Pacific Science 56:299-315 Smith JE, Hunter CL, Smith CM. 2010. The effects of top–down versus bottom–up control on benthic coral reef community structure. Oecologia 163:497-507 Stevens JG, Cook ML. 1971. Latent herpes simplex virus in spinal ganglia of mice. Science 173:843-845 Tomita M, and, Kami K. 2012. Systems biology, metabolomics, and cancer metabolism. Science 336:990-991 Umbach JL, Kramer MF, Jurak I, Karnowski HW, Coen DM, Cullen BR. 2008. MicroRNAs expressed by herpes simplex virus 1 during latent infection regulate viral mRNAs. Nature 454:780-783 Van Houtan KS, Hargrove SK, Balazs GH. 2010. Land use, macroalgae, and a tumor-forming disease in marine turtles. PLoS ONE 5:e12900 Van Houtan KS, Kittinger JN. 2014. Historical commercial exploitation and the current status of Hawaiian green turtles. Biological Conservation 170:20-27 Van Houtan KS, Schwaab EC. 2013. Today’s fisheries challenges require creative, interdisciplinary problem solving. In: Taylor WW, Lynch AJ, Leonard NJ, eds. Future of fisheries: perspectives for emerging professionals. Bethesda, MD: American Fisheries Society. 153-158 Vega Thurber RL, Barott KL, Hall D, Liu H, Rodriguez-Mueller B, Desnues C, Edwards RA, Haynes M, Angly FE, Wegley L. 2008. Metagenomic analysis indicates that stressors induce production of herpes-like viruses in the coral Porites compressa. Proceedings of the National Academy of Sciences of the United States of America 105:18413-18418 Vega Thurber RL, Burkepile DE, Fuchs C, Shantz AA, McMinds R, Zaneveld JR. 2014. Chronic nutrient enrichment increases prevalence and severity of coral disease and bleaching. Global Change Biology 20:544-554 https://peerj.com/articles/602/ Page 11 of 12 Eutrophication and the dietary promotion of sea turtle tumors [PeerJ] 1/27/20, 946 AM

Related research

Closing the gap: mixed stock analysis of three Pulmonary function of the green sea turtle, foraging populations of green turtles (Chelonia Chelonia mydas mydas) on the Great Barrier Reef R. N. Gatz et al., Journal of Applied Physiology, PeerJ 1987

Further evidence of Chelonid herpesvirus 5 Dysregulation of l-arginine metabolism and (ChHV5) latency: high levels of ChHV5 DNA bioavailability associated to free plasma heme detected in clinically healthy marine turtles. F. Omodeo-Salè et al., American Journal of PeerJ Physiology - Cell Physiology, 2010

Genomic evolution, recombination, and inter- Arginine de novo and nitric oxide production in strain diversity of chelonid alphaherpesvirus 5 disease states from Florida and Hawaii green sea turtles with Yvette C. Luiking et al., American Journal of fibropapillomatosis Physiology - Endocrinology and Metabolism, 2012 PeerJ New insights into the regulation of inducible nitric The effect of CO2 enrichment on net oxide synthesis photosynthesis of the red alga Furcellaria S. M. Morris et al., American Journal of lumbricalis in a brackish water environment Physiology - Endocrinology and Metabolism, 1994 PeerJ Suppression of Myeloid Cell Arginase Activity First record of hybridization between green leads to Therapeutic Response in a NSCLC Chelonia mydas and hawksbill Eretmochelys Mouse Model by Activating Anti-Tumor Immunity imbricata sea turtles in the Southeast Pacific. PeerJ Juan J. Miret et al., Jitc, 2019

Powered by

PeerJ Job Listings List & find academic jobs on PeerJ for free. Learn more ∠

https://peerj.com/articles/602/ Page 12 of 12 Marine Pollution Bulletin 103 (2016) 63–71

Contents lists available at ScienceDirect

Marine Pollution Bulletin

journal homepage: www.elsevier.com/locate/marpolbul

Identifying nitrogen sources to thermal tide pools in Kapoho, Hawai'i, U.S.A, using a multi-stable isotope approach

Tracy N. Wiegner a,⁎, Ambyr U. Mokiao-Lee b, Erik E. Johnson b,1 a Marine Science Department, University of Hawai‘i at Hilo, 200 W. Kawili St., Hilo, HI 96720, USA b Tropical Conservation Biology and Environmental Science Graduate Program, University of Hawai‘i at Hilo, 200 W. Kawili St., Hilo, HI 96720, USA article info abstract

Article history: Nitrogen (N) enrichment often results in coastal eutrophication, even in remote areas like Hawai‘i. Therefore, de- Received 23 June 2015 termining N sources to coastal waters is important for their management. This study identified N sources to tide Received in revised form 10 December 2015 pools in Kapoho, Hawai'i, and determined their relative importance using three stable isotopes (δ15N, δ18O, δ11B). Accepted 23 December 2015 Surface waters and macroalgal tissues were collected along 100-m onshore-offshore transects in areas of high Available online 6 January 2016 groundwater input for three months at low tide. Water samples from possible N sources were also collected. δ15 Keywords: Mixing model output, along with macroalgal N values, indicated that agriculture soil (34%) was the largest an- fi Coastal waters thropogenic N source followed by sewage (27%). These ndings suggest that more effective fertilizer application Fertilizers techniques and upgrading sewage treatment systems can minimize N leaching into groundwater. Overall, our Macroalgae multi-stable isotope approach for identifying N sources was successful and may be useful in other coastal waters. Sewage © 2015 Elsevier Ltd. All rights reserved. Stable isotopes Water quality

1. Introduction assuming each source has a distinct isotopic signature (Chang et al., 2002; Seiler, 2005; Hunt, 2006; Table 1). These isotopes can also be – Worldwide, human activities have increased the delivery of nitrogen used to determine whether the NO3N has been transformed by microbi- (N) to coastal waters often resulting in eutrophication (Nixon, 1995; al processes (e.g., denitrification) since leaving the source (Hunt and 11 − Smith, 2003). This can lead to phytoplankton and macroalgal blooms Rosa, 2009). δ B is used as a co-migrating discriminator of NO3 sources and resultant decreases in dissolved oxygen (Nixon, 1995; Smith, because it is not affected by microbial transformations like N and O in − 2003; LaPointe et al., 2005; Smith and Smith, 2006). Eutrophication NO3 . It has been used to trace sewage pollution in groundwater and may alter and cause negative effects to coastal, pelagic, and benthic ma- surface waters as B is found in fabric whiteners, a common constituent rine communities, food webs, fisheries, and microbial communities of domestic wastewater (Leenhouts et al., 1998; Seiler, 2005; Hunt, (Nixon, 1995; Smith, 2003). One of the main sources of anthropogenic 2006). The δ15N signature of macroalgae tissues is used to determine N leading to eutrophication is sewage, which is also a human health if macroalgae are taking up sewage N, or N from other sources, and in- hazard in near-shore waters (Boehm et al., 2010). Monitoring and man- corporating it into their tissues (Umezawa et al., 2002; Savage and aging the effects of sewage in coastal waters has become a major envi- Elmgren, 2004; Lin et al., 2007). Often, opportunistic macroalgal species ronmental challenge globally (Prüss-Üstun et al., 2008; Corcoran et al., are used as bioindicators as their growth is stimulated by increased N 2010). and their tissues reflect the isotopic composition of the N they con- There are several approaches that are used to determine if sewage is sumed (Duarte, 1995; Cohen and Fong, 2006). Macroalgae are favored present in coastal waters; one of them includes measuring stable iso- over phytoplankton as bioindicators in this respect because they mini- 15 18 − 11 14 15 topes, specifically δ Nandδ OofNO3 and δ Binthewatercolumn mally discriminate between N and N during uptake compared to and δ15N in benthic macroalgae and animals tissues (e.g., corals, oys- phytoplankton (Costanzo et al., 2005; Savage, 2005; Swart et al., 15 15 18 − 11 ters) (Hunt, 2006; Risk et al., 2009; Dailer et al., 2010). The δ N and 2014). The combined use of δ Nandδ OinNO3 , δ B dissolved in 18 − − 15 δ OofNO3 can be used to identify sewage and other NO3 sources, water, and δ N in macroalgae tissues allows for identification of an- thropogenic N sources to coastal waters because sewage and fertilizers have dissimilar isotopic compositions from one another, and often from ⁎ Corresponding author. − natural sources, such as soil and ocean NO3 , too. E-mail addresses: [email protected] (T.N. Wiegner), [email protected] Stable isotope studies identifying N sources to fresh- and coastal wa- (A.U. Mokiao-Lee), [email protected] (E.E. Johnson). 1 Present address: University of Hawai‘i at Hilo Analytical Laboratory, 200 W. Kawili St., ters have primarily been conducted in temperate environments using a Hilo, HI, U.S.A. 96720 two isotope approach (Aravena et al., 1993; Pardo et al., 2004), with a

http://dx.doi.org/10.1016/j.marpolbul.2015.12.046 0025-326X/© 2015 Elsevier Ltd. All rights reserved. 64 T.N. Wiegner et al. / Marine Pollution Bulletin 103 (2016) 63–71

Table 1 examine their spatial variability within a system. If this approach is suc- Range of reported stable isotope values (‰) for different environmental sources. cessful in identifying N sources to this tide pool system, it could be 15 − 18 − 11 Source δ N-NO3 δ O-NO3 δ B adopted elsewhere for watershed management purposes. Sewage +7 to +20a,b,c,d −5to+11a,e,f −7to+32h,i Fertilizers −5to+5a,b,c,d +17 to +25a,b,d −8to+17i,j Soils −10 to +15a,b,c,d −10 to +16a,d −5to0k 2. Materials and methods Groundwater −4to+15d +1 to +6e −8to+60k,l Seawater +2 to +7c,g 0to+5g +40i,l a 2.1. Site description Atmospheric N2 0 − – a Atmospheric O2 – +23 – The study site was the Wai'Ōpae tide pools in Kapoho, located on the aKendall, 1998; bHunt, 2006; cDerse et al., 2007; dXue et al., 2009; eAravena et al., 1993; east side of Hawai'i Island, about 30 km southeast of Hilo, Hawai'i fXue et al., 2012; gCasciotti et al., 2008; hWidory et al., 2005; iVengosh et al., 1994; 2 jSeiler, 2005. (Fig. 1). Kapoho's watershed is ~6 km , with an elevation and popula- tion of 183 m and 8,587 people, respectively. Land use is primarily agri- culture (80% of total land area), consisting of papaya, orchid, and few recent ones having used their data in mass-balance mixing models anthurium farms, with the remainder of the land comprised of forests to assign percent contributions to N sources (Deutsch et al., 2006; Voss (8%), barren land (11%), and urban areas (1%). Sewage inputs from cess- et al., 2006). Unfortunately, this latter approach does not take into ac- pools are concerning at the Wai'Ōpae tide pools, where the land is sub- − count isotopic fractionation, isotope ratio variability of NO3 ,andoften siding (0.24 m since 1975) and most homes adjacent to the tide pools − there are too many NO3 sources to an environment for the mass- are serviced by cesspools and a few by septic systems (Engineering balance mixing model to be solved exactly (Xue et al., 2009). It is recom- Concepts, Inc., 2010). Hawai'i Department of Health (HDOH) records mended that three or more stable isotopes be used, as well as a mixing were incomplete about the use of individual wastewater systems model that can account for these uncertainties, like the Bayesian ones (IWS) at Vacationland Estates; however, their data indicate that on- now currently being used for food web analyses (Xue et al., 2009; site units consist of cesspools (33%), septic systems (21%), and aerobic Parnell et al., 2010). Presently, there are a few studies that have used treatment units (5%) (Engineering Concepts, Inc., 2010). Because of 15 18 − 11 − δ Nandδ OofNO3 in combination with δ BtodistinguishNO3 these conditions, Kapoho was assigned the highest risk score for cess- sources in fresh- and coastal waters (Leenhouts et al., 1998; Seiler, pool contamination of coastal waters in a recent report to Hawai'i 2005; Hunt, 2006; Xue et al., 2014). In the last three years, several stud- State (Whittier and El-Kadi, 2014). The area is also designated as a Spe- ies have used Bayesian isotope mixing models to determine percent cial Management Area (SMA), a Critical Wastewater Disposal Area − 15 18 − contribution of different NO3 sources with δ N- and δ O–NO3 data (CDWA), and a Marine Life Conservation District (MLCD), so there is a (Xue et al., 2012, 2014; El Gaouzi et al., 2013; Yang et al., 2013; Ding strong desire and mandate to ensure good water quality. et al., 2014), but to our knowledge, only one study to date has used Additionally, the Wai'Ōpae tide pools are located downslope this approach in coastal waters–temperate coastal waters (Korth et al., of Hawai'i's most active volcano, Kīlauea, and their shoreline is 2014). Lastly, no study yet has employed these three stable isotopes relatively unique to the Hawaiian Islands due to the abundance of with a mixing model and δ15N measurements in macroalgal tissues to geothermally-heated groundwaters entering the coast (Juvik and trace sewage pollution. Together, these measurements will allow for Juvik, 1998). Groundwater is the primary freshwater source entering − the dominant NO3 sources to a water body to be identified with greater and transporting N to the tide pools, as there are no perennial streams confidence, as well as provide a more holistic view of N pollution to and in the region. The Wai'Ōpae tide pools and their watershed also receive, within a system. on average, 200 cm of rain near-shore and 400 cm of rain upslope annu- There is a great need to use an approach like this in tropical coastal ally (Juvik and Juvik, 1998). As a result of the conditions at the Wai'Ōpae waters, especially those with coral reefs. Most coral reefs are located tide pools, the community and county are concerned that sewage may in developing countries where poorly treated or raw sewage enters be transported by groundwater at low tide and impacting the tide the coastal waters (Shahidul Islam and Tanaka, 2004), and sewage is pool ecosystem, which supports an abundance of corals and marine or- one of the primary threats to coral reefs worldwide (Wear and Vega ganisms and is used recreationally by the local population and tourists. Thurber, 2015). Even in developed nations where coral reefs are located (i.e., Hawai'i, U.S.A.), sewage treatment is inadequate (Whittier and El- Kadi, 2014). With the projected growth in the global population by 2 bil- lion people over the next 35 years (Gerland et al., 2014), and preferen- tial growth in tropical coastal regions (Neumann et al., 2015), the amount of sewage entering nearshore waters will increase in the ab- sence of significant intervention. An ideal place to use this approach to determine sewage inputs to tropical coastal waters is Hawai'i because it has the highest usage of gang and domestic cesspools of any state in the United States (USEPA, 2011), permeable substrate, high rainfall, large amounts of groundwa- ter entering coastal areas, and high recreational use of the ocean (Oki et al., 1999). In particular, Hawai'i Island is a good location for this type of research because it is estimated that 77% of the county's popula- tion is serviced by cesspools, which is approximately 50,000 units, com- prising 56% of those in the state (Whittier and El-Kadi, 2014). Objectives of this study were to identify the potential sources of N to a tide pool ecosystem located on Hawai'i Island, determine their relative percent − 15 18 − contributions to tide pool water NO3 using δ Nandδ OofNO3 and δ11B in a mixing model, and assess which of these N sources are being used by benthic macroalgae by measuring δ15N in their tissues. This study is the first to use this combination of isotopes together with a mixing model to determine N inputs to tropical coastal waters and Fig. 1. Location of the Wai'Ōpae tide pools, Kapoho, Hawai'i, U.S.A. T.N. Wiegner et al. / Marine Pollution Bulletin 103 (2016) 63–71 65

2.2. Study approach For this study, the two regions with the highest groundwater influ- ence were selected for 100-m onshore-offshore transects for water The goal of this study was to determine the relative percent contri- and macroalgal tissue sample collection. Transect 1 was located outside butions of different N sources entering the Wai'Ōpae tide pools using of the MLCD boundary and transect 2 was located within MLCD bound- stable isotope signatures of surface waters and benthic macroalgae. To ary (Fig. 2). Tide pool water and macroalgal tissues were collected at accomplish this, potential N sources (sewage, agriculture soil, ground- low tide, once monthly during July, August, and September 2010. One- water, ocean water) were sampled within the watershed and analyzed liter water samples were collected in acid cleaned, triple sample- 15 18 − 11 for δ Nandδ OinNO3 ,andδ B, and used for identification of N rinsed high density polyethylene containers with sample water from sources to the tide pools' surface waters and benthic macroalgal tissues. the surface (~5 cm depth), where freshwater floats due to density strat- Next, groundwater seeps were identified using high-resolution spatial ification, for a total of eight designated distances along the transect line surface water salinity and temperature mapping, and once the areas at 0, 5, 10, 20, 25, 50, 75, and 100 m. These distances were chosen in were identified, onshore-offshore transects were established for surface order to map the spatial extent of sewage N within the tide pools. water and benthic macroalgae sample collection. The relative percent Water samples for nutrient and isotope analyses were filtered in the contributions of the different N sources to tide pool waters were deter- field through pre-combusted GF/F and sterile 0.22-μm cellulose acetate 15 18 − mined using a mixing model which utilized the δ N- and δ O–NO3 filters, respectively, transported on ice to the laboratory, and stored fro- and δ11B isotopic signatures of the tide pool waters and N sources. A zen until analysis. mixing model could not be utilized to determine N source contributions For benthic macroalgae analysis, three replicate tissue samples were to macroalgal tissues because tissues were only analyzed for δ15N and collected from the same eight distances (0 through 100 m) along the the model requires two or more isotopes for analysis when assessing transect line used for water quality sampling, resulting in 24 samples multiple sources (Parnell et al., 2010). Thus, δ15N of macroalgal tissues per transect. Triplicate macroalgal tissue samples were collected to as- were compared to the δ15N of the N sources to determine where they sess the δ15N variability among macroalgal species, distance along tran- obtained their N from (Derse et al., 2007). sect, and transect locations. Pilot surveys indicated that a variety of algal species were present along the benthic substrate of the tide pools. The macroalgae were identified visually with an Olympus™ CH30 micro- 2.3. Sample collection scope using published identification books (Abbott, 1999; Abbott and Huisman, 2004; Huisman et al., 2007). Macroalgal tissue samples col- Possible N sources to the Wai'Ōpae tide pools are sewage (septic lected in the field were sorted and placed on ice for transport to labora- tank sludge, n = 8N,O, 4B), agriculture soil (collected from papaya tory, where tissues were dried in an oven at 60 °C to a constant weight, farms, n = 7N,O, 4B), groundwater (collected from drinking water ground, and homogenized using a Wig-L-Bug grinding mill. Approxi- wells upland of agriculture lands, n = 14N,O, 7B), and ocean water mately, 2 mg of the ground sample were placed in 4 × 6 mm tin capsules (n = 9N,O, 3B). Three sites per source within the Wai'Ōpae tide pools' for elemental and isotope analyses. Vouchers collected at the time of watershed were sampled January, July, August, and September 2010 sampling were preserved in 4% formaldehyde and used for identifica- during dry conditions for dissolved nutrient concentrations and stable tion purposes. Algal slides were created with 1% aniline blue stain, 1% isotopic composition; ocean water was not sampled in January 2010 HCL (used to set stain), and 25% Karo and phenol mixture to seal the and only three septic systems were sampled due to limited accessibility cover slip to the slide. of private property. Each groundwater and agricultural soil sample was collected several kilometers apart from one another, while ocean water 2.4. Sample analyses samples were collected from three locations evenly dispersed across the tide pool area. Fertilizers [14–14–14 (N:Phosphorus (P):Potassium (K)), Nutrients in water samples were measured on a Pulse Technicon™ II 16–16–16 (N:P:K)] used on papaya farms (obtained from a local suppli- autoanalyzer using standard autoanalyzer methods and reference ma- − − er) and agricultural soil samples from various locations within the wa- terials (NIST; HACH 307-49, 153-49, 14242-32, 194-49): NO3 +NO2 −1 + tershed were collected. For each soil sample, 10 g of dried (at 60 °C for [Detection Limit (DL) 0.1 μmol L , USEPA 353.4)], NH4 (DL −1 3− −1 48 h or until constant weight was obtained) soil was suspended in 1.0 μmol L , USGS I-2525), PO4 (DL 0.1 μmol L , USEPA 365.5), −1 100 mL of deionized water, shaken overnight, and filtered through a H4SiO4 (DL 1.0 μmol L , USEPA 366), and total dissolved phosphorus pre-combusted (500 °C for 6 h) GF/F filter (Whatman™) and a sterile (TDP) (DL 0.5 μmol L−1, USGS I-4650-03). Total dissolved nitrogen 0.22-μm cellulose acetate filter (Whatman™) for nutrient and stable (TDN) (DL 5.0 μmol L−1, ASTM D5176) analyses were measured on a isotope analyses, respectively. The extract was representative of the Shimadzu™ TOC-V CSH, TNM-1 analyzer and Low Carbon Water and − leachable soil NO3 after one rain event and this approach has been Deep Seawater Reference Material were used (University of Miami, D. used in previous studies (Derse et al., 2007). Hansell laboratory). Nutrient analyses were conducted at the University Groundwater seeps to Wai'Ōpae tide pools were identified during of Hawai'i at Hilo Analytical Laboratory. 15 18 − the lowest tidal heights of the year by creating high-resolution Analyses of δ Nandδ OinNO3 were conducted on a Thermo- spatial surface water salinity and temperature maps. Mapping was Finnigan™ Delta Plus isotope ratio mass spectrometer (IRMS) with conducted from April through November 2009 at low tide when data normalized to U.S. Geological Survey (USGS) standards (USGS32, groundwater influence was greatest and easiest to detect. High- USGS34, USGS35). IAEA-NO3 was used as a check standard. These anal- resolution spatial surface water mapping was conducted with a YSI™ yses were conducted at the Northern Arizona University Stable Isotope 6600 V2 multi-parameter sonde, attached to a YSI™ 650 MDS data Laboratory. Analyses of δ11B were conducted on a Finnigan™ Triton logger, and a Garmin™ E-Trex global positioning system (GPS), multi-collector thermal ionization mass spectrometer with data nor- which recorded data every three s, resulting in thousands of data malized to NSB951 boric acid. These analyzes were conducted at the points during a survey. The mapping methodology was modified from University of Calgary, Canada, Isotope Science Laboratory. Macroalgal Madden and Day (1992). First, the outer edge of the sampling area tissues were analyzed on a Thermo-Finnigan™ Delta V Advantage was delineated with the YSI sonde, data logger, and GPS attached to a mass spectrometer with a Conflo III interface and a Costech™ ECS floatation device, and then they were towed around in a ‘zig-zag’ fash- 4010 Elemental Analyzer with data normalized to USGS standard NIST ion within the delineated boundary. Data were then uploaded into a 1547 at the University of Hawai'i at Hilo Analytical Laboratory. Isotopic mapping program (Golden Software Surfer 9.0™), overlaid onto an ae- signatures are expressed as standard (δ) values, in units of parts per mil rial map, interpolated, and used to create salinity and temperature maps (‰), and calculated as = [(Rsample − Rstandard)/Rstandard]×1000,where (Fig. 2). R=15N/14N,18O/16O, or 12B/11B. 66 T.N. Wiegner et al. / Marine Pollution Bulletin 103 (2016) 63–71

Fig. 2. Map of surface water (a) salinity (ppt) and (b) temperature (⁰C) at Wai'Ōpae tide pools, Kapoho, Hawai'i, U.S.A. The highest water temperatures were found in conjunction with lowest salinity values, indicating that geothermally heated groundwater was entering the tidal pools. The site was mapped using a YSI 6600 V2 multi-parameter water quality sonde from March to October 2009. Lines and numbers on salinity map indicate transect locations (transects 1 and 2).

15 18 − 2.5. Data analyses regression analysis of our δ N- and δ O–NO3 data revealed that these isotopes were not being enriched in a 2:1 ratio as expected with 15 18 − 11 The δ Nandδ OinNO3 ,andδ B of sources were compared using denitrification (Kendall, 1998). Hence, a denitrification enrichment fac- a one-way analysis of variance (ANOVA). Dissolved nutrient concentra- tor was not incorporated into SIAR. Lastly, data were analyzed by tran- 15 18 − 11 tions and δ Nandδ OinNO3 ,andδ B of tide pool waters were com- sect and distance by averaging isotopic values across dates. Due to the pared among distances (0, 5, 10, 20, 25, 50, 75, and 100 m) and transect few data points from the 75-m distance, this location was excluded as + location (transect 1 and transect 2) using a two-way ANOVA. NH4 , a distance in the mixing model. Percent contributions are reported as 15 18 − H4SiO4,andδ Nandδ OinNO3 data were log-transformed to ensure the 50% Bayesian credibility interval which allows for a wider range of they met the assumptions of parametric statistics. The δ15N values of isotopic variability to be presented (Parnell et al., 2010; Atwood et al., macroalgal species that could be identified and the ones which had at 2011). least three or more samples collected were compared among species using a one-way ANOVA. Macroalgal tissue δ15N were compared 3. Results among distances and between transects using a two-way ANOVA. Post-hoc analyses were conducted using the Tukey HSD multiple com- 3.1. Sources parisons test. All statistical analyses were performed in SYSTAT™ soft- ware package v. 11 at α level of 0.05. There were significant differences in nutrient concentrations and + Mixing models in the stable isotope analysis in R (SIAR) v. 4.0 pro- isotopic signatures among N sources. Sewage had the highest NH4 , 3− − gram were used to determine the relative percent contributions of the TDN, PO4 , TDP, and H4SiO4 concentrations and the NO3 was signifi- different N sources to tide pool waters (Parnell et al., 2010; Atwood cantly enriched in 15Nand18O compared to agriculture soils and et al., 2011; Xue et al., 2012). N sources used in this model included sew- groundwater (p b 0.0001 and p b 0.0001, respectively) (Table 2). Agri- − − age, agricultural soil, and groundwater. Both their measured stable iso- culture soil had the highest NO3 +NO2 and lowest H4SiO4 concentra- tope values and concentrations were used in the model. Ocean water tions compared to other N sources, while ocean water had the lowest − − + 3− was not included in the model as originally intended because nitrate NO3 +NO2 ,NH4 , TDN, PO4 , and TDP concentrations. Ocean water concentrations in seven of the nine samples were below detection limits was the most enriched in 11B compared to all other sources, with for the stable isotope analysis (N2 μmol L−1; Coplen et al., 2012), and groundwater, sewage, and agricultural soil having successively less 11B therefore, its contribution to the nitrate pool in tide pool water was con- (Table 2). The δ11B of groundwater was significantly higher compared sidered to be negligible. The SIAR mixing model program uses Bayesian to agriculture soil (p = 0.033), but similar to sewage (Table 2). While − methods and is able to reflect natural variation and uncertainty within a ocean waters were analyzed for stable isotopes of N and O in NO3 , system by allowing numerous sources of variability to be incorporated these data were not statistically analyzed because their values could (Parnell et al., 2010). The model estimates the probability distribution not be accurately determined due to their nitrate concentrations being for the relative contribution of each source to the mixture (tide pool below detection limits (Coplen et al., 2012). Fertilizers were also not sta- water) taking into account uncertainty associated with multiple sources tistically analyzed in the above ANOVAs because only two samples were and their isotopic compositions. SIAR assumes that the stable isotopic taken in this study. − composition of the NO3 sources are normally distributed, and Ryan Joiner test (similar to Shapiro–Wilk test) was used assess whether this 3.2. Tide pool water 15 18 − assumption was met with our data. δ N- and δ O–NO3 data were nor- 18 − mally distributed for all sources, except for three δ O–NO3 values, one There were significant differences in dissolved nutrient concentra- − − for soil and two for groundwater. Although denitrification was possible tions from tide pool waters. Dissolved NO3 +NO2 , TDN, TDP, and in the tide pool waters at some locations on certain dates (dissolved ox- H4SiO4 concentrations were significantly different among distances ygen concentration range 1.80–5.08 mg L−1, Wiegner unpubl. data), (p = 0.005, 0.004, 0.042, and 0.015, respectively); however, only T.N. Wiegner et al. / Marine Pollution Bulletin 103 (2016) 63–71 67

Table 2 Average (S.E.) and range of dissolved nutrient concentrations (μmol L−1) and stable isotope signatures (‰) for N sources to Wai'Ōpae tide pools, Kapoho, Hawai'i, U.S.A. n = number of samples; Ag. = Agriculture; bdl = below detection limits (see methods for more details).

− − + −3 15 − 18 − 11 Source n NO3 +NO2 NH4 TDN PO4 TDP H4SiO4 δ N–NO3 δ O–NO3 δ B Sewage 8 9.4 (8.3) 2456 (502) 2627 (493) 176.8 (44.8) 153.0 (28.3) 263 (105) 12.30 (1.72) 16.03 (2.21) 0.26 (3.28)a bdl to 67.6 1 to 4202 72 to 4218 2.2 to 402.0 2.1 to 230.0 19 to 781 11.49 to 16.98 15.51 to 20.00 −8.70 to 7.00 Fertilizers 2 –––– – –−0.85 (0.55)b –– –––– – –−1.40 to −0.30 –– Ag. Soil 8 64.8 (24.8) 225 (158) 584 (245) 30.9 (16.7) 43.1 (21.9) 9 (4) 1.33 (2.16)c 7.06 (1.82)d -12.08 (5.08)d 1.5 to 192.0 bdl to 1320 23 to 2069 1.0 to 136.0 4.5 to 158.0 bdl to 31 −4.68 to 9.30 −0.48 to 13.87 −21.30 to 2.00 Groundwater 20 15.0 (1.1) 1 (0) 21 (1) 2.7 (0.2) 3.1 (0.2) 160 (31) 4.32 (0.68)e 3.57 (0.96)e 10.59 (3.75)f 8.5 to 27.3 bdl to 7 16 to 29 0.9 to 3.9 1.2 to 3.9 31 to 436 1.25 to 7.96 −0.55 to 10.26 −4.70 to 38.00 Ocean 9 0.0 (0.0) 0 (0) 7 (1) bdl bdl bdl bdl bdl 40.67 (0.64)g bdl to 0.2 bdl to 1 bdl to 11 39.50 to 41.70

a n =4. b For δ15N analyses, solid samples of 14–14–14 (N–P–K) and 16–16–16(N–P–K) fertilizers were used. c n =7. d n =4. e n =14. f n =8. g n =3.

− − 11 NO3 +NO2 and TDN concentrations significantly differed between 22–24%). Without δ B data, groundwater (38%, 25–53%) was identified − transect location (p = 0.020 and p = 0.004, respectively), with transect as the largest NO3 source, followed by agriculture soil (34%, 22–49%), 1 having values that were three times higher than transect 2. Dissolved and then sewage (27%, 1–40%) (Table 4). − − + NO3 +NO2 ,NH4 , TDN, TDP, and H4SiO4 concentrations along transect 1 showed a decreasing trend in values from 0 m to 100 m. A similar − − trend in NO3 +NO2 , TDN, and TDP along transect 2 was observed; 3.4. Macroalgae + however, NH4 and H4SiO4 concentrations showed a slight increase in 15 18 − 11 values from 0 m to 100 m. The δ Nandδ OinNO3 and δ B signatures The two most common macroalgae found along transect 1 were of tide pool waters were not significantly different among distances or Valonia sp. and Cladophora sp. (Fig. 4). The three most abundant 15 18 − transects sampled. The δ Nandδ O values of NO3 in tide pool waters macroalgae found along transect 2 were Amansia glomerata, from all distances along both transects have values near agriculture soil Dichotomaria marginata,andGalaxaura rugosa (Fig. 4). Other macroalgal and groundwater (Fig. 3). The δ11B signatures of tide pool waters, from species found included Ceramium (transect 1, δ15N = 1.70‰, n = 1), all distances, along both transects, have values near ocean water Pterocladiella (transect 1, δ15N=2.50‰, n = 1; transect 2, δ15N= (Tables 2 and 3). 1.90‰, n = 1), and unknown (transect 1, δ15N(average±S.E.)= 2.45‰ ±0.05,n = 2). The δ15N of macroalgal tissues significantly dif- 3.3. SIAR mixing model fered among species (p b 0.0001) (Fig. 4); Valonia sp. and Cladophora sp. were significantly more enriched in 15N than Amansia sp., Percent contributions of the N sources were similar when analyzed D. marginata,andG. rugosa. Across all species, δ15N ranged between by transect location in the SIAR mixing model. Therefore, the transect −0.90‰ (D. rugosa) to 4.90‰ (Valonia sp.), with an average of data were combined and re-analyzed for a more robust analysis. The 1.58‰ ± 1.13. Studies that have sampled the δ15N of numerous SIAR mixing model was run with and without δ11B data. With the δ11B algal species have analyzed their data by species if all species could data, the model identified groundwater (average: 41%, range: 34–43%) be identified and there were replicate samples of each (Cole et al., − as the largest NO3 source to the tide pool waters among all distances, 2004; Savage, 2005; Dailer et al., 2010). In contrast, studies that com- followed by sewage (36%, 34–42%), and then agricultural soil (23%, piled δ15N data of different species into a composite value did so be- cause they determined no differences in isotopic values among species and/or the availability of individual species during sample collection were variable (Costanzo et al., 2001; Derse et al., 2007; Dailer et al., 2010). Although macroalgal species from our study had different δ15N tissue values, we grouped all algal species together in order to conduct a two-way ANOVA examining the effect of distance and transect loca- tion on δ15N of macroalgal tissues. Algal species were grouped together because all specimens collected could not be positively identified, and common algal species were not found at the two transect locations. The average δ15N of macroalgal species found along transect 1 was significantly higher (2.50‰ ± 0.08) than species from transect 2 (0.97‰ ± 0.08) (p b 0.0001, Fig. 5). Within each transect, macroalgal tissue δ15Nsignificantly differed among distances (p b 0.0001). For tran- sect 1, macroalgae at the 0-m distance was the most enriched with re- spect to 15N. In contrast, macroalgae at the 5-m distance was the most depleted with respect to 15N along transect 2. When macroalgal tissue δ15N values were plotted by transect against distance from shore and δ15 δ15 – − δ18 – − ‰ Ō compared with the N values of the N sources from our study, all Fig. 3. Biplot of average N NO3 and O NO3 ( ) in tide pool waters from Wai' pae, δ15 δ15 Kapoho, Hawai'i, U.S.A. Average stable isotope values for potential N sources to the tide N of the compiled macroalgal tissues had N values similar to agri- pools, as well as the range in their values (boxes) are shown on figure. Error for average culture soil, groundwater, and fertilizers (Fig. 5, Table 2), with most fall- values can be found in Tables 2 and 3. ing within the agriculture soil range. 68 T.N. Wiegner et al. / Marine Pollution Bulletin 103 (2016) 63–71

Table 3 −1 15 18 − 11 Average (S.E.) salinity (ppt), dissolved nutrient concentrations (μmol L ), as well as δ N- and δ O–NO3 ,andδ B values(‰) for Wai'Ōpae tide pool waters, Kapoho, Hawai'i, U.S.A. 3− Water samples are reported by transect location (Transects 1 and 2) and distance from shore (0, 5, 10, 20, 25, 50, 75, and 100 m). (*) indicates where PO4 concentrations were detectable. bdl = below detection limits (see methods for more details). Sample size (n) is three for all transects and distances, except for 75-m distance for both Transect 1 and 2, where n =2.

− − + 15 18 11 Transect Distance Salinity NO3 +NO2 NH4 TDN TDP H4SiO4 δ N δ O δ B 1 0 16.50 (1.46) 14.3 (6.2) 4 (2) 25 (5) 0.7 (0.3)* 77 (33) 5.94 (1.14) 4.63 (1.24) 40.05 (0.92) 1 5 16.93 (0.78) 14.4 (1.6) 1 (1) 23 (4) 0.8 (0.3)* 120 (40) 6.91 (2.19) 7.71 (4.80) 39.71 (1.32) 1 10 19.13 (1.73) 11.6 (2.2) 2 (2) 20 (3) 0.7 (0.3)* 89 (22) 5.57 (0.83) 4.41 (1.10) 39.25 (0.81) 1 20 21.27 (0.74) 14.0 (3.0) 1 (1) 21 (1) 0.4 (0.0)* 78 (23) 6.40 (1.14) 5.32 (1.81) 39.47 (0.81) 1 25 23.27 (2.00) 9.5 (1.3) bdl 18 (1) 0.4 (0.0) 38 (3) 5.96 (1.31) 4.87 (1.70) 40.43 (0.35) 1 50 26.43 (1.47) 9.2 (1.1) bdl 16 (1) 0.2 (0.0) 57 (13) 5.70 (1.78) 5.16 (1.88) 39.71 (1.32) 1 75 31.00 (0.20) 3.0 (0.3) bdl 12 (0) bdl 26 (1) 7.55 (3.19) 8.86 (5.62) 42.73 (0.90) 1 100 33.90 (1.01) 2.1 (1.6) 1 (1) 11 (3) bdl 24 (11) 9.12 (2.00) 10.97 (2.95) 41.77 (0.85) 2 0 22.17 (0.18) 5.6 (1.0) bdl 16 (0) 0.8 (0.5) 47 (15) 6.77 (1.14) 6.69 (1.36) 38.80 (2.00) 2 5 18.27 (1.64) 12.0 (3.0) 1 (1) 18 (2) 1.0 (0.7)* 120 (29) 6.78 (1.40) 6.40 (1.27) 38.40 (2.62) 2 10 21.10 (1.24) 9.0 (2.8) bdl 15 (2) 0.2 (0.1) 102 (35) 7.29 (1.33) 7.54 (1.43) 38.53 (0.79) 2 20 22.30 (1.80) 7.7 (3.2) bdl 16 (2) 0.3 (0.1) 85 (23) 6.17 (0.86) 6.63 (0.83) 39.57 (0.84) 2 25 22.90 (2.08) 6.7 (2.1) bdl 14 (2) 0.3 (0.1) 39 (7) 6.51 (1.07) 6.85 (1.28) 39.70 (0.90) 2 50 27.60 (30.10) 3.9 (1.5) bdl 12 (1) 0.1 (0.1) 48 (19) 7.60 (1.18) 8.11 (2.00) 40.03 (0.78) 2 75 30.40 (6.00) 3.5 (3.5) bdl 11 (2) 0.1 (0.1) 54 (54) 7.95 (2.24) 10.85 (4.78) 41.91 (0.17) 2 100 28.90 (3.86) 3.1 (2.0) 1 (1) 13 (3) 0.2 (0.1) 50 (26) 7.22 (2.25) 8.40 (3.84) 39.80 (2.34)

4. Discussion four fertilizers (ammonium nitrate, urea, and phosphate-based prod- ucts) ranged from −2‰ to 14.8‰, and are attributed to the B mineral 4.1. N Sources used in fertilizer production (Komor, 1997, Wieser et al., 2001). Foliar applications of borax and ground applications of elemental B fertilizers In regions impacted by N pollution, identifying the dominant sources are recommended for papayas to prevent and reduce B deficiencies is important for pollution mitigation and land use management. This that cause bumpiness or deformities in the fruit. Due to a large range can be accomplished, in addition to determining the percent contribu- in isotopic compositions of B in the literature that are derived from var- tion of the different N sources, with stable N isotope measurements ious B sources, further investigations of δ11BforBsourcesareneededto and a mixing model. However, to do this successfully, the different N better utilize δ11B values to trace pollution and determine their relative sources must have unique isotopic signatures. The use of two or more contributions. stable isotope tracers can help overcome this limitation. In our study, we used stable O and B isotopes in addition to those for N to identify 4.2. Tide pool water N pollution sources to the Wai'Ōpae tide pools, and each N source exam- 15 18 − 11 15 18 − 11 ined had significantly different δ Nandδ OinNO3 ,andδ B. Our δ Nandδ OinNO3 and δ B along upland-lowland gradients have 15 − δ NinNO3 for sewage, agricultural soil, and groundwater fall within been used to determine N sources to groundwater in several studies the range reported in the literature (Tables 1 and 2). With regards to (Leenhouts et al., 1998; Seiler, 2005; Hunt, 2006, 2014; Xue et al., 15 − the δ N–NO3 values in the agriculture soil we sampled, they had the 2014). However, to our knowledge, our study is the first to apply this widest range and this result could be due to timing of fertilizer applica- multi-isotope approach to detect N sources to coastal waters along an tions and the fact that soil samples were collected from different farms. onshore–offshore gradient. Dissolved nutrient concentrations (except We also analyzed δ15N from four fertilizers used in the Kapoho region; however, only two of them (14–14–14 and 16–16–16) had measurable − 15 NO3 concentrations, and their δ N values were depleted (0.0‰ and −1.3‰, respectively) compared to those in the agriculture soil. The 18 − δ OinNO3 values of agriculture soil and groundwater from our study fell within the range of literature values (Tables 1 and 2), while the sewage value was more enriched in 18O compared to literature values (Tables 1 and 2). The δ11B values of sewage, groundwater, and ocean water that were measured in our study fell within the range of lit- erature values (Table 1), while agriculture soil exhibited highly depleted values which could be due to the source of B in the various fertilizers ap- plied on upland papaya farms (Wieser et al., 2001). The δ11B values from

Table 4 Average (range) relative percent contributions of N sources [sewage, agriculture soil (Ag. Soil), and groundwater] to Wai'Ōpae tide pool waters, Kapoho, Hawai'i, U.S.A, from 0, 5, 10, 20, 25, 50 and 100 m from shore. Percent contributions were determined using SIAR v 4.0 and values are reported as the 50% credibility interval. Data from distances along transects were combined for this analysis.

Distance (m) Sewage Ag. Soil Groundwater

0 26 (14–38) 35 (24–48) 40 (26–53) Fig. 4. Average (S.E.) δ15N(‰) of macroalgal species from Wai'Ōpae tide pools, Kapoho, 527(3–34) 35 (24–49) 39 (27–52) Hawai'i, U.S.A. Results from a one-way ANOVA are shown on figure (α =0.05). 10 26 (1–41) 34 (23–49) 39 (26–52) Differences among macroalgal species were examined with a Tukey's test and bars 20 26 (2–38) 35 (23–49) 40 (27–53) with different letters are significantly different from one another. Other macroalgal 25 27 (2–40) 35 (23–49) 39 (25–51) species found include Ceramium (transect 1, δ15N=1.70‰, n =1),Pterocladiella 50 28 (4–38) 34 (23–48) 38 (25–51) (transect 1, δ15N=2.50‰, n =1;transect2,δ15N=1.90‰, n = 1), and unknown 100 33 (16–40) 33 (22–47) 34 (25–49) (transect 1, δ15N = 2.45‰ ±0.05,n =2). T.N. Wiegner et al. / Marine Pollution Bulletin 103 (2016) 63–71 69

from 23 to 93 μmol L−1 (Leenhouts et al., 1998). Hence, from this point forward, we will be discussing the SIAR results without δ11B data. We found that agriculture groundwater, soil, and sewage source contributions to tide pool waters were 38%, 34%, and 27%, respectively. The combined percent contribution of anthropogenic N to tide pool wa- ters from agriculture soil and sewage was 61%, while N from groundwa- − − ter was 38%. High NO3 +NO2 concentrations near-shore in low salinity waters further suggest that agriculture soil was the main nutri- − ent source to tide pool waters, as NO3 is often associated with agricul- ture activities and its concentration was highest in our agricultural soil source (Addiscott et al., 1991; Bruland and Mackenzie, 2010; Bishop et al., 2015). Output from the mixing model is important for land man- agers, and ours suggest that more effective fertilizer usage and timing of applications will help minimize leaching into the groundwater and im- + prove water quality. Additionally, high NH4 concentrations at 0 m on transect 1 suggest that sewage may be entering the tide pools at this lo- cation as there was a higher concentration of homes adjacent to this Fig. 5. Average (S.E.) δ15N(‰) of macroalgal tissues found along transects 1 and 2, from transect. While sewage contributed a smaller percentage of the total N Ō δ15 Wai' pae tide pools, Kapoho, Hawai'i, U.S.A., with respect to NoftheNsources compared to agriculture soil, the residential area contributes two orders average (S.E.) from this study. Transect 1 is denoted by squares and transect 2 by triangles. of magnitude more N than agriculture fields on an aerial basis. It is projected that population growth and nutrient loads will increase in the future; these factors could have profound impacts on the tide + NH4 ) and isotopic compositions of tide pool waters did not significantly pool ecosystem by potentially altering the N sources and forms, and in- vary between transect location, but did vary with distance from shore creasing the quantity. These changes in the N flux could stimulate ex- suggesting that groundwater discharging along the shoreline was the cess macroalgal growth, possibly resulting in a phase-shift from a primary source of these constituents; other studies have observed sim- coral- to an algae-dominated ecosystem, like observed on Caribbean, ilar patterns (Hunt and Rosa, 2009; Bruland and Mackenzie, 2010). Iso- Indonesian, and Australian reefs (Hughes, 1994; McManus and topic signatures of tide pool waters were similar among distances and Polsenberg, 2004). Increases in agriculture could also contribute to between transect location suggesting that there was minimal microbial these types of conditions, and therefore, it is important for land man- 15 18 − N transformation affecting the δ Nandδ OinNO3 with distance from agers to have information on percent N source contribution, like those 15 − shore. The δ NinNO3 values from surface waters were similar in iso- from the mixing model, to make informed decisions. topic composition to agriculture soil and groundwater (Fig. 3), suggest- ing that agriculture soil from the watershed is the main anthropogenic 4.3. Macroalgae − − N source to tide pool waters. Dissolved NO3 +NO2 concentrations de- creased with distance from shore suggesting either uptake by phyto- Worldwide, δ15N values in macroalgal tissues have been shown to be plankton and/or denitrification; however, it appears that this latter effective bioindicators of N pollution in water bodies (Cohen and Fong, 18 − process was not occurring in the tide pool waters as the δ OinNO3 2006; Savage, 2005; Smith et al., 2005; Risk et al., 2009; Dailer et al., did not significantly vary within the tide pools with increasing distance 2010). In this study, there were significant differences in δ15N tissue − 15 from shore, and denitrification leaves the residual NO3 isotopically values among macroalgal species. Differences in δ N of macroalgal tis- enriched in 18O(Aravena et al., 1993; Hunt, 2006; Hunt and Rosa, sues may be due to light conditions, nutrient availability, and plant 2009). The δ11B values were similar among distances and between tran- physiology (Cole et al., 2004; Smith and Smith, 2006; Bannon and sect locations, indicative of a well-mixed system, and were close in Roman, 2008; Carballeira et al., 2014; Swart et al., 2014). Water depth value to the ocean water end member, suggesting that ocean water along transects did not exceed ~2 m which allowed for ample light to was the largest B source to tide pools. reach the benthic macroalgae and light levels were consistent among Traditionally, mixing models have been used to assess the percent distances and transects. Water column stratification was also assessed, contributions of food sources to the diets of organisms within food and there was no horizontal (8.16‰ at 0-m, 8.73‰ at 100-m) or vertical 15 − webs (Phillips and Gregg, 2003; Parnell et al., 2010, Atwood et al., (8.52‰ surface, 8.36‰ bottom) differences in δ N–NO3 . Also, nutrient 2011). Here, we used mixing models to determine the percent contribu- concentrations and isotopic signatures of surface waters were similar tion of N sources to a coastal water body along an onshore-offshore between transect location. All three of these findings suggest that light 15 18 − 11 transect using δ Nandδ OinNO3 and δ B. While this approach and nutrient availability were not factors contributing to significant dif- 15 18 − 15 has been used with δ Nandδ OinNO3 in several freshwater studies ferences in the δ N among macroalgal species, but that differences in (Xue et al., 2012, 2014; El Gaouzi et al., 2013; Yang et al., 2013; Ding plant physiology played a role. et al., 2014) and a single coastal water one (Korth et al., 2014), no There are many physiological attributes that could lead to differ- study to date has used it with δ11B. When comparing the SIAR results ences in isotopic values of algal tissues, including: nutrient uptake, with and without d11B data to δ15N values of the macroalgae (Fig. 5), growth rates, part of the plant sampled, age of plant tissue, phylum, tax- it appeared that the model overestimated the contribution of the sew- onomic and functional-form groups, and successional stages (Lobban − 15 age to the tide pool NO3 , as none of our macroalgal samples had δ N and Harrison, 1994; Carballeira et al., 2014). The algal species found values within the range of sewage (Tables 1 and 2). A bi-plot of δ15N along transect 1 consisted of green and red algae, while only red algae 18 − and δ OofNO3 also shows that the tide pool waters fall within the was found along transect 2 (Fig. 4); however, most algal species in our range of groundwater and agricultural soils, not sewage (Fig. 3). A re- study are late-successional species which are not typically used as cent study reported that B isotopes did not work well as a tracer for sew- bioindicator species in isotope studies (Umezawa et al., 2002; Cohen age in brackish waters (Hunt, 2014). They found that in water samples and Fong, 2006; Dailer et al., 2010). Early-successional opportunistic containing 15–20% saltwater, the B pool was dominated by seawater, algae, such as Ulva sp., Gracilaria sp., and Cladophora sp., are typically masking contributions from sources with lower B concentrations used in these studies because of their rapid growth rates in response (Hunt, 2014). Seawater has an approximate B concentration of to increased nutrients (Ryther et al., 1981; Peckol and Rivers, 1995; 398 μmol L−1, while municipal sewage has a concentration ranging Cohen and Fong, 2006); only one of these species was found 70 T.N. Wiegner et al. / Marine Pollution Bulletin 103 (2016) 63–71

(Cladophora sp.) in our study, and only on one transect. However, late- T. Gregg, M. Takabayashi, J. Burns, N. Rozet, P. Tanoue, C. Kryss, successional algal species are able to incorporate a larger record of nutri- R. Most, K. Carlson, W. Sako, M. Kaleikini, and Kapoho Kai Water Associ- ent sources over time and should be considered as potential study spe- ation. We would also like to thank K. McDermid and R. MacKenzie cies for future isotope studies instead of early-successional ones. The whom provided guidance in the finaldesignoftheprojectandthought- macroalgal genera reported in this study exhibit varying physiological ful comments to improve this manuscript. We also thank the anony- characteristics in their phylum and successional stages, which play mous reviewers whom provided comments on earlier drafts of this vital roles in nutrient uptake and N storage (Lobban and Harrison, manuscript. This material is based upon work supported by the National 1994). These factors may be responsible for the significant differences Science Foundation (NSF) under Grants No. 0833211, administered by in the δ15N of the algal tissues; however, without further testing, the the University of Hawai'i. exact physiological characteristics that affect the δ15N of macroalgal tis- sues from Wai'Ōpae tide pools cannot be conclusively determined. Many studies have effectively used δ15N in macroalgal tissues to References trace N pollution because their tissues reflect the source and availability Abbott, I.A., 1999. Marine Red Algae of the Hawaiian Islands. Bishop Museum Press, Ho- of nutrients over time (Umezawa et al., 2002; Smith et al., 2005; Derse nolulu, HI, p. 465. et al., 2007; Dailer et al., 2010). The δ15N of macroalgal tissues in this Abbott, I.A., Huisman, J.M., 2004. Marine Green and Brown Algae of the Hawaiian Islands. − Bishop Museum Press, Honolulu, HI, p. 259. study fell within the range reported for soil NO3 impacted by fertilizers δ15 ‰ Addiscott, T.M., Whitmore, A.P., Powlson, D.S., 1991. Farming, Fertilizers, and the Nitrate (Table 1), with an average N across all species of 1.58 ±1.13 Problem. Oxford University Press, New York, NY, p. 176. (Fig. 4), and no offshore gradient in signatures was observed. Surface Aravena, R., Evans, M.L., Cherry, J.A., 1993. Stable isotopes of oxygen and nitrogen in waters collected from the same distances along both transects that source identification of nitrate from septic systems. Groundwater 31, 180–186. Atwood, T.B., Wiegner, T.N., Mackenzie, R.A., 2011. Effects of hydrological forcing on the macroalgal tissues were collected also suggest that agriculture soils structure of a tropical estuarine food web. Oikos 121, 277–289. are the dominant anthropogenic N source to tide pool waters. Similar Bannon, R.O., Roman, C.T., 2008. Using stable isotopes to monitor anthropogenic nitrogen work utilizing δ15N in macroalgal tissues as bioindicators of N source inputs to estuaries. Ecol. Appl. 18, 22–30. δ15 Bishop, J.M., Glenn, C.R., Amato, D.W., Dulaiova, H., 2015. Effects of land use and ground- pollution typically report enriched N values in tissues which are at- water flow path on submarine groundwater discharge nutrient flux. J. Hydrol. Reg. tributed to sewage N pollution (Umezawa et al., 2002; Costanzo et al., Stud. http://dx.doi.org/10.1016/j.ejrh.2015.10.008 (in press). 2005; Bannon and Roman, 2008; Dailer et al., 2010). Results from stud- Boehm, A.B., Yamahara, K.M., Walters, S.P., Layton, B.A., Keyner, D.P., Thompson, R.S., ies implicating sewage as the dominant N source to macroalgal tissues Knee, K.L., Rosener, M., 2010. Dissolved inorganic nitrogen, soluble reactive phospho- rus, and microbial pollutant loading from tropical rural watersheds in Hawai'i to the have been in areas primarily serviced by sewage treatment facilities coastal ocean during non-storm conditions. Estuar. Coasts 34, 925–936. with either injection wells or outfalls, and not necessarily from on-site Bruland, G.L., Mackenzie, R.A., 2010. Nitrogen source tracking with δ15N content of coastal wetland plants in Hawai'i. J. Environ. Qual. 39, 409–419. units such as cesspools or septic systems. On-site treatment units, like δ15 Ō Carballeira, C., Rey-Asensio, A., Carballeira, A., 2014. Interannual changes in Nvaluesin those at Wai' pae, are spatially diffuse compared to injection wells Fucus vesiculosus L. Mar. Pollut. Bull. 85, 141–145. and sewage outfalls and this may lead to their inputs not being detected Casciotti, K.L., Trull, T.W., Glover, D.M., Davies, D., 2008. Constraints on nitrogen cycling at or their contributions to the N pool being small compared to other the subtropical North Pacific Station ALOHA from isotopic measurements of nitrate and particulate nitrogen. Deep-Sea Res. II 55, 1661–1672. sources. A recent study conducted on Maui found this to be the case Chang, C.C.Y., Kendall, C., Silva, S.R., Battaglin, W.A., Campbell, D.H., 2002. Nitrate stable − where NO3 from agriculture overwhelmed smaller contributions from isotopes: tools for determining nitrate sources among different land uses in the Mis- cesspools and septic tanks (Bishop et al., 2015). sissippi River Basin. Can. J. Fish. Aquat. Sci. 59, 1874–1885. Cohen, R.A., Fong, P., 2006. Using opportunistic green macroalgae as indicators of nitrogen supply and sources to estuaries. Ecol. Appl. 16, 1405–1420. 5. Conclusion Cole, M.L., Valiela, I., Kroeger, K.D., Tomasky, G.L., Cebrian, J., Wigand, C., McKinney, R.A., Grady, S.P., Carvalho da Silva, M.H., 2004. Assessment of a δ15N isotopic method to in- dicate anthropogenic eutrophication in aquatic systems. J. Environ. Qual. 33, N delivery to coastal waters is increasing worldwide with the 124–132. growing human population and expanding urbanization, industrializa- Coplen, T.B., Haiping, Q., Kinga, R., Casciotti, K., Hannon, J.E., 2012. Determination of the tion, and agriculture, and the result is widespread eutrophication δ15N and δ18O of nitrate in water; RSIL lab code 2900, chap. 17 of stable isotope- (Nixon, 1995). Eutrophication has led to phytoplankton and macroalgal ratio methods, sec. C of Revesz, Kinga, and Coplen, T.B. eds, Methods of the Reston Stable Isotope Laboratory; U.S. Geologic Survey Techniques and Methods, book 10, blooms, decreases in dissolved oxygen, changes in pelagic and benthic 35 pp. communities and habitats, food webs alterations, and collapse of fisher- Corcoran, E., Nellemann, C., Baker, E., Bos, R., Osborn, D., Savelli, H. (Eds.), 2010. Sick ies in many places (Nixon, 1995). Therefore, knowing where the N is Water? The Central Role of Wastewater Management in Sustainable Development. A Rapid Response Assessment. United Nations Environment Programme, UN- coming from within watersheds is paramount for better management HABITAT, GRID-Arendal (Report 978-82-7701-075-5). and sustainability of our coastal waters. We found that stable isotopes Costanzo, S.D., O'Donohue, M.J., Dennison, W.C., Loneragan, N.R., Thomas, M., 2001. Anew − approach for detecting and mapping sewage impacts. Mar. Pollut. Bull. 42, 149–156. of N and O in surface water NO3 and macroalgal tissues, combined Costanzo, S.D., Udy, J., Longstaff, B., Jones, A., 2005. Using nitrogen stable isotope ratios of with the use of a stable isotope mixing model was successful in macroalgae to determine the effectiveness of sewage upgrades: changes in the extent determining the N sources to surface waters and benthic macroalgal tis- of sewage plumes over four years in Moreton Bay, Australia. Mar. Pollut. Bull. 51, sues at the Wai'Ōpae tide pools. Mixing model results served as 212–217. Dailer, M.L., Knox, R.S., Smith, J.E., Napier, M., Smith, C.M., 2010. Using δ15Nvaluesinalgal an important tool in determining the relative percent N source contri- − tissue to map locations and potential sources of anthropogenic nutrient inputs on the butions to the tide pool NO3 and showed that agriculture soil was the island of Maui, Hawai'i, USA. Mar. Pollut. Bull. 60, 655–671. dominant anthropogenic N source to tide pool waters and benthic Derse, E., Knee, K.L., Wankel, S.D., Kendall, C., Berg Jr., C.J., Paytan, A., 2007. Identifying sources of nitrogen to Hanalei Bay, Kaua'i, utilizing the nitrogen isotope signature macroalgal tissues. This approach may be useful for other coastal areas in macroalgae. Environ. Sci. Technol. 41, 5217–5223. experiencing increased N loading, especially in areas where knowledge Deutsch, B., Mewes, M., Liskow, I., Voss, M., 2006. Quantification of diffuse nitrate inputs is lacking on the source(s) of N within the watershed. With this type of to a small river system using stable isotopes of oxygen and nitrogen in nitrate. Org. – information, communities, land managers, and policy makers can make Geochem. 37, 1333 1342. Ding,J.,Xi,B.,Gao,R.,He,L.,Liu,H.,Dai,X.,Yu,Y.,2014.Identifying diffused nitrate informed decisions regarding land-use that may help improve coastal sources in a stream in an agricultural field using a dual isotope approach. Sci. Total water quality and allow coastal systems to recover from nutrient Environ. 484, 10–18. pollution. Duarte, C.M., 1995. Submerged aquatic vegetation in relation to different nutrient re- gimes. Ophelia 41, 87–112. El Gaouzi, F.-Z.J., Sebilo, M., Ribstein, P., Plagnes, V., Boeckx, P., Xue, D., Derenne, S., Acknowledgments Zakeossian, M., 2013. Using δ15N and δ18O values to identify sources of nitrate in karstic springs in the Paris basin (France). Appl. Geochem. 35, 230–243. Engineering Concepts, Inc., 2010. Kapoho beach lots, farm lots, and vacationland estates We thank L. Loching, J. Gartner, B. Nishimoto, T. Sakihara, J. Michaud, wastewater feasibility report. Prepared for Hawai'i County Department of Environ- J. Adolf, L. Mead, T. Holitzki, B. Moan, M. Wieser, S. Taylor, T. Soares, mental Management, Hilo, HI. T.N. Wiegner et al. / Marine Pollution Bulletin 103 (2016) 63–71 71

Gerland, P., Raftery, A.E., Ševčíková, H., Li, N., Gu, D., Spoorenberg, T., Alkema, L., Fosdick, Prüss-Üstun, A., Bos, R., Gore, F., Bartram, J., 2008. Safer water, better health: Costs, bene- B.K., Chunn, J., Lalic, N., Bay, G., Buettner, T., Heilig, G.K., Wilmoth, J., 2014. World pop- fits and sustainability of interventions to protect and promote health. World Health ulation stabilization unlikely this century. Science 346, 234–237. Organization, Geneva. Hughes, T.P., 1994. Catastrophes, phase shifts, and large-scale degradation of a Caribbean Risk, M.J., LaPointe, B.E., Sherwood, O.A., Bedford, B.J., 2009. The use of δ15N in assessing coral reef. Science 265, 1547–1551. sewage stress on coral reefs. Mar. Pollut. Bull. 58, 793–802. Huisman, J.M., Abbott, I.A., Smith, C.M., 2007. Hawaiian Reef Plants. Hawai'i: University of Ryther, J.H., Corwin, N., DeBusk, T.A., Williams, L.D., 1981. Nitrogen uptake and storage by Hawai'i Sea Grant College Program Rep. No. UNIHI-SEAGRANT-BA-03-02, Honolulu, the red alga Gracilaria tikvahiae (McLachlan, 1979). Aquaculture 26, 107–115. HI. p. 264. Savage, C., 2005. Tracing the influence of sewage nitrogen in a coastal ecosystem using Hunt, C.D., 2006. Groundwater nutrient flux to coastal waters and numerical simulation stable nitrogen isotopes. Ambio 34, 145–150. of wastewater injection at Kihei, Maui, Hawai'i. USGS Scientific Investigations Report Savage, C., Elmgren, R., 2004. Macroalgal (Fucus vesiculosus) δ15N values trace decrease in 2006-5283. sewage influence. Ecol. Appl. 14, 517–526. Hunt, C.D., 2014. Baseline water-quality sampling to infer nutrient and contaminant Seiler, R.L., 2005. Combined use of δ15Nandδ18O of nitrate and δ11B to evaluate nitrate sources at Kaloko-Honokōhau National Historic Park, Island of Hawai'i, 2009. USGS contamination in groundwater. Appl. Geochem. 20, 1626–1636. Scientific Investigations Report 2014-5158. Shahidul Islam, M., Tanaka, M., 2004. Impacts of pollution on coastal and marine ecosys- Hunt, C.D., Rosa, S.N., 2009. A multi tracer approach to detect wastewater plumes from tems including coastal and marine fisheries and approach for management: a review municipal injection wells in nearshore marine waters in Kihei and Lahaina, Maui, and synthesis. Mar. Pollut. Bull. 4, 624–649. Hawai'i. USGS Scientific Investigations Report 2009-5253. Smith, V.H., 2003. Eutrophication of freshwater and coastal marine ecosystems: a global Juvik, S.P., Juvik, J.O., 1998. Atlas of Hawai'i. third ed. University of Hawai'i Press, Honolulu, problem. Environ. Sci. Pollut. Res. 10, 126–139. HI, p. 352. Smith, C.M., Smith, J.E., 2006. Algal blooms in North Kihei: an assessment of patterns and Kendall, C., 1998. Tracing nitrogen sources and cycling in catchments. In: Kendall, C., processes relating nutrient dynamics to algal abundance. Prepared for County of McDonnell, J.J. (Eds.), Isotope Tracers in Catchment Hydrology. Elsevier, Maui, Lahaina, HI. Amsterdam, Netherlands, pp. 519–576. Smith, J.E., Runcie, J.W., Smith, C.M., 2005. Characterization of a large-scale ephemeral Komor, S.C., 1997. Boron contents and isotopic compositions of hog manure, selected fer- bloom of the green alga Cladophora sericea on the coral reefs of west Maui, Hawai'i. tilizers, and water in Minnesota. J. Environ. Qual. 26, 1212–1222. Mar. Ecol. Prog. Ser. 302, 77–91. Korth, F., Deutsch, B., Frey, C., Moros, C., Voss, M., 2014. Nitrate source identification in the Swart, P.K., Evans, S., Capo, T., Altabet, M.A., 2014. The fractionation of nitrogen and oxy- Baltic Sea using its isotopic ratios in combination with a Bayesian isotope mixing gen isotopes in macroalgae during the assimilation of nitrate. Biogeosciences 11, model. Biogeosciences 11, 4913–4924. 6147–6157. LaPointe, B.E., Barile, P.J., Littler, M.M., Littler, D.S., 2005. Macroalgal blooms on southeast Umezawa, Y., Miyajima, T., Yamamuro, M., Kayanne, H., Koike, I., 2002. Fine-scale map- Florida coral reefs II. Cross-shelf discrimination of nitrogen sources indicates wide- ping of land-derived nitrogen in coral reefs by δ15N in macroalgae. Limnol. Oceanogr. spread assimilation of sewage nitrogen. Harmful Algae 4, 1106–1122. 47, 1405–1416. Leenhouts, J.M., Bassett, R.L., Maddock III, T., 1998. Utilization of intrinsic boron isotopes USEPA, 2011. Pacific Southwest, Region 9, Underground Injection Control. http://www. as co-migrating tracers for identifying potential nitrate contamination sources. epa.gov/region9/water/groundwater/uic-hicesspools.html. Ground Water 36, 240–250. Vengosh, A., Heumann, K.G., Juraske, S., Kasher, R., 1994. Boron isotope application for Lin, H.J., Chen, Y.W., Kao, S.J., Meng, P.J., 2007. Mapping anthropogenic nitrogen through tracing sources of contamination in groundwater. Environ. Sci. Technol. 28, point sources in coral reefs using δ15N in macroalgae. Mar. Ecol. Prog. Ser. 335, 1968–1974. 95–109. Voss, M., Deutsch, B., Elmgren, R., Humborg, C., Kuuppo, P., Pastuszak, M., Rolff, C., Schulte, Lobban, C.S., Harrison, P.J., 1994. Seaweed Ecology and Physiology. Cambridge University U., 2006. Sources identification of nitrate by means of isotope tracers in the Baltic Sea Press, New York, NY, p. 384. catchments. Biogeochemistry 3, 663–676. Madden, C.J., Day, J.W., 1992. An instrument system for high-speed mapping of chloro- Wear, S.L., Vega Thurber, R., 2015. Sewage pollution: mitigation is key for coral reef stew- phyll a and physiochemical variables in surface waters. Estuaries 15, 421–427. ardship. Ann. N. Y. Acad. Sci. 1355, 15–30. http://dx.doi.org/10.1111/nyas.12785. McManus, J.W., Polsenberg, J.F., 2004. Coral–algal phase shifts on coral reefs: ecological Whittier, R.B., El-Kadi, A., 2014. Human health and environmental risk ranking of on-site and environmental aspects. Prog. Oceanogr. 60, 263–279. sewage disposal systems for the Hawaiian Islands of Kauai, Molokai, Maui, and Ha- Neumann, B., Vafeidis, A.T., Zimmermann, J., Nicholls, R.J., 2015. Future coastal population waii. Final report prepared for State of Hawai'i Department of Health. Safe Drinking growth and exposure to sea-level rise and coastal flooding — a global assessment. Water Branch. PLoS ONE 10, e0118571. Widory, D., Petelet-Giaraud, E., Négrel, P., Ladouche, B., 2005. Tracking the sources of ni- Nixon, S.W., 1995. Coastal marine eutrophication: a definition, social cause and future trate in groundwater using coupled nitrogen and boron isotopes: a synthesis. Envi- concerns. Ophelia 41, 199–219. ron. Sci. Technol. 39, 539–548. Oki, D.S., Tribble, G.W., Souza, W.R., Bolke, E.L., 1999. Groundwater resources in Kaloko- Wieser, M.E., Iyer, S.S., Krouse, H.R., Cantagallo, M.I., 2001. Variations in the boron isotope Honokōhau national historical park, island of Hawai'i, and numerical simulation of composition of Coffea arabica beans. Appl. Geochem. 16, 317–322. the effects of ground-water withdrawals. USGS Water Resources Investigative Report Xue, D., Botte, J., De Baets, B., Accoe, F., Nestler, A., Taylor, P., Van Cleemput, O., Berglund, 99-4070. M., Boeckx, P., 2009. Present limitations and future prospects of stable isotope Pardo, L.H., Kendall, C., Pett-Ridge, J., Chang, C.C.Y., 2004. Evaluating the source of methods for nitrate source identification in surface- and groundwater. Water Res. streamwater nitrate using δ15Nandδ18O in nitrate in two watersheds in New Hamp- 43, 1159–1170. shire, USA. Hydrol. Process. 18, 2699–2712. Xue, D., De Baets, B., Van Cleemput, O., Hennessy, C., Berglund, M., Boeckx, P., 2012. Use of Parnell, A., Inger, C.R., Bearhop, S., Jackson, A.L., 2010. Source partitioning using stable iso- a Bayesian isotope mixing model to estimate proportional contributions of multiple topes: coping with too much variation. PLoS ONE 5, e9672. nitrate sources in surface waters. Environ. Pollut. 161, 43–49. Peckol, P., Rivers, J.S., 1995. Physiological responses of the opportunistic macroalgae Xue, D., De Baets, B., Van Cleemput, O., Hennessy, C., Berglund, M., Boeckz, P., 2014. Clas- Cladophora vagabunda (L.) van den Hoek and Gracilaria tikvahiae (McLachlan) to en- sification of nitrate polluting activities through clustering of isotope mixing model vironmental disturbances associated with eutrophication. J. Exp. Mar. Biol. Ecol. 190, outputs. J. Environ. Qual. 42, 1486–1497. 1–16. Yang, L., Han, J., Xue, J., Zeng, L., Shi, J., Wu, L., Jiang, Y., 2013. Nitrate source apportionment Phillips, D.L., Gregg, J.W., 2003. Source partitioning using stable isotopes: coping with too in a subtropical watershed using a Bayesian model. Sci. Total Environ. 463-464, many sources. Oecologia 136, 261–269. 340–347. Marine Environmental Research 56 (2003) 225–243 www.elsevier.com/locate/marenvrev

Impacts of stormwater discharges on the nearshore benthic environment of Santa Monica Bay

Kenneth Schiff*, Steven Bay Southern California Coastal Water Research Project, 7171 Fenwick Lane, Westminster, CA 92688, USA

Abstract Although large loads of potentially toxic constituents are discharged from coastal urban watersheds, very little is known about the fates and eventual impacts of these stormwater inputs once they enter the ocean. The goal of this study was to examine the effects of storm- water discharges on the benthic marine environment of Santa Monica Bay. Sediment samples were collected across a gradient of stormwater impact following significantly sized storm events offshore Ballona Creek(a predominantely developed watershed) and Malibu Creek(a predominantely undeveloped watershed). Sediments offshore Malibu Creekhad a greater proportion of fine-grained sediments, organic carbon, and naturally occurring metals (i.e., aluminum and iron), whereas sediments offshore Ballona Creekhad higher concentrations of anthropogenic metals (i.e., lead) and organic pollutants (i.e., total DDT, total PCB, total PAH). The accumulation of anthropogenic sediment contaminants offshore Ballona Creek was evident up to 2 km downcoast and 4 km upcoast from the creek mouth and sediment concentrations covaried with distance from the discharge. Although changes in sediment tex- ture, organic content, and an increase in sediment contamination were observed, there was little or no alteration to the benthic communities offshore either Ballona or Malibu Creek. Both sites were characterized as having an abundance, species richness, biodiversity and benthic response index similar to shallow water areas distant from creekmouths throughout the Southern California Bight. There was not a preponderance of pollution tolerant, nor a lackor pollution sensitive, species offshore either creekmouth. # 2003 Elsevier Science Ltd. All rights reserved. Keywords: Stormwater; Infauna; Sediment chemistry; Santa Monica Bay

* Corresponding author. Tel.: +1-714-372-9203; fax: +1-714-894-9699. E-mail address: [email protected] (K. Schiff). URL: http://www.sccwrp.org.

0141-1136/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0141-1136(02)00332-X 226 K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243

1. Introduction

Santa Monica Bay is the tale of two distinct types of watersheds. The northern half of the Bay has subwatersheds that are predominantely open lands, consisting of mostly rural communities and amongst the largest National Forests in the Los Angeles region. The southern half of the Bay is intensely urbanized and home to metropolitan Los Angeles, the largest city on the west coast of the United Sates. Watersheds in the northern half of the Bay average 5% imperviousness while watersheds in the southern half of the Bay average 20% imperviousness (Wong, Strecker, & Stenstrom, 1997). While North Bay watersheds may be far from pristine, South Bay watersheds are built almost to saturation. Due to the intense development, nearly all the freshwater input to the Bay has been modified. This modification occurs directly, as in the case of treated municipal effluents, which comprised approximately 1.6 1012 l/year in 1997 (Raco-Rands & Steinberger, 2001). This treated effluent is piped five miles offshore and discharged in depths of 60 m. Similarly, Santa Monica Bay receives treated effluents from other direct ocean discharges including three power generating stations and an oil refinery. This modification of freshwater inputs can also occur indirectly, as in the case of stormwater runoff. Although North Bay watersheds may run dry during the arid summers, streambeds are relatively natural. In contrast, south Bay watersheds are almost entirely lined with concrete. While this infrastructure supports an impress- ively effective mechanism for flood control, it was not designed to enhance water quality; virtually all wet weather runoff that is discharged to the Bay is untreated. Moreover, rainstorms are relatively infrequent enabling a longer period for pollu- tant build-up, often times coupled with very intense rainfall thereby increasing the efficiency of pollutant transport to the ocean (Tiefenthaler, Schiff, & Leecaster, 2001). Runoff from the Santa Monica Bay watersheds during 1991 for example, which was a year of median rainfall, was 1.3 1012 L rivaling the discharge volumes of point source discharges. In the end, the loads of pollutants in stormwater runoff to Santa Monica Bay are large, exceeding all other sources except for the City of Los Angeles municipal waste treatment plant (Weisberg & Dojiri, this volume). Not only are pollutant inputs from Santa Monica Bay watersheds large, they have the potential to generate impacts in the nearshore ecosystem. For example, the contaminants in both wet and dry weather flows from selected subwatersheds have elicited toxic responses in marine organisms (Bay, Greenstein, Lau, Strenstrom, & Kelley, 1996). These organisms include local marine species such as the giant kelp (Macrocystis pyrifera), red abalone (Haliotis refuscens), and purple sea urchins (Strongylocentrotus purpuratus). Although large loads of potentially toxic constituents are discharged from Santa Monica Bay watersheds, very little is known about the fates and eventual impacts of these stormwater inputs once they enter the ocean environment. Benthic environ- ments, in particular, are at risksince they serve as an integrator of storm-discharged particulates. Accumulations of stormwater inputs may composite an entire storm, series of storms, or entire seasons. No one, though, has examined the accumulation of storm-discharged contaminants in offshore environments and the impact they K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 227 may have on benthic communities. The goal of this study was to examine the effects of stormwater discharges on the benthic marine environment of Santa Monica Bay.

2. Methods

2.1. General approach

This project was designed to measure the effects of stormwater impacts offshore two watersheds in Santa Monica Bay. The first watershed, Malibu Creek, is located in the North Bay and the second watershed, Ballona Creek, is located in the south Bay (Fig. 1). Both watersheds are similar in size and, when combined, encompass over half of the entire Santa Monica Bay drainage area (Wong et al., 1997). The Ballona Creekdrainage basin is highly urbanized; 83% of the watershed is devel- oped and comprised of predominantly residential land use. Almost the entire chan- nel is concrete-lined. Conversely, Malibu Creekis predominantly undeveloped; 88% of the watershed is open land and the channel is almost entirely earthen. These dif- ferences in watershed characteristics, in addition to localized diversity in rainfall, lead to large variations in flow and pollutant loading to the ocean, even for the same storm event (LACDPW, 2000). By comparing impacts associated with each water- shed type, we hope to distinguish between effects arising from urban and non-urban stormwater runoff.

Fig. 1. Map of the Santa Monica Bay watershed including Ballona and Malibu Creeks. Sampling sites are indicated. 228 K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243

The study design had three major elements. First, a grid-based spatial survey of sediment characteristics was used to determine the best sampling locations offshore each creekmouth. Second, sediment contamination and benthic community assem- blages were measured directly offshore each creekmouth in the area of greatest potential impact identified from the grid-based survey. Third, sediment contami- nation and benthic community assemblages along an isobath extending from the most impacted location (identified in the second design element) to an area outside of the stormwater plume effect were measured to evaluate the gradient of benthic impact.

2.2. Sampling

The benthic sampling was conducted following significantly-sized (>0.25 in) storm events. The storm-discharged particulates were given a short amount of time to settle before field deployment, usually 3–6 days, but as many as nine days were sometimes required. A total of six events were sampled offshore Ballona Creek during the 1995/1996 and 1996/1997 storm seasons; five events were sampled off- shore Malibu Creek. The five storms sampled offshore Malibu Creek were the same storm sampled offshore Ballona Creek. Sediment samples were obtained using a 0.1 m2 modified Van Veen grab. For contaminant analysis, only surficial sediments (top 2 cm) from undisturbed, repre- sentative grabs were collected. Sediments not in contact with the wall of the grab were placed in separate pre-cleaned containers for grain size, total organic carbon/ total nitrogen (TOC/TN), trace organics, and trace metals analyses. Samples were transported on ice and then frozen (<4 C) prior to trace contaminant analysis or refrigerated (4 C) prior to grain size analysis. For benthic invertebrate infaunal community analysis, entire sediment grab samples were gently washed through a 1.0 mm mesh stainless steel screen on the boat. The organisms retained on the screen were ‘‘relaxed’’ using MgSO4 (Epsom salts) in seawater. After 30 min the sample was fixed with 10% borax buffered formalin and returned to the laboratory. After 24 h, samples were rinsed with freshwater to remove formalin and preserved in 70% ethanol.

2.3. Analytical chemistry

2.3.1. Grain size analysis Grain size analysis was performed using a Horiba Model LA-900 Laser Scattering Particle Size Distribution Analyzer (Dalkey & Leecaster, 2001). Individual sediment samples were first homogenized at room temperature and a representative aliquot introduced to the instrument’s sample reservoir. The sample was then dispersed and circulated through the measuring cell and the various particle sizes were determined by detection of scattered (refracted and reflected) laser light. Data were reported as frequency (%) of particles for 74 different diameters between 0.88 and 1.024 mm. Significant interferents included scratches or bubbles in the measuring cell or parti- cles greater than 1000 mm. During this study, no samples contained fractions greater than 1000 mm so no corrective actions were necessary. K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 229

2.4. Total organic carbon and total nitrogen

Total organic carbon and total nitrogen (TOC/TN) analysis was performed using a Carlo Erba 1108 CHN Elemental Analyzer equipped with an AS/23 Autosampler. A detailed description of the method can be found in SCCWRP (1992). Frozen sediments were thawed and homogenized at room temperature, then dried at 60 C overnight. After taring, an aliquot of each sample was digested with concentrated HCl vapors to remove inorganic carbon. The acidified sample was again dried and weighed, then crimped in a tin boat. The Carlo Erba CHN Analyzer oxidizes each sample boat in a quartz combustion chamber and, using PoropakQS packedcol- umn, reaction products were separated and then quantified using a thermal con- ductivity detector. Acetanilide was used as the external standard. Acetanilide and cyclohexanone were used for QC checkstandards. The Certified Reference Material was PACS-1 (3.69% C, National Research Council).

2.4.1. Trace metal analysis Sample preparation for major and trace element analysis followed US EPA Methodology (US EPA, 1996). Approximately 2 g of oven-dried, fine-ground sedi- ment was digested using 5:2, trace metal grade nitric acid:hydrochloric acid. The acidified samples were placed in a SEMS-MS 1000 Microwave Oven Extractor. Samples were brought to a uniform volume with reagent grade water, and the solids removed by centrifugation or settling overnight. The supernatant with sample digest was transferred to a new polyethylene bottle prior to analysis. Inductively coupled plasma–mass spectroscopy (ICP–MS) was used to determine concentrations of inorganic constituents (aluminum, antimony, arsenic, beryllium, cadmium, chromium, copper, iron, lead, mercury, nickel, selenium, silver, and zinc) from sample digest solutions using a Hewlett Packard Model 4500 following US EPA Methodology (US EPA, 1991). The internal standard solution included rho- dium and thulium. Major interferents included argon, sodium, and magnesium. Instrument blanks were run to identify sample carry-over. A spiked sample of known concentration was used as the laboratory control material.

2.4.2. Pesticides (DDT) and polychlorinated biphenyls (PCB) Chlorinated pesticide and polychlorinated biphenyl analyses were conducted using procedures established by US EPA (1983, 1986). Six DDT isomers and metabolites (o,p0-DDT, p,p0-DDT, o,p0-DDE, p,p0-DDE, o,p0-DDD, p,p0-DDD) and 27 PCB Congeners (Congeners 8, 18, 28, 29, 44, 50, 52, 66, 77, 87, 101, 104, 105, 118, 126, 128, 138, 153, 154, 170, 180, 187, 188, 195, 201, 206, 209) were quantified. Specific methodological details used by the laboratory can be found in Zeng and Khan (1995). Samples for DDT and PCB analysis were first thawed and homo- genized at room temperature. 20–30 g of sample were then centrifuged to remove pore water. Sediments were dried using anhydrous sodium sulfate, and then solvent extracted in triplicate with methylene chloride utilizing a roller table that maximizes sediment–solvent contact time. Extracts were cleaned-up by removing sulfur with activated copper and passed through a 2:1, Alumina:Silica packed column. 230 K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243

Extracts were eluted with 1:6, hexane:methylene chloride and concentrated to 1 ml prior to instrumental analysis using a Hewlett Packard Model 5890 II gas chroma- tograph equipped with a 60 m 0.25 mm ID (0.25 mm film thickness) DB-5 fused silica capillary column and a 63Ni electron capture detector (GC-ECD) for quantifying analytes.

2.4.3. Polynuclear aromatic hydrocarbons Twenty-eight different polynuclear aromatic hydrocarbons (PAH) analyses were conducted using procedures established by the US EPA (1983, 1986). Specific methodological details used by the laboratory can be found in Zeng and Khan (1995). Samples for PAH analysis were first thawed and homogenized at room temperature and 20–30 gm of sample were centrifuged to remove pore water. Sedi- ments were dried using anhydrous sodium sulfate, and then solvent extracted in tri- plicate with methylene chloride utilizing a roller table that maximizes sediment– solvent contact time. Extracts were cleaned-up by removing sulfur with activated copper and passed through a 2:1, Alumina:Silica packed column. Extracts were eluted with 1:6, hexane:methylene chloride and concentrated to 1 ml prior to instrumental analysis on a Hewlett Packard Model 5890 II gas chromatograph equipped with a DB-5 column (60 m 0.25 mm ID 0.25 mm film thickness) and a Hewlett Packard Model 5870 Mass Selective Detector in electron impact ionization mode for quantifying analytes.

2.5. Biological analysis

Biological sample analysis included three major steps: (1) sorting the sample under a dissection microscope into six different taxonomic groups—annelids, molluscs, arthropods, ophiuroids, miscellaneous echinoderms, and ‘‘other phyla’’; (2) biomass measurements of each group of organisms; and (3) taxonomic identification and enumeration. A minimum of 10% of each sample was re-sorted to determine if organisms were missed in the original sort. If sorting efficiency was less than 95%, then the entire sample was resorted. After sorting, each group was weighed to the nearest 0.01 gh and reported to the nearest 0.1 gh (wet weight). Taxonomic identi- fication and enumeration is by far the most difficult of the three analytical steps. The goal was to identify each organism to species level (or lowest taxon possible). Ten percent of all samples were re-identified and enumerated by a second taxonomist for quality assurance. A specific procedure was established for discrepancies in identifi- cation (Bergen et al., 2001). A voucher collection was initiated and is maintained by the analytical laboratory.

2.6. Data analysis

Assessment of sediment contamination consisted of examining mean concentra- tions of specific pollutants offshore each creekmouth from multiple surveys. Com- parisons between watersheds were accomplished using non-parametric Mann– Whitney T-tests. Assessments of biological impairment were approached in three K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 231 distinct fashions. First, infauna data were examined for community condition such as abundance, biomass, and number of species. This was followed by community measures such as diversity, evenness and dominance. Next, the infauna results were evaluated using the Benthic Response Index (BRI), a scale of pollutant impact using community assemblage parameters (Bergen et al., 2000). Finally, a species level approach was used for examining data for ‘‘indicator’’ species. Indicator species included both pollution-associated taxa and taxa characteristic of reference com- munities (Bergen et al., 2001).

2.6.1. Benthic Response Index (BRI) The BRI quantifies the response of benthic communities to environmental dis- turbances in the Southern California Bight. The BRI is an example of direct gra- dient analysis and, for southern California, represents a pollution gradient from reference to impact at specified depth intervals derived from ordination analysis. The BRI is calculated as the abundance-weighted average of each species position on the pollution gradient. The BRI value for each sample is:

Pn pffiffiffiffiffi 3 Pi Ni ¼ i¼1 BRI Pn pffiffiffiffiffi 3 Ni i¼1 where: n=number of species in the sample Pi=gradient position for the ith species Ni=abundance count of the ith species in the sampling unit. The BRI is scaled from 0 to 100 and thresholds were developed for reference conditions (425) and four levels of community response. These thresholds included marginal deviation from reference (25 72). Threshold levels are accurate to withinthree index values.

3. Results

3.1. Spatial survey

A discernible plume footprint, sampled prior to the rainy season, was observed offshore Ballona Creekbased upon measurements of fine-grained sediments ( Fig. 2). The plume footprint at Ballona Creekdistinctly increased in percent fine-grained material directly offshore the creekmouth relative to sediments collected at similar depths either upcoast or downcoast. The footprint extended between two and four km upcoast, but less than two km downcoast. At 25 m depth, the proportion of sediment fines doubled in the heart of the footprint (40 vs. 22% fines). The footprint also reached to 40 m depth roughly five km offshore. At depths of 10 m (approx. 1 km offshore) the footprint was not evident; there was no change in grain size as one 232 K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243

Fig. 2. Gradients in sediment grain size offshore Ballona Creekprior to the wet season. Transects exten- ded 2 km downcoast to 4 km upcoast at depths of 10, 25 and 40 m. moved longshore. Presumably, this is a result of increased mixing/redistribution or differential settling of runoff particles at this depth. Although more complicated, a plume footprint was also present at Malibu Creek (data not shown). Part of this complication was that background sediments offshore Malibu Creekcontained much more silt and clay than offshore Ballona Creek. Interestingly, the heart of the footprint was offset from the Malibu Creekmouth by 2 km in the downcoast direction. However, the extent of the footprint offshore Malibu Creek remained the same as offshore Ballona Creek; at least 2 km upcoast and 4 km downcoast from the most impacted location. In addition, the strongest sediment signal was at the 25 m depth contour, similar to offshore Ballona Creek. Empirical evidence also showed that watershed inputs were present since bits of terrestrial organic debris were commonly observed in the Malibu Creeksediment samples consistent with the plume footprint. Since relatively distinct signals in the plume footprint were evident, a transect line for future benthic sampling was established along the 25 m isobath offshore each watershed.

4. Sediment quality

Sediments offshore Malibu Creekhad a greater proportion of fine-grained materials and had greater organic enrichment than sediments offshore Ballona Creek( Table 1). The sediment from directly offshore Malibu Creekhad 68% more silt+clay, 62% K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 233

Table 1 Mean concentrations (95% confidence intervals) of sediment characterisitics and constituent concen- trations directly offshore Ballona (n=6) and Malibu (n=5) Creeks following storm events between 1995 and 1997 (all samples collected at 25 m depth)

Constituent Units Ballona CreekMalibu Creek

Mean 95% CI Mean 95% CI

Fines % dry 31.6 1.3 53.1 4.6 TOC % dry 0.594 0.155 0.963 0.162 TN % dry 0.059 0.008 0.078 0.008

Aluminum mg/dry g 11,492 2496 17,280 5091 Arsenic mg/dry g 5.1 0.6 5.6 1.0 Cadmium mg/dry g 0.5 0.1 0.7 0.1 Chromium mg/dry g 40.7 2.5 52.6 9.5 Copper mg/dry g 12 1 13 1 Iron mg/dry g 14,997 1628 21,720 1866 Lead mg/dry g 26.4 1.3 10.3 0.6 Mercury mg/dry g 0.18 0.03 0.08 0.03 Nickel mg/dry g 14.3 0.8 27.8 2.3 Selenium mg/dry g 0.46 0.17 0.68 0.05 Silver mg/dry g 0.95 0.07 0.31 0.03 Zinc mg/dry g 54 2 56 4

Total DDT ng/dry g 25.6 3.7 15.5 3.1 Total PCB ng/dry g 21.5 7.3 3.0 1.4 Total PAH ng/dry g 240.6 109.2 56.2 21.9

Bold indicates significantly different concentration. more total organic carbon, and 32% more total nitrogen than sediments directly offshore Ballona Creek. Of the 12 trace metals measured in sediments offshore each creekmouth, Malibu Creekhad higher concentrations of naturally-occurring elements such as aluminum and iron (Table 1). In contrast, Ballona Creekhad higher concentrations of anthro- pogenic trace metals such as lead, silver, and mercury. Sediment concentrations of other trace metals, such as copper and zinc, were similar offshore the two watersheds. All of the organic constituents measured were more concentrated in sediments directly offshore Ballona Creekthan Malibu Creek( Table 1). The sediments directly offshore Ballona Creekhad 40% more total DDT, 86% more total PCB, and 77% more total PAH than sediments offshore Malibu Creek. In addition, sediments directly offshore Ballona Creekconsistently detected 18 of the 27 PCB congeners measured compared to only four offshore Malibu Creek. Similarly, sediments directly offshore Ballona Creekconsistently detected 10 of the 24 PAH compounds measured compared to only five directly offshore Malibu Creek. The gradient of plume influence was easily distinguished offshore Ballona Creek (Figs. 3 and 4). The sediment texture and constituent concentrations were con- sistently highest directly offshore the Ballona Creekmouth and then decreased in both the upcoast and downcoast directions. For example, fine-grained materials and 234 K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243

Fig. 3. Grain size and trace metal concentrations (mean95% confidence intervals) measured in sedi- ments across the gradient of stormwater influence offshore Ballona and Malibu Creeks. All samples were collected at 25 m depth following significantly-sized storms between 1995 and 1997.

TOC were 65 and 85% higher, respectively, directly offshore the Ballona Creek mouth compared to 4 km downcoast. Similarly, trace metals were between 36% (for zinc) and 136% (for lead) higher directly offshore the creekmouth relative to 4 km upcoast. The difference among organic constituents was even more extreme, differ- ing by a factor of two (for Total DDT) and 10 (for Total PAH). K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 235

Fig. 4. Total organic carbon and organic constituents concentrations (mean95% confidence intervals) measured in sediments across the gradient of stormwater influence offshore Ballona and Malibu Creeks. All samples were collected at 25 m depth following significantly-sized storms between 1995 and 1997.

The gradient of plume influence was less obvious for sediments offshore Malibu Creekcompared to Ballona Creek( Figs. 3 and 4). Fine-grained sediments and TOC differed by only 19 and 40%, respectively, between the most influenced site 2 km downcoast of the Malibu Creekmouth compared to the least influenced site 4 km upcoast. No substantial differences can be observed across the gradient of plume 236 K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 influence for lead, zinc, Total DDT, Total PCB, and Total PAH. In fact, con- centrations of total PCB and Total PAH were nearing the limit of detection for this study.

4.1. Infaunal communities

The benthic community directly offshore Malibu Creekcontained more organ- isms, more species, greater diversity, and greater biomass than the benthic commu- nity directly offshore Ballona Creek( Table 2). Between 1995 and 1997, the average abundance directly offshore Malibu Creekwas 33% greater than directly offshore Ballona Creek. Both creeks shared a large population density; wet season abun- dances averaged 316 and 238 individuals per 0.1 m2, respectively. Species richness was significantly greater offshore Malibu Creekcompared to offshore Ballona Creek. Similarly, Malibu Creek exhibited higher diversity than Ballona Creek. Community assemblages were not dominated by a small proportion of the large number of species found offshore each watershed, however, as indicated by the relatively high evenness values at both sites.

Table 2 Mean (95% confidence intervals) of biological community parameters directly offshore Ballona (n=6) and Malibu (n=5) Creeks following storm events between 1995 and 1997 (all samples collected at 25 m depth)

Biological Units Ballona CreekMalibu Creek parameter Mean 95% CI Mean 95% CI

Total abundance (No./0.1 m2 ) 237.5 50.8 316.0 55.4 Species richness (No./0.1 m2 ) 74.8 5.8 91.2 7.8 Diversity (Shannon–Wiener H0) 1.65 0.02 1.73 0.04 Evenness (Pielou’s J) 0.88 0.02 0.88 0.02

Benthic Response Index (BRI) 24.0 1.7 30.5 0.7

Total biomassa (gm/0.1 m2 ) 2.1 0.4 6.1 2.2 Annelida 1.1 0.3 2.6 0.7 Arthropoda 0.3 0.1 0.3 0.1 Ophiuroida 0.3 0.3 2.0 1.3 Echinodermata 0.0 – 0.0 – Mollusca 0.2 0.1 0.6 0.8 Other taxa 0.1 0.1 0.5 0.2

Abundance/biomass (No. individuals/g) 120.7 33.8 60.0 23.9 Ratio

a Wet weight with outlier organisms removed (i.e. large molluscs with shells, large epifaunal echino- derms, etc.). K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 237

The mean wet season BRI value directly offshore Malibu Creekwas significantly higher than offshore Ballona Creek( Table 2). The mean BRI offshore Ballona Creek (24.0) was less than 25 indicating that this site was similar to reference conditions and within the 90th percentile of other reference sites in the SCB with respect to community assemblage and relative abundance. The BRI value offshore Malibu Creek(30.5) was greater than 25 indicating that this site marginally deviated from reference condition. It was therefore outside the 90th percentile of reference sites from the SCB, but still maintained a community that sustained at least 75% of the expected species. This indicated that there was little, or no, loss of biodiversity in communities offshore Malibu Creek. The community assemblages measured directly offshore Ballona and Malibu Creeks at 25 m contained taxa characteristic of shallow (10–30 m) reference communities measured throughout the SCB (Table 3). All 19 of the characteristic reference assemblage taxa that are consistently found (> 60% occurrence) and in notable abundance (average > 2 per 0.1 m2) were also found directly offshore Ballona Creek; 17 of 19 were found directly offshore Malibu Creek. The relative abundance of these reference community taxa comprised between 30 and 43% of the total abundance offshore Malibu and Ballona Creeks, which was similar to the mean relative abundance throughout the SCB (37% of total abundance). Except for the BRI, there was no trend in community assemblage parameters along the 25 m monitoring transect that was spatially correlated with the influence of Ballona or Malibu Creekdischarges ( Fig. 5). Total abundance, species richness, and diversity directly offshore either creekmouth was not significantly lower (or higher) than sites upcoast or downcoast from the discharge. The BRI did show a spatial pattern that could be spatially correlated with Ballona and Malibu Creek discharges. The highest BRI values from Ballona Creekwere measured directly off- shore the mouth, then decreased upcoast and downcoast from the discharge. The highest BRI values offshore Malibu Creekwere 2 kmdowncoast from the creek mouth. Offshore each of the respective watersheds, the lowest BRI values were at the sites furthest from the creekdischarges, located 4 kmupcoast. Over-interpretation is cautioned, however, since there were no statistically significantly differences among sites within watersheds. Furthermore, all sites offshore Ballona Creekwere below the reference threshold. Pollution-associated indicator taxa were relatively low in abundance along the 25 m monitoring transect offshore Ballona and Malibu Creeks (Fig. 6). Capitella capi- tata, an opportunistic species indicative of degraded environs, rarely occurred off- shore either creekmouth. Euphilomedes carcharodonta, is another species often times associated with moderately impacted locations. This species abundance was also low, but did show some spatial pattern relative to discharges offshore both Malibu and Ballona Creeks. Although the greatest E. carcharodonta densities were observed where the highest BRI measurements were found, none of the average wet season abundance measurements were statistically significantly different among Ballona Creekstations or among Malibu Creekstations. Taxa characteristic of shallow reference communities were in greater abundance rela- tive to pollution-associated indicator species along the 25 m monitoring transect offshore Ballona and Malibu Creeks (Fig. 6). Spiophanes missionensis and Amphideutopus oculatus 238 K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243

Table 3 Mean abundance (No./0.1 m2+95% confidence interval) of taxa found directly offshore Ballona Creek (n=6) and Malibu Creek( n=5) which were commonly found in reference locations of the Southern California Bight (SCB)a

Taxon Abundance (No./0.1 m2)

Ballona Malibu SCB

1995/1997 95% 1995/1997 95% 1994 ave. CI ave. CI ave.

Amphideutopus oculatus 10.2 5.8 10.4 9.8 11.6 Paraprionospio pinnata 5.2 3.1 7.0 5.0 10.7 Spiophanes missionensis 6.5 4.1 0.0 – 10.7 Maldanidae 11.2 2.8 13.2 1.9 9.9 Spiophanes bombyx 0.2 0.3 0.0 – 9.5 Glottida albida 0.2 0.3 0.8 0.7 7.7 Mediomastus sp 12.8 7.4 7.2 3.8 7.4 Tellina modesta 7.7 3.1 0.6 0.8 5.5 Macoma yoldiformis 1.3 1.1 2.2 1.0 5.3 Owenia collaris 3.5 3.8 0.2 0.4 5.3 Ampelisca cristata 9.0 3.9 2.4 0.5 5.2 Lumbrineris sp 23.7 8.9 2.6 1.7 4.9 Apoprionospio pygmaea 1.8 0.5 0.6 0.8 4.8 Carinoma mutabilis 0.7 0.8 1.8 1.8 3.4 Phoronida 1.5 1.1 23.4 18.5 3.4 Ampelisca brevisimulata 6.3 2.6 7.0 5.2 3.2 Ampharete labrops 1.0 0.9 0.4 0.5 2.7 Amphicteis scaphobranchiata 0.3 0.4 0.2 0.4 2.4 Rhepoxynius menziesi 0.8 1.1 14.8 3.4 2.1

Percent of total station abundance 42.6 4.5 29.8 3.6 37.3

a From Bergen et al. (2001). are all species associated with reference communities of the SCB (Table 3). In general, stations offshore Malibu Creekhad greater densities of S. missionensis and A. oculatus. Some spatial patterns across the gradient of stormwater discharge could be observed in the relative abundance of these species, but not in the negative direction. The lowest densities of all three reference community indicator species were not observed directly offshore the discharge, nor at the site with the highest BRI values. Mid-depth (30–115 m) reference indicator taxa were also present in significant proportions offshore Malibu, but not Ballona Creek( Fig. 7). Monticellina dorso- branchialis, Cossura sp., and Amphiodia sp. are all taxa typical of mid-depth regions that normally have greater TOC and fine-grained sediments than shallow habitats. The presence of these species in depths less than 30 m is rare in the SCB. Densities of these taxa were much greater at Malibu Creekcompared to Ballona Creekand also appeared to mimic the spatial pattern indicative of the stormwater gradient offshore Malibu Creek. Amphiodia, in particular, is a pollution sensitive species and is rarely found in locations of moderate anthropogenic impact in mid-depth regions. K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 239

Fig. 5. Infaunal biological commiunity parameters (mean95% confidence intervals) measured across the gradient of stormwater influence offshore Ballona and Malibu Creeks. All samples were collected at 25 m depth following significantly-sized storms between 1995 and 1997.

5. Discussion

Stormwater discharges appeared to alter the benthic habitat in Santa Monica Bay. Changes in sediment texture, organic content, and contaminant composition off- shore Ballona Creek extended 3 km offshore and up to 4 km alongshore were measured during this study. These alterations have been observed offshore urban 240 K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243

Fig. 6. Abundance (mean95% confidence intervals) of pollution tolerant taxa (Capitella, Euphilomedes) and reference community taxa (Spiophanes, Amphideutopus) measured across the gradient of stormwater influence offshore Ballona and Malibu Creeks. All samples were collected at 25 m depth following significantly-sized storms between 1995 and 1997. watersheds throughout the southern California Bight (Schiff, 2000). For example, concentrations of trace metals were enriched nearly twice as frequently near the 12 largest river and creekmouths that drain to the Bight compared to shallow water areas distant from creekmouths. Not only were sediment texture and organic content altered offshore Ballona Creek, but changes in grain size and TOC K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 241

Fig. 7. Abundance (mean95% confidence intervals) of pollution sensitive taxa indicative of deeper depths (>30 m) measured across the gradient of stormwater influence offshore Ballona and Malibu Creeks. All samples were collected at 25 m depth following significantly-sized storms between 1995 and 1997. were also observed offshore the rural watershed of Malibu Creek. It appears that naturally derived terrestrial material also contributes organic particles to offshore areas. Although changes in benthic habitat were observed, stormwater discharges did not appear to be degrading the resident benthic community. Relatively healthy 242 K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 communities were measured offshore both creeks and they were dominated by characteristic reference site taxa. Any negative impacts that could be observed were very subtle, such as the mild response of Tellina modesta, a pollution sensitive mollusk. This study represents one of the first applications of the BRI to stormwater impacted areas. While stormwater areas were a location that had the potential to degrade benthic communities in the development of the BRI (Bergen et al., 2000), this study directly addressed this question. Interestingly, BRI values were generally low and the rural watershed, Malibu Creek, actually had a higher BRI than the urban watershed. However, it was not the presence of shallow water pollution tol- erant species that altered the BRI offshore Malibu Creek. Rather, it was the pres- ence of mid-depth pollution sensitive species (e.g. Amphiodia sp.) in shallow water samples that altered the BRI. Species such as these are likely recruited to the fine- grained organic-rich habitats typically found in deeper waters. It is the presence of these mid-depth species that drove the BRI higher. As the landscape in our water- sheds changes, perhaps it is the lackof fine-grained organic-rich materials that is altering the offshore benthic communities. One possibility why no large deleterious impact to benthic communities was observed is because the sampling sites were not in the right locations. The footprint survey conducted during the dry season, was designed to help address this question. However, it was limited in both space and time. For example, benthic habitat quality and benthic community structure have been examined within the mouth of Ballona Creek, an area inshore of our sampling grid and one that becomes almost entirely freshwater during large storm events (Soule, Oguri, & Jones, 1993). The sediments con- tained relatively high contaminant concentrations and, episodically, found nematodes in great abundance. A second location of potential impact is in deeper waters, beyond the offshore extent of our sampling grid. In these locations, storm discharged particulates may commingle with particles from other sources, such as municipal wastewater efflu- ents. Regardless, the fate and transport of runoff derived materials is not well under- stood in the marine environment and continues to be an area of research need.

Acknowledgements

The authors wish to acknowledge the assistance of the R/V Seaworld and R/V Seawatch for sampling assistance. Also, several individuals participated in the col- lection and preparation of samples including Dario Diehl, David Tsukada, and Liesl Tiefenthaler. Staff chemistry personnel helped to analyze samples including Charlie Yu, Kim Tran, and Eddie Zeng. Sample grain size analysis was conducted by the City of San Diego. Trace metal analysis was conducted by CRG Marine Labora- tories, Inc. Benthic laboratory analysis was conducted by MEC Analytical Systems, Inc. A special note of gratitude is given to Mary Bergen and Richard Gersberg. Portions of this study were funded by the Los Angeles County Department of Public Works and conducted in collaboration with the Natural Resources Defense Council and University of Southern California Seagrant. K. Schiff, S. Bay / Marine Environmental Research 56 (2003) 225–243 243

References

Bay, S., Greenstein, D., Lau, S., Stenstrom, M., & Kelley, C. (1996). Toxicity of dry weather flow from the Santa Monica Bay watershed. Southern California Academy of Sciences, 95, 33–45. Bergen, M., Cadien, D., Dalkey, A., Montagne, D. E., Smith, R. W., Stull, J. K., Velarde, R. G., & Weisberg, S. B. (2000). Assessment of Benthic infaunal condition on the mainland shelf of Southern California. Environmental Monitoring and Assessment, 64, 421–434. Bergen, M., Weisberg, S., Smith, R., Cadien, D., Dalkey, A., Montagne, D., Stull, J., Velarde, R., & Ranasinghe, J. A. (2001). Relationship between depth, sediment, latitude, and the structure of benthic infaunal assemblages on the mainland shelf of southern California. Marine Biology, 138, 637–647. LACDPW. (2000). Los Angeles County 1994–2000 integrated receiving waters report. Alhambra, CA: Los Angeles County Department of Public Works. Dalkey, A., & Leecaster, M. (2001). Comparison of sediment grain size analysis among two methods and three instruments using environmental samples. In S. Weisberg, & D. Elmore (Eds.), Southern Cali- fornia coastal water research project, annual report 1999–2000 (pp. 289–295). Westminster, CA: South- ern California Coastal Water Research Project. Raco-Rands, V. n., & Steinberger, A. (2001). Characteristics of effluents from large municipal wastewater treatment facilities in 1997. In S. Weisberg, & D. Elmore (Eds.), Southern California coastal water research project, annual report 1999–2000 (pp. 28–44). Westminster, CA: Southern California Coastal Water Research Project. SCCWRP (1992). Total organic carbon and total nitrogen in marine sediments, sediment trap particles, municipal effluents, and surface runoff. In: Southern California coastal water research project 1990–91 and 1991–92 annual report (pp. 50–60). Westminster, CA: Southern California Coastal Water Research Project. Schiff, K. (2000). Sediment chemistry on the mainland shelf of the Southern California Bight. Marine Pollution Bulletin, 40, 267–276. Soule, D., Oguri, M., & Jones, B. 1993. The marine environment of Marina Del Rey, marine studies of San Pedro Bay, California—part 21. Report to the Department of Beaches and Harbors, County of Los Angeles from Harbors Environmental Project, University of Southern California, Los Angeles, CA 90089-0371. Tiefenthaler, L., Schiff, K., & Leecaster, M. (2001). Temporal variability patterns of stormwater con- centrations from an urban watershed. In S. Weisberg, & D. Elmore (Eds.), Southern California coastal water research project, annual report 1999–2000 (pp. 52–62). Westminster, CA: Southern California Coastal Water Research Project. US EPA. (1983). Analysis of water and wastes. Cincinnati, OH: US Environmental Protection Agency, Environmental Monitoring and Systems Laboratory [EPA-600/4-79-020]. US EPA. (1986). Test methods for evaluating solid waste, physical/chemical methods (3rd ed.). Office of Solid Waste and Emergency Response, Washington, DC: United States Environmental Protection Agency [SW-846]. US EPA. (1991). Methods for the determination of metals in environmental samples. Washington, DC: Office of Research and development [EPA/600/4-91/010]. US EPA. (1996). Addendum to SW-846: Test Methods for evaluating solid waste, physical/chemical methods (3rd ed.). Office of Solid Waste and Emergency Response, Washington, DC: United States Environ- mental Protection Agency. Wong, K. M., Strecker, E. W., & Stenstrom, M. K. (1997). GIS to estimate storm-water pollutant mass loadings. Journal of Environmental Engineering, 123, 737–745. Zeng, E., & Khan, A. (1995). Extraction of municipal wastewaters effluent using 90-mm C-18 bonded disks. Journal of Microcolumn Separations, 7(5), 529–539.