Ecology. 73(4), 1992, pp. 1182-1 193 8 1992 by the Ecological Society of America EFFECTS OF FLOODING ON ROOT AND SHOOT PRODUCTION OF BALD CYPRESS IN LARGE EXPERIMENTAL ENCLOSURES1 J. PATRICK MEGONIGAL~ Savannah River Ecology Laboratory, Aiken, South Carolina 29801 USA FRANK P. DAY Department of Biology, Old Dominion University, Norfolk, Virginia 23529 USA Abstract. Effects of hydroperiod on the root production of bald cypress (Taxodium distichum) saplings were determined in large (8.0 m2 x 1.5 m deep) watertight enclosures over three growing seasons. Our objectives were to determine the effect of continuous and periodic flooding regimes on biomass production, carbon allocation to roots and shoots, and root-system morphology. The effect of the flooding treatments on plant biomass was different for 1-yr-old seedlings and 3-yr-old saplings. After one growing season, root and shoot biomass was highest in the periodically flooded (PF) treatment. After three growing seasons there were no significant differences in total biomass but there were differences in root-to-shoot ratios. Improved growth in the continuously flooded (CF) treatment began in the second growing season and coincided with morphological adaptations to flooding. Such adaptations include the production of water roots, development of intercellular air spaces, and distinctly different root-system morphologies. Periodically flooded cypress allocated more carbon to roots than did continuously flooded cypress and developed deeper root systems. A relatively deep rooting zone may have provided the P F saplings access to water and dissolved nutrients within the water table (50-60 cm deep during summer). Continuously flooded plants had low root-to-shoot ratios and shallow root systems. A relatively shallow rooting zone with ample water and nutrients allowed CF cypress to allocate relatively more biomass to leaves. After 3 yr, total productivity in the two treatments was not significantly different, yet belowground production was greater in periodically flooded saplings (P = .05) and there was a tendency for higher aboveground production in continuously flooded saplings (P = .14). Without the belowground production estimates we might have concluded that CF plants were more productive than PF plants. Most plants can respond to changing resource availabilities by shifting the allocation of carbohydrates to roots or shoots. Because resource availability in freshwater forested wetland ecosystems can be highly variable, studies of production should include estimates of root production. Key words: adventitious roots; aerenchyma; carbon allocation; flooding; productivity; rhizotron; rooting depth; root production; roots; root : shoot; root system morphology; Taxodium distichum. The magnitude, duration, and timing of flood events are important factors regulating primary productivity in wetland forests. The effects of flooding on primary productivity have been investigated in many wetland forest ecosystems. Few such studies have considered root production, despite evidence that it comprises a significant portion of primary production in forest eco- systems (Hams et al. 1980, Megonigal and Day 1988, Raich and Nadelhoffer 1989). Carbon is allocated to roots at the expense of leaves, stems, reproductive tis- sues, and secondary compounds (Mooney 1972, War- ing and Schlesinger 1985, Tilman 1988). Factors that promote the allocation of energy and nutrient resources I Manuscript received 19 November 1 990; revised 22 Au- gust 199 1; accepted 1 1 September 199 1. Present address: Department of Botany, Duke University, Durham, North Carolina 27706 USA. to roots necessarily reduce potential aboveground pro- duction. A thorough understanding of the factors af- fecting the productivity of wetland forests requires knowledge of their effect on carbon allocation to roots (Brinson et al. 198 1, Vogt et al. 1986). It is difficult to assess root production without se- verely disturbing the soil (Bohm 1979). Root produc- tion studies in wetland forests are particularly prob- lematic because it is difficult to manipulate the water table and flooding depth. Greenhouse or growth-cham- ber experiments offer control over flooding depth, but usually they are of short duration and suffer from a limited soil volume. In greenhouse studies, most tree species respond to continuous flooding with reduced rates of growth rel- ative to periodic flooding treatments (Kozlowski 1984); however, extremely flood-tolerant tree species such as bald cypress (Taxodium distichum [L.] Richard) oc- casionally have shown improved growth under contin- August 1992 CARBON ALLOCATION IN WETLAND RHIZOTRONS 1183 TABLE 1. Biomass of flooded bald cypress seedlings relative to the driest treatment in the study. Flooding increased (+), decreased (-), or had no effect (0) on biomass. Flooding treatments are relative to the soil surface. All differences are significant at P 5 .05. Effect relative to the driest treatment Driest treatment Treatment for com~arison Shoot Root Total Citation Field capacity 1 cm above surface, 0, + 0 Dickson and Broyer 1972 1 cm above surface, no 0, + - Dickson and Broyer 1972 10 cm below surface At the surface 5 cm below surface 0 McLeod and Sherrod 198 1 0 McLeod and Sherrod 1 98 1 Drained, watered 2 cm above surface - - - Shanklin and Kozlowski 1985 6 cm below surface At the surface 0 - - McLeod et al. 1986 6 cm above surface - - - McLeod et al. 1986 6 cm below surface At the surface 0 - Donovan et al. 1988 6 cm above surface 0 - Donovan et al. 1988 Drained, watered 6 cm above surface - - - Donovan et al. 1989 12 cm above surface - - - Donovan et al. 1989 uous flooding (Table 1). Field studies have shown that bald cypress production decreases on extremely wet or extremely dry sites, but most do not consider be- lowground production. Thus, it is reasonable to ask: Is the net primary production (shoots + roots) of ex- treme hydrophytes such as bald cypress comparable on continuously flooded and periodically flooded sites? The objectives of our study were to determine ex- perimentally the effects of periodic and continuous flooding on (1) biomass allocation to roots and shoots, (2) vertical root distribution, and (3) root system struc- ture in young bald cypress trees over a 3-yr period. MATERIALS AND METHODS Rhizotron cell design Two wetland rhizotron cells (enclosures for root- growth experiments) were constructed during the win- ter and spring of 1986 in the Rhizotron Facility at the University of Georgia's Savannah River Ecology Lab- oratory (Day et al. 1989). The two cells were directly adjacent, and there was no differential shading or pro- tection from wind. The cells were in full sunlight. Each cell was 2.83 x 2.83 x 1.5 m deep, impermeable to water, and open to the atmosphere. A pipe in the bot- tom of each cell allowed soil water to drain at a con- trolled rate. The cells were leached with water and checked for leaks before they were filled with soil. The reconstructed soil profiles consisted of 100 cm of a sapric Histosol underlain by 10 cm of sand and 10 cm of gravel. The soil was from a former cypress wetland near Jacksonboro, South Carolina. Two oxygen cham- bers (Carter et al. 1984) and three fused platinum- copper redox probes (Faulkner et al. 1989) were in- stalled in the soil profile of each cell at 20 cm depth intervals. This interval was considered adequate to de- scribe the soil profile. Rainfall was measured adjacent to the cells. Because hydroperiod is the most important factor affecting wetland structure and function, we gave spe- cial attention to the hydraulic design of the experiment and to the hydroperiod treatments. One cell was con- tinuously flooded and one was periodically flooded in a manner typical of many southeastern forested wet- lands (Fig. 1). Natural seepage and turnover of soil water was simulated by allowing -20 L/d to drain from each cell. Based on output and storage volumes (Table 2), the soil water pool turned over at roughly 9-mo intervals. Creek water was pumped from a polyethyl- ene storage tank and sprinkled onto the soil surface to replace water draining from the bottom of the cells. A system of wells, floats, solenoid switches, and pumps prevented the water table from falling below a pre- scribed depth. The prescribed depth was +20 cm (above the soil surface) in the CF cell and variable in the PF cell according to the following schedule: - 20 cm from March to May and October to December, - 50 cm from June to September, and + 20 cm from January to Feb- ruary. However, the water table often was higher than the scheduled level because of rainfall events (Fig. 1). The hydroperiod treatments were initiated on 23 May 1986 and maintained until 20 October 1988. Experimental design Bald cypress seeds were germinated in January 1986. The seedlings were grown in a greenhouse and con- ditioned to above-surface flooding for 1 mo before they were transplanted to the rhizotron cells on 9 May 1986. Twenty-five seedlings were randomly assigned to each cell using a random numbers table. Another 25 plants were used for initial biomass estimates. There was no significant difference in height (mean -t 1 SD = 34 k 5 cm) among the three groups (ANOVA, P = .23). To minimize competitive effects, the seedlings were reg- ularly spaced at 5 1 -cm intervals in a matrix of 5 col- umns and 5 rows; seedlings on outside rows were 51 cm from the wall. J. PATRICK MEGONIGAL AND FRANK P. DAY Ecology, Vol. 73, No. 4 M J J A S O N D J F M A M J J A S O N D J F M A M J J A S O 1986 1987 1988 FIG. 1. Water table depth and rainfall volume for the 30-mo study period Soil oxygen content and redox potential (Eh) were determined at 20 cm depth intervals both prior to the initiation of hydroperiod treatments and at 2-wk intervals after treatment initiation. A sample of plants (n = 8 in 1986 and 1988; n = 9 in 1987) was harvested each October, at the end of the growing season but prior to leaf abscission. Plants were harvested in a pattern designed to maximize inter-plant distances among the remaining individuals. Shoots were sepa- rated into leaves, branches, and stems. Roots and soil were removed from a 30 x 30 cm plot, centered on the bole of the plant, to a soil depth of 30 cm. Exca- vation was by hand with a small trowel. The holes were back-filled with fresh soil from the original source. Samples were frozen and stored until they could be processed; freezing had no apparent effect on the color or tensile strength of the roots. After thawing, the roots were washed, separated from organic debris, air-dried, and separated into size classes. Soil samples were washed gently with tap water through a standard no. 10 (2 mm-mesh) sieve. Care was taken during the washing process to recover (with tweezers) most of the fine roots; however, an unknown fraction of the 1-2 mm root size class was undoubtedly lost. Dead roots were separated from a subset of the samples taken in the third year. Dead roots occurred in the fine size class TABLE 2. Total inputs and outputs of water and selected nutrients for the two rhizotron cells (enclosures for root-growth experiments) during the 30-mo study. CF = continuously flooded, PF = periodically flooded. Input Output Storage* Component of soil solutiont CF PF CF PF CF PF * Storage is the product of average nutrient concentration and total soil water mass as determined from saturated bulk density cores. t Total N was determined by Kjeldahl digestion. Concentrations of the other nutrients were determined after filtration (0.45-pm mesh filter) by inductively coupled plasma emission spectroscopy (Ca, Mg, Fe, and Mn) or colorimetric absorbance (NH,-N and PO,-P). See Day et al. (1989) for details on nutrient analyses. August 1992 CARBON ALLOCATION IN WETLAND RHIZOTRONS 1185 800 - 40 crn 600 -- Pf X / 80 crn 0-OCONTINUOUS FLOODING I 600 1 .-.PERIODIC FLOODING - 2 0 0 1 1 I I I 1 I : : : : : : : : l MAY SEP JAN MAY SEP JAN MAY SEP 1986 1987 1988 MAY SEP JAN MAY SEP JAN MAY SEP 1986 1987 1988 FIG. 2. Redox potential and oxygen content at three depths. Redox potentials below 200-400 mV (-) generally reflect anaerobic conditions (Bohn 197 1). = periodic flooding; 0 = continuous flooding. (< 1 mm) only and accounted for an insignificant frac- tion of the total fine root mass. All samples were oven- dried at 65?C and cooled in a desiccator before weigh- ing. In October 1988 the cells were completely excavated to determine the total root biomass of each plant. The soils were removed in three depth increments (0-30 cm, 30-60 cm, and 60-100 cm). Roots were not sev- ered until they were entirely exposed so that all root biomass estimates could be associated with individual plants. The roots of plants removed in 1987 were iden- tified by their location, color, and flexibility. They were collected as a single batch sample. These roots were included in the production estimates but not the bio- mass estimates. Fine root biomass in the surface 30 cm was estimated from 20 random soil cores (9.6 cm diameter). Random sampling coordinates were chosen with a pseudo-random number generator. Statistical analyses A limited number of treatment means were com- pared with t tests (SAS 1985). Analysis of covariance (ANCOVA) and Type I sums of squares were used to test for homogeneity of slopes and differences in the Y intercept. Differences in rates of production and rela- tive growth are based on the time x treatment inter- action term in an Analysis of Variance (ANOVA) as described by Poorter and Lewis (1986). RESULTS Environmental conditions The water level was consistently 10-20 cm above the soil surface in the continuously flooded (CF) treat- ment. The water table varied in the periodically flooded (PF) treatment according to the schedule of maximum water-table depth and rain events (Fig. 1). Redox po- tential in the CF treatment fell to an average of - 18 mV at 20 cm depth for the entire study. The PF treat- ment had an average redox potential of 5 1 1 mV at 20 cm (Fig. 2). Redox potentials lower than 200-400 mV generally reflect aerobic conditions (Bohn 197 1). Ox- ygen concentrations at 20 cm averaged 1% in the CF cell and 16% in the PF cell. Thus, the CF treatment was nearly always anaerobic and reduced near the sur- face, while the PF treatment was nearly always aerobic (or hypoxic) and oxidized. J. PATRICK MEGONIGAL AND FRANK P. DAY Ecology, Vol. 73, No. 4 ABOVEGROUND BIOMASS haBRANCH 1986 * BOLE 0 LEAF T O T A L CONTINUOUS PERIODIC FLOODING FLOODING FIG. 3. Aboveground biomass (means and 1 SE) of con- tinuously flooded and periodically flooded bald cypress seed- lings for three growing seasons. Percentages are the compo- nent biomass relative to total biomass within a treatment. Significant differences (P 5 .05) in component biomass be- tween flooding treatments are marked with an asterisk (*) placed above the bar with the higher mean. Over the course of the study, the CF cell received e2.5 m3 more creek water than the PF cell; water output from the CF cell was e 0 . 2 m3 less (Table 2). Total inputs and outputs of nutrients were similar. The generally greater pool of soil solution nutrients in the CF cell suggests greater leaching or mineralization in the continuously flooded soil. Shoot biomass Shoot growth of bald cypress was very rapid in both cells. Three-yr-old saplings were 2 m tall and up to 8.4 cm in diameter (measured at 20 cm above the soil surface). After one growing season the shoot biomass of PF seedlings was approximately three times that of CF seedlings (P 5 .0001). However, the difference in aboveground biomass narrowed during subsequent growing seasons (Fig. 3). At the end of the second grow- ing season, average shoot biomass of the PF cypress was 0.5 times greater than the CF cypress (P = ,008). At the end of the third growing season there was no significant difference in shoot biomass between treat- ments. Biomass allocation to leaves, branches, and boles differed between years and treatments. During the first growing season, continuously flooded plants allocated more shoot biomass to leaves than periodically flooded plants (67% vs. 5S0/o, P 5 ,000 1 , Fig. 3). Periodically flooded plants allocated more to wood (P 5 ,0007). This pattern did not persist in subsequent growing sea- sons, when CF plants allocated less biomass to leaves and branches and more biomass to boles than PF plants ( P 5 ,0005 for each comparison). BELOWGROUND BIOMASS 1988 . COMPLETE HARMST TOTAL 40.. 20.- 400nr 2 n fi 0 CONTINUOUS PERIODIC ha (2 rn rn 1986 E32-5 rnrn 0) 5 rnrn T O T A L FLOODING FLOODING 24% 10% 10% :% 0- m m L 91% * FIG. 4. Belowground biomass (means and 1 SE) of the continuously flooded bald cypress seedlings. Bars give totals from 30 x 30 x 30 cm excavations following each growing season. Square symbols in the 1988 frame are total be- lowground biomass inclusive of roots recovered in a complete excavation of each rhizotron cell. Asterisks (*) mark signifi- cant differences at P 5 .02 in between-treatment comparisons. August 1992 CARBON ALLOCATION IN WETLAND RHIZOTRONS 1187 TABLE 3. Annual biomass increment of bald cypress grown in large rhizotron cells. Estimates are based on annual differences in mean bi0mass.t CF = continuously flooded, PF = periodically flooded. Annual production 1986 1987 1988 (g.plantrl.yr-I) CF PF CF PF CF PF Shoot total* 22.0 64.7*** 308.4 420.9* 1,026.8 718.7 NS Root total 11.4 3 1.7*** 129.6 179.0* 448.8 739.8* Grand total 33.4 96.4*** 438.0 599.9* 1,475.6 1,458.5 NS RGRP 3.1 1 4.00*** 2.60 1.96* 1.42 1.13 NS t Statistical comparisons were made between treatments within years for each measurement of growth. Significant differences are indicated with asterisks (* = 5.05, *** = 5.001, NS = not significant). *The 1986 estimate accounts for the biomass of the seedling stock (leaf = 0.54, bole = 0.62, branch = 0. root = 0.52 &plant). 8 Relative growth rate = log,,(final mass) - log,,(initial mass). Root biomass Root biomass estimates based on 30 x 30 x 30 cm excavations show a temporal pattern similar to that of shoot biomass (Fig. 4). The average total root biomass was 2.5 times as large in the PF treatment as the CF treatment after one growing season, and 1.5 times as large after two growing seasons (P 5 ,003, Fig. 4). At the end of the third growing season, there was no sig- nificant difference in root biomass between the treat- ments. Although useful for making comparisons, the 30 x 30 x 30 cm plot samples underestimated root biomass for 2- and 3-yr-old plants. The complete excavation at the end of the study showed that average total root mass was considerably higher under periodic flooding (mean i SE = 95 1 k 133 vs. 590 i 72 g/plant, P = .04, Fig. 4). The 1987 harvest (2-yr-old plants) excluded roots outside the sample plots. These roots were recovered in a single batch sample during the 1988 destructive harvest (root size classes L 2 m m diameter). Their av- erage biomass was 17 g/plant (CF) and 100 &plant (PF). Random cores of the upper 30 cm taken in 1988 showed more fine root biomass ( < 2 mm diameter) in the PF treatment than in the CF treatment (mean i 1 SE = 33 k 2 g/m2 and 21 i 4 g/m2, P = .01). Root and shoot increment Estimates of the annual biomass increment to shoots and roots were calculated as the difference between the mean biomass of each component in successive years (Table 3). In 1986 and 1987 the biomass increment of PF cypress was greater than that of CF cypress (P = .0001 and .03, respectively). In 1988 the total produc- tion was nearly the same under continuous and pen- odic flooding (CF = 1,476 &plant and PF = 1,459 &plant; P = .96), but there was a difference in biomass allocation. Average root increment in the PF cell was greater by 29 1 &plant (P = .05), while average shoot increment was lower by 308 &plant (P = .14). In the second year, relative growth rates of CF plants were greater than PF plants (P = ,001). There was no sig- nificant difference in relative growth rates in the third growing season (P = .34). Root size class and depth distributions Based on the 30 x 30 x 30 cm excavations, CF plants allocated a greater proportion of belowground mass to fine roots ( < 2 mm diameter) than PF plants during each year of the study (Fig. 4, P 5 .003). PF plants allocated more mass to structural roots (> 5 mm diameter) in 1986 and 1987 (P 5 .0001). A complete excavation of the cells after 3 yr of treat- ment revealed that PF cypress saplings had deeper root systems than CF saplings. The proportion of total root mass below 30 cm was 30% in the P F cell and 6% in the CF cell. Some roots in the PF cell reached to the bottom of the 100-cm soil profile. Soil water roots (sensu Hook 1984) emerged from the soil surface of the CF cell early in the second grow- ing season. They were initiated on the lateral roots and grew upward into the water column. Some eventually reentered the soil. They began secondary thickening during the second growing season. After three growing seasons the soil water roots accounted for 37% of the fine root ( < 2 mm diameter) biomass and 19% of the medium root (2-5 m m diameter) biomass of CF plants. Primary soil water roots were e 2 mm in diameter, white, and succulent, with many fibrous secondary roots. Roots > 5 mm diameter comprised roughly 3% of the water root biomass after 3 yr. Root-to-shoot ratios Regressions of root biomass on shoot biomass were significant for the CF cell after 1 yr and for both cells after 3 yr (P < .002; Fig. 5). There was no difference in slope or Y intercept for 1-yr-old seedlings (P r .2 1). Regression lines for 3-yr-old saplings had a common slope but different Y intercepts (P = .01). Thus, for a given amount of shoot biomass, PF plants had more root biomass than CF plants. The CF plants allocated 28% of total dry mass increment belowground while P F plants allocated 46% belowground (Table 3). There 1188 J. PATRICK MEGONIGAL AND FRANK P. DAY Ecology, Vol. 73, No. 4 1 -YR-OLD BALD CYPRESS 4.5 7 Root system structure PERIODIC FLOODING CONTINUOUS --.--- FLOODING 3.5 I- 0 - 2 . 5 0 a C - 1.51 I n (SHOOT MASS) /S FLOODING -- CONTINUOUS F7 A I FIG. 5. Plots of In(root mass) vs. ln(shoot mass) for the first and third growing-season harvests. Regression models were significant for the continuously flooded treatment in the first year (P = ,0004) and for both treatments in the third year (P 5 ,002). 2.5 3.0 3.5 4.0 4.5 5.0 I n (SHOOT MASS) 7.5 s 3-YR-OLD BALD CYPRESS was surprisingly little variation in the biomass of 3-yr- old CF plants (Fig. 5). Variability in the water table may have promoted the expression of genotypic vari- ation in PF plants (Lewontin 1974). Stem and root anatomy Hand-cut sections of root and stem tissue were stained with toluidine blue and studied with light microscopy. Intercellular spaces were apparent in the secondary phloem of both the continuously and periodically flooded plants. However, the air spaces were consis- tently much larger, far more numerous and more con- tinuous in the phloem of CF plants. A layer of spongy tissue characterized by loosely packed parenchyma cells and schizogenic intercellular spaces occurred in the outer l/4 of the phloem of PF plants but often extended to near the cambial layer in the phloem of CF plants. Air spaces occurred both within the rays and in the phloem tissue between the rays. In radial section the air spaces were longitudinally continuous in the tissue of CF plants. There was a total absence of air space in the xylem. Intercellular spaces cannot account for the relatively large, buttressed bases of the CF stems. Root system morphology after three growing seasons was studied by removing the top 30 cm of soil while leaving roots 2 2 mm diameter in place. The two flood- ing regimes resulted in strikingly different root systems (Fig. 6). Continuously flooded roots were diageotropic or negatively geotropic, highly tapered, succulent, and restricted to a relatively confined area around the stem. The root systems of individual stems showed relatively iittle overlap. Periodically flooded roots were posi- tively geotropic (disappearing below the 30 cm soil plane in a short distance), longer, and less tapered. The root systems of PF stems tended to overlap. In this 3-yr study we found striking differences in growth and allocation by bald cypress exposed to con- tinuous and periodic flooding. Because each flooding regime was represented by a single experimental unit, caution in interpretation is warranted due to the lack of true replicates (Hurlbert 1984, but see Hawkins 1986). We suggest, however, that the observed differ- ences in plant response are most parsimoniously in- terpreted as the effects of differential flooding patterns. Biomass production Biomass production by bald cypress was greater with periodic flooding than with continuous flooding during the first growing season. Similar flooding effects have been reported for the seedlings of a variety of woody, hydrophytic species in short-term microcosm studies (Keeley 1979, Sena Gomes and Kozlowski 1980a, 1988, Newsome et al. 1982, Norby and Kozlowski 1983, Peterson and Bazzaz 1984, Day 1987, Donovan et al. 1988). Studies of bald cypress seedlings, however, have reported variable growth responses to continuous in- undation (Table 1). In the present study, treatment effects on growth rate varied with age. Rates of relative growth in the continuously flooded (CF) cell were great- er than or equal to the periodically flooded (PF) cell in the second and third growing seasons. The relative difference between cells in total biomass and biomass increment decreased over the three years of the study (Figs. 3 and 4, Table 3). Increased relative growth rate in the CF cell during the second growing season suggests that the plants had acclimated to the flooding through morphological and physiological ad- aptations (e.g., water roots and intercellular spaces). Because plants generally meet the demands of leaf and root production before investing in wood (Waring and Schlesinger 1985: 34), relatively little allocation to stems in the first (1986) growing season suggests that the CF plants were stressed. Increased allocation to wood in the CF cell coincided with improvements in growth rate in 1987 and 1988. It is likely that relatively slow growth in the CF cell during the first growing season was due to transplant August 1992 CARBON ALLOCATION IN WETLAND RHIZOTRONS CONTINUOUSLY FLOODED PERIODICALLY FLOODED metre FIG. 6 . Composite photographs of the root systems of the two treatments. The top 30 cm of soil was removed, leaving all roots 2 2 mm in diameter in place. The rhizotron cells are 2.83 m long, 2.83 m wide, and 1.5 m deep. shock. Despite a continuous turnover of soil water, the CF cell was always anoxic and reduced (study mean = - 18 mV at 20 cm, Fig. 2). Although few microcosm studies report soil redox potential or oxygen status, it is possible that flooded microcosms remain in a par- tially aerated condition. Jones et al. (1989) reported an average redox potential of 178 mV in the center of waterlogged flower pots, a value that suggests an an- aerobic but relatively oxidized environment compared to our cells. Flood-conditioning our seedlings in mi- crocosms may not have been sufficient to acclimate them to the relatively reduced environment of the rhi- zotron cells. Relatively slow growth of CF plants in 1986 may have been the result of direct injury to the root system. Another factor contributing to transplant shock and subsequent recovery is flooding depth rel- ative to plant height. At 40 cm in height, newly planted CF seedlings were 50% submerged, while 3-yr-old sap- lings were only 5% submerged. As an alternative explanation, rapid early growth in the PF treatment may have led to competition for light, nutrients, and water. Competition may have been less important in limiting the growth of the CF plants, which were relatively small. However, we believe that com- petition is an unlikely explanation for the change in relative growth rates after one growing season. The fall harvests removed l/3 of the individuals (roots and shoots) in 1986 and '12 of the individuals in 1987. Under the competition scenario, thinning should have caused a burst of growth in the PF treatment at the beginning of the 1987 and 1988 growing seasons. Diameter mea- surements taken regularly during the 1988 growing sea- son demonstrate that this was not the case. Following thinning, both treatments grew at the same relative rate for the first 3 mo of the growing season (P = .69; mean [k 1 SE] increment in relative basal area = 0.92 +- 0.06 [CF] and 0.86 i- .08 [PF]). Stem and root anatomy Production of adventitious roots and intercellular spaces are common responses to waterlogged soils and are frequently associated with flooding tolerance (Crawford 1983, Hook 1984). Adventitious roots are believed to confer tolerance to flooding (Sena Gomes and Kozlowski 1980a, Tsukahara and Kozlowski 1985, Kozlowski et al. 199 1). Initiation of adventitious roots has been correlated with stomata1 reopening in Frax- inus pennsylvanica and with improved growth in Eu- calyptus camaldulensis (Sena Gomes and Kozlowski 1980a, b). In the present study, the appearance of soil water roots on CF plants coincided with increased growth in the second growing season. These roots com- prised a significant fraction of fine and medium root biomass in 1988 (37% and 19%, respectively). A number of potential functions have been suggested for adventitious roots: they (1) allow water absorption upon loss of the original root system (Sena Gomes and Kozlowski 1980~) ; (2) allow rapid uptake of dissolved 1190 J . PATRICK MEGONIGAL A N D FRANK P. DAY Ecology, Vol. 73, No. 4 oxygen and nutrients from the relatively oxic water column; (3) are sites of accelerated alcoholic fermen- tation and provide a compensatory energy source dur- ing periods of anaerobiosis (Hook et al. 197 1, Keeley and Franz 1979); and (4) release toxins such as CO,, ethanol, and ethylene (Crawford 1983). Soil water roots may have contributed to the recovery of normal phys- iological function in the CF plants. In trees the ability to transport oxygen from above the water surface to the root system is key to long-term success in flooded environments. Hook et al. (197 1) demonstrated that there was oxygen transport through the stem to the root system of flooded Nyssa sylvatica var. biyora. Histological studies of the roots revealed intercellular spaces in the phloem and cortex (Hook et al. 1970a, b). Schizogenous formation of intercellular spaces and oxygen transport through the stem was demonstrated in Pinus serotina, a moderately flood- tolerant gymnosperm (Topa and McLeod 1986). Fisher and Stone (1990) measured high values of air perme- ability in the roots of Taxodium ascendans. It is likely that well-developed intercellular spaces in the phloem of CF bald cypress saplings permitted oxygen transport to the root system and contributed to high production rates in the second and third years of the present study. The fact that schizogenous intercellular spaces were less developed in the P F plants is consistent with the rel- atively short flooding treatment. Root system structure Several microcosm studies have shown that upon flooding hydrophytic plant species replace much of the original root system with a new, morphologically dis- tinct root system (Hook 1984). The adventitious roots and soil water roots of continuously flooded seedlings are typically larger in diameter, less branched, and more succulent than those grown in the absence of flooding. Flooded roots often exhibit diageotropism. Continu- ously flooded roots in the present study had these char- acteristics. Keeley (1 979) observed that soil water roots on continuously flooded swamp tupelo (Nyssa sylvatica var. bijlora) seedlings were replaced within 1 yr by roots resembling those from a "drained" treatment. Our data suggest that soil water roots become the primary root system of flood-tolerant tree species as proposed by Hook (1984). Indeed, Harms et al. (1980) observed water roots on mature bald cypress and swamp tupelo 3 yr after the construction of a reservoir caused an increase in the depth and duration of flooding. Carbon allocation Root-to-shoot ratios commonly decrease in response to prolonged flooding in microcosms (Dickson and Broyer 1972, Keeley 1979, Sena Gomes and Kozlowski 1980a, Kane 198 1, Newsome et al. 1982, Norby and Kozlowski 1983, Peterson and Bazzaz 1984, Tang and Kozlowski 1984, Topa and McLeod 1986, Day 1987, but see Shanklin and Kozlowski [I9851 for an excep- tion). Similar observations have been made in the field (Megonigal and Day 1988). At least three possibilities explain this phenomenon: (1) ample water and nutri- ents in the rooting zone permit a shift in carbon allo- cation from roots to shoots (Tilman 1988); (2) flooding causes extensive root mortality, restricts translocation of carbon to roots, or limits root cambial activity; or (3) flooding decreases root biomass but increases root turnover, and root production is relatively unaffected (Nadelhoffer et al. 1985). It is likely that transplant shock caused root-system mortality in the CF treat- ment at the beginning of the study. By the end of the first growing season, however, CF plants had root to shoot ratios comparable to those of P F plants. Appar- ently, the CF plants had made morphological adjust- ments to flooding stress by the end of the first growing season (Fig. 5). Root-to-shoot ratios for 3-yr-old cypress saplings were clearly lower in the CF treatment than the P F treatment (Fig. 5). The high growth rates of the CF plants make it appear unlikely that reduced root bio- mass was due to toxic soil effects. Higher root turnover is unlikely because there were few dead roots, and de- composition rates were relatively slow in the CF treat- ment (Day et al. 1989). Rather, the data suggest a dif- ference in carbon allocation. Abundant water and dissolved nutrients in the rooting zone of the CF treat- ment afforded these plants adequate belowground re- sources at a minimum investment in root biomass. They were essentially growing in a hydroponic solu- tion. A high relative proportion of fine roots provided increased water- and nutrient-absorbing area. In con- tinuously flooded environments, low root-to-shoot ra- tios may confer a competitive advantage by allowing an increase in leaf biomass (Tilman 1988). The PFplants had high root-to-shoot ratios and deep root systems (to 100 cm). A relatively deep rooting zone may have provided the P F saplings access to water and dissolved nutrients within the water table (50-60 cm below the soil surface during summer; Fig. 1). A common observation is that rooting depth is re- lated to flooding frequency in swamp forests. Lieffers and Rothwell (1987) found a strong relationship be- tween rooting depth and the depth to water table for Larix laricina and Picea mariana on a peat soil. Armstrong et al. (1 976) found a strong correlation be- tween soil oxygen flux and root mass for Picea sitch- ensis. Root mass was positively correlated with the force required to topple the tree. Our data (present study) contribute experimental evidence for a positive relationship among rooting depth, root biomass, and the depth to water table. Flood and drought tolerance Although highly tolerant of flooding, bald cypress exhibits adaptations characteristic of less tolerant tree species, such as buttressed bases, soil water roots, and intercellular air spaces. These adaptations promote gas August 1992 CARBON ALLOCATION IN WETLAND RHIZOTRONS 1191 transport to the root system and oxidation of the rhi- zosphere, allowing the plant to carry on aerobic res- piration (Armstrong 1964, Hook et al. 1971, Keeley and Franz 1979, Topa and McLeod 1988). Thus, the optimum environment for cypress is probably an aer- obic one. Why then can bald cypress grow equally well in continuously flooded (but not stagnant) and peri- odically flooded environments? Highly flood-tolerant trees are generally drought sen- sitive. Dickson and Broyer (1972) showed that low soil moisture caused internal moisture stress and reduced growth in bald cypress. Under drought stress, cypress shoots were irreparably damaged in 3-4 hr. The pe- riodically flooded plants in the present study showed signs of water stress (e.g., red and yellow colored fo- liage), especially in the third year of growth. Optimum growth probably could have been achieved by holding the water table close to the surface during the summer (at -20 cm rather than at - 50 cm), thereby permitting both adequate aeration and soil moisture. Field studies have usually found continuous or ex- cessive flooding associated with reduced aboveground production in bottomland forests (Conner and Day 1982, Malecki et al. 1983) and cypress swamps (Schle- singer 1978, Mitsch and Ewel 1979, Brown 1981). Nat- ural and human-made impoundments have been linked to decreased growth (Mitsch et al. 1979) and increased mortality (Harms et al. 1980) in mature cypress swamps. Likewise, field studies have found that periodic flood- ing is associated with improved aboveground produc- tion in bottomland forests (Mitsch and Ewel 1979, Brown 198 1, Conner and Day 1982, Megonigal and Day 1988, but see Brown and Peterson [I9831 for an exception). Stem growth in cypress domes was best on sites that were neither extremely wet nor extremely dry (Mitsch and Ewel 1979, Marois and Ewel 1983). Bald cypress may be more sensitive to inadequate moisture than excessive moisture. Indeed, sensitivity to drought stress may help explain why they occur only in the most flooded habitats. Conclusions Continuous flooding is usually considered more physiologically stressful to hydrophytes than periodic flooding, but for bald cypress and other extreme hy- drophytes there is a trade-off between flood tolerance and drought tolerance (Keeley 1979). If dominated by hydrophytes, a soil that is continuously flooded with shallow water may support higher productivity than a periodically flooded soil that frequently dries to the wilting point. Donovan et al. (1 988) noted that growth ofwater tupelo is not consistently affected by saturation or flooding. This is also the case with bald cypress (Table 1). Bald cypress and water tupelo may have a narrow range of tolerance for wet and dry extremes. The varied responses of these species to flooding treat- ments in microcosm studies may reflect the wide va- riety of conditions that researchers consider "flooded" or "drained" (Table 1). The relatively short duration of these studies may also have contributed to variations in the growth response. Most field studies concerning the effects ofhydrology on primary production are based solely on above- ground estimates (Schlesinger 1978, Brown 198 1, Con- ner and Day 1982). Yet plants are generally plastic in partitioning carbohydrates to roots or shoots (Tilman 1988). Although aboveground productivity is related to total productivity on a broad scale, our data dem- onstrate the importance of measuring root production when relating production to environmental variables. For example, without the belowground production es- timates we would have concluded that continuously flooded plants tended to be more productive than pe- riodically flooded plants in the third growing season. There are an estimated 13 million ha of palustrine forested wetlands in 10 states of the southeastern Unit- ed States (Hefner and Brown 1985). Many of these forests are dominated by bald cypress and water tupelo. Palustrine forests are an important source of organic carbon for southeastern aquatic ecosystems. Leaf litter and other forms of allochthonous carbon are the pri- mary source of energy for coastal plain rivers in the region (Mulholland 198 1, Meyer and Edwards 1990). Changes in hydroperiod-e.g., with global sea-level rise-that affect either forest productivity or carbon allocation to roots and shoots can potentially reduce leaf litter export. A better understanding of the role of wetland forests in linking upland and aquatic ecosys- tems will require increased attention to root production and other belowground processes. Lyndon Lee was actively involved in the design and ini- tiation of this research. Rebecca Sharitz kindly arranged most of the financial support. Special thanks to John Bailey who assisted in every phase of the work and processed most of the root samples. Jennifer Barnes, Kent Bryant, Tom Cerravolo, Richard DeBolt, Keith Dyer, Jerry Gamin, Sally Landaal, Joy Lewis, Brian Maddox, Margaret Shea, and Fred Stone con- tributed to various phases of the project. Me1 Turner assisted in the interpretation of the tissue sections. Bill Schlesinger, Michael Scott, Deborah Clark, and two anonymous reviewers provided comments on early drafts of the manuscript. This research was supported by the National Environmental Re- search Park program of the Savannah River Ecology Labo- ratory (SREL) and the Division of Wetlands Ecology under contract DE-AC09-76SR00-8 19 between the United States Department of Energy and the University of Georgia. F. P. Day's travel to SREL was partially funded by an Oak Ridge Associated Universities Faculty Research Participation Trav- el Contract (S-3221). Armstrong, W. 1964. Oxygen diffusion from the roots of some British bog plants. Nature 204:801-802. Armstrong, W., T. C. Booth, P. Priestley, and D. J. Read. 1976. The relationship between soil aeration, stability and growth of sitka spruce (Picea sitchensis (Bong.) Carr.) on upland peaty gleys. Journal of Applied Ecology 13:585- 591. 1192 J. PATRICK MEGONIGAL AND FRANK P. DAY Ecology, Vol. 73, No. 4 Bohm, W. 1979. Methods of studying root systems. Spring- er-Verlag, New York, New York, USA. Bohn, H. L. 1971. Redox potentials. Soil Science 112:39- 45. Brinson, M. M., A. E. Lugo, and S. Brown. 1981. Primary productivity, decomposition and consumer activity in freshwater wetlands. Annual Review of Ecology and Sys- tematics 12:123-161. Brown, S. I98 1 . A comparison of the structure, primary productivity, and transpiration of cypress ecosystems in Florida. Ecological Monographs 51:403-427. Brown, S., and D. L. Peterson. 1983. Structural character- istics and biomass production of two Illinois bottomland forests. American Midland Naturalist 110: 107-1 17. Carter, V., D. Shultz, M. K. Garrett, and W. E. Webb. 1984. A technique to measure oxygen in the root zone of saturated and unsaturated soils. Selected Papers in Hydrologic Sci- ence 1-5. Conner, W. H., andJ. W. Day. 1982. Theecology offorested wetlands in the southeastern United States. Pages 69-87 in B. Gopal, R. E. Turner, R. G. Wetzel, and D. F. Whigham, editors. Wetlands: ecology and management. International Scientific Publications, Jaipur, India. Crawford, R. M. M. 1983. Root survival in flooded soils. Pages 257-283 in A. J. P. Gore, editor. Ecosystems of the world. Elsevier Scientific, Amsterdam, The Netherlands. Day, F. P. 1987. Effects of floodingand nutrient enrichment on biomass allocation in Acer rubrum seedlings. American Journal of Botany 74: 1541-1 554. Day, F. P., J . P. Megonigal, and L. C. Lee. 1989. Cypress root decomposition in experimental wetland mesocosms. Wetlands 9:263-282. Dickson, R. E., and T. C. Broyer. 1972. Effects of aeration, water supply, and nitrogen source on growth and devel- opment of tupelo gum and bald cypress. Ecology 53:626- 634. Donovan, L. A,, K. W. McLeod, K. C. Sherrod, and N. J. Stumpff. 1988. Response of woody swamp seedlings to flooding and increased water temperatures. I. Growth, bio- mass, and survivorship. American Journal of Botany 75: 1181-1 190. Donovan, L. A., N. J. Stumpff, and K. W. McLeod. 1989. Thermal flooding injury ofwoody swamp seedlings. Journal of Thermal Biology 14: 147-1 54. Faulkner, S. P., W. H. Patrick, and R. P. Gambrell. 1989. Field techniques for measuring wetland soil parameters. Soil Science Society of America Journal 53:883-890. Fisher, H. M., and E. L. Stone. 1990. Air-conducting po- rosity in slash pine roots from saturated soils. Forest Science 36: 18-33. Harms, W. R., H. T. Schreuder, D. D. Hook, and C. L. Brown. 1980. The effects of flooding on the swamp forest in Lake Ocklawaha, Florida. Ecology 61 : 14 12-1 42 1. Harris, W. F., D. Santantonio, and D. McGinty. 1980. The dynamic belowground ecosystem. Pages 1 19-1 29 in R. H. Waring, editor. Forests: fresh perspectives from ecosystem analysis. Oregon State University Press, Corvallis, Oregon, USA. Hawkins, C. P. 1986. Pseudo-understanding of pseudorepli- cation: a cautionary note. Bulletin of the Ecological Society of America 67:184-185. Hefner, J. M., and J. D. Brown. 1985. Wetland trends in the southeastern United States. Wetlands 4: 1-1 1. Hook, D. D. 1984. Adaptations to flooding with fresh water. Pages 265-294 in T. T. Kozlowski, editor. Flooding and plant growth. Academic Press, Orlando, Florida, USA. Hook, D. D., C. L. Brown, and P. P. Kormanik. 1970a. Lenticel and water root development of swamp tupelo un- der various flooding conditions. Botanical Gazette 131:2 17- 224. Hook, D. D., C. L. Brown, and P. P. Kormanik. 1971. In- ductive flood tolerance in swamp tupelo (Nyssa sylvatica var. biflora (Walt.) Sarg.). Journal of Experimental Botany 22:78-89. Hook, D. D., 0 . G. Langdon, J. Stubbs, and C. L. Brown. 1970b. Effect of water regimes on the survival, growth, and morphology oftupelo seedlings. Forest Science 16:304- 71 1. -. .. Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field exveriments. Ecological Monographs 54: 187-21 1. Jones, R. H., R. R. Sharitz, and K. W. McLeod. 1989. Effects of flooding and root competition on growth of shaded bot- tomland hardwood seedlings. American Midland Natural- ist 121:165-175. Kane, E. 1981. Swamp forest root systems: biomass distri- bution, nutrient content, and aspects of growth and mor- tality. Thesis. East Carolina University, Greenville, North Carolina, USA. Keeley, J. E. 1979. Population differentiation along a flood frequency gradient: physiological adaptations to flooding in Iv'yssa sylvatica. Ecological Monographs 49:89-108. Keeley, J. E., and E. H. Franz. 1979. Alcoholic fermentation in swamp and upland populations of Iv'yssa sylvatica: tem- poral changes in adaptive strategy. American Naturalist 113:587-592. Kozlowski, T. T. 1984. Responses of woody plants to flood- ing. Pages 129-162 in T. T. Kozlowski, editor. Flooding and plant growth. Academic Press, New York, New York, USA. Kozlowski, T. T., P. J. Kramer, and S. G. Pallardy. 1991. The physiological ecology ofwoody plants. Academic Press, New York, New York, USA. Lewontin, R. C. 1974. The analysis of variance and the analysis of causes. American Journal of Human Genetics 26:4OO-4 1 1. Lieffers,V. J., and R. L. Rothwell. 1987. Rootingofpeatland black spruce and tamarack in relation to depth of water table. Canadian Journal of Botany 765:8 17-82 1. Malecki, R. A,, J. R. Lassoie, E. Rieger, and T. Seamans. 1983. Effects of long-term artificial flooding on a northern bottomland hardwood forest community. Forest Science 29535-544. Marois, K. C., and K. C. Ewel. 1983. Natural and manage- ment-related variation in cypress domes. Forest Science 29: 627-640. McLeod, K. W., L. A. Donovan, N. J. Stumpff, and K. C. Sherrod. 1986. Biomass, photosynthesis and water use efficiency of woody swamp species subjected to flooding and elevated water temperatures. Tree Physiology 2:341- 346. McLeod, K. W., and K. C. Sherrod. 1981. Bald cypress seedling growth in thermally altered habitats. American Journal of Botany 68:9 18-923. Megonigal, J . P., and F. P. Day. 1988. Organic matter dy- namics in four seasonally flooded forest communities of the Great Dismal Swamp. American Journal of Botany 75: 1334-1343. Meyer, J. L., and R. T. Edwards. 1990. Ecosystem metab- olism and turnover of organic carbon along a blackwater river continuum. Ecology 71:668-677. Mitsch, W. J., and K. C. Ewel. 1979. Comparative biomass and growth of cypress in Florida wetlands. American Mid- land Naturalist 101:4 17-426. Mitsch, W. J., C. L. Dorge, and J. R. Wiemhoff. 1979. Eco- system dynamics and a phosphorus budget of an alluvial cypress swamp in southern Illinois. Ecology 60: 1 1 16-1 124. Mooney, H. A. 1972. The carbon balance of plants. Annual Review of Ecology and Systematics 3:3 15-346. August 1992 CARBON ALLOCATION IN WETLAND RHIZOTRONS 1193 Mulholland, P. J. 198 1 . Organic carbon flow in a swamp- stream ecosystem. Ecological Monographs 51:307-322. Nadelhoffer, K. J., J. D. Aber, and J. M. Melillo. 1985. Fine roots, net primary production, and soil nitrogen availabil- ity: a new hypothesis. Ecology 66: 1377-1 390. Newsome, R. D., T. T. Kozlowski, and Z. C. Tang. 1982. Responses of Cl'mus amerlcana seedlings to flooding of soil. Canadian Journal of Botany 60: 1688-1 695. Norby, R. J., and T. T. Kozlowski. 1983. Flooding and SO2 stress interaction in Betula papyrifera and B. nigra seed- lings. Forest Science 29:739-750. Peterson, D. L., and F. A. Bazzaz. 1984. Photosynthetic and growth responses of silver maple (Acer saccharinum L.) seedlings to flooding. American Midland Naturalist 112: 262-272. Poorter. H.. and C. L. Lewis. 1986. Testing differences in relative growth rate: a method avoiding curve fitting and pairing. Physiologia Plantarum 67:223-226. Raich, J. W., and K. J. Nadelhoffer. 1989. Belowground carbon allocation in forest ecosystems: global trends. Ecol- ogy 7011346-1354. SAS. 1985. SAS procedures guide for personal computers, Version 6 edition. SAS Institute, Cary, North Carolina, USA. Schlesinger, W. H. 1978. Community structure, dynamics and nutrient cycling in the Okefenokee cypress swamp- forest. Ecological Monographs 48:43-65. Sena Gomes, A. R., and T. T. Kozlowski. 1980a. Growth responses and adaptation of Fraxinus pennsjdvanica seed- lings to flooding. Plant Physiology 66:267-27 1. Sena Gomes, A. R., and T. T. Kozlowski. 1980b. Effects of flooding on Eucalyptus camaldulensis and Eucalyptus glob- ulus seedlings. Oecologia (Berlin) 46: 139-1 42. Sena Gomes, A. R., and T. T. Kozlowski. 1988. Physio- logical and growth responses to flooding of seedlings of Hevea brasiliensis. Biotropica 20:286-293. Shanklin, J., and T. T. Kozlowski. 1985. Effect of flooding of soil on growth and subsequent responses of Taxodium distichum seedlings to SO,. Environmental Pollution 38: 199-2 12. Tang, Z. C., and T. T. Kozlowski. 1984. Water relations, ethylene production, and morphological adaptation of Fraxinuspennsylvan~ca seedlings to flooding. Plant and Soil 77: 183-192. Tilman, D. 1988. Plant strategies and the dynamics and structures of plant communities. Princeton University Press, Princeton, New Jersey, USA. Topa, M. A,, and K. W. McLeod. 1986. Responses of Pinus clausa, Pinus serotina and Pinus taeda seedlings to anaer- obic solution culture. I. Changes in growth and root mor- phology. Physiologia Plantarum 68:523-53 1. Topa, M. A,, and K. W. McLeod. 1988. Promotion ofaeren- chyma formation in Pinus serotina seedlings by ethylene. Canadian Journal of Forest Research 18:276-280. Tsukahara, J., and T. T. Kozlowski. 1985. Importance of adventitious roots to growth of flooded Platanus occiden- talis seedlings. Plant and Soil 88: 123-1 32. Vogt, K. A,, C. C. Grier, and D. J. Vogt. 1986. Production, turnover, and nutrient dynamics of above- and be- lowground detritus of world forests. Advances in Ecological Research 15:303-377. Waring, R. H., and W. H. Schlesinger. 1985. Forest eco- systems. Academic Press, New York, New York, USA.