263 CYPRESS ROOT DECOMPOSITION IN EXPERIMENTAL WETLAND MESOCOSMS Frank P. Day, Jr. Department of Biological Sciences, Old Dominion University Norfolk, Virginia 23529 J. Patrick Megonigal Savannah River Ecology Laboratory Drawer E, Aiken, South Carolina 29801 Lyndon C. Lee Office of Wetlands Protection U.S. Environmental Protection Agency and University of Georgia 40t M Street SW, Washington, DC 20460 Abstract: Two forested wetland mesocosms containing organic soil (a sapric histosol) and cypress seedlings were constructed in the rhizotron facility at the University of Georgia's Savannah River Ecology Laboratory. A different hydroperiod (continuous versus periodic) was designed and implemented for each mesocosm. A vertical litter bag technique was used to measure cypress root decomposition rates in the mesocosms, Rapid leaching losses followed by 6 months o fno mass loss occurred in both mesoco sms. During the final 9 months of the study, roots in the periodically flooded mesocosm began to significantly decompose while those in the continuously flooded mesocosm remained unchanged. Inhibited decay was correlated with reduced redox potential, oxygen, and pH, but the delayed response to changes in these factors suggests that the coupling between root decomposition and the abiotic environment is not tight. The prescence of living cypress roots could explain more rapid decay near the surface of the periodically flooded mesocosm. In general, nutri en ts were higher in concentration and amount in the periodically flooded decomposing roots but lower in the soil water. Net accumulations of N occurred in the periodically flooded roots. These patterns indicate that periodically flooded ecosystems should have more conservative nutrient cycles thancontinuously flooded systems. In general, mesocosms appear to provide a useful experimental approach to the study of belowground ecosystem processes. Key Words: anoxia, cypress, decomposition, forested wetland, hydroperiod, mesocosm, root. I N T R O D U C T I O N Fine roots represent a large and dynamic portion of belowground biomass, soil delritus, and nutrient capital; however, there have been few studies in which root 264 WETLANDS, Volume 9, No. 2, 1989 decomposition or turnover have been measured (Waid 197,~, McGinty 1976, Hackney and de la Cr~ 1980, Harris et al. 1980, Brinson et al. 1981, McClaugherty et al. 1982, Ellison et al. 1986, Hackney 1987, Tupacz 1988, Gallagher 1988). In a periodically flooded wetland, fluctuating water levels are likely to have significant effects on belowground ecosystem dynamics (Dickson and Broyer 1972, Reddy and Patrick 1975). The waterlogged portion of the soil profile is expected to be quite different chemically and physically from the non-waterlogged portion, and these differences, in turn, should produce varied root decomposition rates. Brinson et al. (1981) stated that".., detritus produced by root mortality is almost universally excluded (from wetiand studies)." This paucity of knowledge is directly related to the difficulty of obtaining belowground measurements without severely disturbing the soil (Bohm 1979), and these problems are especially severe in flooded soils. It is also difficult to test hypotheses regarding the effects of flood- ing using conventional approaches. Field measurements are subject to the vagaries of the natural environment, and highly controlled microcosm experiments lack re- alism. There is a need for new approaches for studying belowground dynamics. The rhizotron facility at the University of Georgia's Savannah River Ecology Laboratory (SREL) provided a unique opportunity for a mesocosm ap- proach to the study of belowground processes. Mesocosms are experimental units that hold much promise for manipulative ecological studies (Odum 1984), A meso- cosm fills the gap between the difficult-to-manipulate natural field plot and the unrealistic microcosm. A mesocosm approach allows for a significant amount of realism because of the large size of the experimental unit, but it also provides a high degree of control so that hypotheses can be tested. These units still have some over- simplified and artificial characteristics, but they are more realistic than smaller units. In this study, two forested wetland mesocosms containing an organic soil and cypress (Tax.odium distichum) seedlings were constructed in the rhizotron facility at the Savannah River Ecology Lab. A different hydroperiod treatment (continuous flooding versus periodic flooding) was incorporated into each meso- cosm to evaluate the effects of varied flood patterns on several system processes. S i nce treatment effec Is co uld not be direc tl y tested because of the lack of mesoco sm cell replication, the pretreatment conditions were carefully monitored in both ceils to establish their initial states. This paper presents the results of the hydroperiod manipulations on cypress root decomposition and offers an interpretation of the results as they relate to ecosystem processes. Environmental conditions in the mesocosms were monitored throughout the soil profile for the duration of the study. Results from a greenhouse study of root decay in pots exposed to different hydrop- eriods arc also reported and related to the mesocosm study. Day et al., CYPRESS ROOT DECOMPOSITION 265 METHODS The Mesocosms Two mesocosms were constructed during the winter and spring of 1985- 86 in the rhizotron facility at the Savannah River Ecology Laboratory. The soil profiles consisted of about 1 m ofa sapric histosol underlain by layers of sand and gravel (Figure 1). The soil was obtained by a commercial peat mining company from a former cypress wetland near Jacksonboro, S.C. Oxygen chambers (Carter et al. 1984), rcdox probes, and thermisters were installed, and rain collectors were attached to the rhizotron walls. Twenty-five cypress seedlings were planted with even spacing in each mesocosm. Vertical movement of water through the meso- cosms was allowed to prevent complete stagnation. For the duration of the experiment, the valves from the sumps were adjusted to allow an output flow of Wiring to w a L ? ' - _ lO cm ~-'~,"1 10 cm ~" A Cypr c,5 s SoedJIngs/Saplings ltu""?,:'??', ' San,~. Grave~ Soil Water I~= Output l /~#u/at lon - C o p p e t ~ v / i f ? Redox OxygBn Probes Chambers \ 1 \ i i j - j f - j , ~ . f J 1 1 f l , 1 1 " - Tygon lut~/ng C o p p e r fublng Figure 1. Cross sectional diagram of a cypress wetland mesocosm. The dimen- sions of each unit are 2.83 m wide, 2.83 m long, and 1.5 m deep (1 m of soil). 266 WETLANDS, Volume 9, No. 2, 1989 about 114 L per day into calibrated storage tanks. A system of floaLs, solenoid switches, and pumps delivered water (collected from Upper Three Runs Creek) from a storage tank into the cells through a sprinkler hose on the soil surface so that prescribed water levels were maintained. Flow meters were used to measure the volume of input. Since structure and function in wetland systems are highly dependent on the timing and extent of flood events (Brinson e t al . 1981, Wharton e t al. 1982, Day e t al. 1988), a different hydroperiod was initiated in each mesocosm on May 23, 1986 and maintained for the duration of the study. One mesocosm was continuously flooded and the other simulated a periodic flood cycle typical of many southeastern forested wetlands (Figure 2). Seasonally flooded forested wetlands commonly exhibit minimal surface flooding but extensive saturation of the rooting zone. 20 ~ 1O L I L I J . - - I m 0 ,,( r e " I , I LI- O --20 - r O. ILl --3"0 Z ,,( LLI >- --t,0 1- Z O I - 5 0 - - 6 t" 1 S o d ...... I _ / . . . . / u S u r f a c e 1 I I i I I l - - i ~ "L -~" I I I ~ / 1, - - C o n t i n u o u s l y F l o o d e d M e s o c o s m ,, P e r i o d i c a l l y F l o o d e d M e s o c o s m K I L 1 I I t ,1 B I I I K M J J A S O N 0 J F N A M 1 9 8 6 % I I | I "~ J J A S 0 1 9 8 7 Figure 2. Hydroperiods of the two mesocosms during the root decomposition study. Vertical bars represent + 1 SE. Day et al., CYPRESS ROOT DECOMPOSITION 267 Soil oxygen content, pH, redox potential (Eh), and temperatures were de- termined at 20 cm depth intervals prior to the initiation of hydroperiod treatments and every two weeks thereafter. Gas-water sample chambers were used to de- termine oxygen content and take soil water samples (Carter et al. 1984). Each cham- ber consisted of a 6 cm d_iameter, 20 cm length of PVC pipe capped on each end (Figure 1). The sides of the pipe were slotted and covered with a fine mesh stainless steel screen to allow movement of soil water into the chamber. A copper tube that perforated the top cap, but did not extend below it, allowed us to sample the gas trapped under the cap (Palrick 1977). Another tube that extended to the bottom of the chamber allowed us to sample the soil pore water. Oxygen was measured as a pea'cent,age of almospheric concentration with an oxygen meter from paired cham- bers at each depth. Redox probes were constructed by fusing a I cm length of pure 22 gauge platinum wire to a length of 12 gauge insulated copper wire (Figure 1 ). The junction of the wires was sealed with epoxy so that only the platinum tip was exposed. The probes were checked for accuracy with a saturated solution of guinhydrone in pH 7 buffer and rejected if they deviated 10 mV (the accuracy of our meter). Potentials were measured against a calomel reference probe with a pH meter and corrected for the referez~e potential by adding 244 inV. There were three replicate probes at each depth. Temperature was determined with a thermistor. Soil water was collected once a month at five depths in duplicate from each mesocosm for determinations of total NI-I,-N, PO 4 -P, Ca, Mg, K, Mn, and Fe. Water in the sample chambers was emptied 24 h before collection to ensure a fresh sample. The pH of each sample was determined within 4 h of collection. The samples were then passed through 0.45 gm filters using acid-washed filter flasks and pre~d~ed with concentrated HCL. Filter blanks were substracted from each sample. An un- preserved subsample was used to determine total NH4-N on a LaChat autoanalyzer within 48 h of collection (Crooke and Simpson 1971). Total PO -P, Ca, Mg, K, M_n, and Fe were determined on a Jarrell-Ash Model 965 inductively coupled Plasma Emission Spectrograph (PES) (Method 200.7, U.S. Environmental Protection Agency 1983). Quality control for PES samples was achieved with spikes and splits of 10% of the samples. Percent recovery on the spikes averaged 86% (PO~-P), 99% (K), 93% (Ca), 87% (Mg), 80% (Fe), and 98% (Mn). The split samples showed equal p~zision. The same elements were analyzed in soil samples. Microenviron- menial variation between mesocosms and among sample dates and depths was tested by A N O V A and Duncan's Range Test (p < 0.01). Treatment effects could not be directly tested because hydroperiod treatments were not replicated. How- ever, we can infer that differew,.es between m ~ s were due to hydroperiod effects if the cells were the same prior to treatment and the only diffetetw.e afterward was in hydroperiocL Both mesocosms were almost identical at the start of the study (Table 1), and they were Created the same for the duration of the study except for the hydroperiod. 2 6 8 W E T L A N D S , V o l u m e 9, N o . 2 , 1 9 8 9 bO o t .~ o _~ ~ --.a ~ ~ -..a --.a ~ ka Ga ,..,, - . , tO o U, t,~ t-~ "3 ax _~ ~?~ ~?~ 3 ~? , -q o " O e-~ O ~ . o O O P Day et al., CYPRESS ROOT DECOMPOSITION 269 Root Decomposition Measurements Just prior to the start of the hydroperiod treatments, vertical 1 mm mesh bags (60 cm long with 20 cm segments) filled with preweighed cypress roots were inserted into the soil in the mesocosms (30 bags each). Vertical bags were used so decomposition could be integrated over a depth range with which detailed micro- environmental conditions could be associated (Tupacz 1988). The cypress roots were collected from a forested wetland located at the Savannah River Plant. These were washed and air dried. Approximately 6 - 8 g of 2 - 5 mm diameter roots were placed in each bag segment. Air dry mass/oven dry mass conversion factors were determined from subsamples. Five bags were retrieved from each mesocosm approximately every 2 months; the final collection intervals were extended. The roots were washed, oven dried for 48 h at 60 C, and weighed. Decomposition was estimated by the percentage of the original mass remaining. The roots were also analyzed for total N, total P, Ca, Mg, K, Mn, and Fe by previously described methods. Differences between mesocosms and among sample dates and depths were tested by ANOVA and Duncan's Range Test (p < 0.01). Cross correlations (Pearson's) were computed among all environmental parameters and mass loss rates to identify significant relationships. Greenhouse Study The effects of hydroperiod on root decomposition rates were also exam- ined in a greenhouse experiment at Old Dominion University. The experiment had a complicating factor (Acer rubrum seedlings in the pots), but the results provide some insight into the patterns observed in the mesocosms. Tree roots were obtained in March, 1985 at a single site in the Great Dismal Swamp, Virginia. No attempt was made to separate the roots by species, but the dominants on the site were Acer rubr~m, Nyssa aquatica, and Nyssa sylvatica var. biflora. Preweighed (1 - 2 g), unconfined root bundles were prepared by tying together three root segments, 7 cm long and 2-5 mm diameter. 'Care was taken to make the bundles as uniform as possible in root type, size and general appearance. Air dry mass/oven dry mass conversion factors were determined. The experimental designinvolved 192 clay pots (15 cm diameter) that contained 1-yr old Acer rubrurn seedlings and soil from the maple-gum site where the roots were obtained. The pots were placed in 33 cm x 33 cm x 20.3 cm deep watertight plastic boxes (4 pots per box). Sixteen boxes were assigned to each of three flood treatments (no flooding, continuous flooding, and periodic flooding). One root bundle was inserted vertically into the soil in each pot. The treatments were initiated in April, 1985 (Day 1987). The flooded pots were inundated to 5 cm above the soil surface with deionized water. Theperiodically flooded pots were alternately drained and reflooded to 5 cm above the soil every 2 weeks. The unflooded pots were watered two to three times a week with equal volumes of deionized water. 270 WETLANDS, Volume 9, No. 2, 1989 In June, August, and October, 1985, root bundles were extracted from five randomly chosen pots from each treatment, washed, oven dried at 70 C for 48 h, and weighed. Disappearance of mass was used as a quantitative indicator of decompo- sition. ANOVA and Duncan's Multiple Range Test (p < 0.01) were used to test for significant flooding effects. RESULTS Cypress Root Decomposition in Mesocosms Three distinct phases of root decay were apparent (Figure 3). A rapid 12 - 19% mass loss occurred during the first 2 months of the study in both mesocosms and at all depths. During the following 6 months, little or no decomposition was observed in either mesocosm. No significant differences (ANOVA, p < 0.05) between mesocosms or among depths were detected during these first two phases. tn the 15th month (all depths) and 17th month (top 20 cm only) of the study, the cypress roots in the periodically flooded mesocosm had significantly (ANOVA, p < 0.05) less mass remaining than the roots in the continuously flooded mesocosm. During the final 9 months of the study, roots in the top 20 cm of the periodically flooded mesocosm began to decompose rapidly while those in the continuously flooded mesocosm still showed essentially no mass loss (Figure 3). A slight trend of slower decay with increasing depth was observed, but the only statistically significant relationship was detected in the periodically flooded mesocosm on the final sample date (ANOVA, p = 0.022). Nutrient Dynamics in Decomposing Roots In general, nutrients were found in higher concen~ations (A_NOVA, p < 0.05) and higher absolute amounts in the periodically flooded roots compared to the continuously flooded ones (Figure 4). Nitrogen was accumulated in amounts greater than the initial content in the periodically flooded roots but generally showed a net loss from the continuously flooded roots (Figure 4). Phosphorous (Figure 4), Ca, and K (not shown here) all decreased in content in both mesocosms as decomposition progressed. Iron was usually above our detection limits, and Mn behaved similar to P. Magnesium showed little loss from the decomposing roots and no temporal dynamics. There were no consistent differences in nutrient dynamics among depths. Day et al., CYPRESS ROOT DECOMPOSITION 271 5" U.I rr" t f ) U3 I,-- Z LLI (..3 t 'r ' IJ.I 13. t ~ Z Z I...U ?'r" U') I-,- : Z I..;.I I3C ILl 13_ 100 O z ! \ 0 - - 2 0 cm DEPTH CONTINUOUS FLOODING 80 ~ F L O O D I N G 100 . . . . 20 - -&O cm DEPTH 90 IL Figure 3. cosms. 80 i -., "., 70 - ~ 100 z /,0--60 cm DEPTH Z ~ 90 W m 80 < i - - / \ 1 , i z 7 0 [ "If LI.I / O nr" tt.I 13- . . . . . I ! I 1 ,,, I I JUL $ E P N O V J A N AUG OCT 1986 1987 Mass loss from cypress roots in vertical litter bags in the two meso- Bags were implanted on May 24, 1986. Vertical bars are + 1 SE; N=5. 272 WETLANDS, Volume 9, No. 2, 1989 110 gO 70 5O QJ 30 C O C..) 110 ~- 90 Z ~ 7O ?. .-~ 50 L,. O N 30 i I t0 o_ 90 70 50 20 N 0 - 2 0 cm ? \ N 20--/.0 cm . / \ , . , , ~ I \ . . . . . / , 0 - 6 0 cm Conl inuous ly I looded - ' - Per iodicol ly I looded I t i I I I . , J S D J A O 1986 1987 P P 2 0 - 4 0 cm P & 0 - 6 0 cm I I I I I I J S D J A O 1986 1987 gO 7O 50 30 10 90 7D 50 3O I0 gO 70 50 3O 1o Figure 4. Total nitrogen and phosphorous content (expressed as percent of origi- nal content) in decomposing roots. Microenvironment of Mesocosms The continuously flooded mesocosm had redox potentials in the anoxic range throughout the study at all depths, and the oxygen measurements reflected this pattern (Figure 5). There were no significant differences (ANOVA, p < 0.05) in Day et al., CYPRESS ROOT DECOMPOSITION 273 i "7 - .r" / l . _ / - ~ E R C E N T ~XYOEN i < {:] '~ E \ < / P I o o ~r IE Q z 1 4w o ~ o o o ?:~ | I REOOX P O T E N T I A L rmv1 4 ~E Z I [ C~IO O O O ! Figure 5. Redox potential (Eh) and percent oxygen at three depths in the two mesocosms. The anoxic boundary is not a constant but only approximates the level at which anaerobic conditions develop. 274 WETLANDS, Volume 9, No. 2, 1989 redox or oxygen among depths in the continuously flooded mesocosm. The peri- odically flooded mesocosm remained aerobic in the top 20 cm of soil throughout the study (Figure 5), but had progressively lower redox and oxygen levels with increas- ing depth (Duncan's, p < 0.01). The duration of oxygen deficit, which occurred during the peak flood period (Figure 2), increased with increasing dept h (Figure 5). The periodically flooded mesocosm had significantly higher redox potentials and oxygen levels than the continuously flooded mesocosm at all depths. The pH of the top 40 cm of soil was significantly (ANOVA, p < 0.05) greater in the periodically flooded mesocosm (4.4 - 5.6) than the continuously flooded mesocosm (4.2 - 5.2) (Table 2). In both mesocosms, pH decreased with in- creasing depth. There were no significant (ANOVA, p < 0.05) differences in soil temperatures between mesocosms or among depths. Concentrations of nutrients in the soil water were generally greater in the continuously flooded mesocosm (ANOVA,p < 0.05) (Table 2). In the continuously flooded mesocosm, all elements increased in concentration with increasing depth (ANOVA, p < 0.01) (Table 2). This pattern was significant only for Fe in the periodically flooded mesocosm. Seasonal patterns were apparent in the concentra- tion data. Nutrient concentrations peaked during late summer and early fall and declined to lows in the spring. Peak concentrations in the water were considerably lower in the second year of the study compared to the first. Environmental Correlations The data from the two mesocosms were combined for correlation analysis because some data were missing from the periodically flooded cell when water was not present, and the continuously flooded cell exhibited reduced variation because it was flooded all of the time. Since the greatest variation in nutrient concentrations was seasonal and all nutrients showed similar seasonal patterns, all nutrients correlated significantly (p < 0.01) with each other (Table 3). Temperature is, of course, also seasonal, and it correlated positively with all nutrients. Decreases in Eh were highly related to decreases in O z (r=- 0.86, p < 0.01). Redox potential and O~ were, in turn, positively correlated with pH and negatively correlated with the duration of flooding or soil saturation and concenlrations of NH4-N, PO4-P, and Fe (Table 3). Reduced root decomposition, as indicated by higher mass remaining or lower decay rates, was associated with increased NH~-N, PO4-P, duration of flooding and temperature, and decreased Eh, O 2, and pH (Table 3). However, most of the relationships between environment and decay rates only became apparent near the end of the study when significant variation in decay was observed. Day et al., CYPRESS ROOT DECOMPOSITION 275 O 0 E 1 E ! ed ? q ! ~ ,t~ ,--- t ~ ~ ~O ~ ~ ~ O~ r ~ O ~ .., r:_ ~ ,6 ? m ~ ~5 ~5 o o ~ ~ o ~ ~ ~ ' ~ O~ ( 'q ~'~ Cq O~ ~ I "~ ~,~ ,-~ t ~ ~-- ~ ~ 6 ~ -T ~ o , " r" r~ ~ ~ ~ 5 ~ o d 6 ~ ~ E ---- 276 WETLANDS, Volume 9, No. 2, 1989 A L ~ ..~_.~ - ~ ,~., ~ ,o'., ~ . . q v ,o',. v ~ v ~ ~ " q~ , . ~ e " ~ . ~ ' = ~ t--.. " ~ o , O e - L . k",a ~ I 1 , . ~ < - ---.~8 g= ,_.~' N ' ~ " t ~ Day et al., CYPRESS ROOT DECOMPOSITION 277 Root Decomposition in Greenhouse Root mass decreased significantly during the study in all treatments (Figure 6). The analysis showed a significant (ANOVA, p = 0.0002) flooding effect on decomposition by the end of the study, as the non-flooded roots showed greater mass loss than the continuously flooded and periodically flooded roots. The continuously flooded and periodically flooded roots had 70% (+1.4) and 68% (+2.5) of their original mass remaining after 6 months, while the non-flooded roots had 54% (+4) mass remaining. Visual inspection revealed substantially more fragmen- tation of the root bundles by actively growing roots of the red maple seedlings in the non-flooded pots. DISCUSSION Decomposition of Roots The decay of belowground roots is similar in some ways but dissimilar in others to the process of aboveground litter decay. The initial, relatively rapid loss of mass in the first few weeks or months is likely a result of leaching (Tupacz 1988). This fLrst stage of root decay is quite comparable to what occurs in aboveground leaf litter (Day 1983). The differences in hydroperiod in the present study seemed to have no effect on initial leaching rams. Following the initial leaching loss, decomposition was arrested for the duration of the study in the continuously flooded mesocosm and for at least 6 months in the periodically flooded mesocosm. The late initiation of decay in the periodically flooded cell at 20 cm depth and the results of the correlation analysis suggest that the abiotic environment is influencing root decomposition but the coupling between the two is not tight. Divergence of the two major treatments is not apparent until the fifteenth month of the study. Just as aboveground litter decomposition is strongly influenced by the chemical and structural nature of the fitter (Day 1982), our results support Tupacz's (1988) suggestion that the highly lignaceous property of roots accentuates this influence belowground. However, near the end of the study, abiotic and/or biotic environ- mental differences in the two mesocosms were beginning to have an effect on root decomposition. Abiotic Environment The correlation results indicate that reduced root decomposition is associ- ated with increased duration of flooding or soil saturation and decreased redox potential, oxygen levels, and pH. Chronic anoxic conditions in the continuously flooded mesocosm could explain greater inhibition in the constantly flooded unit. 278 WETLANDS, Volume 9, No. 2, 1989 If) Z Z "