Ecological Studies of the Sunken Forest,
Fire Island National Seashore, New York

NPS Scientific Monograph No. 7
NPS Logo


In forest ecosystems there may be a close relationship between the distribution of the organic compartment and the distribution of inorganic nutrients. Factors which affect the biomass and production of forested ecosystems usually also affect the patterns of nutrient distribution, particularly in intrasystem nutrient cycles (Ovington 1968; Art 1974). The influence of the organic compartment on the distribution of nutrients is probably at a maximum in the moist tropics where a majority of the exchangeable nutrients in the ecosystems are held in the living biomass (Greenland and Kowal 1960).

The distribution of biomass, primary production, and vegetative surface area within the organic compartment of the Sunken Forest undoubtedly have extremely important effects upon the biogeochemical processes of the ecosystem. A close relationship between organic matter and nutrient distributions would be expected in this ecosystem since the siliceous sands which comprise the soil and rock mineral compartment are highly weathered.

The distribution of primary production between perennial tissues (wood and bark) and annual tissues (leaves and fruit) will necessarily influence the patterns of intrasystem nutrient transfers. The distribution of vegetative surface area in the maritime forest may directly influence not only the intrasystem but also the intersystem cycles since impaction of salt-spray aerosols may be a major nutrient input.

The biomass, net primary production, and vegetative surface area of the Sunken Forest were analyzed not only to discern the relationships between nutrient and organic matter distributions but also to establish the position of the maritime forest in the spectrum of forest ecosystem biomasses and productivities (Art and Marks 1971).


The biomass, productivity, and dimensional relationships of the Sunken Forest were investigated in a 20 X 30-m plot chosen to reflect the relatively undisturbed, mature holly-shadbush-sassafras-blackgum vegetation. This plot was also used for analysis of the nutrient relationships in the Sunken Forest. The plot was located in the eastern portion of the forest 15 m north of the secondary dune crest (Fig. 14). The topography of the plot was fairly regular with an average slope of 17% and aspect of 125°. Mean elevation of the plot was 3.6 m above m.s.l., with a maximum relief of 3.6 m. There were no boggy areas within the plot; however, there was a fresh-water bog 5 m to the east.


The dbh of all 188 trees in the plot were measured to the nearest millimeter. The distance from the top of the crown to the ground was measured to the nearest 0.1 m. The heights of all shrubs in the plot were measured to the nearest centimeter and the diameter at ground level was measured to the nearest millimeter. The herbaceous layer cover was sampled by stratifying the 600 m2 plot into six 10 X 10-m units and randomly sampling six 1 X 1-m plots in each. The percent species cover in each of the 36 herb plots was estimated and the density of liana stems was measured.

Living Biomass

Volume, surface area, and weight relations of the living biomass and net primary production of the vegetation in the plot were determined through various dimension analysis techniques. The methods used were basically modifications of the dimension analysis regression techniques of Whittaker (1961, 1962, 1965, 1966) and Whittaker and Woodwell (1967, 1968, 1969).

Modifications of the regression technique were necessitated largely by the morphology of trees in the maritime forest whose heights are limited by salt-spray impaction and whose stems exhibit frequent forking. Curves relating tree heights to stem diameters exhibit an abrupt truncation between 6 and 7 m of tree height (Fig. 38). Apparent inconsistencies in the relations of tree diameters and stem dimensions made the estimation of biomass, production, and surface areas solely from allometric equations related to dbh a questionable approach. The selection of sample trees which would have accounted for the high variability in stem morphology would have been a difficult if not impossible procedure. Furthermore, sample trees over the entire size ranges of the species in the plot were not available for harvesting outside the Sunken Forest.

Fig. 38. Tree diameter and height relationships.

Thus, the methods were dictated by limitations of time, morphology of the trees, and regulation of harvesting trees in a forest protected by the National Park Service. Due to the relatively small land area and unique character of the Sunken Forest, I felt it would be wholly inappropriate to harvest entire trees or shrubs within its confines. Only detailed measurements of a nondestructive nature were made on the trees and shrubs in the Sunken Forest plot, while individuals from another site on Fire Island were harvested for estimation of variables such as stem production, root biomass, and specific gravity, which could not be measured by nondestructive methods.

Nondestructive Methods

The dimension analysis technique ultimately adopted involved detailed stem surface and volume measurements of the 118 trees in the plot. Sixty-seven of these trees had main stems that forked at least once, while the other 51 had unforked stems. Measurements made on trees with forked stems differed slightly from those made on unforked individuals (Art 1971).

On each unforked tree stem, diameters were measured at meter intervals from the ground surface to the apex to calculate surface area and volume of the stem. Basal diameters of all branches arising from the stem were measured just beyond the branch basal swell and these measurements were used in conjunction with regression formulae to estimate branch dimensions.

On the 67 forked trees the stem dimensions were measured in a similar manner up to the point where the stem forked. At this point the diameters of newly arising stem segments were measured, but only one was randomly selected for further measurement of stem dimensions. This procedure was repeated at each new fork. The stem and branch dimensions of those stem segments not directly measured were estimated from a series of regressions based on analyzed stem segments, relating stem segment basal diameter to stem surface area, stem volume, and branch dimensions.

Harvesting Techniques

In midsummer 1968, series of 15-19 sample branches per species were harvested from Ilex opaca, Amelanchier canadensis, and Sassafras albidum trees outside but in the vicinity of the analysis plot. Regression equations of branch dimensions (age, surface area, numbers of twigs, current and old leaves, and dry weights of wood plus bark, twigs plus leaves, dead wood, old leaves, and fruits) on branch basal diameter were calculated from measurements made on these sample branches (Whittaker 1965; Whittaker and Woodwell 1968).

From each sample branch, three to seven twigs were randomly selected for estimation of mean twig and leaf weights, surface areas, and insect consumption (Appendix II) following procedures of Whittaker (1968) and Whittaker and Woodwell (1968). Leaf areas (one side of leaf) were measured photometrically.

Surface areas of the sample branches were calculated according to general equations developed by Whittaker and Woodwell (1967) which estimate branch surface area from branch basal diameter, number of current twigs, branch length, and mean diameter of current twigs.

Harvesting of whole trees was carried out in July 1969, in the Talisman area on Fire Island, approximately 4 km east of the Sunken Forest. Due to limitations of tree height, the estimation of specific gravities of wood and bark, wood/bark volume ratios, and stem production was from individuals of a smaller range of sizes than in the Sunken Forest plot (Appendix II). Five Ilex opaca, four Sassafras albidum, and four Amelanchier canadensis sample trees were harvested, cut into 0.5-m logs, and measured for dimensions and production in the manner outlined by Whitttker and Woodwell (1968).

Root systems of the 13 sample trees were excavated by hand, washed, weighed fresh, and sub-sampled for dry weight estimation. Root losses in excavation were estimated from a tally of the broken root ends and from regressions of root weights on the basal diameters of completely excavated sample roots (Whittaker and Woodwell 1968).

Tree Biomass and Production Estimation

Branch dimensions were estimated by applying branch regression equations (Appendix II) to basal diameters of branches measured along analyzed stem segments. The volumes and surface areas of analyzed stem segments were calculated from mensurational equations for the frustum of a right cone. The stem and branch dimension estimates were then summed to give totals for each unforked tree and subtotals for each of the forked trees.

The biomass contribution of forked tree stem segments not directly measured was estimated for each species from regressions based on 13-18 analyzed stem segments. The stem surface area, stem volume, and branch dimensions of these analyzed stem segments were calculated and regressions were derived relating these dimensions to the basal diameter of each stem segment (Appendix II). These regressions were applied to the basal diameters of segments not analyzed and the estimates of stem and branch dimensions were added to the subtotals of analyzed segments of the forked trees to give totals.

Stem biomass was estimated by multiplying the total stem volume by the stem specific gravity. The volume and biomass of stem wood and stem bark were estimated by multiplying the total stem volume by the bark volume:stem volume ratio and then multiplying the resultant bark and wood volumes by the bark and wood specific gravities.

The root biomass was estimated for trees in the plot from regressions of root dry weight on tree diameter measured just above the basal swell (Appendix II). Using these regressions, trees in the same clump of sprouts were treated as individual trees each with a separate root system.

Using the 14-18 sample branches, regressions of branch age on branch basal diameter and weight of branch wood plus bark on branch age were calculated to estimate branch production (Appendix II). Production of branch wood plus bark was estimated by first estimating the age of each branch on analyzed stem segments from regression equations. Then the weights of the branch wood plus bark for both the estimated age and the estimated age minus one year were calculated using another set of regression equations. Net branch wood plus bark production was estimated as the difference between these two weights. Branch production of stem segments not analyzed above the basal measurement was estimated from regressions relating total branch production of the stem segment to stem segment diameter of unforked, analyzed stem segments (Appendix II).

Stem wood net primary production was calculated by multiplying the total weight of the wood by the ratio of cross-section area of the current periodic annual increment (mean of last 5 years) to total wood cross-section area at the middle of each log and then summing for all logs in the stem (Whittaker and Woodwell 1968). Stem bark production was estimated using the annual wood weight increment:total wood weight ratio and the bark dry weight of each log and summing for each stem. Regressions relating stem wood and bark production to tree diameter at breast height were then calculated (Whittaker and Woodwell 1968). Compared to the stem weight estimation based on measured stem volume and stem specific gravity, regressions of total stem weight on tree diameter were found to be systematic underestimates. The regressions, based on trees of smaller height and diameter, underestimated the stem biomass of the largest individuals on the plot by as much as 50% compared to the volume X specific gravity estimate. Since the regressions of stem production, like stem biomass, are based on stem weight relations of the sample trees, it is assumed that the stem wood and stem bark production of individual trees is also underestimated by the application of regressions based on the harvested trees.

To correct for this underestimation, a correction coefficient was calculated for each tree by dividing the stem biomass calculated from the product of stem volume and stem specific gravity by stem biomass calculated by the regression formula. Estimates of stem wood and stem bark production based on regressions were multiplied by the correction coefficient. The inability to age roots and to account for losses and turnover of the finer root fractions makes it difficult to measure and estimate tree root productions (Lieth 1968; Newbould 1967, 1968; Bray 1963). Apart from the few studies in which the annual root loss was estimated, net primary production estimates of the total root systems are usually calculated by multiplying the root biomass by the shoot production to shoot biomass ratio, yet there is little evidence to support the assumption that the ratio accurately estimates root production (Newbould 1968). Whittaker (1962) hypothesized root production was underestimated by multiplying the root biomass by the woody shoot production:woody shoot biomass ratio and was overestimated by multiplying the root biomass by the total shoot production (woody and current twig):woody shoot biomass ratio. In the present study, root production was estimated as an average of the two calculations.

To account for minor contributions, the biomass, dimension, and production relations of Nyssa sylvatica tree individuals were estimated by direct stem measurements and application of the Sassafras albidum tree branch, root, stem, and nonanalyzed stem segment regression equations. Direct analysis of N. sylvatica was not undertaken due to the presence of only four trees in the plot and the absence of harvestable individuals at the Talisman site. The Prunus serotina and Pyrus arbutifolia trees in the ecosystem analysis plot were analyzed for biomass and dimension relations by the application of the shrub regressions (Appendix II). Production relations of the branches, stem wood, stem bark, and roots for the trees of these two species were estimated from the mean production:biomass ratios of Ilex opaca, Sassafras albidum, and Amelanchier canadensis.

Shrub and Herb Layers

Shrub layer biomass was estimated from regressions of dimensions on shoot ground-level diameters, following the procedures outlined by Whittaker (1961, 1962, 1963, 1966) and Whittaker and Woodwell (1968). Three shrub-size individuals of Ilex opaca, Amelanchier canadensis, Sassafras albidum, Pyrus arbutifolia, Prunus serotina, Viburnum dentatum, and Vaccinium corymbosum were harvested at Talisman. The root systems of these 21 shrubs were excavated and the root losses estimated. The measurements made on the three individuals, which spanned the shrub diameter size range for each species measured on the plot, served as the basis for regression equations of whole shrub dimension relations on shrub basal diameter (Appendix II). No estimates were made for production of stem wood, stem bark, branch wood and bark, or roots, since only three sample shrubs were harvested for each species and the shrub layer in the ecosystem analysis plot was relatively sparse. No analysis of Rhus radicans or lIex glabra shrubs was undertaken due largely to time limitations.

Dimensions and production of the herbaceous layer were analyzed by the excavation of five individuals of each species within the Sunken Forest. The percent cover of each sample plant was estimated before harvesting. The root systems were washed with distilled water, herbs were separated into roots and shoots, dried at 85°C and weighed. Tree seedlings and other small herbs were dried, weighed, and analyzed whole. A cover:weight ratio was then calculated for each species.

Aerial biomass and leaf area of Smilax rotundifolia lianas were sampled from five individuals carefully pulled out of the tree canopy. The leaves were separated from the stems and the total leaf area of each stem was measured. Leaves and stems were then dried and weighed. Biomass was expressed on a land area basis by multiplying the mean Smilax weight by the density of individuals in the plot.


The basic validity of these data is suggested because the results fall within a range expected from other studies, and because the litter fall data closely corroborate the leaf biomass estimates obtained by dimension analysis. Furthermore, most of the stem volume of each tree in the plot was actually measured rather than estimated from regression formulae based on a small number of sample trees. However, errors associated with the application of dimension analysis methods to the plot were frequently impossible to estimate. The regressions based on sample branches generally had multiple correlation coefficients (r) of 0.95 and above, except for fruits, dead wood, and old leaves.

The regressions used to estimate nonanalyzed stem segment parameters had r's above 0.93, excepting Sassafras albidum branch dimensions which all had r's below 0.40, probably resulting from the great variability noted in the vigor of large trunks and branches of this species. The nonanalyzed stem segment estimates comprised 18%, 8%, and 13% of the stem volumes of Ilex opaca, Sassafras albidum, and Amelanchier canadensis, respectively, and 20%, 26% and, 25% of the total branch weights of these species.

All regressions for dimensions and production for stems and roots were calculated from a relatively small number of sample trees (4-5) due to limitations of time. The regressions of stem wood and bark production on dbh of the sample trees had r's ranging from 0.63 to 0.96, while the regressions for total stem weight on dbh had r's of 0.92 and 0.99. The extrapolations beyond the size ranges of the sample trees compound the errors in the regressions of stem and root biomasses and productions. The estimates of stem bark and stem wood production are not exact, but are as accurate as could be expected under the restraints discussed above.

While extrapolated estimates of stem wood and stem bark production could be corrected for by the ratio of observed stem weight to predicted stem weight, root production estimates are based on root biomass, shoot biomass, and the corrected shoot production estimates. The root biomass estimates for many of the trees are extrapolations for which there was no known correction factor. It is also not known how the root biomass and production may be affected by the salt-spray restrictions of the canopy. Any error in biomass estimation will probably be amplified in the production estimates. Further sources of error in production estimates are the imperfect relationships between tree age and diameter (Table 10) and possible unmeasured site differences between the Sunken Forest ecosystem analysis plot and the Talisman area in which trees and shrubs were harvested.

In light of all the factors contributing to errors, particularly of stem bark, stem wood, and root production, the results of this study must be considered as reasonable approximations of the dimension and production parameters in the Sunken Forest ecosystem analysis plot.

Organic Debris

The lengths and mid-diameters of dead trees and large fallen branches within the plot were measured to estimate volumes. Biomass was estimated by multiplying the volume by live stem specific gravity, a procedure which leads to overestimation since portions of fallen trees have rotted.

Dead twigs associated with live branches were estimated by the separation of dead material from the harvested sample branches used in the regression estimation of living branch biomass. Regressions relating dead material to living branch basal diameter were applied to the branch diameters of analyzed stem segments (Appendix II). Another series of regressions estimated the dead wood on live branches for the nonanalyzed stem segments on forked trees. No attempt was made to estimate entire dead branches in the canopy since their great variability precluded regression estimates.

The soil organic matter was sampled in July 1968 at two points in each of six 10 X 10-m units within the Sunken Forest plot. Soil samples were taken from the upper 15 cm (including litter) and from the 15-30-cm layer using a 3-inch diameter soil sampling tube. The 24 samples were air dried and passed through a 2-mm sieve which retained a fraction that was entirely organic. The >2 mm fraction was dried at 85°C and weighed, while the <2 mm fraction was estimated from the loss on ignition by heating samples to 500°C for 24 hours in a muffle furnace. The loss on ignition technique was used with the awareness that material other than organic matter is lost in the heating of soil samples. This method therefore overestimates soil organic matter. Subsamples of all fractions of the 24 samples were retained for chemical and physical analysis.

Bulk density was measured by excavating five 15 X 15 X 15-cm samples from the surface and subsurface layers, by drying at 85°C and by weighing. Litter fall was sampled in 12 baskets (two in each of six 10 X 10-m units) as described previously.

Cation exchange capacity was determined by the standard ammonium acetate-potassium chloride technique (Wilde et al. 1964).

<<< Previous <<< Contents >>> Next >>>

Last Updated: 21-Oct-2005