May 17, 1994 - West Palm Beach, Florida ... Analyses of water chemistry data from Lake Okeechobee, Florida, suggest .... described in Lim no-Tech (1993 ).
Arch. Hydrobioi./Suppl. 107 (Monographische Beirr;ige)
71-88
Stuugart, jJnuary 1995
Historical trends in the Lake Okeechobee ecosystem IV. Nitrogen: phosphorus ratios, cyanobacterial dominance, and nitrogen fixation potential By VAL H. SMITH, VICTOR}. BIERMAN, Jr., BRADLEY L. }ONES and KARL E. HAVENS Department of Research, South Florida Water Management District, West Palm Beach, Florida With 8 figures in the text
Abstract Analyses of water chemistry data from Lake Okeechobee, Florida, suggest that in-lake total nitrogen: total phosphorus (TN :TP) ratios have declined significantly during the past 20 years, and that these changes have been accompanied by a trend towards increasing nitrogen-limitation of phytoplankton growth in the lake. These analyses suggest that both the potential for cyanobacterial dominance and the potential for planktonic nitrogen fixation have increas ed over this interval, leading to an increased likelihood of nuisance cyanobacterial blooms. It is recommended that if additional external nutrient loading controls are considered and implemented, water quality management strategies that minimize P loading, maximize N: P loading ratios, and maximize in-lake TN :TP ratios, may help minimize the frequency and intensity of N-fixing cyanobacterial blooms in Lake Okeechobee.
Introduction Lake Okeechobee is a large eutrophic lake that serves as a vital water resource for south central florida. Although it is very shallow (1973-1992 average mean depth 2.7 meters), the surface area averages 1730 km 2, and it has an average hydraulic residence time of 1.2 years. The last 20 years have included both very wet and extremely dry periods, which caused the surface area of the lake to vary by as much as 50 percent. This variation in lake level appears to have influenced horizontal mixing and phosphorus dynamics within the lake, as well as to have impacted the littoral vegetation, wading bird populations, and other aquatic fauna (RICHARDSON & I-IAMOUDA 1995, SMITH et al. 1995, WARREN et al. 1995). The water quality in Lake Okeechobee also is strongly influenced by drainage from its agricultural watershed (FLAIG & HAVENS 1995). Dairy farms and beef cattle operations are concentrated north of the lake, and uncontrolled runoff from these 0341-2881/95/0107 -0071 $4.50 {) l'J'JS E. SchwcizerbJn'sch(: VerbgsbuddunJiung, 0-70176 StuttgJ. n
l
72
areas is phosphorus-rich. To the south is the Everglades Agricultural Area (EAA), where the dominant crop is sugar cane grown on organic muck soil; runoff from these fields is high in nitrogen and moderately high in phosphorus (FLAIG & HAVENS 1995). However, most of the nutrient-rich .drainage from the EAA has been diverted away from the lake since 1979 (AUMEN 1995). The water quality in Lake Okeechobee also has apparently deteriorated during the past 20 years, as indicated by rising total phosphorus concentrations in the
.L;v'
~
..,N,.... IS
""'
water column (CANFIELD & HOYER 1988, jANUS et a!. 1990, jAMES et a!. 1995b). For lakes in general, one of the most consistent effects of increasing nutrient availability is an increase in the frequency and intensity of summer blooms of cyanobacteria {PICK & LEAN 1987, REYNOLDS 1987). Such blooms have been associated worldwide with nuisance water quality conditions such as surface scums, filter clogging, taste and odor problems, and summer fish kills (HORNE 1979, GREGOR & RAST 1981, REYNOLDS 1987). In addition, cyanobacteria also can produce potent toxins that have resulted in the poisoning of livestock and domestic animals (CARMICHAEL 1981 , PERSSON eta!. 1984). Furthermore, because cyano-
73
Nitrogen: phosphorus ratios
V.H. SMITH, Y.J. BIERMAN, Jr., B.L. jONES and K .E. HAVENS
a LOOJ
Lake •' .,oobkeechobee
00 '
. l006
bacteria are generally thought to be a less desirable food source for zooplankton than other taxa (HANEY 1987, KOHL & LAMPERT 1991 ), it has been suggested that their d o minance in the phytoplankton can restrict the flow of energy and carbon from primary producers to consumers, and potentially reduce yields of ecologically and commercially valuabl e fish species (NICHOLLS eta!. 1986). The potential for nuisance cyanobacterial blooms, particularly those of buoyant nitrogen-fixing species, has been defined as a key issue in the water quality management of Lake Okeechobee {MACEINA & SOBALLE 1989a, 1989b, AUMEN & GRAY 1995 ). A decline in the lakewater total nitrogen: total phosphorus ratio UANUS et a!. 1990, ]AMES et al. 1995b), and evidence for secondary nitrogen limitation of phytoplankton growth (HAVENS 1994 a, b), together have suggested that the potential for nuisance cyanobacterial growth currently is increasing. HAVENS et a!. {1995) confirmed the presence of more frequent and more intense blooms of nitrogen-fixing cyanobacteria in Lake Okeechobee in recent years, and the relative biomass of cyanobacteria in the phytoplankton now greatly exceeds that previously measured {CICHRA et a!. 1995). In this paper, a 20-year record of Lake Okeechobee water quality data is analyzed to evaluate historical trends in nitrogen: phosphorus ratios in the lake, and the potential role of nutrients in regulating cyanobacterial dominance and rates of nitrogen fixation. A quantitative nutrient-based index for nitrogen fixation p\)tential in lakes is proposed, and historical trends in this index for Lake Okeechobee are quantified.
Materials and methods Lake Okeechobee (Fig. I) has 32 inflow and outflow channels, most of which have been regularly monitored by the South Florida Water Management District (SFWMD) at 2-4 week intervals since 1973. Daily discharge data are available for
Fig. I. Map of Lake Okeechobee, Florida (U.S.A.), showing the eight sampling locations.
most inflow and outflow points, allowing the construction of comprehensive nutrient budgets. Physical, chemical, and biological parameters also have been sampled at eight limnetic sites at 2-4 week intervals since 1973. The water chemistry samples reported here were analyzed for total Kjeldahl nitrogen (TKN, mg N L-1), ammonium nitrogen (NH4-N, mg N L-I), nitrate + nitrite-nitrogen (N02 + NOJ-N, mg N L- 1), total phosphorus (TP, mg P L - 1), and soluble reactive phosphorus (SRP, mg P L- 1). Total nitrogen (TN, mg N L- 1), and dissolved inorganic nitrogen (DIN, mg N L- I) were calculated from the measurements of nitrate+nitrite, ammonium, and total Kjeldahl Nitrogen. In addition, measurements also were made of Secchi disk transparency These data were checked both for accuracy and reliability (SFWMD 1992), and stored in the SFWMD water quality database WQDORA. Data collection methods and nutrient budget calculations are described further in JAMES et a!. (1995 ). All data manipulations and statistical analyses were performed using SYSTAT (WILKINSON 1989). The principal dataset analyzed in this study was comprised of seasonallyaveraged values for each of the eight limnetic stations (see below), and estimates of total nitrogen and total phosphorus loading from each year of record UAMES et al. 1995a). Unless indicated otherwise, all analyses were conducted using seasonal (May- November) mean values from a unified seasonal water quality data file as
//
74
Y.H. SMITH,
V.J.
Nitrogen : phosphorus ratios
BIERMAN, Jr., B.L. J ONES and K.E. HAVENS
described in Lim no-Tech (1993 ). This time interval was chosen for analysis because
75
40y-------------------------------------------,
historically this had been considered to be the period in which cyanobacterial blooms had most frequently occurred. In some cases these seasonal means were based on fewer than seven monthly averages due to missing nitrogen and I or
0
~
a: a... 1:-:
phosphorus data.
35
30 25
z
Results and discussion Trends in nutrient limitation in Lake Okeechobee Nutrient ratios have been used as an objective criterion for inferring nutrient limitation of freshwater phytoplankton growth at least since PEARSALL (1932), who used inorganic N: P ratios as an indic ator of N - versus P-limitation of algal growth in lakes of the E nglish Lake District. More detailed quantitative criteria for the assessme nt of N- and P-limitation in lakes were later developed by SAKAMOTO (1966), w ho proposed that nitrogen limitation of algal biomass in Japanese lakes
f-
a: w
~
::.:::
:s
20 15 10+-.-------------------------------------~
5" N-limitation 0+----r---.----~---r---.----~---r---.----~--~
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
YEAR
was most likely when the epilimnetic total nitrogen: total phosphorus (TN :TP) ratio was < 10: I by weight. He concluded that algal biomass was limited by
Fig. 2. Time trends in water column TN: TP mass ratios.
phosphorus when the epilimnetic TN :TP ratio was > 17, and that either N or P may be lim iting at intermediate TN :TP ratios ; these criteria were subsequently
by P. Nitrogen and N +P co-limitation were most common at two western and one
confirmed independently with algal bioassays by FORSBERG & RYDING (1980). SMITH (1982) later demonstrated that the effects of TN on algal biomass decreased
southern stations, and the frequency of N -limitation increased in each of the three years of their study. The analyses of HAVENS (1994a, b) and ALDRIDGE et al. (1995)
progressively as the TN :TP ratio increased in temperate zo ne lakes, and also
thus independently support the changes in nutrient limitation that are suggested by
developed a multiple regression model for Florida lakes. Similar multiple regression
the trends in TN :TP ratios shown in Fig. 2.
models for chlorophyll a were also developed independently by RYDING (1981), and for Florida lakes by CANFIELD (1983). In the case of Lake Okeechobee, temporal and spatial patterns of water column nitrogen: phosphorus ratios suggest that both N and P may potentially be
Nutrient criteria for cyanobacterial bloom potential assessment One objective approach for assessing the potential for cyanobacterial proliferation in lakes is to define quantitative criteria which identify the conditions under
limiting at different times and locations within the lake (SCHELSKE 1989,
which nuisance blue-green blooms may be expected to occur. Cyanobacteria in
ALDRIDGE et al. 1995). However, based on the TN :TP ratio c~iteria of SAKAMOTO
ge neral appear to ha ve a strong competitive advantage und er conditions of nitrogen-
(1966) and FORSBERG & RYDING (1980), historical trends in water column TN:TP
limitation (TILMAN 1982, SMITH 1983 a, b), and if sufficient phosphorus is available,
rat ios suggest that phytopl an kton growth in Lake Okeechobee may have changed
nitrogen-fixing cyanobacteria can proliferate even in the absence of sources of
from being primarily phosphorus-limited before 1982, to being jointly N- and
dissolved inorgani c nitrogen (HORNE 1979). Nitrogen: phosphorus ratios correlate
P- limited afterwards (Fig. 2). Although the results of early bioassays are somewhat eq uivocal (SCHELSKE
MCCAULEY 1992). Ecological resource-ratio theory (TILMAN 1982, TILMAN et al.
well with the likelih ood of N-limitation (FORSBERG & RYDING 1980, DOWNING &
1989), analyses of nutrient limitation by BREZONIK et al. (1979) suggested
1982) suggests that cyanobacteria would be expected
cons iderab le P-limitation of phytoplankton grow th in Lake Okeechobee in the
plankton under conditions of lowTN :TP load ing _ratios (SCHINDLER 1977, FLE"IT
1970s. In contrast, comparative analyses of N- and P-based trophic state indices have suggested a lake-wide transition from P-limitation to N-limitation between
et al. 1980), and of low TN: TP ratios in the water column (SMITH 1983 a, b, 1986, 1990, 1992).
to
dominate in the phyto-
the 1970s and 1980s (HAVENS 1994 b). Moreover, recent expe rimental nutrient
Because non-nutrient factors also may influencecyanobacterial dominance in
enrichment bioassays performed in 1990-1992 by ALDRIDGE et al. (1995) have
Florida lakes (CANFIELD et al. 1989), the empirical model devel oped by SMITII
suggested that when nutrient limitation occurred, phytopl ankton growth at five
(1986) was used to predict the proportion of cyanobacteria (by biovolume) in Lake Okeechobee:
pelagic locations in the lake during this time period was limited by N, rather than
76
VH. SM ITH, V.J. BIERMAN, Jr., B.L. J ONES and K.E. HAVENS
(I) legit(% BG) = 2.358-1.297 log TN+ 0.692 log TP- 2.058 log SD + 0.538 log Zm, where legit (% BG) is the legit- transformed relative biomass of cyanobacteria in the phytoplankton . SMITH (1986) concluded from his analysis that in addition to TN and TP concentrations, light avail ability in the water column as estimated from Secchi disk transparency (SD) and the mixed depth (Zm), strongly influences the relative biomass of cyanobacteria (% BG) in lakes. Annual eight-station mean values of TN, TP1 and SD from each year of record were used to pred ict % BG from Eq. I, and because Lake Okeechobee is well mixed, the ann ual average value for mean depth was used as an estimate of the thickness of the mixed layer. The predictions of Eq. I were then compared to actual measurements of cyanobacterial relative biomass measured during 1974 at the eight SFWMD sampling stations (MARSHALL 1977; cf. fig. 9 and 10 in SCHELSKE 1989), and to recent measurements made from 1988-1990 by CICHRA et al. (1995) at sites located near seven of the eight SFWMD stations. Because of the large spatial and interannual variation that exists in Lake Okeechobee phytop lankton species composition (CICHRA et al. 1995), no attempt to predict detailed annual trends in % BG from Eq. I was made. Rather, focus was on the question whether Eq. 1 would predict a general historica l trend towards increasing cyanobacterial dominance in the lake. This empirical model indeed predicted a potential 50%
N itrogen : phosphorus rat ios -
601
('J
50
0
en 0
Fig. 1 and 2). Although the above resu!ts have confirmed a historica l increase in overall cyanobacterial dominance, it is blooms of buoyant, nitrogen-fixing cyanobacteria which have been identified as a major management issue in Lake Okeechobee (MACEINA & S0BALLE 1989a, b). In order to address this more specific concern, we sought to develop new graphical criteria for the potential dominance of N-fixing cyanobacteria that cou ld be applied to the lake. SMITH (1983 a) defined a TN :TP ratio criterion for the dominance of both N-fixing and non -N-fixing cyanobacteria, and a similar analysis was made here using his data for N-fixing cyanobacteria alone. When replotted, the data of SMITH (1985) showed an apparent sharp boundary at a TN :TP ratio of ca. 22: I by mass, at or below which dominance by N -fixers occurs (Fig. 3). It should also be noted that the data shown in Fig. 3 were growing season (typically May-September) means, because in this ana lys is we were primari ly interested in long-term trends, and not in instantaneous dai ly phenomena. It is clear from previous analyses of phytoplankton data from individual samp ling dates (e.g. CANFIELD eta!. 1989) that such plots exhibit much greater scatter than do p lots of seasonal or annual means; such variance stems in large part from day-to-day climatic and/ or hydrometeorologica l variat ions which
•
X
u::
suggests that nutrient conditions favorable for blooms of N-fixing cyanobacteria
~
indeed have increased in recent years in Lake Okeechobee, and it is important to note that this marked increase in favorability also corresponds closely to the apparent step-change in ecological state observed by jAMES et al. (1995) in their
0)
0
-4~>--~----~--~----~--~--~~a-4 0 200 400 600 800 1000 1200 1400
analysis of overall trends in water quality in Lake Okeechobee.
TOTAL PHOSPHORUS, mg m-3 Nutrient criteria for the assessment of nitrogen-fixation potential Measurements made by BREZONIK et al. (1979) suggest that areal rates of nitrogen fixation in Lake Okeechobee in the late 1970s were modest when compared with similar data from other north temperate, tropical and sub-tropical lakes. Their early estimates of annual nitrogen fixation are relatively low, even when compared with other lakes having similar TP concentrations (Fig. 5). However, the decreases in water column TN :TP ratios shown in Fig. 2, and the apparent increase in favorability for N-fixing cyanobacteria suggested by Fig. 4,
Fig. 5. Relationship between annual areal rates of nitrogen fixation and water column total phosphorus concentrations in temperate lakes (open squares}, tropical lakes (closed squares}, and Lake Okeechobee (closed triangle} (data from SMITH 1990 and BREZONIK et al. 1979).
I.
SRP concentrations of 2: 0.0 I 0 mg P L -I, and
2.
DIN (NH4+NOx) concentrations of .,;0.100 mg N L- 1•
together indicate that the nitrogen-fixation potential of phytoplankton in Lake
When values of SRP and DIN from the Lake Okeechobee water chemistry database indicated that both of the above nutrient conditions were met, a dummy
Okeechobee may have increased significantly in recent years. This hypothesis was tested empirically using quantitative nutrient criteria derived from the Lake
variable coding for N-fixation potential (NDUMMY) was created and assigned a value of unity; otherwise, NDUMMY was set to zero. It is important to note that
Okeechobee water chemistry database.
the joint occurrence of these two conditions also corresponds to a average DIN:
In developing objective nutrient criteria for the determination of nitrogen fixation potential, it is important to consider the actual concentrations of nitrogen
SRP ratio < 10: I by mass, and that the NDUMMY criterion defines only the
and phosphorus in addition to their ratios (PAERL 1988). HORNE & COMMINS
consider that NDUMMY provides information concerning the nutrient potential for nitrogen fixation, and recognize explicitly that this potential may not actually
(1987), for example, have explicitly proposed that any criteria for the assessment of N-fixation potential should include information about the availability of both inorganic nitrogen and phosphorus. They suggested that when soluble reactive p)10sphorus (SRP) concentrations exceed cellular requirements, then concentrations of dissolved inorganic nitrogen (DIN) potentially may regulate rates of N-fixation; under conditions of phosphorus sufficiency, nitrogen fixation was expected to occur when DIN concentrations consistently fall below 0.05-0.1 mg L -I. Based on the suggested criteria of HORNE & COMMINS (1987), we concluded that under optimal temperature, light intensity, and hydrometeorological conditions, planktonic nitrogen fixation in lakes is most likely to occur when the following nutrient conditions are met simultaneously in the water column:
nutrient conditions which are thought to be favorable for N-fixation. We thus
be realized at a given site when other physical, chemical, or ecological conditions inhibit or prevent this biogeochemical process from occurring (HOWARTH et al. 1988 a, b, PAERL 1988). However, prior to the use of NDUMMY in assessing historical trends in nitrogen fixation potential in the lake, it was necessary to verify whether values of NDUMMY =I did indeed correspond to nutrient conditions favorable to enhanced nitrogen fixation in the water column. We thus compared predicted seasonal trends in nitrogen fixation potential based on the NDUMMY criterion with actual nitrogen fixation measurements with data collected from PHLIPS & IHNAT (1995).
80
VH. SMITH, V.J. BIERMAN, Jr., B.L JONES and K.E. HAVENS
Nitrogen : phosphorus ratios
Concentrations of DIN and SRP from the SFWMD water chemistry database were used initially to calculate daily values of NDUMMY during 1989-1991 for all stations except L003, where no corresponding measurements of acetylene reduction were made by PHLIPS & IHNAT (1995). These data predicted a pronounced seasonal pattern in the nutrient potential for nitrogen fixation, with maximal rates occurring in the fall, and minimal rates occurring in the spring (data not shown). There was a general agreement of the predicted and observed results, with the NDUMMY-predicted pattern (Fall > Summer > Spring > Winter) corresponding closely to the actual observed trends (Fall = Summer > Spring > Winter: PHLIPS &
90
II
70
:J
60
0
z fz w
IHNAT 1994 ). The best correspondence between predicted and measured trends
0
w
fixation potential were examined. In this analysis, the greatest weight was given to conditions of chronic nitrogen stress, rather than brief episodes of N-deficiency which may persist for too short a time for physiological adaptation or for phytoplankton species shifts to occur. Thus May-November averages of DIN and SRP were used to calculate a single seasonal value for NDUMMY at each of the eight permanent SFWMD sampling stations for each year of record . A plot of these data (Fig. 6) suggests that the frequency of nutrient conditions favorable for nitrogen fixation has indeed increased over the past 20 years (Kendall's tau, p < 0.01 ). In particular, between 1973-1981, the frequency of NDUMMY = 1 was less than 50% for eight of the nine years for which data are available. In contrast, after 1981 the frequency of favorable conditions was~ 50%
80
>-
:2: :2:
also was found to occur at stations far removed from the central mud zone of the lake basin, where light limitation from wind-resuspended sediments frequently prevents the nutrient-driven nitrogen fixation potential from being fully realized. Following validation of the NDUMMY criterion, historical trends in nitrogen
81
100
50 40
a:
30
a...
20 10 0· 1972
1974
1976
1976
1980
1962
1964
1966
1966
YEAR
1990
1992
Fig. 6. Frequency of nurriem conditions favorable for nitrogen fixation in Lake Okeechobee, as inferred from NDUMMY (see text).
apparent step-reduction in lakewide TN:TP ratio that occurred after 1981 0AMES
The relationship between NDUMMY and water column TN :TP ratios at all stations for all years of record was compared next. When both nutrient criteria were met (NDUMMY =1), the .m edian water column TN :TP ratio was 22:1 by mass, a value that is remarkably consistent with the in-lake TN :TP S: 22: I criterion developed independently in Fig. J from the data of SMITH (1985). In contrast, when the nutrient criteria were not met (NDUMMY =0), the median water column TN:
eta!. 1995). A further analysis of the time trends for NDUMMY at each of the eight
TP ratio was higher (JJ: I by mass), and a Wilcoxon 2-sample test revea led a highly significant differen ce (p = 0.0001) between the two sets of data.
individual SFWMD stations for all years (data not shown) revealed that each station has exhibited some years where both criteria were met (NDUMMY =1),
Possible mechanisms involved in declining in-lake N: P ratios
for nine of the eleven years analyzed. These results also are consistent with the
and that in almost every year more than one samp ling station met these nutrientbased criteria for N-fixation potential. Favorable conditions for N-fixation were most common at stations near the northern end, and least common at stations near
The evidence summarized above suggests that recent decreases in water column TN :TP ratios in Lake Okeechobee have been accompanied by significant increases both in overall cyanobacterial dominance and in planktonic N-fix ation
the southern end. This predicted spatial pattern of N -fixation potential is
potential. The mechanisms invo lved in the observed decline of lakewater TN: TP
consistent with the spatial distribution of nutrient loadings, in which phosphorus loadings are proportionately higher in the north (FLAIG & HAVENS 1995, JAMES et a!. 1995 ). Thus, despite the fact that non-nutrient factors such as inorganic
ratios are not yet clear, however. Although a strong positive correlation between
turbidity or mixing may inhibit or prevent N-fixation in the water column at varying dates and locations, the NDUMMY analysis provides an objective, nutrient-based estimate of the long-term patterns in N-fixation potential in the water column.
in-lake TN :TP ratios and TN :TP loading ratios can be seen in other warm-water lakes (Fig. 7), no corresponding correlation between in-lake and input TN :TP ratios was found in Lake Okeechobee itself (Fig. 4). The data in Fig. 4 thus suggest that recent reductions in water column TN :TP ratios in Lake Okeechobee have been driven largely by in -lake processes. In order to make a preliminary examination of internal processing, the difference between in - lake and loading TN :TP ratios was calcu lated, and hi storica l
VH. SMITH, V.J. BIERMAN, Jr., B.L. }ONES and K.E. HAVENS
82
en ---en
1000~--------------------------~
al
E
1:1
>. .D -........
0 f-
•
Tropical 0
Florida
... Okeechobee
• 100