Eutrophication and macroalgal blooms in temperate and tropical coastal waters: nutrient enrichment experiments with Ulva spp.

Receiving coastal waters and estuaries are among the most nutrient-enriched environments on earth, and one of the symptoms of the resulting eutrophication is the proliferation of opportunistic, fast-growing marine seaweeds. Here, we used a widespread macroalga often involved in blooms, Ulva spp., to investigate how supply of nitrogen (N) and phosphorus (P), the two main potential growth-limiting nutrients, influence macroalgal growth in temperate and tropical coastal waters ranging from low- to high-nutrient supplies. We carried out N and P enrichment field experiments on Ulva spp. in seven coastal systems, with one of these systems represented by three different subestuaries, for a total of nine sites. We showed that rate of growth of Ulva spp. was directly correlated to annual dissolved inorganic nitrogen (DIN) concentrations, where growth increased with increasing DIN concentration. Internal N pools of macroalgal fronds were also linked to increased DIN supply, and algal growth rates were tightly coupled to these internal N pools. The increases in DIN appeared to be related to greater inputs of wastewater to these coastal waters as indicated by high δ15N signatures of the algae as DIN increased. N and P enrichment experiments showed that rate of macroalgal growth was controlled by supply of DIN where ambient DIN concentrations were low, and by P where DIN concentrations were higher, regardless of latitude or geographic setting. These results suggest that understanding the basis for macroalgal blooms, and management of these harmful phenomena, will require information as to nutrient sources, and actions to reduce supply of N and P in coastal waters concerned.


Introduction
Nutrient inputs to coastal waters have increased in coastal environments worldwide as a direct consequence of the growing human population and increased settlement and use of coastal areas (Nixon et al., 1986;. These changes in nutrient availability lead to increased eutrophication, a growing threat facing coastal ecosystems (National Research Council (NRC), 2000;Bricker et al., 2008). One common symptom of eutrophication is profuse blooms of marine seaweeds, or macroalgae (Lavery et al., 1991;Sfriso et al., 1992;Valiela et al., 1997;Morand & Merceron, 2005;Fox et al., 2008;Fig. 1a), a feature that has received wide press and public notice (e.g. New York Times, July 1, 2008; Naples Daily News, March 15, 2006;Boston Globe, September 27, 2001), and is widespread along the coasts of the world (Table 1; Raffaelli et al., 1998;Morand & Merceron, 2005).
It has been argued that N supply is a main control on macroalgal growth in temperate coastal areas (Nixon & Pilson, 1983;Oviatt et al., 1995;Howarth et al., 2000). In tropical latitudes carbonate sediments derived from coral reefs may sequester phosphate (PO 4 3À ) and may lead to P limitation of macroalgal growth (Lapointe et al., 1992;McGlathery et al., 1994), but other studies show exceptions to this general pattern (Larned, 1998;Fong et al., 2001;Elser et al., 2007), and much research is still needed to better understand the processes driving coastal eutrophication and strategies for water quality management (NRC, 2000;Smith & Schindler, 2009).
External N supply interacts with internal N pools in macroalgae (Fujita, 1985;Bjornsater & Wheeler, 1990;Fong et al., 2003;Teichberg et al., 2008) to create different growth responses (Fujita, 1985;Pedersen & Borum, 1996;Teichberg et al., 2008). The sources of internal N pools have been assessed with stable isotopic methods  (Hayden et al., 2003). (McClelland & Valiela, 1998;Aguiar et al., 2003;Savage & Elmgren, 2004;Cole et al., 2005;Teichberg et al., 2007Teichberg et al., , 2008. Some macroalgae reflect N isotopic signatures of their source with little fractionation, making them potential indicators of anthropogenic nutrient inputs (Savage & Elmgren, 2004;Deutsch & Voss, 2006;Thornber et al., 2008). No comprehensive study of algal blooms and algal physiological responses has been done on a global scale across latitudes and oligotrophic to eutrophic conditions to understand nutrient limitation and the potential for macroalgal blooms worldwide. This is particularly important at this time as nutrient additions to estuaries and coasts increase, and eutrophication becomes one of the greatest threats to our coasts and estuaries (NRC, 2000;Howarth, 2008). Macroalgal blooms involve relatively few taxa (Valiela et al., 1997;Morand & Merceron, 2005) that are widely distributed throughout the coasts of the world. In particular, species of Ulva can be found in many coastal waters (Table 1), thus providing a useful biological model to make geographical comparisons as to the nutrient controls on growth of bloomforming macroalgae.
In this study we examined the relationship of macroalgal growth, internal nutrient pools, and natural stable isotopes of Ulva spp. to different ambient nutrient supplies and tested whether additional N or P above ambient nutrient regimes increased macroalgal growth, increasing likelihood of blooms. We measured growth responses and nutrient content of fronds of local Ulva spp. in each of seven coastal systems, including three subestuaries within one estuarine system, for a total of nine sites where ambient nutrient supplies differed. From these experiments, we compared the relative growth responses across latitudes under oligotrophic to eutrophic conditions and under additional N and P enrichment.

Study sites
We chose coastal systems ( Fig. 1) in which the level of ambient nutrient concentrations and algal productivity span the range of levels found in oligotrophic to eutrophic coastal waters. In North America, our work was done in Waquoit Bay, MA, USA (4115 0 N and 7015 0 W), an estuarine complex with a wet, temperate climate, in which we chose three subestuaries, Sage Lot Pond, Quashnet River, and Childs River, that receive different land-derived nitrogen loads, creating widely differing nutrient concentrations and occurrences of macroalgal blooms (Valiela et al., 1997;Fox et al., 2008). In Europe, we worked in Mondego Estuary, Portugal (4011 0 N and 815 0 W), and Palude della Rosa, Venice Lagoon, Italy (4515 0 N and 1214 0 E), eutrophic sites that are also in wet, temperate climates with high macroalgal biomasses (Flindt et al., 1997;Martins et al., 2001). Venice lagoon is historically one of worst-case scenarios against which to compare all other sites (Sfriso et al., 1992;Flindt et al., 1997;Sfriso & Marcomini, 1997). To represent dry, tropical climates, we worked in Jobos Bay, Puerto Rico (1719 0 N and 6612 0 W), a site with oligotrophic waters and low macroalgal biomass (Bowen & Valiela, 2008) and Urias estuary, Mexico (2311 0 N and 10612 0 W), a site with significant nitrogen enrichment and high macroalgal biomass (Ochoa-Izaguirre et al., 2002). Flamengo Sound, Brazil (2313 0 S and 4511 0 W) represents a site with low nutrient loads and a diverse macroalgal community (Corbisier et al., 2006) in a wet, tropical climate, while San Antonio Bay, Argentina (4017 0 S and 6419 0 W) is an eutrophic site (P. Martinetto et al., unpublished results) in an arid, temperate climate.

Experimental layout
To test the impact of ambient nutrient supply on growth of Ulva spp., we carried out in situ cage experiments in the nine sites. We used Ulva lactuca Linnaeus in our experiments in Waquoit Bay, Jobos Bay, Urias Estuary, San Antonio Bay, and Mondego Estuary; Ulva fasciata Delile in Flamengo Sound; and Ulva laetevirens J. E. Areschoug in Venice Lagoon. All macroalgal fronds were collected from the water near the sediment surface, as either floating or attached forms. Incubations were carried out once during the peak of the growing season in each site, except for in Waquoit Bay subestuaries where incubations were carried out several times throughout the growing season. Ulva fronds were incubated in experimental cages in waters at each of the nine sites. The cages were 15 cm Â 20 cm Â 20 cm and constructed of four acrylic plastic sides to allow for light penetration and two mesh sides with 1 mm mesh openings to allow for horizontal water flow through the units, but also to restrict access to cages by grazers. The four acrylic sides of the cages consisted of two sides of 20cm Â 20 cm and a top and bottom of 20 cm Â 15 cm. The two mesh sides were 20 cm Â 15 cm. The cages were individually attached to a line 3 m apart. The line was anchored approximately 0.3 m from the sediment surface with concrete blocks. Additionally, the cages each had a buoy attached by a rope to allow the cage to float and rotate freely with the water flow. This design allowed for each cage to be suspended within the macroalgal canopy at 0.1-0.3 m from the sediment surface.
To discern the effect of additional N or P enrichment in coastal waters, we measured Ulva spp. growth rates after 10-13 days in situ in control cages with no nutrients added and in cages where we experimentally added nitrate (NO 3 À ) or PO 4 3À to the water within the cages.
Experimental cages were randomly distributed under the three nutrient treatments, unenriched controls, added NO 3 À , and added PO 4 3À , with n 5 4 per treatment.
To deliver N and P to macroalgal fronds in cages, NO 3 À or PO 4 3À additions were prepared by dissolving KNO 3 or KH 2 PO 4 in a 3% agar solution, making a 2 M KNO 3 or 1 M KH 2 PO 4 solution, and by setting the agar in a perforated PVC pipe at the center of the unit to allow for slow release over the course of the 10-13-day experimental run. In situ nutrient delivery methods were previously tested for effectiveness (Teichberg, 2007;Teichberg et al., 2008). We used NO 3 À for our N supply in our enrichments because NO 3 À was by far the greatest contributor to total dissolved inorganic nitrogen (DIN) in most of our sites. Water depth ranged between 0.5 and 9 m among sites, with the highest tidal ranges in Mondego Estuary (up to 3.5 m) and San Antonio Bay (up to 9 m). Cages were submerged at all times in all of the sites, except for Mondego Estuary, where cages were out of the water for short periods during low tide. To measure maximum growth response of Ulva spp., we carried out experiments during peak growing season in each site.

Analysis of water nutrient concentrations
To measure ambient and experimental nutrient concentrations to which Ulva spp. fronds were exposed during the incubation period, ambient water and water within the experimental cages was sampled at the beginning, middle, and end of the experiment using a 60 mL syringe and sipping water from a thin tube inserted permanently in the center of each cage without disturbing the cages. Water was filtered through a GF/F glass fiber filter into 60 mL plastic sample bottles at the field site and then frozen.

Response of macroalgae
To assess the growth response of macroalgae to ambient nutrient supply and experimental nutrient enrichment, initial and final wet weights of Ulva spp. fronds were measured before and following incubation by blotting algae dry with paper towel. Growth rates were calculated as % growth day À1 using a linear growth calculation to compare the relative % change in initial and final biomass within each cage over the course of the incubation period. To assess relationships between Ulva spp. growth and annual nutrient concentrations in the water among the sites, we used regression analysis. Data that were not normally distributed were log-transformed before regression analysis.
To test whether there were significant growth responses to N or P enrichment within each site, we used a one-way-ANOVA, and post hoc Tukey test. To directly compare macroalgal growth responses to the enriched nutrients among the sites, we needed to standardize the growth rates based on the relative response to the enrichment from the controls within a site. We did this by calculating a relative response to NO 3 À and PO 4 3À by subtracting the mean growth rate of the control treatments from the mean growth rates of the nutrient enrichments within each site, and propagating the error. Then, to assess the relationships between a relative response to NO 3 À and PO 4 3À and ambient nutrient concentrations among the sites we used regression analysis. Data that were not normally distributed were logtransformed before analysis.
To determine internal nutrient pools and N isotopic signatures of fronds, samples of Ulva spp. were collected from each site, cleaned, dried at 60 1C, and ground to a fine powder with mortar and pestle. Dried macroalgal tissue was analyzed for total N content, and N stable isotopes at the University of California, Davis, Stable Isotope Facility. Regression analysis was used to assess relationships between Ulva spp. N content and N isotopes, and water nutrient concentrations among the sites. Data that were not normally distributed were logtransformed before regression analysis.

Analysis of nutrient concentrations
The concentrations of DIN and PO 4 3À differed substantially among the study sites (Fig. 2). In some cases ambient concentrations measured in this study during the incubation periods differed from the annual nutrient concentrations reported (supporting information, Table SA1). This variation in short-term concentrations from the annual means may be influenced by seasonal differences in nutrient concentrations. The incubations were carried out during the peak of the growing season, and therefore, the high biomass of macroalgae in some of the sites may have rapidly depleted the nutrients from the water column.
The experimental nutrient enrichments of NO 3 À or PO 4 3À , both successfully increased nutrient concentrations in the cages above ambient nutrient concentrations during the incubations (Fig. 2), as shown by the points above the 1 : 1 line. Variation in the degree of nutrient enrichment from each site depended on the range in ambient nutrient concentrations as well as the hydrodynamics of each system. For example, the sites with faster flow rates, larger tidal exchange, or wave action (Flamengo Sound, Mondego Estuary, and San Antonio Bay), and sites with higher ambient N concentrations (Mondego Estuary, Venice Lagoon, and San Antonio Bay) showed less of an increase in nutrient concentrations relative to the ambient in spite of the experimental enrichment. To better display the increases with enrichment, we calculated the difference between mean concentrations of NO 3 À and PO 4 3À in the cages that were enriched and corresponding control cages (Fig. 2, insets). In all cases, the data show that treated cages maintained higher concentrations of the enriched nutrient. Concentrations of NO 3 À , NH 4 1 , and PO 4 3À inside control cages were similar to the ambient water concentrations (Teichberg, 2007).

Response of macroalgae to ambient conditions
The growth rates of Ulva spp. that were simply incubated with no experimental nutrient addition differed considerably among the sites ( Table 2). Rates of growth of fronds of Ulva spp. were significantly higher in coastal waters with larger mean annual concentrations of DIN (Fig. 3, top) and PO 4 3À (Fig. 3, bottom). The logarithmic relationship of growth to DIN concentrations was significant (Fig. 3, top), while that of PO 4 3À was driven largely by the high concentration in one site, San Antonio Bay (Fig. 3, bottom). There were no obvious patterns in growth responses in correlation with latitude, with the exception that the highest growth rates were all from temperate systems (Table 2, Fig. 3).
In the Waquoit Bay subestuaries, where we measured ambient growth responses in different times of the year, we found a seasonal growth response of U. lactuca (Table 2). Growth was higher in summer than fall in all subestuaries, with July being the peak time of growth of U. lactuca ( Table 2).
The internal N content in fronds differed among the study sites and increased logarithmically as annual DIN concentrations increased (Fig. 4, top). Macroalgal fronds from all sites were above the minimum nutrient requirements (lower dotted line), and macroalgae from more than half of the sites were higher than the N requirements reported for maximum growth rates of Ulva spp. (upper dotted line, Fig. 4, top). The growth response of Ulva spp. significantly increased with increasing % tissue N (Fig. 4, bottom), with the exception of fronds from Mondego Estuary, supporting the notion that growth is closely linked to both external nutrient supply and internal nutrient pools.
The d 15 N of the macroalgae from the study sites differed, ranging from 4 to 16% and increased logarithmically with increasing annual DIN concentration (Fig.  4, middle). Macroalgae from sites with oligotrophic waters (Sage Lot Pond, Flemengo Sound, and Jobos Bay) had substantially lower d 15 N values than those from sites with eutrophic waters (Fig. 4, middle).

Relative response of macroalgae to nutrient enrichment
There was a marked difference in relative response of Ulva spp. to nutrient enrichment in the different sites ( was restricted by supply of PO 4 3À (Fig. 5). The other sites showed less conspicuous trends. The relative response to NO 3 À and PO 4 3À , however, were not correlated to latitude (F 5 0.01, P 5 0.93 and F 5 1.1, P 5 0.32, respectively; Fig. 6). The magnitude of the relative response to either nutrient, however, was generally greater in higher latitudes (Fig. 6).
To ascertain whether the differences in responses to the nutrient enrichments were related to differences in ambient nutrient regimes in the different coastal waters, we plotted the growth response under enrichment vs. mean ambient nutrient concentrations in each site during the time of the experiments (Fig. 7). Relative response of fronds enriched with NO 3 À decreased as ambient NO 3 À increased (Fig. 7, top), demonstrating that fronds responded to NO 3 À addition when NO 3 À was in limited supply, and did not respond to NO 3 À addition in environments where NO 3 À was plentiful. We note that NO 3 À was such a dominant contributor to the DIN that we could find a similar relationship of growth of Ulva spp. to DIN (Fig. 7, top, inset). There was a significant increase in the relative growth of algal fronds to PO 4 3À in relation to the ratio of DIN to DIP (Fig. 7, bottom). This relationship was largely a function of NO 3 À supply, because growth of macroalgal fronds was not significantly related to PO 4 3À concentration (data not shown). In Venice Lagoon, San Antonio Bay, and Mondego Estuary, where ambient DIN concentrations were highest, Ulva spp. increased growth in response to PO 4 3À addition.

Response of macroalgae to ambient conditions
This paper takes advantage of the presence of a model alga across many shores of the world, as well as a species that can be used in experimental manipulations, to provide evidence of the ecological response to a broad range of increasingly eutrophic conditions across the world. The growth response of Ulva spp. was correlated to long-term ambient supply of nutrients in our study sites, suggesting that increased nutrient availability stimulates macroalgal growth. Ulva spp. in the control treatments were particularly driven by increases in annual DIN concentrations, but also responded to increased P supply. DIN rather than P, however, was in far greater supply in our study sites, as is the case in many coastal waters worldwide (Howarth, 2008). Our data suggest that increases in DIN concentrations lead to increased growth and abundance of macroalgae that could in turn lead to more frequent occurrences of macroalgal blooms.
The variability in growth rates in Waquoit Bay estuaries demonstrate that growth may differ throughout the growing season within a site, due to variability in For all treatments, n 5 4 in each site, with the exception of Urias and Mondego estuaries where n 5 3. Significant differences within a site across nutrient treatments are indicated by F and P values from the results of one-way ANOVA. ambient nutrient supply, temperature, and light intensity during the year. The variation in seasonal growth responses of U. lactuca in the subestuaries of Waquoit Bay is supported by studies on macroalgae in temperate sites (Peckol et al., 1994;Pedersen & Borum, 1996;Balducci et al., 2001). Thus, seasonality of growth responses must play a role in the occurrence of macroalgal blooms. Seasonality in responses, however, may be less marked in macroalgae from tropical sites, and blooms may be triggered by different factors, such as changes in freshwater inputs in wet season relative to dry season or cooler temperatures (Ochoa-Izaguirre et al., 2002;Piñ on-Gimate et al., 2008), where shifts in algal assemblages may occur. The internal N pools of macroalgal fronds were also linked to external DIN supply and were in the range of percent tissue N found in other studies of Ulva spp. (Fujita, 1985;Wheeler & Bjornsater, 1992;Pedersen & Borum, 1996;Lourenco et al., 2005). All values of percent tissue N were above the critical N requirements for minimum growth, and fronds from some sites exceeded the requirements for maximum growth (Pedersen & Borum, 1996). Thus, percent tissue N values of Ulva  spp. above 2.2% in this study likely represent excess N storage, which can reach up to 5.8% in U. lactuca (Pedersen & Borum, 1996). Judging from the size of internal N pools, growth of macroalgae in these waters may not be N-limited, or may be able to sustain maximum growth during times of lower nutrient availability. The ambient growth rates of Ulva spp. increased with internal N content, suggesting that both external and internal tissue N are strongly linked to growth and may be used as indicators of macroalgal growth response. Macroalgal growth rates in Mondego Estuary were an exception to this pattern, suggesting that growth may be limited by something other than nutrient supply. A recent study by Martins et al. (2007) showed through a spatially dynamic model that the productivity of green macroalgae varied considerably in Mondego Estuary, due to the hydrodynamics of the system and differences in temperature and light conditions at different depths of intertidal and subtidal areas. Therefore, it is important to consider that some systems may respond differently under varying environmental conditions. Despite these exceptions, it seems that growth of Ulva spp., and other fast growing algae, can be predicted by external and internal N history. The range in d 15 N of Ulva spp. from this study is similar to that of macroalgae from other studies exposed to heavy DIN sources (Table 3). These studies suggest that d 15 N of macroalgae are heavier where DIN concentrations are higher. The heavy d 15 N signatures of the fronds in our study sites were linked to high DIN concentrations, suggesting that the increased N loads   were associated with wastewater rather than other N sources (McClelland & Valiela, 1998;Cole et al., 2005), particularly so in Urias Estuary and San Antonio Bay. If fertilizer or atmospheric deposition was dominant in the sites where we did our study, d 15 N of the fronds would have been near 0% (McClelland & Valiela, 1998). In many other case studies, the heaviest d 15 N signatures were found in Ulva spp. compared with other macroalgal taxa (Table 3). Thus, Ulva spp. are a sensitive sentinel for detecting wastewater inputs to coastal waters, and might help predict the potential for future macroalgal blooms, because concentrations of wastewater-derived NO 3 À in the water column determined Ulva spp. growth responses in our study sites. For the coastal waters we studied, therefore, the d 15 N data suggest that wastewater was the major contributor to increases in DIN concentrations, and thus, likely to be the main cause of macroalgal blooms.

Relative response of macroalgae to nutrient enrichment
The relative growth response of Ulva spp. to experimental N or P enrichment varied across temperate and tropical sites and was not linked to geographic or latitudinal differences as had been previously suggested (Nixon & Pilson, 1983;Lapointe et al., 1992;Howarth, 2008). Rather, we demonstrated that N and P limitation of growth was linked directly to nutrient availability. In the case of macroalgae of the genus Ulva, DIN supply, primarily in the form of NO 3 À , was a dominant controlling feature in all coastal waters studied, temperate or tropical: if DIN supply was low, it restricted macroalgal growth, and if DIN was high, PO 4 3À became limiting. Although NO 3 À has been previous found as a driving factor in macroalgal growth, no existing study before this study has simultaneously conducted experiments on macroalgal growth responses to nutrients on a global scale to test how widespread is this phenomenon.
The results of our study support the meta-analysis of Elser et al. (2007), which showed no consistent pattern in N or P limitation across latitudes. Although Elser et al. (2007) showed some correlation with N limitation being more common in temperate sites, and P limitation being more common in tropical sites, these correlations were very weak (R 5 0.22 and 0.34, respectively) and there were no references to ambient nutrient conditions of these studies sites, which could be more important than latitudinal differences alone.
We are unsure how general our findings may be for all macroalgal species, because we know that in macroalgae there are diverse, species-specific responses to nutrient supply (Fujita, 1985;Fong et al., 2001;Teichberg et al., 2008). Ulva spp., however, are rather widespread, and are the dominant protagonists of many instances of harmful macroalgal blooms across the world. It is probable that other fast-growing macroalgae show similar limitation in proportion to the ambient supply of these two nutrients in the water-column as Ulva spp. Barile (2004) found that in subtropical coastal waters of eastcentral Florida, where water N : P were on average 8 : 1, that macroalgae with high uptake affinities for DIN (U. lactuca, Chaetomorpha linum, Enteromorpha intestinalis, Caulerpa spp., and Gracilaria tikvahiae) were N limited. In Southeastern Brazil, growth of macroalgae was P limited where water N : P were greater than 16 : 1 (Lourenco et al., 2005). In Waquoit Bay estuaries, where water N : P was approximately 3 : 1 during the growing season, G. tikvahiae were N limited (Teichberg et al., 2008), along with U. lactuca, as reported in this study. Although there is variability in growth responses, uptake affinities, and tissue N and P storage capacities among macroalgal taxa (Fujita, 1985;Pedersen & Borum, 1996;Fong et al., 2003), it is likely that water N : P may be useful to predict nutrient limitation in bloom forming species, as they were in our sites across the world. Thus, if the responses we report here are at all representative, the results have implications for many estuaries and coastal waters, tropical or temperate, where we are certain that human activities are increasing nutrient loads and changing N : P (Sfriso et al., 1992;Valiela et al., 1997;Larned, 1998;Lapointe et al., 2004). The results of this study demonstrate that nutrient supply strongly influenced growth and other features of Ulva spp., common macroalgae frequently involved in harmful macroalgal blooms. The comparisons of experimental results highlight the fact that as coastal waters increasingly undergo nutrient enrichment, blooms of algae will increase, and that the limiting element supporting the growth, initially nitrogen, shifts to phosphorus in those waters subject to the highest nitrogen loadings. It appears that understanding the specific mechanisms supporting macroalgal growth and managing the occurrence of blooms, regardless of temperate or tropical location, will require consideration of ambient supplies of N and P, as well as information on the magnitude of wastewater-related nutrient loads and the relative role of N and P in the specific locality involved. This result has additional applied relevance, because we have shown that ambient DIN concentrations and N : P are good predictors of macroalgal growth responses to nutrient enrichment.
Managers of coastal water quality are always facing the dilemma of not knowing explicitly what might be done to predict future time trends, maintain, or restore a coastal resource. If we know current loading rates, and we can predict future trajectory of nitrogen and phosphorus availability in a coastal water body, we might be able to use the information on N : P in water and macroalgal tissues to anticipate the formation of macroalgal blooms. In addition, knowing whether externally supplied nitrogen or phosphorus inputs might be restricting macroalgal growth may be a welcome asset, in particular because control of nitrogen is harder and more costly. These monitoring tools in combination with isotopic techniques may be used to detect the main nutrients responsible for bloom occurrences and point to specific sources to target for management.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Table SA1. Mean ambient water nutrient concentrations AE standard error averaged from annual measurements from other studies and during the experiment incubation period in this study for each site.
Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.