SALT MARSH RECOVERY FROM A CRUDE OIL SPILL: VEGETATION, OIL WEATHERING, AND RESPONSE

Rebecca Z. Hoff and Gary Shigenaka National Oceanic and Atmospheric Administration

Hazardous Materials Response and Assessment Division 7600 Sand Point Way, NE Seattle, Washington 98125

Charles B. Henry, Jr. Institute for Environmental Studies

Atkinson Hall, Room 42 Louisiana State University

Baton Rouge, Louisiana 70803

ABSTRACT: Responding to oil spills in marshes is always problematic, since inappropriate response activities can easily add to the overall damage to the marsh. When a spill of Prudhoe Bay crude oil covered a fringing Salicornia virginica marsh in Fidalgo Bay, Washington (north-ern Puget Sound) in February 1991, response personnel used several low-impact techniques to remove oil from the marsh, and minimized access by cleanup workers. Following the response, we established a monitoring program to track marsh recovery, and to document the effectiveness of the response techniques used and their impacts on the marsh.

Through monthly sampling over a 16-month period, we monitored vegetative growth and tracked the chemical degradation of remaining oil. Sampling was conducted along transects located in four areas af-fected in different ways by the spill, including an oiled, trampled section; an oiled, vacuumed section; and an oiled, washed, and vacuumed section. In addition, a control transect was established in an unoiled adjacent marsh.

The study included both biological and chemical components. Bio-logical measurements included percent cover of live vegetation (sampled monthly) and below-ground plant biomass (sampled at the beginning of each growing season in April 1991 and April 1992). Sediment samples included surface sediment (monthly) and core samples collected at the beginning and end of the growing seasons. Sediment samples were analyzed using gas chromatographylmass spectroscopy, and indicator compounds were tracked to determine rates of oil degradation.

Results from 16 months of post-spill monitoring show that foot tram-pling was most detrimental to marsh plants, while washing with vacuum-ing removed the most oil and minimized adverse impacts to vegetation. Dense clay substrate helped prevent oil from penetrating the sediment, thus minimizing acute toxic effects from oil exposure to marsh plant rootstock. By the second growing season post-spill, Salicornia and other marsh plants were growing in all areas except one heavily oiled patch. The monitoring program will be continued to determine if any delayed impacts occur in the marsh.

In late February 1991, a pump failed during offloading at a Texaco Refinery near Anacortes, Washington, resulting in a spill of North Slope crude oil. Approximately 30,000 gallons of oil entered adjacent Fidalgo Bay, a shallow bay characterized by broad mud flats and large subtidal eelgrass beds. In an effort to keep oil away from the eelgrass

beds and the herring spawn that was occurring at the time of the spill, oil was contained along the southern shoreline of the bay using booms. This containment, concurrent with a high tide cycle and northerly winds, resulted in heavy oiling of the fringing salt marsh along the southern edge of the bay.

This fringing marsh, dominated by Salicornia virginica and Distichlis spicata, is not pristine (it is adjacent to roadbed fill of a state highway). Nonetheless, the marsh provides increasingly scarce wetland habitat for aquatic organisms and shorebirds. During the cleanup, efforts were made to control access and detrimental impact to the marsh by cleanup workers. Several low-impact cleanup techniques were used in an at-tempt to remove oil without causing further damage to the marsh. A simple monitoring program was established to document the effec-tiveness of these response techniques and the recovery of marsh vegetation.

Study design and methods

In April 1991, four transects were established in the fringing salt marsh along the south end of Fidalgo Bay for the purpose of monitor-ing vegetative recovery in oiled areas (Figure 1). These transects represented areas affected in different ways by the spill. Transect 1 was a control, located in an unoiled area; transect 2 was located in an area that was relatively lightly oiled but subjected to trampling from exten-sive foot traffic; transects 3 and 4, both heavily oiled and protected from trampling by the placement of boards over the surface of the marsh, were cleaned in two different ways. Transect 3 was vacuumed with a high-powered industrial vacuum to remove surface oil, while transect 4 was flushed with low-pressure, ambient temperature sea-water, and vacuumed similarly to transect 3.

Tidal elevations for each transect were measured using standard survey methods, and all sample quadrats were found to be within 0.15 meters of each other (elevations ranged from 2.60 m to 2.76 m). Because of its relatively high elevation, the marsh is flooded infre-quently, during exceptionally high tide cycles only.

The narrow width of the marsh limited the area available for sam-pling. In addition, we selected monitoring sites where we were certain that particular treatments had taken place. As a result of these limita-tions, the study design does not include replicate samples, nor were sample transects or quadrats selected randomly. Our results are there-

307

308 1993 OIL SPILL CONFERENCE

\ \

STUDY C*AREA •

\ \ TRANSECT \ / ^ ^ ^ c o n t r o l

\ ^ "t' ^

H Salt marsh

F / d o / g o B a i;

TRANSECT! oi led, heavy foot traff ic

TRANSECT 3 oi led, vacuumed m

^~\ \

/ l / TRANSECT 4 1 / o i l e d , washed, vacuumed a

1

/ ^^ Á^ —^ X S T ^ ' K « , G H w A Y TT"-J

Q

< 0

Z

0 £

X u

<

Figure 1. Map of the southern portion of Fidalgo Bay with the location of sample transects

fore qualitative, rather than quantitative, and quadrat samples should be interpreted as indicator locations that are monitored over time, rather than as statistical representatives of the entire marsh.

Photo quadrats (0.25 m2) were established at fixed locations along each transect. Each quadrat was photographed monthly and a sedi-ment sample collected for chemical analysis of oil weathering. Biolog-ical core samples were collected at the beginning of the growing season of both years (in April) for analysis of below-ground plant biomass. Chemistry core samples (diameter 7 cm and length greater than or equal to 10 cm) were collected at all transects at the beginning of the study (April 1991), toward the end of the first growing season (October 1991), and at the beginning of the second growing season (April 1992).

Quadrat photo slides were projected over a standard grid of 360-25 mm squares to estimate vegetative cover. Grid squares that contained any portion of green (live) plant material were counted as covered. The total number of squares designated as covered was divided by the total number of squares in the grid (360) resulting in the estimate of percent cover. Thus, this estimation method could result in similar values of percent cover for plants that are actually growing at different densities.

Below-ground biomass was measured from biological core samples collected with a hand-held aluminum coring device (10.7 cm diameter) sampling a sediment volume of approximately 1.1 liter. Biological core samples were analyzed by rinsing the contents through a 2.0 mm sieve, separating out the vegetative material and sorting live and dead mate-rial. Live and dead materials were dried separately at 60° C until repeated weighing resulted in the same weight (drying time was ap-proximately 48 hours).

Chemistry analysis consisted of extraction of the oil from sediment samples and characterization of the oil using gas chromatography/mass spectroscopy (GC/MS). The extraction process involved three sepa-rate sonication extractions using hexane as the primary solvent. Surro-gate standards were added to the samples to monitor extraction recov-eries. The GC/MS was operated in multiple ion detection mode using a method developed to source-fingerprint and monitor compositional changes in oil from weathering.910

Oil weathering was tracked by identifying indicator compounds including pristane and phytane for the first seven months of the study. By October 1991, pristane and phytane had largely disappeared from the sampled oil. This rate of degradation is comparable to those measured in other studies conducted in marshes.4 Overall patterns in weathering were compared with a reference sample of fresh Prudhoe Bay crude oil. Chemistry core samples were used to characterize subsurface oiling and measure depth of oil penetration. Cores were divided into 2.5 cm sections and analyzed separately by section.

Results

Vegetative cover. Patterns of vegetative cover between oiled and control sites show clear differences over time (Figure 2; Table 1). Dormant Salicornia plants at transect 1 (control quadrat) began bud-

ding in May 1991 and the quadrat was 100 percent covered by June. In contrast, at oiled transects 2 and 3, plants began budding in May, filled in slowly over the next two months, and approached 100 percent cover by September. Transect 4 exhibited a pattern more closely resembling the control site than the other two oiled sites. By July, this quadrat was 100 percent covered.

In 1992, the growth patterns were much more similar among tran-sects (Figure 2; Table 1). Budding began earlier than the previous year (probably because of unusually warm temperatures), and quadrats at all transects were approaching 100 percent cover by June. Values for percent cover were not available for April for transects 2, 3, and 4, because poor contrast in the quadrat photographs made it impossible to differentiate accurately live buds from dormant plant stems.

Below-ground biomass. Biological core data from April 1991 and 1992 show some interesting patterns that relate well to the percent cover data (Figures 2 and 3). Though all samples were collected in the same manner with the same coring tool, volume of sediment collected was not always equal for each core. Therefore, live below-ground biomass was presented as a percent of the total biomass in the core (Figure 3, Table 2).

Percent live biomass shows marked differences between transects (Figure 3). Though all transects had a smaller percent live biomass in 1992 than in 1991, transects 1 (control) and 4 (washed and vacuumed), had approximately equal percentages for both seasons. In contrast, at transects 2 and 3 percent live biomass in 1992 is greatly reduced from the biomass measured at the same locations in 1991. This difference is especially noticeable at transect 2 (trampled).

Though the vegetative cover at all transects in 1992 appears to be similar, the below-ground biomass indicates that much of this above-ground growth may originate from only a few plants, and that the surface vegetation is not always supported by a dense below-ground root mass, especially at transects 2 and 3.

These patterns could reflect several possibilities: • Transect 2 shows more variability in general (see oil weathering

results), probably an artifact of the trampling • Below-ground roots and rhizomes were damaged at transects 2

and 3 during the spill, but did not die until after the first growing season; and

• Plants that survived at transects 2 and 3 are growing under sub-optimal conditions (the plants are stressed) and are directing a greater proportion of energy into above-ground vegetation rather than into below-ground structures.

Sediment chemistry and oil weathering. Chemistry cores showed that oil did not penetrate deeply into the sediments, probably because of the dense clay sediments. At all transects, most of the oil was found in the top 2.5 cm. Maximum oil penetration was measured at transect 2, at greater than 5 cm below the surface.

Oil weathering is represented by ratios of C18/phytane for transects 2, 3, and 4 (Figure 4).7 Transect 2 exhibited the most variability in chemistry results. This fits with its treatment history and with the biological data discussed above, since trampling would be expected to result in patchiness (both in vegetation and oil distribution). Other

Table 1. Estimates of percent vegetative cover for all transects during 1991 and 1992 growing seasons!

Transect 1991

1 2 3 4

1992 1 2 3 4

April

0 0 0 0

35.0 NA2 NA2 NA2

May

51.7 5.6

22.2 70.3

98.9 90.6 92.8 88.9

Vegetative percent cover June

98.1 20.8 51.7 95.3

100 100 92.5 99.4

July

100 43.3 65.0

100

100 100 96.9

100

August

100 89.4 82.5 99.2

Sept

100 99.7 88.6

100

Oct

98.6 99.7 93.3

100

1. Samples were collected only through July 1992 2. NA: data not available due to poor quality of the sample photo-graphs

FATE AND EFFECTS 309

TRANSECT 2

Apr May Jun Aug Sep Oct Apr May Jun Jul Aug Sep Oct

TRANSECT 3 lOO lOO

<D

u 5 40 u In

Cu 2 0

TRANSECT 4

Apr May Jun Jul Aug Sep Oct Apr May Jun Jul Aug Sep Oct

1991 COVERAGE � 1992 COVERAGE �

Figure 2. Estimates of percent vegetative cover plotted over time for each transect for 1991 and 1992

transects showed more consistent patterns over time, with weathering occurring at a steady rate from April 1991 through October 1991. By April 1992, these compounds had disappeared from the sediment samples.

The overall pattern of weathering is illustrated by extracted ion chromatograms (M/E 85, characteristic of alkanes) shown in Figure 5,

Figure 3. Below-ground biomass from biological sediment cores plot-ted as percent live biomass (y axis) by transect (x axis) for 1991 and 1992

from samples collected at transect 4. A reference North Slope crude oil is compared with chromatograms from April 1991, August 1991, and April 1992 (shown at relative scales). As suggested by the C18/phytane plots (Figure 4) the majority of alkanes present in oil residues in August 1991 show significant weathering. By April 1992, the majority of the remaining alkanes are not from crude oil, but rather from natural marsh organic material. Transect 3 followed a pattern similar to transect 4, while transect 2 exhibited more variable weathering. Though these chromatograms show weathering of alkanes, poly-aromatic hydrocarbons (PAH) were still present in August 1992.

Table 2. Data from below-ground biomass from biological core samples

Below-ground biomass

Transect Percent live biomass!

(%) Live biomass

(g) Total biomass2

(g) 1991

1 2 3 4

1992 1 2 3 4

29.09 47.31 29.94 19.93

20.93 2.99

13.16 15.97

10.66 15.68 4.53 6.28

7.72 1.17 2.12 7.51

36.64 33.14 15.13 31.51

36.88 39.13 16.11 47.02

1. Percent of total biomas represented by live fraction 2. Varies by volume of sample obtained from the core (not always equal)

310 1993 OIL SPILL CONFERENCE

TRANSECT! 2.5

CO

CC <D C CO

>> ç ÛU

-«* 00

Ü

2.0

1 5

1 ()

0.5

2.5

A M J J A S O N D J F M A 1991 1 9 9 2

TRANSECT 3

TRANSECT 4

A M 1991 —

J A S O N D J F 1 9 9 2 —

M

Figure 4. Oil weathering measured from surface sediment samples plotted over time for each transect; weathering is shown by the ratio of the indicator compound phytane, divided by carbon 18 (y axis) by month (x axis)—Transect 2 shows no value for July 1991, because the sample collected did not contain enough oil to conduct an analysis.

Discussion

that transect 2 (the trampled area) shows the most severe impacts 16 months after the spill reinforces this fact.

Separate from any response activities, oil itself is toxic to marsh plants to varying degrees. The specific effect depends on factors such as oil type, thickness of oil on the marsh surface, season in which the spill occurred, species of plants involved, and the percentage of the individual plant that is covered with oil.1,2,5,8 Most studies on this topic have been conducted in Spartina marshes, so we were unsure whether the SalicornialDistichlis marsh would follow similar patterns or re-spond differently.

Several black patches with heavy layers of residual crude oil remain in the Fidalgo Bay marsh (at locations outside of our initial sampling transects). Sixteen months after the spill, these areas are still unable to support any live vegetation, though they are surrounded by living plants. Since this spill caused initial heavy oiling over the marsh, the option to leave the marsh to recover without attempting to remove oil probably would not have produced positive results for recovery of marsh vegetation.

Given that it was desirable to remove the heavy layer of oil from the Fidalgo Bay marsh, the ideal response technique would have been one that removed the most oil with the minimum disturbance to vegetation and marsh sediment. The use of low-pressure, ambient-temperature washing with vacuuming proved to be the optimal response technique in this situation. This technique appears to have been effective because it removed the most oil (oil was lifted off the surface of the marsh with the water and removed with the vacuum) without uprooting plants or trampling sediment.

A key component of this response was restricting cleanup workers to board walkways to prevent trampling of marsh sediments. The re-growth of vegetation in the first growing season after the spill is testimony to the fact that SalicornialDistichlis impacted during their dormant period can survive short-term oiling, providing physical im-pacts (trampling) are avoided. The results from transect 2 illustrate that trampling probably caused worse damage to the vegetation than any direct impacts from the light oiling this area received.

Natural weathering of crude oil on surface sediments occurred over time at all transects. Although we were unable to quantify the amount of oil remaining on each transect because of the limited number of sediment samples, we do know that rates of weathering differed by transect (Figure 4). In general, where less oil was present, it degraded more rapidly. Since weathering takes place at the surface interface of the oil, the thicker the layer of oil, the slower the overall weathering. Most of the weathering we have discussed relates to degradation of straight-chain hydrocarbons. Though most of these components were degraded after 16 months, PAHs still persist at all transects.

The large decrease in live below-ground biomass in 1992 at transects 2 and 3 indicates that rootstock in these areas was damaged in some way, and though live below-ground biomass was present the first spring after the spill, many of these roots and rhizomes did not survive through the second year. This decline in live below-ground biomass in 1992 is disturbing since it could indicate the potential for delayed effects on vegetative growth in future years. Several studies have documented the importance of the below-ground root and rhizome system to the above-ground productivity in marsh plants.2'3 A study conducted in salt marshes in Oregon with species similar to Fidalgo Bay found that root and rhizome production was more than 10 times that of the areal plant production. Continued monitoring through the 1993 growing season and beyond will determine whether Salicornia plants begin to allocate energy to rebuild root and rhizome systems, or increase density of above-ground growth, or both.

Several factors are important to consider in evaluating the informa-tion collected in our monitoring program. First, the spill occurred during the dormant season for Salicornia and Distichlis, both perennial species whose surface vegetation dies back during the winter. Studies on Spartina alterniflora show that this species is better able to survive oiling when it occurs during a dormant, or low-growth season, rather than during the main growing season.1

Our study confirms what many other researchers have documented: physical disturbance of salt marshes, especially of surface sediment, have detrimental and long-lasting effects on the marsh ecosystem and on marsh vegetation in particular.8, "•12 A key element in protecting the Fidalgo Bay marsh during spill response was keeping the response personnel and equipment from trampling the marsh surface. The fact

Acknowledgments

We thank Sharon Christopherson, NOAA Scientific Support Coor-dinator for her insight in using low-impact response methods during spill cleanup and in helping set up the study. Gini Curl made the figures and illustrations on short notice and with good humor. Michelle May-field of Louisiana State University conducted many of the chemical analyses; Alan Fukuyama of Pentec Environmental did the analysis of biological sediment cores. We could not have conducted the study without the cooperation and assistance of the Texaco Puget Sound refinery.

FATE AND EFFECTS 311

I o n 8 5 . 0 0 a m u

1 . 2E4-1

1 0 0 0 0 - ]

8000

6000H

4000-

2000-

from V3 : MS

NSC

u J L ^ , 20 40 60

Ion 85.00 amu. from V3 : MS

1 5 0 0 -

1 0 0 0 -

5 0 0 "

r?\- ___jjjjyw£

T4 4/91

iUiJiULLli,. 20 40 G0

I o n 2 0 0 0 "

1 5 0 0 -

1 0 0 0 -

5 0 0 -

0 -

8 5 . 0 0 a m u . f r o m V 3 : MS

T4 8/91

J L P Ï0K^ , i i i , i i i i

2 0 4 0 6 0

Ion 85.00 amu 1 . 2E4-

1 0000-j

8000

6000

4 0 0 0 '

2 0 0 0 H

0

f r o m V 3 : MS

T 4 4 / 9 2

...i^^i^M^ 2 0 4 0 6 0

Figure 5. Chromatograms from surface sediment samples collected from transect 4 illustrate the weathering of alkanes. North Slope crude oil (NSC) is shown as an unweathered reference; the other plots are for samples from April 1991, August 1991, and April 1992.

References

1. Alexander, S. K. and J. W. Webb, Jr., 1985. Seasonal response of Spartina alterniflora to oil. Proceedings of the 1985 Oil Spill Con-ference, American Petroleum Institute, Washington, D.C., pp355-357

2. Alexander, S. K. and J. W. Webb, Jr., 1987. Relationship of Spartina Alterniflora growth to sediment oil content following an oil spill. Proceedings of the 1987 Oil Spill Conference, American Petroleum Institute, Washington, D.C., pp445-449

3. Gross, M. F. , M. A. Hardisky, P. L. Wolf, and V. Klemas, 1991. Relationship between aboveground and belowground biomass of Spartina alterniflora (smooth cordgrass). Estuaries, vl4, ppl80-191

4. Hershner, C. and J. Lake, 1980. Effects of chronic oil pollution on a salt-marsh grass community. Marine Biology, v56, pp 163-173

5. Hershner, C. and K. Moore, 1977. Effects of the Chesapeake Bay oil spill on salt marshes of the lower bay. Proceedings of the 1977 Oil Spill Conference, American Petroleum Institute, Washington, D.C., pp529-533

6. Hoffnagle, J. R., 1980. Estimates of vascular plant primary pro-duction in a west coast saltmarsh-estuary ecosystem. Northwest Science, v54, pp68-79

7. Kennicutt, M. C., 1988. The effects of biodégradation on crude oil

bulk and molecular composition. Oil and Chemical Pollution, v4, pp89-112

8. Mendelssohn, I. A., M. W. Hester, and C. Sasser, 1990. The effect of a Louisiana crude oil discharge from a pipeline break on the vegetation of a southeast Louisiana brackish marsh. Oil and Chemical Pollution, v7, ppl-15

9. Michel, J., M. O. Hayes, W. J. Sexton, J. C. Gibeaut, and C. B. Henry, 1991. Trends in natural removal of the Exxon Valdez oil spill in Prince William Sound from September 1989 to May 1990. Proceedings of the 1991 International Oil Spill Conference, Ameri-can Petroleum Institute, Washington, D. C , pp 181-187

10. Michel, J., C. B. Henry, W. J. Sexton, and M. O. Hayes, 1990. The Exxon Valdez winter monitoring program results. Proceed-ings for the Conference on Oil Spills, Management and Legislative Implications, Newport, Rhode Island

11. Vandermeulen, J. H. and J. R. Jotcham, 1986. Long-term persis-tence of bunker C fuel oil and revegetation of a north-temperate saltmarsh: Miguasha 1974-1985. Proceedings of the Ninth Annual Arctic Marine Oilspill Program Technical Seminar, Environment Canada, Edmonton, Alberta

12. Vandermeulen, J. H., B. F. N. Long, and F. D'Ozouville, 1981. Geomorphological alteration of a heavily oiled saltmarsh (He Grande, France) as a result of massive cleanup. Proceedings of the 1981 Oil Spill Conference, American Petroleum Institute, Wash-ington, D.C., pp347-351