|
|
|
System-wide Monitoring Program
Synthesis of the Water Quality Data
RESULTS
Sampling Interval
A comparison in daily mean, minimum and maximum DO (% saturation) was made between 30-min and 4-h sub-sampled intervals during deployments in July-August 1997 and 1998 at Big Bay Creek and St. Pierre Creek in the ACE Basin NERR. A two-sample t-test indicated no significant difference in mean DO (% saturation) between 30-min and 4-h sub-samples (Table 5). Minimum DO (% saturation) was significantly greater during 4-h sampling intervals at Big Bay Creek in July-August 1998, while maximum DO (% saturation) was significantly greater during 30-min. sampling intervals at St. Pierre Creek in 1997.
Table 5. Results of t-tests to determine significant differences in mean, minimum and maximum DO (% saturation) between 30 min. and 4 h sampling intervals. Analyses are based on homogeneous variances.
| Big Bay Creek |
30 min. |
4 h |
p |
| 1997 |
|
|
|
| mean |
68.86 |
69.07 |
0.94 |
| min. |
36.81 |
46.02 |
0.13 |
| max. |
93.76 |
91.25 |
0.32 |
| 1998 |
|
|
|
| mean |
60.47 |
61.48 |
0.78 |
| min. |
25.72 |
36.56 |
0.03* |
| max. |
80.83 |
81.45 |
0.89 |
| St. Pierre Creek |
30 min. |
4 h |
p |
| 1997 |
|
|
|
| mean |
81.52 |
81.32 |
|
| min. |
40.15 |
48.43 |
|
| max. |
110.04 |
105.56 |
0.01* |
| 1998 |
|
|
|
| mean |
85.73 |
87.38 |
0.69 |
| min. |
53.98 |
59.17 |
0.35 |
| max. |
104.84 |
103.6 |
0.82 |
| * significant at p=0.05 |
|
|
|
|
|
|
|
Effect of deployment period on dissolved oxygen (% saturation)
Hypoxia (% of time DO < 28% saturation) and supersaturation (% of time DO > 120% saturation) were compared among 1, 2, 4, 7 and 14-day deployments at each site during July-August 1997 and 1998. Graphical examination of dissolved oxygen over time consistently revealed a marked increase in hypoxia with increasing deployment duration at 11 sites (Chesapeake Bay VA (Taskinas Creek); Narragansett Bay (T-wharf); Old Woman Creek (State Route 2); Tijuana River (Tidal Linkage); Weeks Bay; Waquoit Bay (Central Basin); Hudson River (Tivoli South); North Carolina (Zeke’s Island); Jobos Bay (Station 9); and both Sapelo Island sites). This suggested decay in dissolved oxygen readings over time due to instrument drift. Independent data to validate this trend were available for relatively few sites. Therefore, regression analysis was used to determine whether a relationship existed between hypoxia and time intervals post-deployment. Regression analysis indicated a significant quadratic relationship between percent of time that DO < 28% saturation and deployment duration for both years (Figure 1). Analysis of variance indicated a significant difference in average hypoxia among deployment duration intervals at sites measuring dissolved oxygen in summer 1997 (df= 4/170, f=0.946, p=0.008) and 1998 (df= 4/180, f=0.186, p=0.02).
Figure 1. Regression of hypoxia (% of time) versus deployment duration.
When the effect of deployment duration on percent of time that DO was supersaturated (>120% saturation) was examined among reserve sites, most sites showed a decrease in supersaturation over deployment duration in one or both years (1997 and 1998). Sites showing decreases in supersaturation over time included Wells (Inlet site), North Inlet-Winyah Bay (Oyster Landing), Elkhorn Slough (Azevedo Pond), Weeks Bay, Apalachicola (East Bay Bottom), Jobos Bay (Station 9), Tijuana River (Tidal Linkage), ACE Basin (Big Bay Creek), and Sapelo Island (Flume Dock). Sites showing an increase in supersaturation with deployment duration included Narragansett Bay (Potter’s Cove), Great Bay, Padilla Bay (Bayview Channel), Delaware Bay (Blackbird Landing), and Hudson River (Tivoli South Bay). Regression analysis indicated a significant relationship between percent of time with supersaturation conditions and deployment duration for both 1997 and 1998 (Figure 2). Analysis of variance indicated no significant difference in mean frequency of supersaturation among deployment periods in July-August 1997 (df = 4/170, f = 0.14, p= 0.967) or 1998 (df= 4/180, f= 0.49, p= 0.737). The examination of percent of time variables in relation to deployment duration at individual sites and in the regression analysis for all sites combined suggests a bias is introduced by instrument drift for percent of time variables. In order to minimize bias due to drift, further analyses using percent of time DO variables included data up to 48 h post-deployment only.
Figure 2. Regression of supersaturation (% of time) versus deployment duration.
Descriptive analyses by reserve site
Water depth (m), water temperature (°C), salinity (ppt), and dissolved oxygen (mg/L and % saturation) data collected between January 1996 and December 1998 were analyzed individually for 44 sites in the National Estuarine Research Reserve’s System-Wide Monitoring Program (Figure 3). In the site summaries that follow, the percent of data for each water quality variable included in analyses are stated accordingly. Percentages were calculated as the total number of actual observations relative to the total number of possible observations (48 observations/day x 365(6) days/year = 52,608 observations possible). Reasons for exclusion of data from analyses included deletion of erroneous data and no data collection.
Spatial and temporal data collection was greatest along the Southeast coast, with full coverage from January 1996 to December 1998 at these reserves (Table 6). Along the West Coast, the Mid-Atlantic Coast and the Gulf of Mexico and Caribbean, 75% of sites collected data in all seasons and almost all months examined. The Northeast Coast and Great Lakes reserves typically only collected data between spring and fall; however, some winter data were collected at sites in the Narragansett Bay, RI, and the Wells, ME, reserves.
Descriptive statistics (mean, min, max, and frequency of occurrence) were used to characterize sites with respect to water quality parameters. Minimum and maximum values observed for each parameter at each site between 1996-1998 are summarized in Table 7. Negative water depths were observed in all five geographic regions, while water depths greater than 5 m were only observed in the Northeast. Water temperature ranged from below 0°C (all regions except for the Gulf of Mexico and Caribbean Sea) to 45.9°C (ACE Basin-St. Pierre Creek) in the Southeast. Minimum and maximum salinity varied from 0 to 53.8 ppt and dissolved oxygen varied from 0-35.2 mg/l and 0-489.1 % saturation. Because temperature, salinity, and pressure affect dissolved oxygen (mg/L) concentration, and in order to compare dissolved oxygen between sites and reserves using the same scale, only percent saturation of dissolved oxygen was discussed in the site summaries.
Deployment-level harmonic regression analysis models were fit separately for each of five water quality variables at each of the 44 sites (Table 8). Mean sums of squares for each of the three model components (12.42 hour cycles, 24 hour cycles, and interaction between these cycles) were computed for each water quality variable at each of the 44 sites (Table 9). Percent of variance attributed to each of these three model components for a given water quality variable is presented for each variable in the site summaries; however, because tidal periods may differ among sites, interpretation of the variance attributed to these cycles is best left to personnel at each site.
Respiration, production, and net ecosystem metabolism were computed for 27 sites from all five geographic regions included in this study (Table 10). All but three sites (Great Bay Buoy; Chesapeake Bay VA-Goodwin Island; and Waquoit Bay-Central Basin) were determined to be net heterotrophic; however, sites exhibited considerable seasonal variability with regards to heterotrophic vs. autotrophic condition. Individual metabolism calculations, as well as a graphic showing temporal tendency towards heterotrophic vs. autotrophic conditions, appear with each site summary
Table 6. Temporal and spatial distribution of water quality sampling effort (months) at NERR sites, 1996-1998.
| Site |
1996 |
1997 |
1998 |
Total |
Seasons |
| ELKAP |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| ELKSM |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| PDBBY |
12 |
11 |
12 |
35 |
wi, sp, su, fa |
| PDBJL |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| SOSSE |
9 |
8 |
2 |
19 |
wi, sp, su, fa |
| SOSWI |
8 |
7 |
7 |
22 |
sp, su, fa |
| TJROS |
12 |
11 |
12 |
35 |
wi, sp, su, fa |
| TJRTL |
0 |
8 |
12 |
20 |
wi, sp, su, fa |
|
|
|
|
|
|
| GRBGB |
8 |
8 |
8 |
24 |
sp, su, fa |
| GRBSQ |
0 |
5 |
8 |
13 |
sp, su, fa |
| HUDSK |
4 |
9 |
9 |
22 |
sp, su, fa |
| HUDTS |
9 |
8 |
9 |
26 |
sp, su, fa |
| NARPC |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| NARTW |
4 |
12 |
0 |
16 |
wi, sp, su, fa |
| OWCSU |
7 |
7 |
7 |
21 |
sp, su |
| OWCWM |
7 |
7 |
7 |
21 |
sp, su |
| WELHT |
9 |
10 |
9 |
28 |
sp, su, fa |
| WELIN |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| WQBCB |
3 |
8 |
5 |
16 |
sp, su, fa |
| WQBMP |
0 |
0 |
2 |
2 |
fa |
|
|
|
|
|
|
| CBMJB |
6 |
5 |
4 |
15 |
sp, su |
| CBMPR |
6 |
5 |
4 |
<
15 |
sp, su |
| CBVGI |
0 |
3 |
12 |
15 |
wi, sp, su, fa |
| CBVTC |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| DELBL |
11 |
12 |
12 |
35 |
wi, sp, su, fa |
| DELSL |
11 |
12 |
12 |
35 |
wi, sp, su, fa |
| MULB6 |
5 |
12 |
12 |
29 |
wi, sp, su, fa |
| MULBA |
3 |
12 |
12 |
27 |
wi, sp, su, fa |
|
|
|
|
|
|
| ACEBB |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| ACESP |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| NIWOL |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| NIWTA |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| NOCMS |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| NOCZI |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| SAPFD |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| SAPML |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
|
|
|
|
|
|
| APAEB |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| APAES |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| JOB09 |
11 |
8 |
7 |
26 |
wi, sp, su, fa |
| JOB10 |
11 |
5 |
6 |
22 |
sp, su, fa |
| RKBBR |
0 |
0 |
11 |
11 |
sp, su, fa |
| RKBUH |
2 |
12 |
12 |
26 |
wi, sp, su, fa |
| WKBWB |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
| WKBFR |
12 |
12 |
12 |
36 |
wi, sp, su, fa |
Table 7. Minimum and maximum values for water quality variables, 1996-1998.
Site
|
Depth
|
Temp.
(°C) |
Salinity
(ppt) |
DO
(%sat) |
DO
(mg/L) |
| ELKAP |
-.1 / 1.1 |
-1.3 / 35.6 |
0.5 / 39.2 |
0/403.6 |
0 / 25.7 |
| ELKSM |
0 / 2.8 |
5.5 / 24.4 |
8.3 / 39.2 |
0/234.6 |
0 / 20.5 |
| PDBBY |
-.1 / 4.3 |
-1.5 / 23.5 |
19.2 / 32.2 |
12.6/225.5 |
1.2 /19.0 |
| PDBJL |
-.2 / 2.0 |
-0.6 / 28.2 |
0 / 28.0 |
0/194.9 |
0 / 15.2 |
| SOSSE |
0 / 2.3 |
0 / 27.0 |
0 / 31.3 |
0/230.2 |
0 / 19.2 |
| SOSWI |
0 / 3.3 |
1.2 / 25.0 |
0 / 32.8 |
10/149 |
0.9 /13.6 |
| TJROS |
.1 / 2.04 |
7.5 / 30.3 |
0.3 / 40.2 |
0/449.6 |
0 / 32.8 |
| TJRTL |
-.1 / 1.7 |
7.5 / 32.1 |
0.4 / 35.2 |
0.2/462.0 |
0 / 32.0 |
|
|
|
|
|
|
| GRBGB |
2.6 / 6.1 |
2.2 / 25.3 |
1.3 / 30.9 |
45.1/168.1 |
3.3 /14.5 |
| GRBSQ |
.4 / 6.1 |
.8 / 27 |
0 / 29.9 |
45.2/147.3 |
3.5 /13.4 |
| HUDSK |
.4 / 1.1 |
.1 / 24.3 |
.1 / 0.2 |
18.9/122.1 |
1.6 / 16.4 |
| HUDTS |
-.1 / 2.2 |
0 / 29.8 |
0 / 0.1 |
8.3/154 |
.7 / 17 |
| NARPC |
.6 / 4.0 |
-1.3 / 25.8 |
13.6 / 32.1 |
3.6/204.4 |
0.3 /16.9 |
| NARTW |
2.4 / 6.9 |
0.7 / 23.8 |
23.7 / 32.1 |
0.2/138.4 |
0 / 15.3 |
| OWCSU |
0.1 / 1.6 |
4.3 / 29.4 |
0.1 / 0.3 |
0.4/178.2 |
0 / 15.8 |
| OWSWM |
0 / 1.5 |
4.1 / 31.7 |
0.1 / 0.3 |
0 / 209.6 |
0 / 18.8 |
| WELHT |
-0.1 / 1.6 |
-0.8 / 26.1 |
0 / 29.8 |
20.4/201.2 |
1.7 /16.1 |
| WELIN |
-0.1 / 5.1 |
-1.8 / 23.1 |
0.5 / 36 |
0.4 /145.3 |
0 / 15.8 |
| WQBCB |
0 / 2.3 |
0.9 / 27.5 |
21.8 / 32.6 |
4.3 /269.6 |
0.3 /19.4 |
| WQBMP |
0.8 / 1.7 |
3 / 10.6 |
27.4 / 31.1 |
70.5/125.6 |
6.5 /12.3 |
|
|
|
|
|
|
| CBMJB |
0 / 1.6 |
4.6 /36.7 |
0 / 0.4 |
0/231.4 |
0 / 18.4 |
| CBMPR |
0 / 3.2 |
7.6 /30.5 |
0 / 0.4 |
0/198.2 |
0 / 15.7 |
| CBVGI |
-0.2 / 1.7 |
2.3 /31.6 |
9.9 / 27.5 |
23.6/195.7 |
2 / 18.2 |
| CBVTC |
-0.3 / 2.1 |
-0.7 /35.1 |
0.1 / 20.3 |
0 / 239.0 |
0 / 22.1 |
| DELBL |
0 / 2.4 |
-0.3 /33.3 |
0 / 7.7 |
0 / 195.9 |
0 / 17.9 |
| DELSL |
0 / 2.8 |
-0.5/31.4 |
0.1 / 36.3 |
0 / 346.9 |
0 / 32.1 |
| MULB6 |
0.7 / 4.5 |
-1.4 / 28 |
13 / 35.4 |
7.8 / 243 |
0.6 /18.7 |
| MULBA |
0.5 / 2.7 |
-0.2 /30.1 |
0 / 15.6 |
4.9 / 240.7 |
0.4 /18.7 |
|
|
|
|
|
|
| ACEBB |
-0.1 / 2.8 |
2.6 /36 |
4.3 / 41.7 |
2.8 /488.7 |
0.2 /32.4 |
| ACESP |
-0.1 / 3.7 |
4.3 /45.9 |
6.6 / 46 |
0.4 / 284 |
0 / 25.8 |
| NIWOL |
0 / 2.8 |
1.6 /34.9 |
0.1 / 38.6 |
0.1 /489.1 |
0 / 35.2 |
| NIWTA |
0.1 / 2.3 |
0.4 / 36 |
0 / 33.4 |
0 / 390.7 |
0 / 30.5 |
| NOCMS |
0 / 3.5 |
-0.4 /34.6 |
7.8 / 37.2 |
2.4 /264.8 |
0.2 /21.5 |
| NOCZI |
0 / 3.5 |
-0.3 /33.1 |
1.6 / 35.4 |
4.3 /232.9 |
0.3 /18.2 |
| SAPFD |
1.2 / 2.8 |
6.7 / 32.8 |
4.3 / 32.2 |
1.5 /204.8 |
0.1 /14.1 |
| SAPML |
1.3 / 2.3 |
7.3 / 31.7 |
1.8 / 37.5 |
3.3 /179.5 |
0.2 /11.6 |
|
|
|
|
|
|
| APAEB |
0.8 / 3.5 |
5.7 / 33.2 |
0 / 32.2 |
2 / 185.4 |
0 / 23.2 |
| APAES |
-0.1 / 3.7 |
2.9 / 33.6 |
0 / 30.6 |
0 / 210.6 |
0 / 16.7 |
| JOB09 |
-0.2 / 0.9 |
20.9/35.4 |
22.7/53.8 |
0 / 221.4 |
0 / 13.1 |
| JOB10 |
1.2 / 0.3 |
24.7/33.4 |
16.7/42.8 |
0 / 498.6 |
0 / 28.0 |
| RKBBR |
1.0 / 2.3 |
15 / 35.4 |
1.4 / 36 |
0.3 / 500 |
0 / 8.7 |
| RKBUH |
0.4 / 1.7 |
13.3 / 34.1 |
.2 / 33.6 |
0 / 285.3 |
0 / 21.3 |
| WKBFR |
1.1 / 3.5 |
6.2 / 33.7 |
0 / 17.6 |
0 / 248.8 |
0 / 22.1 |
| WKBWB |
0 / 1.9 |
1.6 / 35.4 |
0 / 22.0 |
0 / 193.7 |
0 / 18.4 |
Table 8. Number of deployments analyzed in first-stage, harmonic regression analysis.
| NERR site |
Depth |
DO (mg/l) |
DO (% sat) |
Temp. |
Salinity |
| ELKAP |
34 |
28 |
29 |
36 |
35 |
| ELKSM |
36 |
26 |
26 |
36 |
36 |
| PDBBY |
28 |
22 |
25 |
28 |
24 |
| PDBJL |
44 |
35 |
35 |
49 |
49 |
| SOSSE |
13 |
16 |
17 |
26 |
24 |
| SOSWI |
14 |
9 |
9 |
24 |
22 |
| TJROS |
64 |
50 |
51 |
65 |
64 |
| TJRTL |
21 |
26 |
26 |
28 |
27 |
|
|
|
|
|
|
| GRBGB |
25 |
33 |
33 |
34 |
34 |
| GRBSQ |
19 |
18 |
18 |
19 |
19 |
| HUDSK |
31 |
26 |
26 |
30 |
25 |
| HUDTS |
29 |
25 |
25 |
30 |
25 |
| NARPC |
44 |
31 |
32 |
44 |
43 |
| NARTW |
14 |
13 |
13 |
14 |
14 |
| OWCSU |
38 |
34 |
33 |
38 |
38 |
| OWCWM |
30 |
24 |
24 |
30 |
30 |
| WELHT |
25 |
19 |
19 |
23 |
21 |
| WELIN |
34 |
34 |
34 |
34 |
34 |
| WQBCB |
18 |
17 |
17 |
18 |
18 |
| WQBMP |
2 |
2 |
2 |
2 |
2 |
|
|
|
|
|
|
| CBMJB |
11 |
9 |
9 |
12 |
12 |
| CBMPR |
12 |
6 |
6 |
12 |
12 |
| CBVGI |
16 |
18 |
18 |
18 |
18 |
| CBVTC |
38 |
32 |
32 |
41 |
40 |
| DELBL |
65 |
62 |
63 |
64 |
63 |
| DELSL |
60 |
53 |
53 |
60 |
60 |
| MULB6 |
32 |
27 |
29 |
31 |
30 |
| MULBA |
41 |
33 |
34 |
40 |
40 |
|
|
|
|
|
|
| ACEBB |
28 |
24 |
24 |
33 |
33 |
| ACESP |
34 |
22 |
23 |
34 |
29 |
| NIWOL |
61 |
51 |
55 |
63 |
59 |
| NIWTA |
64 |
55 |
56 |
64 |
63 |
| NOCMS |
40 |
33 |
34 |
42 |
41 |
| NOCZI |
46 |
39 |
38 |
46 |
44 |
| SAPFD |
9 |
35 |
36 |
52 |
50 |
| SAPML |
5 |
48 |
48 |
59 |
59 |
|
|
|
|
|
|
| APAEB |
56 |
29 |
30 |
58 |
58 |
| APAES |
41 |
24 |
24 |
37 |
37 |
| JOB09 |
26 |
18 |
17 |
28 |
22 |
| JOB10 |
24 |
20 |
22 |
24 |
19 |
| RKBBR |
20 |
10 |
11 |
20 |
20 |
| RKBUH |
48 |
29 |
29 |
48 |
47 |
| WKBFR |
70 |
60 |
60 |
70 |
68 |
| WKBWB |
64 |
53 |
53 |
65 |
65 |
Table 9. (PDF)Results of harmonic regression analysis for water quality variables (1996-1998) expressed as percent sums of squares for 12.42 hour cycles (SST), 24 hour cycles (SSD) and interactions between these two cycles (SSTD)
Table 10. (PDF)Results of metabolism analyses for 27 of 44 NERR sites from 1996-1998 data. Production (Pg), respiration (Rtot) and net ecosystem metabolism (NEM) measured as gO2/m2/d.
West Coast Reserves
Northeast Coast and Great Lakes Reserves
Mid-Atlantic Coast Reserves
Southeast Coast Reserves
Gulf of Mexico and Caribbean Reserves
Reserve comparisons
Summer
Mean salinity, salinity range, and frequency of hypoxia and supersaturation were compared among sites within a reserve for summer (June 21-Sept. 21) 1997-98 using t-tests. Mean salinity and mean salinity range were significantly different for most sites and reserves (Table 38). No significant differences in mean salinity were observed for the Chesapeake Bay-MD, Jobos Bay, and Narragansett Bay reserves. No significant differences in mean salinity range were observed for the ACE Basin, Apalachicola Bay, Hudson River, Narragansett Bay and North Inlet reserves. Few significant differences were detected for hypoxia or supersaturation. Hypoxia was significantly different among sites at the Delaware Bay, Elkhorn Slough, Padilla Bay and Rookery Bay reserves (Table 39). Supersaturation was significantly different among sites within the Chesapeake Bay-VA, Elkhorn Slough, Jobos Bay, Tijuana River, and Weeks Bay reserves (Table 39).
Seasonal
Seasonal patterns in the percent of time (first 48 hours post-deployment) that hypoxia and supersaturation occurred when data from all reserves were combined. Hypoxia was most prevalent in summer, while supersaturation was most prevalent in spring (Figure 195).
Analysis of frequency of hypoxia revealed few seasonal differences at the reserve level. The only reserves for which significant seasonal hypoxic conditions occurred were in the ACE Basin, North Inlet, Padilla Bay, Rookery Bay and Tijuana River reserves (Table 40). At these reserves, hypoxia was greatest in summer, with the exception of Padilla Bay where hypoxia was most prevalent in fall. Within reserves, significant seasonal differences in the frequency of hypoxia occurred at the Oyster Landing and Thousand Acre sites (North Inlet), Joe Leary Slough (Padilla Bay), Blackwater River and Upper Henderson sites (Rookery Bay) and Oneonta Slough and Tidal Linkage sites (Tijuana River). Hypoxic conditions were most prevalent at both North Inlet and Tijuana sites and at the Blackwater River site in summer. At the Upper Henderson site (Rookery Bay), hypoxia occurred with greatest frequency in spring. Hypoxia was most prevalent at Joe Leary Slough (Padilla Bay) in fall.
Analysis of frequency of supersaturation revealed few seasonal differences at the reserve level. The only reserves where significant seasonal supersaturation occurred were Padilla Bay, South Slough, Tijuana, and Weeks Bay (Table 40). For these reserves, supersaturation occurred with the greatest frequency during spring, except for Weeks Bay where supersaturation was most prevalent in fall. At sites within reserves, significant seasonal differences in supersaturation occurred at East Bay Surface (Apalachicola), Blackbird Landing (Delaware Bay), Bayview Channel (Padilla Bay), Stengstacken Arm (South Slough), Head of Tide (Wells), and Fish River and Weeks Bay (Weeks Bay). Supersaturation was most prevalent at East Bay Surface, Bayview Channel, and Stengstacken in spring, while supersaturation occurred most frequently at Weeks Bay sites in fall. Supersaturation was most prevalent at the Head of Tide site in summer and most prevalent at the Blackbird Landing site in fall and spring.
Analysis of salinity revealed significant differences at the reserve level. Lowest mean salinity was generally observed in winter or spring, depending upon reserve site. Analysis of mean salinity at each site within a reserve indicated that significant differences were observed for all sites except for Tivoli South in the Hudson River NERR (Table 41). Analysis of salinity range indicated significant differences for all reserves except Chesapeake Bay-VA, Hudson River, Jobos Bay, North Inlet, and Waquoit Bay (Table 41). The season when lowest mean salinity range occurred varied by reserve. Seasonal differences in salinity range were found for all sites except for Hudson River sites, Station 10 at Jobos Bay, State Route 2 at Old Woman Creek, and Central Basin at Waquoit Bay (Table 41).
Figure 195. Seasonal patterns of mean hypoxia and supersaturation for all sites, 1997-1998.
Results of statistical analysis to determine significant differences in mean salinity and salinity range between sites within each Reserve (* t-test for unequal variances; ** p < 0.05) (see PDF).
Results of statistical analysis to determine significant differences in percent of deployment (first 48 h) with hypoxia (<28% saturation) or supersaturation (>120% saturation) between sites within a Reserve, Jul-Aug 1997 and 1998 (* t-test for unequal variances; ** p < 0.05) (see PDF).
Results of statistical analyses to determine differences in hypoxia and supersaturation (first 48 h) among sites within a Reserve and between seasons at each site and each Reserve (#statistic used was c2, One-Way ANOVA (F) or *t-test for unequal variances; ** p < 0.05) (see PDF).
Results of statistical analyses to determine differences in salinity and salinity range among sites within a reserve and between seasons at each site and each Reserve (#statistic used was c2, One-Way ANOVA (F), or * t-test for unequal variances; ** p < 0.05)
(see PDF)
Inter-annual
Summer hypoxia (Jul-Aug) varied at reserve sites among years (1997-1998). Few sites experienced hypoxia for more than 20% of the first 48 hours post-deployment. In 1997, four sites (Blackbird Landing, Delaware Bay; Tidal Linkage, Tijuana River; State Route 2, Old Woman Creek; and East Bay Surface, Apalachichola Bay) experienced hypoxia for >20% of the first 48 hours post-deployment (Figure 196). In 1998, the Tidal Linkage site and four additional sites (Oneonta Sough, Tijuana Riveder; Jug Bay, Chesapeake Bay-MD; Oyster Landing, North Inlet-Winyah Bay; and Azevedo Pond, Elkhorn Slough) experienced hypoxia for >20% of the first 48 hours post-deployment. In 1998, one site (Blackwater River, Rookery Bay) experienced hypoxia > 80% of the first 48 hours post-deployment. Correlation of percent of time with hypoxia and supersaturation among years (1997 and 1998) was performed to determine whether consistent annual patterns existed. Analysis using Kendall’s tau indicated no significant correlation (J=0.28, p=0.054) for hypoxia during the first 48 hours post-deployment in summer (July and August) 1997 and 1998. A significant, positive correlation (J= 0.48, p=0.002) was found for increasing hypoxia during the first 14 days post-deployment during these years.
Three sites (Central Basin, Waquoit Bay; Weeks Bay; and Azevedo Pond, Elkhorn Slough) experienced supersaturation for more than 20% of the first 48 hours post-deployment in both Jul-Aug 1997 and 1998. One additional site in 1997 (Taskinas Creek, Chesapeake Bay-VA) and two additional sites in 1998 (Station 9 at Jobos Bay, data only available for August 1998; and East Bay Bottom, Apalachichola Bay) experienced supersaturation for more than 20% of the first 48 hours post-deployment (Figure 197). Results of correlation analysis for supersaturation indicated no significance during the first 48 hours post-deployment (J=0.16, p=0.25) or for data collected during the first 14 days post-deployment (J=0.08, p=0.55). Not surprisingly, the association for hypoxia and supersaturation data taken within 2 days post-deployment was highly correlated with those data taken within 14 days after deployment (Table 42). Hypoxia during the first 48 hours and first 14 days post-deployment was correlated regardless of year, whereas supersaturation during the first 48 hours and first 14 days post-deployment data was only correlated within the same year.
Table 42. Results of Kendall’s tau correlation analysis (J and p) for hypoxia and supersaturation from data taken within 48 h post-deployment and 14 days post-deployment for summer (July and August) 1997 and 1998.
Variable
|
2d Hypox97 |
2d Hypox98 |
2d Supersat.97 |
2d Supersat98 |
| 14d Hypox97 |
0.73(.001) |
0.29(.03) |
|
|
| 14d Hypox98 |
0.38 (.006) |
0.72(.000) |
|
|
| 14d Supersat97 |
|
|
0.74(.001) |
0.07(.58) |
| 14d Supersat98 |
|
|
0.09(.44) |
0.67(.001) |
Figure 196. Percent of first 48 hours post-deployment with hypoxia, Jul and/or Aug 1997-1998. If sites are not listed, no data was available for either Jul or Aug in either year.
Figure 197. Percent of first 48 hours post-deployment with supersaturation, Jul and/or Aug 1997-1998. If sites are not listed, no data was available for either Jul or Aug in either year.
Duration of hypoxia
Hypoxia occurred throughout all geographic regions in the NERR System. Between 1996-1998, 765 hypoxic events were observed during the first 48 hours post-deployment, with at least one event observed at almost all NERR sites. Hypoxia was most frequently observed at West Coast Reserves (40%), followed by Reserves in the Gulf of Mexico and Caribbean (22%). East Coast Reserves (Southeast, Mid-Atlantic, and Northeast/Great Lakes) contributed similarly to overall hypoxic events (10-16%) and represented the least frequent occurrence of hypoxia geographically (Table 43).
Between 1996-1998, 307 hypoxic events were observed at reserves on the West Coast, 60% of which lasted less than 4 hours (Table 43). Exceptional hypoxia at West Coast sites was primarily due to site selection, limited flushing due to tidal gates and deployment in pond systems. Twenty-six percent of hypoxic events lasted 4-8 hours and 11% of hypoxic events lasted 8-12 hours. Only three percent of hypoxic events lasted between 12-20 hours, and no hypoxic events longer than 24 hours were observed. Hypoxia was observed at all sites in this region except for Bayview Channel (Padilla Bay) and South Marsh (Elkhorn Slough).
Ninety hypoxic events were observed at reserves in the Northeast Coast and Great Lakes region, 87% of which lasted less than 4 hours (Table 43). Seven percent of hypoxic events lasted 4-8 hours and four percent of hypoxic events lasted 8-12 hours. Only one percent of hypoxic events lasted as long as 20 hours and one percent of hypoxic events lasted longer than 24 hours. Hypoxia occurred at all sites in this region except for both sites at the Great Bay reserve.
One hundred twenty-five hypoxic events were observed at reserves in the Mid-Atlantic Coast region, 73% of which lasted less than 4 hours (Table 43). Twenty-one percent of hypoxia events lasted 4-8 hours and three percent of hypoxia events lasted 8-12 hours. Only two percent of hypoxic events lasted 12-16 hours and one percent of hypoxic events lasted longer than 24 hours. Hypoxia occurred at all sites in this region except for both sites at the Jacques Cousteau Reserve at Mullica River, NJ.
Seventy-eight hypoxic events were observed at reserves in the Southeast Coast region, 83% of which lasted less than 4 hours (Table 43). Fourteen percent of hypoxic events lasted 4-8 hours and one percent of hypoxic events lasted 8-12 hours. Only one percent of hypoxic events lasted longer than 24 hours. Hypoxia was observed at all sites in this region except for the Marsh Landing site at the Sapelo Island, GA, reserve and the Zeke’s Island site at the North Carolina Reserve.
One hundred sixty-five hypoxic events were observed at reserves in the Gulf of Mexico and Caribbean Sea, 69% of which lasted less than 4 hours (Table 43). Sixteen percent of hypoxic events lasted 4-8 hours and six percent of hypoxic events lasted 8-12 hours. Six percent of hypoxic events lasted between 12-24 hours and three percent of hypoxic events lasted longer than 24 hours. Hypoxia was observed at all sites in this region except for the Weeks Bay site.
Not all reserves collected similar quantities of data during the same time of the year; thus, an index was developed to predict duration and magnitude of hypoxia on an annual basis, according to the frequency and duration of observed hypoxic events at sites, and is presented in Figures 198-202.
Table 43. Frequency of occurrence and duration of hypoxia at NERR sites, 1996-1998.
| Region |
N |
%N |
<4h |
4-8h |
8-12h |
12-16h |
16-20h |
20-24h |
>24h |
| West Coast |
307 |
40 |
0.60 |
0.26 |
0.11 |
0.02 |
0.01 |
0.00 |
0.00 |
| Northeast |
90 |
12 |
0.87 |
0.07 |
0.04 |
0.00 |
0.01 |
0.00 |
0.01 |
| Mid.Atlantic |
125 |
16 |
0.73 |
0.21 |
0.03 |
0.02 |
0.00 |
0.00 |
0.01 |
| Southeast |
78 |
10 |
0.83 |
0.14 |
0.01 |
0.00 |
0.00 |
0.00 |
0.01 |
| Gulf/Carib. |
165 |
22 |
0.69 |
0.16 |
0.06 |
0.02 |
0.03 |
0.01 |
0.03 |
|
765 |
Mean |
0.75 |
0.17 |
0.05 |
0.01 |
0.01 |
0.00 |
0.01 |
Predicted annual occurrence and duration of hypoxia at NERRs (see PDF)
System-level analysis
Depth, temperature, salinity, and dissolved oxygen
Examination of the average conditions experienced at water quality monitoring sites in the National Estuarine Research Reserves revealed interesting patterns among geographic regions. Most of the sites are shallow, with mean depth for 35 of the 44 sites < 2 m (Figure 203). All sites located in the Southeast (NC, SC, and GA), Gulf and Caribbean averaged < 2 m in depth. At two Reserves (Great Bay and Narragansett Bay), water depths greater than 4 m were sampled.
Not surprisingly, temperature extremes at water quality sites were indicative of the climate within the region. Reserves where mean water temperature ³ 25°C persisted for more than 25 % of the year were located in the Southeast, Gulf Coast and Caribbean. At Jobos Bay, Puerto Rico, water temperature was ³ 25°C for more than 95% of the year (Figure 204). Low temperature extremes (£ 10°C) occurred at all reserves. At two sites on the west coast (South Slough and Padilla Bay), two sites in the Northeast (Wells and Narragansett Bay), and three sites in the mid-Atlantic region (Mullica River, Delaware Bay, and Chesapeake Bay-VA), water temperature remained £ 10°C for more than 25% of the year.
Salinity conditions at the NERR water quality sites were indicative of the differences in estuarine morphology, freshwater inflow, evaporative processes, and degree of ocean exchange present in the NERRs. Mean salinity was < 8 ppt at 14 sites, 8-28 ppt at 16 sites, and >28 ppt at the remaining 14 sites. Annual salinity variation at most NERR sites was highly variable, expect for three reserves (Hudson River, Old Woman Creek and Chesapeake Bay-MD) which were characterized by low salinity or freshwater conditions (Figure 205).
Hypoxia and supersaturation occurred throughout the reserve system, regardless of geographic region. When present, hypoxia persisted on average less than 15% of the first 48 hours post-deployment at all sites, except for Oneonta Slough and Tidal Linkage (Tijuana River), Azevedo Pond (Elkhorn Slough), and Blackbird Landing (Delaware Bay). At these four sites, hypoxia persisted between 23-35% of the first 48 hours post-deployment on average. Hypoxia was negatively correlated (R=-0.54, df=44, p=0.000) with latitude, positively correlated (R= 0.57, df=44, p=0.000) with warm water temperature (³ 25°C), and negatively correlated (R= -0.44, df=44, p=0.003) with cold water temperature (£ 10°C). These findings collectively suggest that hypoxia is most likely to be experienced at sites with sustained warm water events and least likely to be experienced at sites with sustained cold water events. All sites that experienced hypoxia also experienced supersaturation. Six sites (Bayview Channel, Padilla Bay; Central Basin, Waquoit Bay; and both sites at the Mullica River and Weeks Bay reserves) experienced supersaturation, but did not experience hypoxia in July-August 1997-98 (Figure 206). Supersaturation was not significantly correlated with water depth, temperature, or salinity (for more information see PDF).
Cluster Analysis
Cluster analysis was used to group the 44 sites in the NERR SWMP according to physical, chemical, geological, and geographical attributes (Tables 44-45). Four site groupings were identified from cluster analysis (Figure 207). Three of the groups appeared to correspond to geographical region and latitude. Group 1 was a large grouping, primarily consisting of sites located on the northeast and mid-Atlantic seaboard. Two West Coast reserves, located at similar latitudes as northeast and mid-Atlantic reserves, were also included in this group. Sites belonging to group 2 were primarily located in the Southeast, except for the Mullica River-Buoy 126 site in New Jersey. The third group was primarily comprised of sites located in the Gulf of Mexico; however, three freshwater sites (both sites at the Chesapeake Bay-MD reserve and State Route 2, Old Woman Creek) also clustered with this group. Group four sites consisted of sites from the two southernmost West Coast reserves (Elkhorn Slough and Tijuana River Estuary), along with the Jobos Bay reserve in Puerto Rico and the Big Bay (ACE Basin) site in South Carolina.
With the exception of eight reserves (Padilla Bay, Wells, Hudson River, Old Woman Creek, Mullica River, North Inlet-Winyah Bay, ACE Basin, and Weeks Bay), sites within a reserve were more similar to each other than to other sites located in other reserves. Within group 1, Bayview Channel in the Padilla Bay reserve was more similar to sites in Waquoit Bay and Narragansett Bay than to the Joe Leary Slough site in Padilla Bay. Joe Leary Slough was most similar to the Sawkill Creek site at Hudson River. Similarly, the Head of Tide site in the Wells reserve was more similar to the Lower Bank (Jacques Cousteau-Mullica River) and the State Route 6 (Old Woman Creek) sites than it was to the Inlet site in the Wells reserve. Within group 2, Buoy 126 (Jacques Cousteau-Mullica River) was most similar to sites in North Carolina. Within group 3, the Weeks Bay site (Weeks Bay) was more similar to sites in group 3 than the Fish River site (Weeks Bay). Within group 4, Big Bay Creek (ACE Basin) was most similar to both sites in the Tijuana River Estuary NERR. Similarity between these sites was primarily due to similar attributes such as high frequency (30% of data) with warm water temperature (> 25°C), similar salinity (>28 ppt), and similar land usage and habitat characteristics (>50% developed and <50% forested). Furthermore, the Big Bay site and both sites in the Tijuana River were located within two minutes of latitude from each other (32°10’N vs. 32°12’N).
The dendrogram of site attributes produced three major groupings (Figure 208). Within group 1, impervious surface and habitat were most similar. Within group 2, hypoxia and warm water temperature (³ 25°C) were most similar, which reinforces the correlative relationship previously noted between hypoxia and warm water temperatures (for more information see PDF).
Periodicity (Harmonic regression analysis)
First stage regression analyses fit most individual deployment series well and typically gave a good fit within the range of the data. The time series for most variables at most sites were strongly periodic. Spectral (Fourier) time series analyses performed previously by Wenner et al. (1998) and by ourselves as preliminaries to harmonic analyses, repeatedly showed periods related to 12.42 hour, 24 hour, 29.5 day, and 365.24 day cycles. These cycles were also visually evident in most of the series through inspection using microscope and plot.week. Further visual inspection revealed a number of other complicating factors in the data such as (1) shifts in mean response at new meter deployments, (2) shifts in cycle amplitude at deployments, (3) meter decay, and (4) unusual events (presumably weather-related) on the order of 1-2 days in duration.
R-square (R2) values and Root-Mean Square (RMSE) values for each of the five water quality variables are summarized in Figures 209-218. Sites within each Reserve whose mean levels were significantly different from each other using the Sidák method are noted with a double asterisk (**).
In each figure, the mean R2 or RMSE value across all deployments is displayed with standard error bars. Mean R2 for depth typically exceeded 0.90, except for four sites (Head of Tide-Wells, Saw Kill-Hudson River, and both Old Woman Creek sites) with R2 around 0.75 (Figure 209). Mean R2 for water temperature (Figure 210) and salinity (Figure 211) were typically greater than 0.80, although for a few sites (Sawkill-Hudson River and State Route 6-Old Woman Creek) R2 for salinity was less than 0.40. Mean R2 for dissolved oxygen was typically 0.70 to 0.90 for most sites (Figure 212-213).
Root mean square for error (RMSE) values were an important complement to the R2 values. The RMSE values demonstrate the predictive capabilities of the regression; about 95% of the actual data values fit the regression curves within a distance of less than twice the RMSE. RMSE values for depth were typically less than 0.10 m (Figure 214). RMSE for water temperature was usually less than 0.75°C, i.e., temperature was generally predictable to within 1.5°C (Figure 215). RMSE for salinity was less than 1.25 ppt, except for 4 sites (both South Slough sites, Joe Leary Slough-Padilla Bay, and Head of Tide-Wells) with RMSE around 2.5 ppt (Figure 216). It is interesting to note that the sites having low R2 values for salinity (Hudson River, State Route 2-Old Woman Creek) also have small RMSE. This agreement suggests that the models predict salinity well, even though they may have low R2 values (e.g., low R2 values reflect little explainable variability at these freshwater sites). For dissolved oxygen, RMSE was typically less than 1 mg/l (Figure 217) or less than 10% saturation (Figure 218).
Variance in water quality variables was expressed as a function of the partial sums of squares for the deployment level regressions due to a 12.42 hour component (SST), a 24 hour component (SSD), and interaction between these components (SSTD). The interaction component (i.e., cross-product terms) allows the signature of one component to change with the signature of the other component for each water quality variable.
Overall, the 12.42 hour component accounted for more than 50% of variance in water depth, with less than 20% of variance attributed to the 24 hour component (Figures 219-220). The 12.42 hour component accounted for more depth variance than the 24 hour component at 75% (n=33) of the NERR sites, except for a few sites which experienced minimal depth variance (i.e., Saw Kill-Hudson River; Azevedo Pond-Elkhorn Slough; Jobos Bay; Old Woman Creek; Padilla Bay; and Weeks Bay).
For water temperature, 24 hour cycles tended to dominate most sites and, on average, accounted for approximately 30- 50% of variance (Figure 221). At most sites, 12.42 hour cycles accounted for less than 20% of water temperature variance (Figure 222). Sites that constituted exceptions to this rule included the Inlet Site-Wells site and the Great Bay-Great Bay site, where the 12.42 hour cycle accounted for 40-60% of temperature variance and 24 hour cycle accounted for less than 20% of temperature variance. Other sites where the variance of the 12.42 cycle accounted for more temperature variance than 24 hour cycle included South Marsh-Elkhorn Slough, T-wharf-Narragansett Bay, and Bouy 126-Mullica River.
Surprisingly, salinity was only strongly influenced (40-80% of salinity variance) by the 12.42 hour cycle at half of the sites (Figure 223). At sites where salinity was not strongly influenced (10-20% of salinity variance) by the 12.42 hour cycle, 24 hour cycles only accounted for 5-35% of salinity variance (Figure 224); thus, interaction between these cycles accounted for the majority of salinity variance. Twenty-four hour cycles accounted for slightly more salinity variance (>5%) than 12.42 hour cycles at sites that experienced only slight variations in annual salinity. These sites included low-salinity sites (Hudson River, Old Woman Creek, Jug Bay-Chesapeake Bay MD, and Joe Leary Slough-Padilla Bay), high-salinity sites (Potters Cove-Narragansett Bay and Jobos Bay), and sites that experienced minimal tidal effects (Weeks Bay, Apalachicola, Azvedo Pond-Elkhorn Slough, and South Marsh-Padilla Bay).
Harmonic regression results for dissolved oxygen were highly diverse across sites, and very similar for both percent saturation and mg/L units (Figures 225-228). Dissolved oxygen variance due to 24 hour cycles was usually between 15 and 40%; however, there were some notable exceptions. Twenty-four hour cycles accounted for 50-70% of dissolved oxygen variance at the following sites: Azevedo Pond-Elkhorn Slough; Central Basin-Waquoit Bay; Potters Cove-Narragansett Bay; Saw Kill-Hudson River; and both Jobos Bay sites. Variance in dissolved oxygen due to 12.42 hour cycles was also quite diverse at the 40 sites analyzed. These cycles (12.42 hours) typically accounted for 5-30% of dissolved oxygen variance; however at 7 sites (Inlet Site-Wells Bay; Squamscott River-Great Bay; T-wharf-Narragansett Bay; Tivoli South Bay-Hudson River; Scotton Landing-Delaware Bay; Taskinas Creek-Chesapeake Bay VA; and Marsh Landing-Sapelo Island), 12.42 hour cycles accounted for 40-60% of dissolved oxygen variance. (for more information see PDF)
Photosynthesis and respiration
Approximately 75% of the data calculated from the SWMP between 1996-1998 were consistent with the assumption that the water masses moving past the sensors were homogeneous. Two major exceptions to this pattern occurred at the Sawkill Creek site (Hudson River) and at Joe Leary Slough (Padilla Bay). At Sawkill Creek, only 31% of the data were included in metabolic rate calculations. At this site, the data sonde was deployed just upstream of a dam in the creek. Physical processes (e.g., advection of different water masses when stream flow was greater than 0.4 m3/s), rather than biological processes, probably enhanced oxygen exchange across the air-water interface and controlled the oxygen dynamics at this site. At Joe Leary Slough, 57% of the observations were used to calculate metabolic rates. This small, intermittently flushed slough drains agricultural fields and pastures. Restricted water flow and intermittent tidal flushing between the Joe Leary Slough and Padilla Bay, due to a dam and one-way tide gates, led to some questionable results. When high tides occurred during the middle of the day, the calculations led to an erroneous estimate of production or respiration. In this case, physical processes, rather than biological processes, controlled oxygen concentrations. Instrument drift was a significant problem at three sampling sites: Apalachicola Bay-Bottom, Rookery Bay-Upper Henderson, and Weeks Bay-Fish River. At these sites, only the first 2 days of each deployment were used to estimate metabolic rates.
Respiration generally was greater than gross production (Figure 229), except at two sites (Goodwin Islands-Chesapeake Bay VA and Central Basin-Waquoit Bay) where gross production exceeded respiration and one site (Hudson River Sawkill) were production and respiration were equal. Average metabolic rates among Reserves were variable: 0.6-15.6 gO2 m-3 d-1 gross production, 0.7 – 19.3 gO2 m-3 d-1 total respiration, and 0.3 to – 9.1 gO2 m-3 d-1 net ecosystem metabolism. The highest rates occurred at the Elkhorn Slough-Azevedo Pond and Chesapeake Bay Maryland-Jug Bay sites. Metabolic rates at these sites were nearly double the rates at the other sites (Figure 229). The lowest gross production and total respiration rates occurred at the Hudson River-Sawkill site, a small freshwater creek. Rates were also generally low at the Great Bay sites, Narragansett Bay-T-wharf and Weeks Bay-Fish River (Figure 229).
Temperature and metabolic rates were significantly correlated at most of the 27 sites evaluated (Waquoit Bay Metoxit Point was not included in the analyses due to the small amount of data available for this site). Gross production and temperature were positively correlated (p=0.05) at all but six sites (and negatively correlated at two sites), respiration and temperature were positively correlated (p=0.05) at all but three sites, and net ecosystem metabolism and temperature were correlated at all but six sites. These findings suggest that biological processes controlled metabolic rates at these sites, with warmer temperatures associated with higher metabolic rates.
In contrast, salinity and metabolic rates were significantly correlated at only about half of the sites. On these occasions, metabolic rates were higher during periods of high salinity. Higher gross production values generally occurred at euhaline sites (25- 30 ppt) that experienced a moderate variation (10 ppt) in salinity range over annual cycles. These sites include Padilla Bay- Bay View Channel, Waquoit Bay-Central Basin, and Narragansett Bay-Potters Cove). High gross production was also observed at the tidal freshwater sites sampled in Chesapeake Bay Maryland (Figure 230). Metabolic rates were not associated with tidal range or DO (% sat) concentration (for more information see PDF).
|
|
|
