Estuaries and Coasts

, Volume 33, Issue 2, pp 448–470

Phytoplankton Patterns in Massachusetts Bay—1992–2007

Authors

    • Battelle
  • David G. Borkman
    • Graduate School of OceanographyUniversity of Rhode Island
  • P. Scott Libby
    • Battelle
  • Richard Lacouture
    • Estuarine Research CenterMorgan State University
  • Jefferson T. Turner
    • Biology Department and School for Marine Science and TechnologyUniversity of Massachusetts Dartmouth
  • Michael J. Mickelson
    • Massachusetts Water Resources Authority
Article

DOI: 10.1007/s12237-008-9125-9

Cite this article as:
Hunt, C.D., Borkman, D.G., Libby, P.S. et al. Estuaries and Coasts (2010) 33: 448. doi:10.1007/s12237-008-9125-9

Abstract

The Massachusetts Water Resources Authority (MWRA) conducts a comprehensive multidisciplinary monitoring program in Massachusetts Bay, Cape Cod Bay, and Boston Harbor to assess the environmental effects of a relocated secondary-treated effluent outfall. Through 2007, 8.7 years of baseline data and 7.3 years of postdiversion data (16 total years), including species level estimates of phytoplankton and zooplankton abundance, have been collected. MWRA’s monitoring program and other studies make this region one of the most thoroughly studied and well-described marine systems in the world. The data show that the diversion of MWRA effluent from the harbor to the bay has decreased nutrients concentrations and improved water quality in the harbor (e.g., higher dissolved oxygen, lower chlorophyll). The diversion also resulted in an increase in dissolved inorganic nutrients (especially ammonium) in the vicinity of the bay outfall, but no obvious impacts such as increased biomass or decreased bottom water dissolved oxygen have been observed. Regional changes in phytoplankton and zooplankton unrelated to the diversion have been seen, and it is clear that the bays are closely connected both physically and ecologically with the greater Gulf of Maine. Direct responses to modifications of the nutrient field within a 10 × 10-km area centered near the midpoint of the 2-km long outfall diffuser in Massachusetts Bay (a.k.a. the nearfield) have not been seen in the plankton community. However, plankton variability in the bays has been linked to large regional to hemispheric scale (NAO) processes.

Keywords

Massachusetts BayBoston HarborNAOTemporal and spatial trendsPhytoplanktonZooplanktonChlorophyllOutfallCeratiumPhaeocystisPhytoplankton time series

Introduction

Coastal phytoplankton populations are under multivariate control with the dominant controlling processes varying both regionally and temporally (Smayda 1998). The industrial era has been marked by human activities that are increasingly influencing plankton communities through various pathways including urbanization of the coasts (Bouwman et al. 2005) and increased fertilizer use (Seitzinger et al. 2002; Glibert et al. 2006) with subsequent increases in nutrient delivery to coastal waters (Howarth et al. 2002), increases in atmospheric nutrient deposition via increased fossil fuel burning (Paerl 1997; Carstensen et al. 2005), and modification of marine food webs (Pauly et al. 1998). These human activities may act to modify the range of variability in forcing functions, such as nutrient concentration, experienced by coastal phytoplankton. Of these controlling processes, land-based processes, such as nutrient input may be especially important determinants of coastal phytoplankton abundance levels and species composition (Clarke et al. 2006; Lancelot et al. 2007). However, variations in “ocean climate” (Dayton et al. 1999), i.e., the degree of oceanic influence, may also influence long-term patterns of coastal phytoplankton abundance and species composition. This increasingly appreciated mode of varying ocean climate influence on coastal plankton is influenced by basin-scale to hemisphere-scale oceanographic and atmospheric variability with concomitant effects on local weather patterns and costal phytoplankton populations (Smayda et al. 2004; Cloern et al. 2007).

Interannual variation in external forcing, via interannual variation in the strength of coastal currents, has been identified as an important driver of Massachusetts Bay phytoplankton abundance and species composition. The strength of the alongshore coastal current in spring time is an important factor in transport of red tide bloom forming Alexandrium fundyense populations into Massachusetts Bay (Anderson et al. 2005, 2007). The interannual variation in the degree of oceanic intrusion into Massachusetts Bay also appears to be an important determinant of the magnitude of the winter–spring bloom (Jiang et al. 2007a). Furthermore, the influence of an ocean basin-scale climatic driver, the North Atlantic Oscillation, has been detected in interannual variation of Massachusetts Bay zooplankton abundance (Turner et al. 2006). The multiple and annually variable determinants of coastal phytoplankton populations make it difficult to separate patterns and changes caused by industrial-age human activities from those related to variation in climate. Long (multidecadal) time series observations of sufficient spatial and temporal resolution are needed to provide the resolution needed to identify changes in phytoplankton abundance and species composition that may be caused by human activities from those that may be ascribed to natural variability (Smetacek and Cloern 2008).

Massachusetts Bay is a relatively open coastal system along the heavily urbanized Massachusetts coast near Boston, MA (Fig. 1). The Massachusetts Water Resources Authority (MWRA) provides secondary treatment for sewage from two million people in the greater Boston metropolitan area; the average discharge is 380 million gallons per day (16.6 m3 s−1). Prior to September 2000, this sewage effluent was discharged into outer Boston Harbor. In an effort to improve water quality in the harbor, an extensive redesign of the MWRA sewage treatment and discharge program, including full secondary treatment and relocation of the outfall to a site 15 km offshore in Massachusetts Bay, was begun in the 1990s (Aubrey and Connor 1993). An extensive water column monitoring program was established by the MWRA to collect ambient water quality data in Massachusetts and Cape Cod Bays for the assessment of potential environmental responses to the relocation of treated sewage effluent discharge. Among the concerns raised when outfall relocation was proposed, and again when it was authorized, was the potential impact of relocating a major nutrient source in Boston Harbor to the plankton communities found in Massachusetts Bay (MWRA 1991, 1997, 2004; Libby et al. 2007). In response to those concerns, the monitoring program was designed and implemented in 1992 (MWRA 1991) to assure regulators and the public that the impact would be as small as predicted by extensive preconstruction environmental assessments (EPA 1988). The program’s primary focus is on the ecological response within a 10 × 10-km region centered on the outfall (a.k.a. the nearfield; Fig. 1), but includes areas farther afield that could potentially be affected or could serve as a control for the nearfield response.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig1_HTML.gif
Fig. 1

Nearfield and farfield water quality stations and regional station groupings sampled in Massachusetts Bay, Cape Cod Bay, and Boston Harbor by the MWRA Outfall Monitoring Program. Generally, “Massachusetts Bay” can refer to the whole system between Cape Ann and Race Point; in this paper, it is the area north of Race Point, but outside Boston Harbor

This paper presents data from a comprehensive, continuous 16-year data set collected by the MWRA since 1992. The monitoring program includes extensive sampling of nutrients, phytoplankton, and zooplankton communities in the bays and at selected locations in Boston Harbor. The design was established to address changes in phytoplankton biomass, phytoplankton production rates, the abundance of nuisance or noxious phytoplankton species, and species composition of phytoplankton or zooplankton in response to the relocation. Data were initially gathered to establish baseline water quality conditions and, after outfall diversion on September 6, 2000, to detect whether significant departures from the baseline condition developed. Baseline monitoring surveys were designed to evaluate water quality on a relatively high-frequency basis (17 times per year) for a limited area in the vicinity of the outfall site and a low-frequency basis (six times per year) over an extended area throughout Boston Harbor, Massachusetts Bay, and Cape Cod Bay. The monitoring has continued for over 16 years and underwent a major optimization in 2004 that reduced the number of nearfield stations and surveys based on 9 years of monitoring data (MWRA 2003, 2004; Hunt et al. 2008). The data set offers one of the most comprehensive, consistent, long-term coastal data sets available globally and includes one of the few long-term plankton data sets supported by a full suite of nutrient, water quality, and meteorological and hydrodynamic parameters to support data interpretation. In this paper, we present portions of this comprehensive data set with a goal of assessing the degree of anthropogenic (via sewage discharge) and climate influences on the observed variation in Massachusetts Bay phytoplankton abundance and species composition during 1992–2007.

Methods

Depending on year, the MWRA monitoring program has sampled between 34 and 48 water quality stations dispersed throughout Boston Harbor, Massachusetts Bay, and Cape Cod Bay (42° N by 71° W; Fig. 1). The stations were distributed in a strategic pattern on the boundaries and interior of the nearfield and includes farfield stations located along the expected direction of net plume transport to the south of the diffuser, stations within Cape Cod Bay, and east of the diffuser near the boundary between the Gulf of Maine and Massachusetts Bay (Fig. 1). This station distribution was designed to develop a comprehensive characterization of the Massachusetts Bay/Boston Harbor region and to support interpretation of any ecological responses that could result from the outfall discharge. Twenty-one nearfield stations located in a grid pattern covering an area of approximately 100 km2 centered on the MWRA bay outfall were sampled through 2003; the station count was reduced to seven in 2004 based on a comprehensive data review (MWRA 2003) and a statistical optimization of the sampling design (Hunt et al. 2008). The 2003 review also led to a reduction in the number of nearfield water column monitoring survey from 17 during 1992–2003 to 12 surveys per year beginning in 2004. Twenty-seven farfield stations located throughout the Boston Harbor–Massachusetts Bay–Cape Cod Bay ecosystem have been consistently sampled six times per year since 1994 and were no part of the 2004 program optimization. These stations are combined with the higher frequency nearfield sampling (Fig. 1) to provide a reasonably synoptic perspective of the system’s characteristic through time and space. Data presented in this paper are available on request from MWRA. The data are synthesized and discussed in a series of reports downloadable from http://www.mwra.state.ma.us/harbor/enquad/trlist.html.

The MWRA sample collection and analysis techniques were performed according to the procedures outlined in Albro et al. (1993, 1998) and Libby et al. (2002, 2005, 2006). Hydrographic data were collected at all nearfield and farfield stations with a conductivity–temperature–depth (CTD) system that included auxiliary sensors for dissolved oxygen, chlorophyll fluorescence, optical beam transmittance, light irradiance (PAR), and in 2005, colored dissolved organic matter (CDOM). During the upcast at each station, a rosette equipped with 9-L Go-Flo/Niskin bottles was used to collect discrete water samples from five depths: bottom (within 5 m), midbottom, mid-depth (adjusted to sample the chlorophyll a maximum when present), midsurface, and surface (1–2 m of surface). Adjusting the mid-depth to sample the chlorophyll maximum ensured that the ocean depth most likely to interact with the effluent plume during summer stratification was monitored and avoided data biasing due to an inflexible target depth. The target depths for sample collection were predetermined based on ecological understanding but fine tuned in the field using high-resolution hydrographic profile data including salinity, temperature, fluorescence, dissolved oxygen, and turbidity. The sample bottles were subsampled for analysis of a range of dissolved and particulate nutrients, plankton, and productivity parameters (Table 1). Whole-water and screened-water phytoplankton samples were collected at preselected stations from the surface and mid-depth/chlorophyll maximum and fixed with Utermöhl’s solution (whole-water) or formalin (screened-water). Zooplankton samples were collected with vertical tows over the upper 25 m with a 0.5-m diameter, 102-μm mesh net equipped with a flow meter at a subset of stations and fixed with buffered formalin.
Table 1

Summary of water column measurements made under the MWRA Harbor and Outfall Monitoring Program

Measurement type

In situ parameter

Laboratory analysis

Physical characterization

Temperature, salinity, dissolved oxygen, beam attenuation

Dissolved oxygen (DO)

Total suspended solids (TSS)

Nutrients

Colored dissolved organic matter (CDOM)

Dissolved inorganic nitrogen (DIN=NH4+NO3+NO2)

Ammonium (NH4)

Nitrate (NO3)

Nitrite (NO2)

Phosphate (PO4)

Silicate (SiO4)

Dissolved organic carbon (DOC)

Total dissolved nitrogen (TDN)

Total dissolved phosphorous (TDP)

Phytoplankton biomass

Fluorescence

Chlorophyll

Particulate organic carbon (POC)

Particulate organic nitrogen (PON)

Particulate phosphorous (PP)

Biogenic silica (BSi)

Productivity

 

Primary productivity

Plankton community structure

 

Taxonomy and abundance of phytoplankton

Taxonomy and abundance of zooplankton

Methods relevant to this paper include chlorophyll/fluorescence, dissolved inorganic nitrogen (DIN) forms, and phytoplankton and zooplankton enumeration. Chlorophyll a samples are collected in duplicate at a subset of stations at each of the five sample depths. Samples were vacuum-filtered on glass fiber filters (Whatman 47-mm-diameter GF/F filters), stored frozen, and analyzed using a Turner Designs fluorometer (Yentsch and Menzel 1963; Arar and Collins 1992). The chlorophyll a data were used to postcalibrate the readings from the in situ fluorometer. The resulting fluorescence values (in micrograms per liter) were integrated using 0.5 m depth binned in situ data over the entire water column to calculate areal chlorophyll (in milligrams per square meter) at each station.

Dissolved inorganic nutrient samples were collected at all depths and all stations. The samples were filtered through a 0.4-μm pore-sized membrane filter (e.g., Nuclepore) and the filtrate was frozen until analysis. The concentrations of ammonium (NH4), nitrate (NO3), and nitrite (NO2) were measured colorimetrically on an autoanalyzer (e.g., Technicon II, Alpchem, and Skalar). The NH4 analysis was based on the technique of Solorzano (1969) whereby absorbance of an indophenol blue complex is measured at 630 nm. Nitrite was measured by the method of Bendschneider and Robinson (1952). The total of NO3+NO2 was determined by reducing all NO3 in the sample to NO2 and analyzing for NO2. The reduction is accomplished using a cadmium column (Morris and Riley, 1963). Dissolved inorganic nitrogen is calculated as the sum of NH4+NO3+NO2.

An 800-mL aliquot of the phytoplankton samples was settled to 50 mL (16:1 ratio) in a graduated cylinder (nominal 5-to-1 height-to-width ratio) and decanted by low vacuum aspiration. A 1-mL aliquot of the concentrate was transferred to a Sedgwick–Rafter chamber and counted under a compound microscope with phase-contrast optics and long-working-distance objectives (but an inverted scope in 1995–1997). Only a portion of the 1-mL Sedgwick–Rafter chamber volume was examined (e.g., 48 field-of-view paths at ×500 magnification). Taxa counting-effort cut offs were established at 75 entities of the top three taxa and 400 cells total count. Abundances were calculated by dividing the number of cells counted by the volume examined in Sedgwick–Rafter chamber and multiplying by the concentration factors, typically 16:1.

The method used for phytoplankton enumeration has changed slightly over the years. During 1992–1994, phytoplankton were counted at a lower magnification (×250) in a Sedgwick–Rafter counting cell, likely resulting in an underestimation of the smallest and most abundant components of the phytoplankton such as microflagellates. During 1995–1997, there was a change to the use of an inverted microscope; the greater magnification and associated high multiplier factors may have resulted in elevated microflagellate abundance during this period. In 1998, there was a change back to counting in a Sedgwick–Rafter cell, but at ×500 for better detection of the small component of the phytoplankton. The differences between the three methodologies employed during 1992–2007 were mainly related to differences in the emphasis (low, high, then medium) given to estimation of the smallest component (i.e., microflagellates) of the phytoplankton community. The abundance of large (>20 μm) dinoflagellates was estimated by using a size fractionation filtering method similar to that of Turner et al. (1995). In this, the >20 μm dinoflagellates in a 4-L water sample were concentrated by filtering through a 20-μm screen. The retained >20 μm aliquot was reduced to a volume of 10 mL for a 400:1 concentration factor. This concentrated sample was counted at ×250 in a Sedgwick–Rafter cell for identification and enumeration of >20 μm dinoflagellates.

Zooplankton community analyses were performed as described in Turner et al. (2006). For enumeration, samples were reduced with a Folsom plankton splitter to aliquots of at least 300 individuals; zooplankton species were identified to the lowest practical taxon. For copepods, adults were identified to species and sex, and copepodites to genus. Copepod nauplii were not identified further, and other zooplankters (chaetognaths, fish larvae, pteropods, etc.) were also not identified beyond major group.

Data Treatment

Data Aggregation

The data presented are based on survey and annual averages for in situ, nutrient, and plankton by area (Fig. 1). Survey averages for in situ data include all stations within an area and are binned by surface and bottom depths. Survey averages for nutrient data include all stations and depths sampled. Phytoplankton and zooplankton survey averages are shown for the nearfield area (stations N04, N16, and N18). The phytoplankton survey averages represent both surface and mid-depth samples.

Statistical Methods

Time series analysis was applied to the dominant phytoplankton and zooplankton groups in the nearfield area to identify long-term abundance trends and cycles from 1992 to 2007. The method employed (Broekhuizen and McKenzie 1995) is robust to strongly seasonal time series, such as those observed in some plankton species. Time series analysis requires serial observations at regular time intervals. The MWRA plankton monitoring data had its highest sampling frequency in the nearfield region with plankton observations in 154 of 192 months (80%) during 1992–2007. For this time series analysis, we binned nearfield observations into monthly means to construct a 192-month time series. There were 38 months lacking observations. There were no observations in January of any year and only six observations per year in the early (1992–1994) years of the time series. January abundance levels were filled with the average of February and the preceding December for 1996–2004. January of other years lacked data to average and so were filled with the average of the 1996–2004 filled Januarys. This yielded a 192-month time series having 80% observed data and 20% filled data. The time series was then deseasonalized with a 12-month lagging average. Smoothing of the deseasonalized series was then considered: Monte Carlo simulations have indicated that data gaps of up to 30% may be successfully interpolated without loss of a signal and that the signal loss is equal to approximately three times the longest data gap (Sturges 1983). The largest filled data gap in the 192-month nearfield plankton time series was 3 months (November 1993–January 1994 and November 1994–January 1995) indicating that, despite the filled data gaps, long-term trends of greater than 9 months should be detected. Because of this data gap and because of the interest in long-term trends, a smoothing window of 19 months, equivalent to ∼10% of the time series length, was chosen to represent the long-term trend. The long-term trend estimated by this method represents a smoothed deseasonalized abundance level about which the actual seasonal pattern fluctuates.

An exploratory multivariate statistical approach was applied to the 1992–2006 portion of the phytoplankton data to identify patterns of long-term phytoplankton community variation. The statistical analysis software Primer (Plymouth Routines In Multivariate Ecological Research; Clarke and Gorley 2001) was used for this multivariate analysis. The MWRA phytoplankton data for 1992–2006 consisting of 2,994 phytoplankton samples was reduced to 1,002 samples by monthly averaging phytoplankton abundance in six regions (boundary, Cape Cod Bay, coast, harbor, nearfield, offshore). Separate surface and chlorophyll maximum depths were retained. The resultant data matrix was 1,002 samples (described above) in length with each sample having the abundance of 249 distinct species or phytoplankton groups. The entire 249 species/groups by 1,002 sample data matrix was first analyzed by calculation of a similarity matrix (Bray–Curtis similarity matrix was used) on log-transformed (x + 1) and standardized data and application of multidimensional scaling (MDS) analysis to identify patterns of all phytoplankton samples. In these two-dimensional MDS plots, the distance between sample points corresponds to the similarity between samples (i.e., close points are similar, distant points are less similar). The goal was to identify dominant patterns of variation (i.e., spatial by region or by depth; temporal by years or months; or by a putative treatment effect such as pre-postoffshore outfall) in phytoplankton community composition.

A third statistical approach was taken to determine if there have been any potential outfall effects on plankton abundance and community composition. A statistical comparison of various phytoplankton taxa was conducted to test the hypothesis that there are differences in the prediversion and postdiversion abundance levels. Phytoplankton data were averaged into regional means [six regions: boundary (F27), Cape Cod Bay (F01 and F02), coastal (F13, F24, and F25), Boston Harbor (F23, F30, and F31), nearfield (N04, N16, and N18), and offshore (F06); see Fig. 1 for locations] by event ID and were further binned by the two depths sampled (near-surface and mid-depth). Data distributions were checked for approximation of the normal distribution prior to analyses with the Kolmogorov–Smirnov test. Most data distributions were skewed and deviated significantly from the normal distribution, so a nonparametric Mann–Whitney U test (i.e., Wilcoxon rank–sum test) was employed for the pre-September 6, 2000 (prediversion) and post-September 6, 2000 (postdiversion) comparisons. The number of observation in the preperiod and postperiod varied regionally and by variable. For phytoplankton, the nearfield was most frequently sampled with 223 observations prior to the offshore diversion and 224 observations after the diversion. The boundary and offshore regions had only 35 (pre) and 37 (post) observations because they were limited to sampling at one station for each region (other stations have been added more recently, but do not provide adequate prediversion data for this comparison). Cape Cod Bay had 106 (pre) and 74 (post) observations and the coastal region and harbor areas had 124 (pre) and 111 (post) observations.

Results

Over the 16 years of MWRA monitoring, a general classic annual sequence of coastal ecological events has become apparent. The patterns are evident even though the timing and year-to-year manifestations of these events are variable. Typically, but not always, a winter–spring phytoplankton bloom occurs as light becomes more available, temperature increases, and nutrients are readily available. Later in the spring, the water column transitions from well-mixed to stratified conditions, cutting off nutrient flux to the surface. The summer is generally a period of strong stratification, depleted surface water nutrients, low biomass, and a relatively stable mixed-assemblage phytoplankton community. In the fall, stratification deteriorates and mixing supplies nutrients to surface waters. This process occasionally contributes to the development of a fall phytoplankton bloom. Dissolved oxygen (DO) concentrations are lowest in the bottom waters prior to the fall overturn of the water column—usually in October and rarely less than 6 mg L−1. By late fall or early winter, the water column becomes well-mixed and returns to winter conditions. In winter, the combination of wind mixing and low light levels serve to inhibit accumulation of phytoplankton biomass until the following year’s winter–spring bloom. These patterns are described in greater detail below.

Temperature and Salinity Temporal Patterns

Water temperature data from the near-surface and near-bottom waters of the nearfield region of Massachusetts Bay show a 15°C and 7°C, respectively, annual range in temperature, typical of latitude (42° N), superimposed with year-to-year variability (Fig. 2a). The peak and minimum temperatures vary across years with some years being warmer and others cooler. Colder winters are indicated by minimum winter temperatures below 3°C as in 1993, 1994, and 1996 and again in 2003 to 2005. Peak surface summer temperatures are about 18°C with some years reaching 20°C (1995, 1999, and 2003). The seasonal thermal stratification is clearly evident in Fig. 2a with bottom temperatures under 10°C in winter in some of the cold years (e.g., 1993, 1994, 2003) and above 12°C in summer in the 1990s (e.g., 1995, 1996, 1997, 1999). Maximum bottom temperatures vary as a result of physical factors such as wind direction, strength, and timing. The onset of thermal stratification is typically in May, but can vary depending on spring weather conditions.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig2_HTML.gif
Fig. 2

Time series of near-surface (blue) and near-bottom (green) temperature (a) and salinity (b) in the vicinity of the bay outfall (averaging the data from nearfield stations N16, N18, and N20)

Salinity in near-surface and near-bottom waters data mirrors inversely the temperature data with lower surface salinity in the spring and early summer during periods of peak river input. The lowest surface salinity was recorded in June 1998 and in the spring of 1993 and 2006 (Fig. 2b) and were associated with recorded freshwater inputs or major meteorological events (e.g., storms) that imparted stochastic signatures on the overall seasonal patterns. The seasonal temperature and salinity variation and storm effects combine to add local and interannual variability to the onset of stratification in the Massachusetts Bay system (Libby et al. 2007). Stratification is examined as the differential of density difference between the surface and bottom water with a difference of ≥1 kg/m3 considered indicative of stratification. Although a well-stratified system is typically present by May and extends through late October or November when storms mix the water column, stratification can set up as early as late March or early April and peaks in August (Libby et al. 2008).

Nutrient Patterns

Nutrient concentrations in the Massachusetts Bay system typically progress through a series of seasonal events that are closely linked with physical and biological factors. DIN concentrations (Fig. 3) are high (10 to 12 μM) in the winter when biological uptake is low and water column mixing is thorough. Concentrations decrease in the surface waters during the winter–spring phytoplankton bloom due to the uptake and the onset of stratification which cuts off the supply of nutrients from deeper waters. As stratification strengthens, nutrients are generally depleted in surface waters and increase at depth in the summer due to remineralization processes (Fig. 3). As fall destratification occurs, nutrient concentrations increase in the surface waters and are often associated with a fall bloom (Libby et al. 2007). Other nutrients (SiO4, PO4) follow a similar pattern of elevated winter concentration and reduced summer concentration.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig3_HTML.gif
Fig. 3

DIN (DIN=NH4+NO3+NO2) seasonal pattern for near-surface and near-bottom waters in the nearfield for 2007

The seasonal and annual mean nutrient concentrations (e.g., NH4, NO3) across distinct regions of the system were also evaluated for patterns and trends. The change in NH4 concentrations in the nearfield is consistent with model simulations which predicted that the transfer of effluent from Boston Harbor to Massachusetts Bay would greatly reduce nutrients in the harbor and increase them locally in the nearfield (Signell et al. 1996, Signell 2007). This change was predicted to have little impact on concentrations in the rest of Massachusetts and Cape Cod Bays. The spatial patterns in NH4 concentrations in the harbor, nearfield, and bays since the diversion in September 2000 have consistently confirmed this prediction (Taylor 2006; Libby et al. 2007, 2008). These spatial changes in NH4 are also manifest in annual mean concentrations for these areas. For example, the annual mean NH4 concentration in Boston Harbor (Fig. 4a) dropped sharply from ∼10 μM in 2000 to ∼1.5 μM in 2001. A sharp decrease was also seen at the coastal stations which are strongly influenced by water quality conditions in Boston Harbor. In contrast, the increase in annual mean NH4 in the nearfield was not as dramatic as the harbor and coastal water decrease. Compared to 1999 (∼1.2 μM), the last full year before the bay outfall came online, annual mean NH4 levels in the nearfield almost doubled in 2001 (∼2 μM; Fig. 4a).
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig4_HTML.gif
Fig. 4

Annual mean NH4 (a) and NO3 (b) concentration in Massachusetts and Cape Cod Bays. Mean of concentrations over depths, stations, and surveys within each region. Dotted line denotes offshore outfall online after 2000

Since 2001, however, NH4 has shown a system-wide decrease and even in the nearfield concentrations are again comparable to the 1999 levels. This decline in NH4 over the past several years can be seen in all of the survey regions and current annual concentrations are comparable to 1992–1999 across the bays. The shift in NH4 from prediversion to postdiversion years can be seen in the contour plots showing the difference between the seasonal means from each period for the entire survey area (Fig. 5a). The reduction in Boston Harbor and coastal water NH4 concentrations can be clearly seen across seasons, as well as the increase in NH4 at the bay outfall location. These differences in NH4 concentrations in Boston Harbor and nearfield area represent significant changes from baseline based on regression analyses (P < 0.05; Libby et al. 2008). The trends in annual mean concentration for other inorganic nutrients are more erratic as seen in the example of NO3 (Fig. 4b), which has actually increased over much of the bays (except Boston Harbor) over the years monitored. An examination of seasonal NO3 means increase in values (1 to 3 μM) over most of the bays during the postdiversion winter–spring and fall seasons (Fig. 5b).
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig5_HTML.gif
Fig. 5

Change in seasonal NH4 (top row) and NO3 (bottom row) concentrations (in micromolar) from baseline to postdiversion. Based on the difference of means calculated over all depths from each station, survey, and season for each period

Chlorophyll Patterns

Seasonal trends in phytoplankton biomass as indicated by chlorophyll and particulate organic carbon (POC) are tied to physical conditions, nutrient availability, and ecosystem dynamics. The phytoplankton biomass seasonal signal in Massachusetts and Cape Cod Bays is dominated by winter–spring and fall blooms. Winter–spring phytoplankton blooms occur due to elevated growth related to increased light availability, nutrient-replete conditions, and seasonal stratification of the physical environment prior to temperature-related increases in mortality due to grazing and nutrient depletion. Typically, the timing of the fall bloom has been tied to decreased stratification and increased inputs of nutrients into the surface waters (Oviatt et al. 2007).

As shown for nearfield survey means (Fig. 6), areal chlorophyll values have been generally consistent between 1992–1998 and 2001–2007. The data for 1999 and 2000 stand out from the rest of the time series as these years were characterized by high chlorophyll concentrations during both winter–spring and fall blooms and, as a result, the annual means were nearly double that of the other years (Fig. 6). The higher chlorophyll levels were present both right before and after the diversion of the outfall to the bay. This period also serves as an apparent transition from a period dominated by February diatom blooms and large fall blooms (1992–1998) to a postdiversion period that has been characterized by substantial April blooms of Phaeocystis pouchetii and relatively minor fall blooms.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig6_HTML.gif
Fig. 6

Chlorophyll (in milligrams per square meter) temporal pattern in the nearfield baseline (blue circles) and postdiversion (green triangles). Data represent survey means and horizontal lines are the annual mean values

The winter–spring increase and fall decrease in chlorophyll is also evident in other areas of the bays. The annual occurrence of moderate to major Phaeocystis blooms since 2000 has resulted in increases in regional chlorophyll and POC levels during the winter–spring postdiversion (Fig. 7). The lack of fall blooms during the postdiversion period is evident in the lower chlorophyll and POC concentrations throughout most of Massachusetts and Cape Cod Bays. These figures also highlight the clear decrease in both areal chlorophyll and POC concentrations at most of the Boston Harbor stations across each season (Fig. 7; Taylor 2006). Overall, the transfer of the outfall from Boston Harbor to the bay has likely contributed to the phytoplankton biomass decrease, as indicated by declining chlorophyll and POC in the harbor. Changes in the bays, however, appear to be related to regional changes in phytoplankton species composition and bloom frequency.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig7_HTML.gif
Fig. 7

Change in seasonal areal chlorophyll (in milligrams per square meter; top row) and POC (in micromolar; bottom row) concentrations from baseline to postdiversion. Based on the difference of means calculated over all depths from each station, survey, and season for each period

Zooplankton Patterns

Zooplankton abundance and community structure in Massachusetts and Cape Cod Bays were generally regionally similar during 1992–2007. The zooplankton community assemblage in the bays is dominated by copepod nauplii, Oithona similis, and Pseudocalanus spp. copepodites throughout the year with subdominant appearances of other copepods such as Calanus finmarchicus, Paracalanus parvus, Centropages typicus, and C. hamatus and sporadic pulses of various meroplankters such as bivalve and gastropod veligers, barnacle nauplii, and polychaete larvae (Turner 1994; Libby et al. 2007). Zooplankton abundance from 1992 to 2007 had seasonal patterns of abundance that generally followed temperature with low levels in winter, rising through spring to maximum summer levels, and declining in the fall (Fig. 8a). The most apparent change in zooplankton has been the lower summer abundance since 2001.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig8_HTML.gif
Fig. 8

Total zooplankton abundance in the nearfield (top). Long-term trend (bottom) obtained by filling, deseasonalizing, and smoothing the data series with the mean (flat line) shown for reference

Time series analyses (1992–2007) demonstrated that there was a substantial long-term decline in the nearfield means for the abundance of total zooplankton during 2001–2006 (Fig. 8b). Total zooplankton declined from levels of 40,000 to 60,000 animals per cubic meter during 1999 to 2001 to levels of ∼25,000 animals per cubic meter during 2004–2005 and then rebounded to 35,000 animals per cubic meter during 2006–2007 (Fig. 8b). Declines in zooplankton groups (total copepods, copepod nauplii) and some species (Oithona spp.) followed a similar long-term declining pattern during 2001–2006. However, not all zooplankton followed this long-term abundance pattern. Calanus finmarchicus exhibited a precipitous drop from 2000 to 2001, followed by a sharp ascent in 2002 to maximum levels that were maintained through 2003–2004, and declined to slightly below mean levels during 2005–2007. The differing long-term zooplankton abundance patterns may reflect species-specific zooplankton responses to interannual variation in wind strength and direction as modulated by a large-scale driver of winter weather in this region (the North Atlantic Oscillation; Turner et al. 2006, Jiang et al. 2007b).

Phytoplankton Seasonal Pattern

Total phytoplankton annual abundance pattern was characterized by levels of less than one million cells per liter during winter, gradually rising to a peak of about two million cells per liter during the summer, and then a decline to lower levels during autumn and into the winter (Fig. 9a). Spring and autumn peaks in the mean annual total phytoplankton cycle reflected the contribution of Phaeocystis blooms (during April) and autumn diatom blooms (during October). Much of the total phytoplankton annual cycle can be explained by the annual pattern of microflagellates (unidentified phytoflagellates <10 μm diameter) which numerically dominated the phytoplankton. Unidentified microflagellates (<10 μm diameter) were usually numerical dominants throughout the year, and their annual abundance pattern generally tracked water temperature, being most abundant in summer and least abundant in winter (Fig. 9b). Total diatom abundance fluctuated between 100,000–500,000 cells per liter during winter and spring with occasional blooms of >1 million cells per liter (Fig. 10a). The annual diatom abundance peak often occurred during summer or early autumn with summer and autumn diatom levels that rival or exceeded those observed during winter–spring (Fig. 10a). Large (>20 μm) dinoflagellates also had a seasonal cycle that usually peaked in summer (Fig. 10b). Large (>20 μm) dinoflagellate abundance generally varied approximately tenfold seasonally, from levels of several hundred cells per liter during the winter to several thousand cells per liter during the summer and autumn.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig9_HTML.gif
Fig. 9

Mean seasonal pattern of a total phytoplankton abundance and b microflagellate abundance

https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig10_HTML.gif
Fig. 10

Mean seasonal pattern of a total diatom and b dinoflagellate abundance (>20 μm)

A seasonal phytoplankton succession in Massachusetts and Cape Cod Bay has been consistently observed. While about 300 phytoplankton taxa (including all size-specific identifiers, i.e., Gymnodinium <10 μm, etc.) were recorded, a much-reduced subset of 30 phytoplankton species or groups dominated the observed phytoplankton annual patterns. The quarterly averaged abundance of these dominant phytoplankton groups and dominant genera and species are summarized in Table 2. The nonmicroflagellate phytoplankton was generally diatom dominated during winter, spring, and autumn, but was occasionally dinoflagellate-dominated during the midsummer stratification period.
Table 2

Quarterly mean abundance (cells per liter) of dominant phytoplankton taxa

 

Species or group

First quarter

Second quarter

Third quarter

Fourth quarter

Diatoms

Chaetoceros spp.

111,400

94,300

20,000

16,400

Chaetoceros socialis

40,100

15,600

1,400

0

Detonula confervacea

1,400

200

0

400

Thalassiosira spp.

76,100

23,600

6,300

73,200

Thalassiosira anguste-lineata

2,100

100

0

0

Thalassiosira nordenskioldii

32,700

2,200

100

1,900

Thalassiosira rotula/gravida

12,400

4,800

500

200

Chaetoceros debilis

11,500

14,100

3,400

1,300

Cerataulina pelagica

200

2,600

14,800

1,700

Dactyliosolen fragilissimus

2,600

6,600

222,200

74,300

Leptocylindrus danicus

200

6,800

72,800

63,800

Leptocylindrus minimus

2,100

1,200

27,000

24,800

Asterionellopsis glacialis

1,500

100

6,700

149,300

Guinardia delicatula

3,000

11,600

19,200

35,600

Pseudo-nitzschia spp.

11,200

7,200

13,900

16,400

Rhizosolenia spp.

300

500

400

5,900

Skeletonema spp.

7,500

48,900

50,500

57,900

Thalassionema nitzschioides

8,300

6,400

9,800

17,500

Dinoflagellates

Dinophysis norvegica

0

500

300

100

Ceratium longipes

200

1,800

1,100

300

Ceratium tripos

300

400

1,000

900

Gymnodinium spp.

14,400

29,400

60,600

40,600

Heterocapsa rotundata

2,600

13,400

18,400

18,300

Prorocentrum minimum

600

4,400

7,800

2,100

Prorocentrum triestinum

0

0

400

400

Protoperidinium spp.

1,100

1,900

2,600

1,800

Scrippsiella trochoidea

0

200

800

200

Ceratium fusus

100

300

500

700

Ceratium lineatum

0

200

200

300

Gyrodinium spp.

1,500

1,700

2,300

2,900

Heterocapsa triquetra

200

1,900

700

2,500

Prorocentrum micans

200

100

700

2,600

Other groups

Phaeocystis pouchetii

349,300

473,600

0

0

Cryptophytes

40,600

112,900

163,700

160,700

Phytoflagellates <10 μm

338,000

666,000

1,087,900

695,300

Total phytoplankton

1,027,700

1,570,600

1,914,400

1,669,700

Data derived by averaging surface and subsurface chlorophyll data from nearfield stations only during 1992–2006. The highest value for each row is in italics

Within the diatoms, a successional pattern from a winter community to a summer community interspersed with transitional spring and autumn communities was evident. The winter (first quarter) diatoms were dominated by Thalassiosira spp. (especially T. nordenskioeldii and T. anguste-lineata), Chaetoceros socialis, Guinardia delicatula, and in some years, Detonula confervacea (Table 2). Other common winter diatoms present in lower abundance levels included Lauderia annulata, Porosira glacialis, and Fragilaria spp. The winter–spring diatom bloom was generally Thalassiosira spp.-dominated initially (February–March) with increasing dominance of Chaetoceros during April. During spring (second quarter), a Chaetoceros- and Thalassiosira-dominated flora persisted with Chaetoceros debilis being characteristic of spring diatoms. During summer, a different diatom assemblage was generally present. The summer diatoms were characterized by Cerataulina pelagica, Dactyliosolen fragilissimus, Leptocylindrus danicus, and Leptocylindrus minimus. Autumn diatoms were characterized by Asterionellopsis glacialis, G. delicatula, Pseudo-nitzschia spp., and Skeletonema spp. A. glacialis was present year-round, but bloomed (maximum observation was 4.8 million cells per liter) during October 1993. Skeletonema spp. were consistent with the morphology of Skeletonema costatum sensu lato as visible via light microscopy at ×250 to ×500 magnification, but likely include other recently described Skeletonema spp. (Zingone et al. 2005; Sarno et al. 2005; Kooistra et al. 2008). While Skeletonema spp. were most abundant during late summer and autumn, they were present year-round, and given the approximately 15°C variation in temperature and concomitant fluctuation in other variables, there is likely presently undetected seasonal variation embedded within Skeletonema spp.

Dinoflagellates also had seasonal succession with relatively low dinoflagellate abundance during winter followed by increased abundance during spring. Dominant spring dinoflagellates were relatively large species such as Dinophysis norvegica and Ceratium longipes. The summer dinoflagellate community was numerically dominated by small species such as Gymnodinium spp. (most were <20 μm diameter), Heterocapsa rotundata, Prorocentrum minimum, and Prorocentrum triestinum. Large Ceratium spp. (C. longipes, C. fusus, and C. tripos) were an important component of the dinoflagellate flora during spring through autumn. In autumn, Gyrodinium spp., Heterocapsa triquetra, and Prorocentrum micans were most abundant.

The predominant feature of the seasonal phytoplankton succession in Massachusetts Bay was the seasonally varying presence of two distinct communities. During winter and spring, a diatom (Thalassiosira- and Chaetoceros-dominated) and Phaeocystis-dominated community was present. During the summer and early autumn months, a different diatom assemblage (D. fragilissimus, Skeletonema spp., Pseudo-nitzschia spp., and L. danicus) and dinoflagellates dominated. Interannual variation in the transition between these winter and summer communities may be related to interannual variation in the vernal onset and autumnal breakdown of water column stratification.

Long-Term Phytoplankton Trends

The long-term mean abundance of total phytoplankton was about 1.4 million cells per liter. Relatively low total phytoplankton (1.1 to 1.3 million cells per liter) was evident during 1992–1994 (Fig. 11). During 1995–1998, total phytoplankton abundance rebounded to near long-term mean levels then declined in 1999 to a relative low point (1.1 million cells per liter). Total phytoplankton abundance then increased progressively during 1999 through 2005 to a peak of 1.9 million cells per liter in early 2005. Total phytoplankton abundance returned to near mean levels (1.4 million cells per liter) in 2006 and increased again in 2007 (1.5–1.7 million cells per liter). The relatively low phytoplankton in the early years of the monitoring program are partially attributed to a methodological issue (counting at low magnification) which likely underestimated microflagellate abundance during 1992–1994 and may have overestimated microflagellate abundance during 1995–1997. This is more clearly seen in the long-term trend in microflagellate abundance which showed three distinct periods: one of relatively low abundance (1992–1994), one of elevated abundance 1995–1997, and a gradual long-term increase from 1998 through 2007 (Fig. 12). The 1992–1994 versus 1995–1997 change corresponds to changes in methodology (different magnification and phytoplankton analysts), suggesting a methodological rather than environmental basis for the trend during this period (Libby et al. 2007). A gradual increase in nearfield microflagellate abundance appears in the data after 1997 from ∼0.55 million cells per liter in 1999 to ∼0.85 million cells per liter during 2004. This increase is not related to any methodological changes as counting techniques were consistent during this period. During 2005–2007, microflagellate abundance declined to near the long-term mean level of about 0.7 million cells per liter.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig11_HTML.gif
Fig. 11

Total phytoplankton abundance in the nearfield (top). Long-term trend (bottom) obtained by filling, deseasonalizing, and smoothing the data series with the mean (flat line) shown for reference

https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig12_HTML.gif
Fig. 12

Unidentified microflagellate (<10 μm) abundance in the nearfield (top). Long-term trend (bottom) obtained by filling, deseasonalizing, and smoothing the data series with the mean (flat line) shown for reference

In contrast to the microflagellates, the nearfield diatom abundance displayed a dramatic long-term decline during 1992–2006 with 2005–2006 levels that were only ∼25% of the peak level observed during 1994 (Fig. 13). In 2007, diatom abundances increased back to the long-term mean level of 0.33 million cells per liter. Within this long-term decline are relative peaks in abundance in 1994, 1998, 2002, and 2007. The relative peaks in diatom abundance roughly correspond with relative nadirs in Phaeocystis abundance (Fig. 14). Correlation analysis of these two trends yielded a Pearson r value of −0.54 (P < 0.0001), indicating that a long-term negative interaction may be occurring between Phaeocystis and diatom abundance in the nearfield. This interaction is likely operative in the winter–spring only, given that it is the time of Phaeocystis presence in the bay and throughout the bays, but it may have lag effects on diatom abundance into the remainder of the year.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig13_HTML.gif
Fig. 13

Total diatom abundance in the nearfield (top). Long-term trend (bottom) obtained by filling, deseasonalizing, and smoothing the data series with the mean (flat line) shown for reference

https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig14_HTML.gif
Fig. 14

Long-term trend for diatom abundance and for P. pouchetii in the nearfield. Pearson r value of two trends was −0.54 (P < 0.0001)

In comparison, the dinoflagellates displayed abundances near long-term mean abundance during 1992–1994, relatively low abundance during 1995–1998, followed by a peak in abundance during 1999–2002, and then a decline to a relatively low abundance period of 500 cells per liter from 2003 to 2007 (Fig. 15). The long-term Ceratium abundance trend followed a similar pattern with the trend in nearfield Ceratium spp. positively correlated with the total dinoflagellate trend (Pearson r = +0.93, P < 0.0001). The relative contribution of Ceratium spp. to total dinoflagellate abundance declined from about 50–90% during 1996–2002 to ∼20% during 2005 and 2007.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig15_HTML.gif
Fig. 15

Long-term trend for total dinoflagellate abundance and for one of the dinoflagellates—Ceratium spp. The two trends were significantly positively correlated (Pearson r of + 0.93, P < 0.0001). Dinoflagellate data are from the >20 μm screened samples

One possible mechanism driving the observed long-term oscillation in Ceratium abundance may be interannual variation in the timing and degree of stratification in Massachusetts Bay. Ceratium are most dominant in Massachusetts Bay and in other temperate coastal seas during the summer stratified period (Cushing 1989). In Massachusetts Bay, there was a moderate positive correlation between degree of stratification and Ceratium abundance with the delta sigma T between surface and near-bottom taken as an indicator of degree of stratification. Application of a 1 month time lag between stratification and Ceratium abundance yielded maximum explanatory power with May stratification explaining approximately 40% of the variance in June Ceratium abundance (Fig. 16). March stratification was also positively correlated with April Ceratium abundance, but no significant stratification–Ceratium correlation was found in other months of the year. Moreover, seasonal peak stratification was not significantly associated with Ceratium abundance. Ceratium’s slow growth (approximately 0.3 divisions per day; Cushing 1989) indicates that a prolonged period of favorable conditions is required for large population accumulation to occur. This may be indicative of a dependence on the establishment of stratification in the late winter/early spring for achievement of Ceratium population development in the spring/early summer.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig16_HTML.gif
Fig. 16

Statistically significant exponential relationship between degree of stratification during May and Ceratium abundance during June in the nearfield region of Massachusetts Bay during 1992–2007. Data points labeled by year

Phytoplankton Spatial Patterns

Phytoplankton abundance and community composition has been consistently monitored at 13 stations since 1994 with two depths (near-surface and a chlorophyll maximum depth) analyzed per station. These stations were located in six distinct regions (Fig. 1) ranging from Boston Harbor to 50 km offshore. Some regional differences in phytoplankton pattern, such as an earlier start to the winter–spring bloom in relatively shallow Cape Cod Bay and regionally differing Phaeocystis abundance have been noted. Also, the development of a distinctive Ceratium-dominated subsurface phytoplankton community occurred during some summers although the degree of spatial (by region and depth) variation in phytoplankton community composition has not been quantified.

The MDS analysis yielded a plot that was of moderate utility in identifying long-term variation, as assessed by the stress value of 0.21. Coding the 1,002 samples by region showed that no distinct regional clusters of samples are apparent (Fig. 17). For example, the nearfield samples (red diamonds), as well as samples of all other regions, may be seen scattered across all areas of the MDS plot. This is indicative of little systematic regional variation in phytoplankton community composition in the monitoring area. The six regions (boundary, Cape Cod Bay, coastal, harbor, nearfield, outfield) sampled by the MWRA monitoring program do not span a large salinity gradient—the harbor phytoplankton samples are relatively close to the mouth of the harbor. Although freshwater forms (Asterionella formosa, Scenedesmus spp.) are occasionally encountered in low numbers at all stations, the monitoring program does not sample a strong estuarine gradient, and a strong regional effect on phytoplankton community composition was not detected by the MDS analysis.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig17_HTML.gif
Fig. 17

Regional coherence of Massachusetts Bay and Cape Cod Bay phytoplankton communities based on community analysis. This multidimensional scaling plot is based on 1,002 regionally averaged samples collected during 1992–2006. Samples coded by sample region (region 1 boundary, 2 Cape Cod Bay, 3 coastal, 4 harbor, 5 nearfield, 6 offshore). Note that no distinct regional clusters of samples are apparent

The monitoring program quantified phytoplankton community composition both near the surface and at a chlorophyll maximum depth (typically 10–15 m depth). Coding the samples by depth showed no distinct clustering pattern, indicating that overall there were no distinct surface or mid-depth phytoplankton communities within the range of stations sampled by the MWRA monitoring program (Fig. 18). Further partitioning of the entire phytoplankton data set into separate seasons and regions may highlight depth differences in the phytoplankton community, such as the mid-depth Ceratium-dominated community that was observed (especially during 1999–2001) in the nearfield and Cape Cod Bay regions during the summer months. However, relative to other sources of variation, there was no long-term, depth-related difference in phytoplankton community composition evident in the MDS analysis. Multivariate community analysis has indicated that seasonal and interannual variation in phytoplankton abundance and species composition was greater than the observed spatial (by region or by depth) variation in phytoplankton abundance and species composition.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-008-9125-9/MediaObjects/12237_2008_9125_Fig18_HTML.gif
Fig. 18

Multidimensional scaling plot for 1,002 phytoplankton samples collected during 1992–2006 coded by sample depth (upright green triangles surface, inverted blue triangles mid-depth). Note that solid blue inverted triangles overlay many of the green upright triangles, and no depth-dependent clusters of samples were evident

Phytoplankton Abundance: Prediversion Versus Postdiversion

Fifteen years (1992–2006), including approximately 9 years of “baseline” monitoring and approximately 6 years of “post” monitoring since the MWRA offshore outfall went online in September 2000, were analyzed for detection of postoffshore outfall diversion changes in phytoplankton abundance using a nonparametric Mann–Whitney U test (or Wilcoxon test). The number of observations in the preperiod and postperiod varied regionally and by variable. For phytoplankton, the nearfield was most frequently sampled with 223 phytoplankton observations prior to the offshore diversion and 224 observations after the September 2000 offshore diversion. The boundary and offshore regions had 35 (pre) and 37 (post) observations while Cape Cod Bay had 106 (pre) and 74 (post) observations. The coastal region and the harbor had 124 (pre) and 111 (post) observations.

These analyses indicated that total phytoplankton had statistically significant (P of Mann–Whitney U test <0.05) pre–post differences in three of the six regions monitored (Table 3). Nearfield total phytoplankton had slight postrelocation increases at both the surface and mid-depth. Surface total phytoplankton increased by 4% (from 1.4 to 1.5 million cells per liter) at the surface and increased by 8% (from 1.5 to 1.6 million cells per liter) at the mid-depth. The boundary station F27 also displayed a large increase of 65% in total phytoplankton, increasing from 1.0 (prediversion) to 1.6 million cells per liter (postdiversion). Cape Cod Bay was the only region displaying a decrease in total phytoplankton with a 23% decline in surface phytoplankton from 1.6 to 1.2 million cells per liter during the postdiversion period.
Table 3

Comparison of mean abundance levels (million cells per liter and cells per liter for 20-μm screened data for total dinoflagellates and Ceratium spp.) of various phytoplankton taxa during the prediversion and postdiversion

Area

Depth

Mean (pre)

Mean (post)

Percent difference

P value

Total phytoplankton

     

 Boundary

C

0.997

1.647

65

0.0114

 Cape Cod Bay

A

1.574

1.209

−23

0.0333

 Nearfield

A

1.446

1.511

4

0.0028

 Nearfield

C

1.5

1.624

8

0.0001

Total diatoms

     

 Cape Cod Bay

A

0.5657

0.1889

−67

<0.001

 Cape Cod Bay

C

0.5592

0.233

−58

<0.001

 Coast

A

0.7543

0.3889

−48

0.0003

 Coast

C

0.6574

0.3543

−46

0.0005

 Harbor

A

0.8755

0.541

−38

0.0009

 Harbor

C

0.7997

0.5699

−29

0.0003

 Nearfield

A

0.5158

0.3295

−36

0.0102

 Offshore

A

0.3866

0.0783

−80

0.0031

 Offshore

C

0.3747

0.0834

−78

0.0172

Total microflagellates

     

 Boundary

A

0.5447

0.6714

23

0.0431

 Boundary

C

0.4603

0.6936

51

0.0016

 Cape Cod Bay

A

0.6329

0.789

25

0.0184

 Cape Cod Bay

C

0.6923

0.8192

18

0.0003

 Coast

A

0.7469

0.8418

13

0.0001

 Coast

C

0.7095

0.7846

11

0.0004

 Harbor

A

0.9601

0.8555

−11

0.0016

 Harbor

C

0.9866

0.8667

−12

0.0005

 Nearfield

A

0.6567

0.8139

24

0.0001

 Nearfield

C

0.605

0.8128

34

0.0001

Phaeocystis

     

 Coastal

A

0.2086

0.316

51

0.0109

 Coastal

C

0.2508

0.4233

69

0.0015

 Harbor

A

0.2094

0.357

70

0.0098

 Harbor

C

0.3237

0.3675

14

0.0024

 Nearfield

A

0.1023

0.1778

74

0.0006

 Nearfield

C

0.2615

0.3136

20

0.0008

Total dinoflagellates

     

 Coast

C

873

450

−48

0.0215

 Nearfield

A

1,425

1,022

−28

0.0038

 Nearfield

C

1,934

1,045

−46

<0.0001

 Offshore

C

1,456

976

−33

0.0391

Ceratium spp.

     

 Coast

C

398

208

−48

0.0026

 Nearfield

A

908

440

−52

0.0001

 Nearfield

C

1,370

586

−57

0.0001

 Offshore

C

802

725

−10

0.0476

Pre and post means compared by Mann–Whitney U test. Only statistically significant differences (P ≤ 0.05) shown. Note that out of the 72 comparisons (6 groups × 6 areas × 2 depths) conducted 51% were significant at P ≤ 0.05 (37 of 72)

A surface, C subsurface chlorophyll maximum depth

Diatom abundance declined in all regions except the boundary region since September 2000 (Table 3). Declines of between 29% (Boston Harbor, mid-depth) and 80% (offshore, surface) were observed and no increases in total diatoms were detected for any region or depth. The greatest declines in diatoms were detected in the offshore region where total diatoms declined from ∼0.4 to 0.08 million cells per liter (80% decline) at both surface and mid-depth. Six diatom taxa (Chaetoceros spp., Thalassiosira spp., Thalassionema nitzschioides, Pseudo-nitzschia spp., Skeletonema spp., and D. fragilissimus) were also analyzed for changes since September 2000 (Table 4). Declines were detected for Chaetoceros spp., Thalassiosira spp., T. nitzschioides, and Pseudo-nitzschia spp.; increases were detected for D. fragilissimus; and both increases and declines were detected for Skeletonema spp. Chaetoceros declines were dramatic, on the order of 80% to 95%, and occurred in all regions except the boundary region. Declines in Thalassiosira spp., T. nitzschioides, and Pseudo-nitzschia spp. were limited to specific areas and depths.
Table 4

Comparison of mean abundance levels (cells per liter) of various diatom taxa during the prediversion and postdiversion

Area

Depth

Mean (pre)

Mean (post)

Percent difference

P value

Chaetoceros spp.

     

 Cape Cod Bay

A

135,000

19,800

−85

<0.0001

 Cape Cod Bay

C

157,8001

20,600

−87

<0.0001

 Coast

A

139,700

15,800

−89

<0.0001

 Coast

C

146,200

17,600

−88

<0.0001

 Harbor

A

217,600

16,300

−93

<0.0001

 Harbor

C

195,300

13,900

−93

<0.0001

 Nearfield

A

65,200

11,800

−82

<0.0001

 Nearfield

C

97,100

15,100

−84

0.025

 Offshore

A

61,300

5,800

−91

0.0104

 Offshore

C

107,100

10,100

−91

0.0126

Skeletonema costatum

     

 Cape Cod Bay

C

75,500

36,200

−52

0.0163

 Nearfield

C

23,300

47,200

103

0.0003

Thalassiosira spp.

     

 Cape Cod Bay

A

68,600

23,100

−66

0.0002

 Cape Cod Bay

C

64,600

29,200

−55

0.0081

 Coast

A

61,900

34,800

−44

<0.0001

 Coast

C

62,200

33,800

−46

<0.0001

 Harbor

A

89,400

37,600

−58

<0.0001

 Harbor

C

76,100

35,800

−53

<0.0001

Thalassionema nitzschioides

     

 Cape Cod Bay

A

17,100

4,800

−72

0.0075

 Cape Cod Bay

A

17,100

4,800

−72

0.0075

 Coast

A

15,200

10,300

−32

0.0013

 Nearfield

A

10,200

7,800

−24

0.0146

Dactyliosolen fragilissimus

     

 Boundary

C

700

1,400

100

0.042

 Coast

C

23,800

95,900

303

0.0131

 Harbor

A

51,500

154,500

200

0.008

 Harbor

C

32,300

149,700

363

0.0295

 Nearfield

A

39,800

106,800

168

0.011

 Nearfield

C

25,900

90,600

250

0.0125

Pseudo-nitzschia spp.

     

 Boundary

A

10,400

5,000

−52

0.0014

 Cape Cod Bay

A

16,300

8,900

−45

0.0295

 Nearfield

A

16,700

7,800

−53

0.0038

 Nearfield

C

13,400

8,800

−34

0.0046

Pre and post means compared by Mann–Whitney U test. Only statistically significant differences (P ≤ 0.05) shown. Note that out of the 72 comparisons (6 groups × 6 areas × 2 depths) conducted 44% were significant at P ≤ 0.05 (32 of 72)

A surface, C chlorophyll maximum depth

Skeletonema spp. displayed a mixed pattern with a doubling—an increase of 103%—observed in the nearfield mid-depth when comparing the prediversion and postdiversion abundance levels (Table 4). However, in Cape Cod Bay, a decline of ∼50% was detected at the mid-depth. No change in Skeletonema abundance was detected at any other region/depth combination. D. fragilissimus was the only diatom to show consistent increases since September 2000, likely due to a large summer bloom of this species in 2006. D. fragilissimus abundance increased significantly in the boundary, coastal, harbor, and nearfield regions. Postdiversion increases were large in these regions with a pre–post difference that doubled at boundary station F27 and increased approximately threefold in Boston Harbor (surface) and at both depths in the nearfield. The largest increases were detected in the mid-depth of the coastal region where mean D. fragilissimus abundance increased from ∼24,000 to 96,000 cells per liter (+300%) and at mid-depth in the harbor where abundance increased ∼360% from 32,000 to 150,000 cells per liter.

Microflagellates displayed increases in four regions (boundary, Cape Cod Bay, coastal, and nearfield) and declined in the harbor (Table 3). No change was detected at the offshore station F06. Only declines were detected in total dinoflagellates with declines of 28% to 48% detected in the coast, nearfield, and offshore regions. Much of this decline appears to be due to declines in Ceratium spp. which declined by ∼50% since September 2000 at both the surface and mid-depth in the nearfield region.

P. pouchetii abundance increased during the postdiversion period in three regions: coastal, harbor, and nearfield. These increases ranged from 14% (harbor mid-depth) to 74% (nearfield surface) and increases occurred at both depths in these three regions (Table 3). A separate analysis was run looking at monthly means rather than annual means for Phaeocystis across the areas/depths. For the monthly results, there were significant increases (P < 0.05) in April Phaeocystis abundance in all the areas and for the May surface water mean in the nearfield. The Phaeocystis increases noted in this study are consistent with the previously observed increase in the frequency of Phaeocystis blooms since 2000 and the recurring presence of Phaeocystis in the nearfield into May.

Discussion

The analyses presented in this study have identified the dominant phytoplankton, their seasonal abundance patterns, and fluctuations in their long-term abundance during 16 years (1992–2007) of monitoring. Several emergent trends and patterns are apparent. The long-term temporal scale (16 years) and the relatively large spatial scale (over 3,500 km2) of the MWRA monitoring program has enabled placing the question of local anthropogenic nutrient enrichment in the context of regional and long-term phytoplankton patterns. The monitoring has indicated that variability in phytoplankton abundance and species composition are generally regionally coherent with no regionally distinctive phytoplankton communities noted in the six regions monitored. For Massachusetts and Cape Cod Bays, the detected postoffshore outfall diversion changes in phytoplankton abundance levels and community composition, including a steady, long-term decline in diatom abundance and a post-2000 increase in Phaeocystis bloom frequency, appear to be part of a long-term regional trend. This suggests that, despite the detection of a localized (nearfield region) nutrient (NH4) signal following offshore outfall diversion, larger scale weather and physical processes with concomitant influence on operative biological processes are likely the most important drivers of the observed phytoplankton variability patterns.

Circulation, water properties, and consequently, much of the biology of Massachusetts and Cape Cod Bays are driven by the general pattern of water flow in the Gulf of Maine (Geyer et al. 1992, Jiang et al. 2007a) as modified by regional and local winds. A coastal current flows southwestward along the Maine and New Hampshire coasts where it may enter Massachusetts Bay at Cape Ann, north of Boston. This current drives an average counterclockwise circulation in Massachusetts Bay and (sometimes) Cape Cod Bay. Water flows out of the bays at Race Point, located at the tip of Cape Cod. Whether the coastal current enters Massachusetts Bay and whether it continues south into Cape Cod Bay depends on the strength of the current and the direction, duration, and speed of the wind. Because the coastal current is strongest during the spring period of high runoff from rivers and streams, the spring circulation pattern is more consistent than that of the summer and fall (Geyer et al. 1992, Jiang et al. 2007a). During the summer, stratification is established with alternating periods of upwelling and downwelling in various locations, depending primarily on the wind direction and strength (Lermusiaux 2001). Water flow and degree of stratification is variable, as the weather patterns change from week to week during summer (Besiktepe et al. 2003).

Variation in Massachusetts Bay phytoplankton abundance, particularly during the winter–spring bloom period, has been related to varying physical (Jiang et al. 2007a) and biological factors (Keller et al. 2001). However, these analyses largely applied a biogeochemical approach using chlorophyll as a proxy indicator of phytoplankton abundance. One strength of the MWRA monitoring program is the monitoring of phytoplankton to species level. Analysis of the 1992–2007 species level and phytoplankton functional group level data has yielded several insights that would have been obscured by analysis of bulk proxy indicators of phytoplankton abundance. A long-term diatom decline of ∼45%, increasing winter–spring Phaeocystis spp. bloom frequency, and cyclical dinoflagellate abundance patterns (lead by changing Ceratium spp. abundance) marked Massachusetts Bay phytoplankton during 1992–2007.

While prolonged water column stratification may limit annual phytoplankton primary production in some years in Massachusetts Bay (Oviatt et al. 2007), differential phytoplankton responses to the degree of seasonal stratification were evident at phytoplankton genus and functional group levels. The impact of interannual variation in water column stability was particularly evident for large Ceratium spp. in Massachusetts Bay phytoplankton during 1992–2007. Ceratium are most dominant in the Gulf of Maine (Bigelow 1924) and in other temperate coastal seas, during the summer stratified period (Cushing 1989). During this period, the large size, high respiration (relative to diatoms), and slow growth rate of the Ceratia may be offset by their strategy of vertical migration across the pycnocline (Holligan 1987, Cushing 1989). Water column stratification is necessary to utilize this strategy, and the degree of seasonal (=summer) stratification has been used to predict long-term variation in Ceratium abundance in the North Sea (Dickson et al. 1992). In Massachusetts Bay, there was a significant positive relation (r2 = +0.40) between degree of stratification and Ceratium abundance with May stratification explaining 40% of the variance in June Ceratium abundance. Ceratium’s relatively slow growth rate indicates that a prolonged period of favorable stratified condition may be required for large population accumulation to occur. This may be indicative of a dependence on the establishment of stratification in the late winter/early spring for the achievement of Ceratium population development in the spring/early summer. Long-term variation in the timing and magnitude of water column stratification may be partially driving the tenfold variation in Ceratium spp. abundance which varied from 200 cells per liter during the mid-1990s and 2003–2008 up to a peak of 2,000 cells per liter during 2000–2001.

Recent laboratory studies have also illustrated the physiological necessity of partially stratified water columns for Ceratium growth. For Ceratium tripos (common in Massachusetts Bay) grown in a laboratory simulation, turbulence of >0.05 cm2 s−3 (equivalent to that generated by a moderate gale in the upper 10 m of the sea) reduced Ceratium growth rate and stopped Ceratium swimming (Havskum et al. 2005). Furthermore, the same turbulence level did not reduce predation on C. tripos by a co-occurring mixotrophic dinoflagellate (Fragilidinium spp.). Thus, turbulence may not only destroy the stratified water column necessary for Ceratium’s vertical migration mode of nutrient and light acquisition, it also appears to have a detrimental effect on Ceratium growth at the cellular level. In Massachusetts Bay, this effect at the cellular level appears to have been propagated up to the population level. Butman et al. (2008) have classified the intensity of northeast storms in Massachusetts Bay during 1990 through 2006. In their analysis, the 2001–2002 period was marked by relatively few northeast winter–spring storms of moderate intensity, while the 1992–1993 and 2004–2005 period was characterized by frequent and intense northeast storms, particularly during the winter and spring (Butman et al. 2008). The relatively calm 2001–2002 period was marked by a mean Ceratium abundance (2,000 cells per liter) that was ten times greater than that recorded during the stormy periods of the early 1990s and 2000–2005 period. The positive effect of these storms on the transport of Alexandrium cells into Massachusetts Bay is well-documented (Anderson et al. 2005). These same spring storms appear to have also negatively impacted the annual establishment of water column stratification and subsequent seasonal development of the Ceratium population in Massachusetts Bay.

The mean phytoplankton primary production in Massachusetts Bay (mean of 220 g C m−2 year−1) during 1998–2005 was marked by little seasonality with approximately equal proportions of this production occurring in spring (March–May), summer (June–August), and autumn (September–November) periods (Hyde et al. 2008). A remarkably similar lack of seasonality in phytoplankton abundance, as estimated by cell counts is evident in these data with similar mean total phytoplankton abundances observed during spring (mean of 1.7 million cells per liter), summer (1.5 million cells per liter), and autumn (2.0 million cells per liter) in Massachusetts Bay during 1992–2007. Winter total phytoplankton abundance based on February observations and limited December observations was 0.6 million cells per liter. This relative consistency in total phytoplankton abundance was maintained by the numerical dominance (mean abundance of 0.7 million cells per liter) of small (<5 μm) unidentified phytoflagellates throughout the year. However, greater seasonal variation was observed within phytoplankton functional groups with diatoms varying about fourfold between their mean winter abundance level of 0.2 million cells per liter and their mean autumn abundance level of 0.8 million cells per liter. Similarly, dinoflagellates varied, on average, fourfold seasonally from 400 cells per liter during winter to 1,700 cells per liter during their annual abundance peak in summer. A pattern of a small phytoflagellate-dominated background superimposed by sporadic seasonal diatom and dinoflagellate blooms characterized Massachusetts Bay phytoplankton. Within this, two distinct seasonally varying phytoplankton communities were apparent: a winter diatom (Thalassiosira spp. and Chaetoceros spp.) and Phaeocystis-dominated community and a summer phytoplankton community dominated by dinoflagellates and a different diatom (D. fragilissimus, Skeletonema spp., Pseudo-nitzschia spp., and L. danicus) assemblage. It is expected that variation in the degree of physical forcing with concomitant effects on phytoplankton growth and loss processes is driving seasonal transitions between the winter assemblage and the summer assemblage.

A known anthropogenic modification was added to the physically dominated Massachusetts Bay system when the bay outfall was activated in September 2000. Comparison of phytoplankton abundance and species composition before and after the offshore diversion of effluent indicate several changes in phytoplankton during the MWRA monitoring period. While changes in total phytoplankton were generally not detected, and detected changes were not apparent in all regions, changes in some phytoplankton functional groups have been more dramatic and more widespread. Diatoms appear to have declined significantly across most regions with the key genera Chaetoceros spp. and Thalassiosira spp. showing declines in most regions. Only two diatoms showed postdiversion increases: D. fragilissimus and Skeletonema spp. D. fragilissimus abundance increased by twofold to fivefold in the boundary, coastal, harbor, and nearfield regions since September 2000. Skeletonema spp. abundance doubled in the nearfield region at mid-depth only. Of the postdiversion phytoplankton changes noted, the Skeletonema spp. increase at mid-depth in the nearfield is the change most likely to be related to offshore outfall diversion.

Dinoflagellates, led by the decline in Ceratium spp., declined in the coast, nearfield, and offshore regions. Simultaneous with this, microflagellates saw generally modest long-term increases in most regions, except in Boston Harbor where microflagellate abundance has declined by 10% since September 2000. April Phaeocystis abundance has increased over all of the regions postdiversion. The relatively minor changes in total phytoplankton abundance mask a series of apparent phytoplankton shifts embedded within the total phytoplankton community. In these, it appears that diatoms (with the exception of Skeletonema spp. and D. fragilissimus) and dinoflagellates have generally declined while microflagellates and P. pouchetii have had relative increases since the September 2000 offshore diversion.

Conclusions

This paper, originally presented as a working document for the recent American Geophysical Union Chapman Conference on Long-Time Series Observations in Coastal Ecosystems: Comparative Analyses of Phytoplankton Dynamics on Regional to Global Scales (Smetacek and Cloern 2008), characterizes the dominant phytoplankton seasonal and long-term trends of Massachusetts Bay phytoplankton during 1992–2007, a period marked by changing anthropogenic influence in this system. Long-term fluctuations in species-specific and phytoplankton functional group-specific abundance that would have been missed by proxy indicators of phytoplankton abundance were identified. Some of these fluctuations (i.e., the long-term variation in Ceratium abundance) appear to be linked to interannual variation in physical forcing. Others, like the long-term decline in diatom abundance and apparent increase in Phaeocystis abundance require further investigation.

The system-specific physical and biological attributes of a marine system act to modify the system-wide response to anthropogenic nutrient enrichment (Cloern 2001). In relatively open Massachusetts Bay, the offshore relocation of a submarine sewage outfall has resulted in an approximate doubling of ambient surface ammonium concentration during the well-mixed winter period. The phytoplankton observations during 1992–2007 indicate that this anthropogenic modification has not been of sufficient magnitude to override the prevailing physical and biological system attributes in the form of changed phytoplankton abundance or species composition patterns. Instead, observed long-term phytoplankton changes in Massachusetts Bay appear to be part of a larger, regional pattern that is largely controlled by oceanic variability.

Acknowledgments

We thank the field scientists from Battelle, MWRA, and MWRA HOM2 contract team for their dedication and commitment to excellence in conducting the MWRA water quality surveys in Massachusetts and Cape Cod Bays. They have performed with distinction under both ideal and challenging weather conditions. Without them, this paper would not be possible. Laboratory scientists from Battelle, the University of Rhode Island, and MWRA are also thanked for their conduct of the nutrient analysis under the HOM study series. Rocky Geyer’s input on the physics of Massachusetts Bay is gratefully acknowledged. One author (DB) acknowledges the support provided by the Environmental Protection Agency’s Science to Achieve Results (STAR) program (Grant RD83244301).

Copyright information

© Coastal and Estuarine Research Federation 2008