1Genetic differences among populations of a keystone species may affect ecosystem functional properties. We tested this by planting Spartina alterniflora from different geographical regions in a newly created salt marsh in Delaware, USA.
2Spartina alterniflora plants from morphologically distinct short-form (back marsh) populations were originally collected from Massachusetts (41°34′ N), Delaware (38°47′ N), and Georgia (31°25′ N) in the USA and vegetatively propagated for 6 years in a salt water-irrigated common garden in Delaware before transfer to a newly created salt marsh.
3The magnitude of the expression of marsh functions in the created marsh, measured over 5 years, remained distinct in patches of each ecotype. End of season aerial biomass, below-ground biomass, root and rhizome distribution, canopy height, stem density, and carbohydrate reserves were closer to values reported for the plants’ native sites than to those typical of Delaware. Thus, many of the plant features characteristic of particular latitudes appear to be under genetic control. Such ecotypic differentiation influences ecosystem function through keystone resource and keystone modifier activities.
4Respiration of the microbial community associated with either dead shoots or the soil varied with plant ecotype in the created wetland and the patterns reflected those reported for their native sites. High edaphic respiration under the Massachusetts ecotype was correlated with the high percentage of sugar in the rhizomes. Edaphic chlorophyll was greater under the canopies of the Massachusetts and Delaware ecotypes than under the Georgia canopy and exhibited a relationship similar to that of algal production rates reported for the native sites. Larval fish were most abundant in pit traps in the Massachusetts ecotype.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Monospecific stands of Spartina alterniflora Loisel. form extensive, almost continuous, salt marshes in low-energy sites along the Atlantic coast of North America from Newfoundland to Florida and along the Gulf of Mexico (Godfrey & Wooten 1979). Thus the S. alterniflora salt marsh system is ideal for determining if plant ecotypic variation plays a role in regulating the intensity of various ecosystem functions over a latitudinal gradient.
Ecotypic variation has been reported in marine organisms in the open ocean, as well as on the shorelines of the seas. The cyanobacterium Prochlorococcus exhibits ecotypic differences related to different light and nutrient regimes associated with depth in the water column (Moore et al. 1998). Gallagher & Alberte (1985) have also seen ecotypic adaptation in an abundant marine diatom, Skeletonema costatum. The existence and distribution of ecotypes clearly enables a species to live over a wider range of habitats, but the impact of ecotypic variation on function at the ecosystem level is unclear and the role ecotypes play in regulating the intensities of functions in open water where water column dynamics are complex is difficult to assess. However, rooted plants in intertidal areas offer stability in space and can be experimentally manipulated, and the regulation of ecosystem function by ecotypic variation within a keystone (Mills et al. 1993) species can be tested.
Although marsh ecosystems depend on appropriate hydrology, they are regulated by vascular plants, directly via carbon and nutrient inputs and also through the impact of their 3-D structure on the resource distribution. Plant shoots fuel the detritus-based food web and act as conduits for the movement of oxygen (Armstrong 1978) and carbon compounds (Gallagher & Plumley 1979) below ground, thereby supporting soil community processes. The depth of rooting determines the distribution of photosynthate below ground and, through the extensive aerenchyma in the tissues, the location and extent of the connection between soil and air. Additionally, vascular plants stabilize the sediment and affect light quantity and quality reaching the soil surface, thus influencing the quantity and timing of edaphic algal growth. Likewise, the plants modify wetland soil temperature (Gallagher 1971) and water table depth (Dacey & Howes 1984). Salt marsh functions critical to living marine resources (sources of food, places of refuge from physical exposure as well as predation, sites for nesting and egg development, and settings for nurseries) are all directly or indirectly related to the vascular plants, whose variation may be genetically controlled and independent of the immediate environment.
Five to seven sods of 20–40 culms each of short-form S. alterniflora were collected haphazardly in 1983 from approximately 1-ha sites on the coasts of Massachusetts, Delaware, and Georgia. Following propagation in the greenhouse for one winter and growth in a common saline water-irrigated garden in Delaware for 6 years, the plants maintained the phenotypic differences characteristic of their latitude. The populations were therefore regarded as ecotypes whose effects on ecosystem functions could be tested in a newly created mid-latitude salt marsh, thereby increasing understanding of the relative importance of plant genetics and environment in the functioning of a salt marsh ecosystem.
We asked whether, when the three ecotypes are transplanted into the field, their morphological characteristics, the amounts of organic matter produced above ground, and the quantity and distribution of that allocated below ground are more like those of natural populations near the transplant site or the site of origin. Aerial primary productivity largely determines the potential quantity of detritus that sets the bottom up control on the aerial and aquatic detritus-based components of the ecosystem. The quantity and distribution of carbon resources translocated below ground determines the quantity and location of fuel available for heterotrophic activity in the soil. Further, non-structural carbohydrate stored in the rhizomes is important for rapid spring growth and the repair of damage to the canopy, whose structure, especially height and stem density, influences access and usage of the marsh surface by the marsh fauna.
We also determined whether other components of the ecosystem were affected by the ecotype of the dominant plant, by comparing edaphic chlorophyll, respiration of the edaphic community, that of the microbial community associated with dead shoots, and fish use within the created wetland. Where possible we also compared those parameters with reports from the sites of origin.
In the fall of 1989, the Halophyte Biotechnology Center (HBC) marsh was formed when 0.2 ha of abandoned farmland was excavated to the level of the intertidal elevation in an adjacent natural marsh (Canary Creek Marsh near Lewes, DE, on the mid-Atlantic coast of the United States). The newly exposed substrate consisted of muddy sand and silty clay layers over coarse sand and gravel (Scotts Corner Formation). A creek-pool system was dug (Fig. 1a) to connect the site, via two tidal creeks, through a narrow upland strip, to creeks of Canary Creek Marsh (Fig. 1b).
Three phenotypically distinct populations of short-form S. alterniflora (originally from Great Sippewissett, MA, 41°34′ N (Fig. 1c), Canary Creek, DE, 38°47′ N, and Sapelo Island, GA, 31°25′ N (Fig. 1d)), were planted in patches in the marsh plain around the pool creek system. Two tall-form populations and three patches of Distichlis spicata populations were also planted in the plain and patches of four populations of Spartina patens were planted along the upper edge of the S. alterniflora patches where the marsh sloped to the upland. The maximum elevation difference within the marsh plain was 15 cm. We focused on short-form S. alterniflora which dominates adjacent sites at this elevation. Populations in the created marsh were designated MA-C, DE-C, and GA-C (for comparison with data from MA-N, DE-N, and GA-N populations in their native locations).
Four replicate patches of each S. alterniflora ecotype were interspersed across the marsh plain with plants at 1 m centres. Fibreglass panels 30 cm deep were placed between patches, such that the top of the panel was level with the soil surface, to prevent vegetative contamination between patches. Also, seedlings were removed each spring to maintain ecotypic purity. Patch size was variable (average 40 m2) but each covered a portion of the marsh plain from the pool–creek system to the S. patens at the marsh–upland border. Although this design was not strictly random, it ensured that tidal water flow and fish movements did not require passage through plants of another population. Over 5 years, we compared ecosystem functions among the short-form ecotypes planted in the HBC marsh. In addition, comparisons were made to literature values for plants at or very near the three original collection sites.
ABOVE- AND BELOW-GROUND ORGANIC MATTER AND MORPHOMETRIC MEASUREMENTS OF THE CANOPy
Plants were harvested from the HBC marsh in late August, 1994 by cutting a previously undisturbed 0.1 m2 plot at the marsh surface in each of the four replicate patches of each ecotype. Plant height, stem density, leaf length, leaf width, internode length, and stem diameter were determined and above-ground biomass was measured after drying at 60 °C to constant weight. Roots and rhizomes were separated from sectioned soil cores (15 cm diameter, 50 cm long), washed, and dry weights were obtained.
Samples of cleaned rhizomes from each of the biomass cores were dried for 1 hour at 100 °C then at 60 °C to a constant weight. Dried samples were ground in a Wiley Mill to pass a 60 mesh screen. Non-structural carbohydrates were determined using the anthrone/sulphuric acid colorimetric method described by Whistler et al. (1962).
EDAPHIC ALGAL CHLOROPHYLL
Two soil cores were taken from under the canopy in each patch (eight cores × three ecotypes). Sampling was conducted three times during the year. Algal growth is very patchy and to avoid bias two 0.5 × 0.5 m subplots in a permanent 1 × 3 m plot were selected using a table of random numbers. The corer (an aluminium pipe 1 cm in diameter and 3 cm long) was tossed into each subplot and the sample taken where its sharpened end landed. Cores were wrapped in aluminium foil and frozen for later analysis. Chlorophyll, used as a measure of edaphic algae, was extracted with 90% acetone from the upper 0.5 cm of each core. Absorption was measured spectrophotometrically at 663 nm before and after acidification with 10% HCl, as described by Gallagher (1971).
DEAD SHOOT MICROBIAL RESPIRATION
Oxygen uptake rates were measured with a Digital Oxygen Meter (YSI Model 58) according to the methods of Gallagher & Pfeiffer (1977). Dead leaves were collected in May from 8 to 12 cm above the soil surface from at least 20 plants scattered throughout the patch. Samples were thoroughly mixed prior to placing a 2-g (fresh weight) subsample into a BOD bottle which was subsequently filled with tidal creek water. Respiration rates were calculated on a dry weight basis.
Carbon dioxide efflux from the soil was measured in late August as in situ emission over 90 min within a 200-mL headspace of a darkened chamber that was sunk 8 cm into the sediment (modified from Howes et al. 1985). Chambers were allowed to equilibrate open for 30 min prior to time-zero sampling. Gas samples were analysed with an infra-red gas analyser, calibrated with a 1% CO2 standard.
ANIMAL USE OF THE MARSH
Pit traps (36 × 28 cm plastic food storage containers) were placed such that their tops were level with the marsh surface to simulate small pools that serve as intertidal refuges for larval fish. We had previously determined that the distance to open water affected catch, therefore each trap was placed 1 m from the edge of the central pool. A Zeiss automatic level and metric stadia rod were used during placement to verify that all traps were at the same elevation (±2 cm). The traps were fished from 5 May to 13 September (the period of larval presence) for 16 night-time high tides during six spring tide periods (STP). Traps were emptied before each high tide and larvae were counted at the subsequent low tide. The mean number of fish caught by a particular trap during a particular STP was calculated and those values for each STP were summed to give a measure of each trap’s fishing success for the period of larval presence.
One-way analyses of variance (anova) were followed by Fisher’s Protected LSD test (α = 0.05) using systat software to determine statistical differences among ecotypes. Data were examined for outliers and were transformed, where appropriate.
ORGANIC MATTER, ABOVE AND BELOW GROUND
In most marshes, biomass reaches within 25 g/m2 of its peak by late August (Kibby et al. 1980) and August numbers are therefore often used as estimates of annual production. In all years, the most southern (Georgia) ecotype produced significantly greater (P ≤ 0.05) above-ground biomass in the HBC marsh than the most northern one (MA), with the DE-C intermediate (1994 shown in Fig. 2a). Relationships recorded at the sites of origin (Gallagher 1975; Gallagher & Howarth 1987) were maintained in the common environment (Fig. 2a).
The amount of below-ground organic matter and its distribution in the HBC marsh reflected patterns typical of the native sites (Figs 2b and 3; because there had been little accumulation of dead macro-organic matter in the relatively new HBC marsh, we made comparisons only to live material reported in the literature studies). Our results suggest that the ecotype is more important than the local soil environment. DE-C and GA-C produced significantly more below-ground biomass than MA-C (Fig. 2b). MA-C did not develop roots below 15 cm, whereas DE-C allocated 12% of its root and rhizome biomass below 15 cm with none below 30 cm, and GA-C allocated 25% between 15 and 30 cm and 8% below 30 cm (Fig. 3).
Percentage sugar in rhizomes of the MA-C ecotype was significantly greater than that in the other two ecotypes, as was total non-structural carbohydrate concentration (Table 1). The pattern found by Gallagher & Howarth (1987), measuring recoverable underground reserves rather than using a chemical extraction, was similar (Table 1). The recoverable reserve method, in which cores are collected from natural plant stands, the above-ground portion of the shoots are removed, the cores are kept in the dark, and regrowth is periodically harvested and summed until no more is produced, correlates well with the chemical extraction method on the same species (Hellings 1990).
Table 1. Carbohydrate reserves in short-form Spartina alterniflora from Massachusetts, Delaware, and Georgia in the HBC marsh and native marsh sites
HBC marsh non-structural (NS) carbohydrates
Native site recoverable reserves
Total NS carbohydrates (%)
g dwt m−2
HBC marsh, a man-made marsh constructed at the Halophyte Biotechnology Center, Delaware, USA. Values are means ± SE, N = 4. If values are followed by the same letter (in columns), they are not significantly different from one another at α = 0.05, using one-way anova and Fisher’s LSD test. Literature values for the native sites: Gallagher & Howarth (1987).
30.1 ± 1.1b
51.4 ± 2.0b
23.1 ± 1.8a
40.4 ± 3.1a
21.9 ± 1.7a
40.4 ± 2.5a
As at their native sites, canopy height of ecotypes in the HBC marsh increased as the source latitude decreased (Fig. 4a), whereas stem density decreased (Fig. 4b). For all four morphological characters (leaf length, leaf width, internode length, and stem diameter) the GA-C ecotype was statistically significantly greater than MA-C and, except for stem diameter, was greater than DE-C as well. Leaf width and stem diameter were significantly greater in DE-C than in MA-C. Table 2 shows the results for a typical year.
Table 2. Plant morphological characteristics for short-form Spartina alterniflora from Massachusetts, Delaware, and Georgia planted in the HBC marsh
Based on chlorophyll measurements, edaphic algae were more abundant under the canopies of MA-C and DE-C ecotypes than under that of GA-C (Table 3). Although chlorophyll data are not available for two of the native sites, algal production values measured in short-form S. alterniflora marshes in Massachusetts, Delaware, and Georgia similarly show a lower value for Georgia than for Massachusetts and Delaware (Table 3). This is probably caused by less light penetrating the taller and more robust plants of the GA-C ecotype (Fig. 4; Table 2), with characteristics such as leaf dimensions increasing and leaf angle becoming less vertical progressively from MA-C to DE-C to GA-C.
Table 3. Edaphic algae under canopies of Spartina alterniflora from Massachusetts, Delaware, and Georgia grown in the HBC marsh and in native natural marsh sites
Microbial decomposers are drivers of the detritus-based food web and decomposition rates for above-ground plant material depend on the ecotype serving as substrate. Respiration rates of the microbial community associated with dead shoots in May were greatest for the MA-C ecotype and least for GA-C (P ≤ 0.05; Fig. 5a). Respiration rates were measured using the same method in two of the native marshes, with rates at DE-N higher, as expected for the more northern ecotype, than at GA-N (Fig. 5a). The GA-N plants showed almost exactly the same rate as the transplanted GA-C ecotype.
The same relationship held true for respiration of the heterotrophic soil community as it did for the dead aerial parts of the plant. Respiration was greatest under the MA-C ecotype. Soil respiration rate under MA-C was approximately twice that for DE-C and GA-C ecotypes (Fig. 5b, repeated with similar results four years later). Reported values for the native sites showed a similar, although stronger, pattern with DE-N and GA-N exhibiting rates considerably less than that of MA-N.
ANIMAL USE OF THE MARSH
More than twice as many Fundulus larvae were caught in traps in MA-C than in those of either of the other two ecotypes, despite being set at the same elevation and having the same submergence time (Table 4).
Table 4. Larval fish (Fundulus heteroclitus) caught in pit traps on the HBC marsh surface under canopies of short-form Spartina alterniflora ecotypes from Massachusetts, Delaware, and Georgia during night spring high tides (May to September)
Values are sums of means of fish caught per trap during each of six spring tide periods (16 tides). If values are followed by the same letter, they are not significantly different from one another at α = 0.05, using one-way anova and Fisher’s LSD test.
90.5 ± 11.7b
37.7 ± 9.0a
43.7 ± 11.0a
Variation in intensity of the expression of the functional properties of ecosystems may be attributed to four driving forces: (i) the biotic and abiotic structure of the overall system and its various sub-habitats; (ii) the climate under which the ecosystem developed and is operating; (iii) synergistic or antagonistic interactions between the ecosystem and the adjacent landscape; and (iv) age of the ecosystem. Spatial heterogeneity in the expression of ecosystem functional properties within an extensive ecosystem is often attributed to environmental dissimilarities or to the influence of keystone species which are crucial in maintaining community organization (Mills et al. 1993). S. alterniflora is a keystone species in the salt marshes along the Atlantic coast of North America.
Figure 6 depicts how the phenotype of Species A, whose genetic potential has been shaped by historic environmental forces also exhibits plastic responses moulded by current environmental forces (italics indicate components in Fig. 6).
The species demonstrates both keystone resource activity, in serving as a resource to fuel other facets of the food web, and keystone modifier activity, in modifying the environment and thereby favouring or inhibiting various other ecosystem biota.
By growing S. alterniflora plants from Massachusetts, Delaware, and Georgia for 6 years in a common garden in Delaware, and subsequently transplanting them to an artificial wetland, many of the variables between the sites of plant origin (abiotic and biotic influences, climate, interactions with adjacent landscapes) were effectively neutralized, as were differences in wetland age. The source of colonizing microbes and animals was also the same, leaving only genetic dissimilarity among the three sources of S. alterniflora to directly or indirectly lead to differences in the intensity of the functions among the marsh patches.
The genetic regulation of the environment-driven plasticity of Species A limits the phenotype of the keystone species. The genetic variation within S. alterniflora, is relatively subtle and covaries with the environment along a latitudinal gradient. Because only intensity, not the type of functions, changes along the latitudinal gradient, the importance of the genetic differences is often not obvious. Differential morphological and physiological plasticity of ecotypes nevertheless affects functioning as a secondary effect of the current environmental forces acting on the genetic potential. Historic environmental forces, as they varied along the environmental gradient, determined the evolution of present-day ecotypes and the consequent genetic regulation of plastic responses, although the interactions are not clear. Depth of rooting, for instance, may have evolved for compatibility with soil differences: in Massachusetts the back marsh soils are high in organic matter and have a relatively shallow rocky base, whereas in Georgia the soils are high in silt and clay and overlay a deep sand base (Gallagher et al. 1977). Therefore the keystone modifier activity of producing a rhizosphere would extend to a greater depth in the southern ecotypes, as shown in Fig. 3.
Keystone resource activity differences among ecotypes may relate to differences in temperature. The growing season in Georgia is much longer than in Massachusetts, where large quantities of carbohydrate are needed to sustain overwintering rhizomes and supply very rapid spring growth. Thus selection pressure will favour more allocation to below-ground storage in the north and less in the south, where large aerial structures maximize photosynthesis, allowing plants to outgrow competitors during the long growing season (Gallagher & Howarth 1987).
Regardless of which historic environmental forces acted to set the bounds for resource and modifier characteristics, some traits exhibit little plasticity and hence are largely independent of the current environment. Morphologically, S. alterniflora populations at HBC more closely resembled plants at the site of their evolutionary origin than those in the adjacent Delaware population. S. alterniflora therefore appears to regulate the intensity of some ecosystem functions quite independently of the local environmental.
The genotype of the plant (MA, DE, GA) had a major impact on the keystone resource activity both above and below ground (Fig. 2). These salt marshes are largely based on a detritus food web, and are thus dependent upon S. alterniflora as a carbon and energy source. The end of season above-ground biomass approximates the detritus production potential, and was significantly lower in MA-C plants than those of GA-C, with DE-C intermediate. In Fig. 6 this is represented as a phenotype-regulated keystone resource activity that regulates the detritus which in turn limits the intensity of expression of the detritus food web.
Detritus quality, which was assessed indirectly by measuring the respiratory rate of the microbes colonizing the degrading grasses, Fig. 5, may also be affected. GA-C, which produces the greatest amount of biomass, has less intense decomposer activity than the MA-C ecotype, which produces the least potential detritus. Thus, genetic regulation of differences in detritus production and quality is a bottom-up factor in the regulation of the intensity of the detritus portion of the food web along the east coast of the United States. Grazing was not quantified, but any differences in shoot quantity and quality in the three ecotypes would affect flux through the shoot grazer portion of the food web (Fig. 6).
The S. alterniflora three-dimensional canopy also exhibits keystone modifier activity (Mills et al. 1993). Light quantity and quality is altered as it passes through leaves which absorb and reflect portions of the spectrum (Pfeiffer et al. 1973) and these changes affect soil temperature which in turn changes evaporation and therefore interstitial moisture and salinity concentrations (Gallagher & Daiber 1974b). Differences among ecotypes thus influence the microenvironment of the edaphic and epiphytic algae (algal habitat structure) and these modified environmental forces, together with differences in the plant surfaces where many algal species grow (Gallagher 1971), are among the factors governing algal production (Gallagher & Daiber 1974b). Shifts in microalgal production and species composition therefore regulate the intensity of the flow through the algal grazer food web, with edaphic chlorophyll greatest under the short MA-C canopy, where larval fish were most abundant, and least under the taller GA-C.
Root and rhizome quantity and quality play similar keystone resource activity roles in the development of the soil community (Gallagher & Plumley 1979; Howes et al. 1984) to those played by the above-ground plant parts for the zone that alternates between aerial and aquatic. Underground parts of S. alterniflora, along with buried leaves and edaphic algae, fuel the soil food web. A strong correlation between below-ground organic matter and the density of infauna has been reported (Minello & Zimmerman 1992). The plant also has a keystone modifier activity (Fig. 6) in these anoxic or hypoxic soils, where rhizomes and roots carry oxygen from the shoots and thus produce oxidized microenvironments in the rhizosphere habitat (Teal & Kanwisher 1966; Armstrong 1978; Mendelssohn & Postek 1982). Hence ecotypic differences in root system distribution regulate energy flow and concurrent mineral cycling.
Edaphic respiration rates suggest that modifier and resource activities influence the soil food web. The MA-C ecotype supports the most soil respiratory activity: although it has 100% of its root and rhizome material in the top 15 cm of the soil, it has no more absolute biomass there than the GA-C ecotype (both approximately 1300 g/m2) and much less than DE-C (almost 2000 g/m2). The amount of carbon available to support heterotrophic activity cannot therefore explain the lower respiration in DE-C and GA-C. The correlation of the high edaphic heterotrophy in stands of MA-C with a high rhizome sugar content indicates that the composition of the underground material may be more important than its quantity (e.g. sugar is a more labile energy source than starch for the edaphic food web). Oxygen distribution and the microdistribution of root or rhizome material are also likely factors.
The canopy provides animal habitat structure, e.g. refuge from predators and sites for feeding and nursery grounds for fish and other salt marsh fauna (Kneib & Stiven 1978; Weisberg & Lotrich 1982; Smith & Able 1994; West & Zedler 2000), as well as determining food web structure and resource distribution (Minello & Zimmerman 1983; Kneib 1984; Kneib 1987; Costa et al. 1994). Plant height, for example, is important in determining appropriate habitat for bird nesting (Zedler 1993), and stem density is a significant factor in providing marsh fauna with protection from predation (Minello & Zimmerman 1983, 1992). Keystone modifier activity is seen, for instance, in Massachusetts, where foraging by Fundulus was more successful in low elevation, low density stands of S. alterniflora than in high elevation, high density Spartina patens (Vince et al. 1976). In the HBC marsh, stem density is higher and internode lengths shorter in the MA than GA and DE ecotypes: the density of larval mummichogs, Fundulus heteroclitus, is approximately twice as great in areas planted with MA-C, confirming its greater refuge value.
If the current environmental forces are beyond those in which Species A is competitive it will be replaced by Species B (Fig. 6), as seen where purple loosestrife (Lythrum salicaria) replaces cattail (Typha species) (Weihe & Neely 1997; Grabas & Laverty 1999) or where S. alterniflora is replaced by Phragmites australis in wetlands (Weinstein & Balletto 1999). A change of this magnitude can have major impacts: changing functions, not just altering their intensity as we have seen in this study. The new keystone species (B) will, of course, have its own genetic potential and plastic responses shaped by historic environmental forces and driven by current ones.
Our examination of the structure and internal functioning of a salt marsh initiated using ecotypes of S. alterniflora from MA, DE, and GA, coupled with information on the source populations, demonstrates that genetic variation in this grass plays a major role in determining functional emphasis within salt marshes along the Atlantic coast. The historic environment at a site has shaped the ecotype present, whereas the current environment influences functional relationships within the plasticity limits of that ecotype. Based on this salt marsh example, the expression of ecosystem functions in other ecosystems is probably manipulated by ecotypic variation in the structure and physiology of the dominant plant.
Support for this research came from the Coastal Ocean Program Office of the National Oceanic and Atmospheric Administration through Grant No. NA90AA-D-SG457 to the University of Delaware Sea Grant Program. We are grateful to Jinglan Wu, Antonia Wijte, Divakar Rao, Pam Morgan, Xianggan Li, and Wendy Carey for their assistance in the field.