Litter decomposition promotes differential feedbacks in an oligotrophic southern Everglades wetland
- First Online:
- Cite this article as:
- Troxler, T.G. & Childers, D.L. Plant Ecol (2009) 200: 69. doi:10.1007/s11258-008-9405-2
The differential accumulation or loss of carbon and nutrients during decomposition can promote differentiation of wetland ecosystems, and contribute to landscape-scale heterogeneity. Tree islands are important ecosystems because they increase ecological heterogeneity in the Everglades landscape and in many tropical landscapes. Only slight differences in elevation due to peat accumulation allow the differentiation of these systems from the adjacent marsh. Hydrologic restoration of the Everglades landscape is currently underway, and increased nutrient supply that could occur with reintroduction of freshwater flow may alter these differentiation processes. In this study, we established a landscape-scale, ecosystem-level experiment to examine litter decomposition responses to increased freshwater flow in nine tree islands and adjacent marsh sites in the southern Everglades. We utilized a standard litterbag technique to quantify changes in mass loss, decay rates, and phosphorus (P), nitrogen (N) and carbon (C) dynamics of a common litter type, cocoplum (Chrysobalanus icaco L.) leaf litter over 64 weeks. Average C. icaco leaf degradation rates in tree islands were among the lowest reported for wetland ecosystems (0.23 ± 0.03 yr−1). We found lower mass loss and decay rates but higher absolute mass C, N, and P in tree islands as compared to marsh ecosystems after 64 weeks. With increased freshwater flow, we found generally greater mass loss and significantly higher P concentrations in decomposing leaf litter of tree island and marsh sites. Overall, litter accumulated N and P when decomposing in tree islands, and released P when decomposing in the marsh. However, under conditions of increased freshwater flow, tree islands accumulated more P while the marsh accumulated P rather than mineralizing P. In tree islands, water level explained significant variation in P concentration and N:P molar ratio in leaf tissue. Absolute P mass increased strongly with total P load in tree islands (r2 = 0.81). In the marsh, we found strong, positive relationships with flow rate. Simultaneous C and P accumulation in tree island and mineralization in adjacent marsh ecosystems via leaf litter decomposition promotes landscape differentiation in this oligotrophic Everglades wetland. However, results of this study suggest that variation in flow rates, water levels and TP loads can shift differential P accumulation and loss leading to unidirectional processes among heterogeneous wetland ecosystems. Under sustained high P loading that could occur with increased freshwater flow, tree islands may shift to litter mineralization, further degrading landscape heterogeneity in this system, and signaling an altered ecosystem state.
KeywordsDecay rate Heterogeneity Accumulation Mineralization Phosphorus Tree islands Linear regression
On a large scale, peatland development and maintenance are influenced by regional climate, geomorphology, site history, and disturbance regimes including fire and drought (Hogg et al. 1992; Almquist-Jacobson and Foster 1995). Just as feedbacks driving differential processes of accumulation and loss are important in maintaining hummock-hollow microtopography within and resilience of peatlands (Nungesser 2003), these differential processes may also be important at the landscape-scale for maintenance of the peatland ecosystem in the landscape (Almquist-Jacobson and Foster 1995). Thus, differential processes of accumulation and loss of carbon and nutrients during decomposition may promote differentiation among wetland ecosystems, and contribute to landscape-scale heterogeneity. Decomposition processes in wetlands are often nutrient limited, and hydrology interacts with nutrient availability, and the extant plant community, to influence peat accumulation or loss (Pastor et al. 2002).
Exogenous sources of nutrients to wetland environments can have deleterious effects on ecological structure and function, and can alter mechanisms for differentiation of wetland ecosystems in the landscape, degrading landscape heterogeneity. In many peatland studies, litter decomposition is cited as an important driver of surficial peat dynamics (Middleton and McKee 2001; Nungesser 2003). In oligotrophic environments, especially the phosphorus (P)-limited Everglades (Noe et al. 2001), litter decomposition studies have demonstrated that P availability promotes net immobilization of P by decomposing litter (Qualls and Richardson 2000; Newman et al. 2001; Davis et al. 2003; Debusk and Reddy 2005; Corstanje et al. 2005; Rubio and Childers 2006). Increased decomposition and P accumulation have been observed in areas where P loading is above oligotrophic levels (Craft and Richardson 1993; Qualls and Richardson 2000; Noe et al. 2001). Thus, if feedbacks controlling differential processes of accumulation and loss between ecosystems are maintained only under conditions of oligotrophy, then external nutrient inputs that change the oligotrophic status of inflowing water to the system may subsequently degrade landscape heterogeneity.
Tree islands are a unique component of the Everglades landscape and an important component of landscape heterogeneity, increasing the diversity of Everglades flora and fauna (Sklar and van der Valk 2002; Troxler Gann et al. 2005; Wetzel et al. 2005). Like other Everglades communities, tree islands have experienced major changes in the last 100 years of human alterations and are expected to undergo further changes as a result of hydrologic restoration. In the Everglades, nutrient loads delivered by actions taken to restore freshwater flow to oligotrophic wetlands may increase nutrient supply, degrade feedbacks that promote differential processes of accumulation and loss, and degrade landscape heterogeneity in the system. In this study, we explored the importance of decomposition processes as drivers of this differential accumulation and loss between tree islands and the adjacent marsh matrix, and the effects of reintroduced freshwater flow to an oligotrophic, heterogeneous, short hydroperiod wetland in the southern Everglades with peatland tree islands. In 1999, we established a landscape-scale, ecosystem-level manipulative experiment to characterize tree island ecosystem response to increased freshwater flow in the southern Everglades. We sought to: (1) determine whether decomposition of a common tree island litter type, Chrysobalanus icaco (cocoplum), varied among treatment levels of water flow rates and bulk nutrient loads, (2) examine differences between two contrasting ecosystems, tree islands and the adjacent sawgrass marsh, and (3) investigate linear relationships between hydrologic factors and decomposition processes in these two contrasting ecosystems. We employed a standard litter bag study and, using a common litter type, controlled for the effects of litter quality, to address these three objectives. We tested the following hypotheses. (1) Overall, tree islands would accumulate C and nutrients while, in marsh ecosystems, mineralization would occur. (2) Increased freshwater flow would result in increased litter decomposition and nutrient accumulation in tree island and marsh ecosystems if surface water P was above oligotrophic levels.
In August 1999 (2 years after levee removal), we selected nine Chrysobalanus icaco-dominated, seasonally flooded tree islands (based on similar island size and vegetation type) downstream of or adjacent to the levee removal segment of the C-111 canal (Fig. 1a). The post-levee increase in freshwater flow to the C-111 Basin provides a unique opportunity to investigate the effects of hydrologic restoration.
We investigated the effects of this increase in freshwater input by selecting tree islands in two different locations relative to the C-111 canal (Troxler Gann and Childers 2006). Six islands were located directly downstream of this canal (i.e., islands with increased freshwater flow) and three islands were in an area west of the canal that had minimal influence of the levee removal (we refer to the latter three islands as “No Flow” islands; Fig. 1a). In six islands near the canal, we identified the three nearest the canal as receiving fully enhanced water flow (“Flow” islands). With the other three islands, we tested the effects of sheetflow by experimentally deflecting surface water flow around the entire islands. We experimentally reduced sheetflow with plastic walls (constructed of 12 mil grade plastic and affixed with PVC posts and cable ties, placed in the marsh upstream from islands; = “Wall” islands; Fig. 1b). See Troxler Gann and Childers (2006) for details of our experimental design and other sampling.
In October 1999, we deployed 126 bags at nine sites with paired tree island and adjacent marsh locations in each site. Paired marsh locations were approximately 5 m downstream from each respective tree island location. Each bag contained approximately 5 g of air-dried C. icaco leaf litter, and was placed on the standing litter surface upon deployment. We collected one bag per location (two bags per site) at 1, 2, 4, 8, 16, 32, 64 weeks. We applied an air-dried/oven-dried weight conversion to determine final oven dry weights for each bag at each collection period. We quantified total mass remaining and loss, N and P concentrations, C:N, C:P, and N:P molar ratios, mass C, N and P for leaf litter at t0 and each collection period thereafter. Decay rates were calculated using the first order, exponential loss decay model, Mt = Moe−kt, where Mt is the dry mass at time t, Mo is original dry mass, and k is degradation coefficient (yr−1; Olson 1963). We calculated total percent loss as (100 − (Mt/M0) × 100) day−1 following Middleton and McKee (2001). Turnover rates (T0.95) were estimated for 95% turnover time as 3 k−1 (Olson 1963). In order to follow nutrient accumulation or loss, absolute mass was calculated as the product of dry mass remaining and C, N or P content of litter following Davis et al. (2003).
After retrieving bags from the field, leaf litter was gently washed of adhering soil (carefully, and to the extent possible, leaving accumulated biofilms intact and attached to the litter), dried to constant weight at 70°C, and weighed. Subsamples from each bag were ground to a homogeneous powder (<500 μm), and analyzed for total (T) N, TP and TC content. Leaf tissue samples were analyzed for TC and TN with a Carlo Erba elemental analyzer. The modified Solorzano and Sharp (1980) method was used to analyze for TP.
From a larger dataset of environmental parameters collected in association with this study, and published elsewhere, we removed auto-correlated hydrologic parameters using Pearson correlation coefficients (Troxler Gann and Childers 2006). Hydroperiod, average water level, wet season water level, water flow rate, annual TP, TN and TOC loads were thus used to determine which of these factors had the greatest influence on leaf litter decomposition. See Troxler Gann and Childers (2006) for details of these data.
We compared treatment means for decay rates and total mass loss, and final values of C and nutrient content, molar ratios and mass C and nutrients with a randomized block analysis of variance (ANOVA) using ecosystem as our blocking factor. Multiple comparisons for treatment means were evaluated with Tukey–Kramer tests. We then compared differences among ecosystems in these same parameters using t-tests. We used the paired t-test to evaluate differences between initial and final C, N and P content, molar ratios and mass nutrients for each ecosystem. We used simple linear regression analyses to quantify relationships between hydrologic factors and mass loss, decay rates, and final nutrient values separately for island and marsh ecosystems. In order to maintain statistical robustness, we only considered island and marsh regression relationships that were significant at a Bonferroni-corrected value of P = 0.05/7 or P ≤ 0.0071 (Zar 1999).
Effects of flow level
Summary of leaf degradation rates (k), turnover times (T) and total percent loss (% loss) of C. icaco litter in tree island and marsh ecosystems with treatment effect of different flow levels
% loss (d−1)
Variation across ecosystem types
We compared decomposition parameters between tree island and marsh ecosystems and between initial and final values for each ecosystem type. Comparing ecosystem types, we found a significant difference in total mass loss, and island and marsh values averaged 27.10 ± 2.78 and 44.27 ± 2.41%, respectively (t = −3.684, P = 0.0025). Ecosystem differences in decay rates were also significant, and these values were 0.241 ± 0.033 and 0.456 ± 0.040 yr−1 in islands and adjacent marsh, respectively (t = −3.472, P = 0.0049). Turnover rates and total percent loss rates followed values for decay rates. Turnover rates (T0.95) were 14.5 ± 1.5 and 7.8 ± 0.8 yr and total percent loss rates were 0.064 ± 0.007 and 0.092 ± 0.007% d−1 in tree islands and marsh, respectively.
Paired differences between initial and final values (viewed horizontally across table) and ecosystem differences between final values (viewed vertically down “final mean” column) of %C, %N, and %P, C:P, C:N, and C:P molar ratios, and mass C, N, and P of leaf litter decomposing in island and marsh ecosystems with t-ratio and P-value shown (P > |t|)
Mass C (g g−1)
Mass N (g g−1)
Mass P (mg g−1)
Comparing final nutrient and C values between tree island and marsh ecosystems, we found significantly lower C concentrations in cocoplum litter decomposing in the marsh, but no difference in N and P concentrations (Table 2). We found significantly higher C:P and N:P ratios in marsh locations, with no difference in C:N ratios. We also found significantly lower C, N, and P mass in marsh locations than in tree island locations after 64 weeks of decomposition (Table 2).
Relationships to explore potential hydrologic controls on litter decomposition
Linear regression relationships between hydrologic variables and decomposition values
Average water level
y = −0.0045 + 0.0033x
y = 325.6 − 21.68x
Wet season water level
y = −0.0162 + 0.0033x
y = 410.0 − 22.65x
y = −0.0313 + 0.0028x
y = −0.0514 + 0.0844x
y = 0.5385 + 0.9820x
y = 60.81 − 25.72x
y = −0.5363 + 1.087x
y = 1.890 − 1.002x
y = −0.1844 + 0.7538x
y = 3.980 + 49.78x
In the marsh, we found that flow rate was the only significant predictor of decomposition values (Table 3). Flow rate explained the most variation in decay rate and mass loss (85 and 82%, respectively, Table 3). Flow rate also described positive relationships with P and N concentrations and final mass P, and negative relationships with C:N molar ratio and absolute mass C (Table 3). Notably, flow rates measured in the marsh varied from 0.87 to 1.23 cm s−1, 0.63 to 0.82 cm s−1, and 0.62 to 0.72 cm s−1 in Flow, Wall and No Flow treatments, respectively (Troxler Gann and Childers 2006).
In wetland landscapes, regardless of latitude, nutrient status or precipitation patterns, heterogeneity commonly emerges as a landscape feature. Landscape heterogeneity occurs with the differentiation of an oligotrophic wetland feature from a more nutrient rich matrix in the case of pocosin wetlands of the SE US (Bridgham and Richardson 1993), raised bogs of New Zealand (Bragazza and Gerdol 2002; Clarkson et al. 2005) and coastal Panama (Troxler 2007) and patterned peatlands of circumpolar latitudes (Glaser and Janssens 1986). Heterogeneity also occurs as the development of a more nutrient rich feature over a nutrient depauperate landscape as in the case of tree islands in alpine tundra (Seastedt and Adams 2001), the Okavango delta (Ellery et al. 1998), seasonally-inundated savannas of Brazil and Bolivia (Ponce and Cunha 1993; Langstroth 1996) and the subtropical Everglades (Troxler Gann et al. 2005; Wetzel et al. 2005).
Regardless of the origin of emergent heterogeneous landscape components, feedbacks driving differential processes of accumulation and loss, including litter decomposition, have been suggested as important mechanisms in the maintenance of wetland heterogeneity at the landscape-scale (Bauer 2004). In many wetland landscapes, heterogeneous features appear to differentiate as C is accumulated at a higher rate relative to the landscape matrix. This process has been linked to differences in hydrology, nutrient availability, standing biomass, and litter quality between the heterogeneous feature and its matrix system (Glaser and Janssens 1986; Bridgham and Richardson 1993; Rietkerk et al. 2004; Couwenberg and Joosten 2005; Belyea and Baird 2006). This study suggested that this process also occurs where peatland tree islands emerge as heterogeneous features of the Everglades wetland landscape.
Previous studies have investigated litter decomposition in heterogeneous elements of wetland landscapes (i.e., Thormann and Bayley 1997). This approach provided an opportunity to explore the importance of ecosystem characteristics and hydrologic interventions on this important process in the oligotrophic Everglades. In wetland environments, leaf litter decomposition rates have been shown to vary from 0.06 yr−1 in a monocot bog in Alaska to 5.11 yr−1 in a riverine mangrove forest of Ecuador (Brinson et al. 1981; Twilley et al. 1997). In this study, degradation rates in peatland tree islands were comparable to, but on average lower than, rates for temperate flooded forests (0.42–1.3 yr−1; Baker et al. 2001), and among the lowest rates reported for wetland ecosystems (this study: 0.23 ± 0.03 yr−1). This is a surprising result given the subtropical climate of south Florida but is likely in part due to the extremely oligotrophic status of the Everglades (Noe et al. 2001).
Despite the relatively small difference in litter decomposition rates between ecosystems and among flow levels, decomposition processes resulted in large differences in C and nutrient concentrations, molar ratios, and absolute mass after 64 weeks (Table 2). Although large scale disturbances like fire and drought are thought to be important drivers in wetlands (Hogg et al. 1992), the differences we found likely serve to maintain differential processes of accumulation and loss between tree islands and adjacent marsh in the periods between disturbance events.
Tree islands in the marsh landscape matrix
Litter quality is often invoked as an important control on litter decomposition rates and the nutrient dynamics of litter decomposition (Brinson et al. 1981; Webster and Benfield 1986; Villar et al. 2001). In this study, we used leaf litter of similar initial quality across all treatment levels and ecosystems so as to evaluate the effects of flow and ecosystem independently of litter quality. Despite the fact that C. icaco litter has slightly lower N:P and C:P ratios than C. jamaicense (Rubio and Childers 2006), the dominant species that would otherwise be decomposing in the marsh, C. icaco litter served as an acceptable substrate with which to evaluate decomposition processes.
Here, the environment created by tree island ecosystems is markedly different from the marsh landscape of the study area. Tree islands maintain higher standing biomass, greater total litter deposition and standing litter cover, deposition of more refractory materials (i.e., wood) with lower light penetration to the forest floor. This results in large variation in biogeochemical characteristics between the two ecosystem types (Troxler Gann et al. 2005). For example, when compared with marsh sites, tree island surface water contains two times greater TN and TOC concentrations (tree island: 39.28 ± 5.71 and 1816.67 ± 250.00 μM l−1, respectively; marsh: 15.00 ± 10.71 and 658.33 ± 42.5 μM l−1, respectively), and an order of magnitude greater TP concentrations (tree island: 2.32 ± 0.21 μM l−1; marsh: 0.09 ± 0.01 μM l−1) in the early wet season (Troxler and Childers, unpublished data). These ecosystem differences likely exert the strongest controls on differences in mass loss, decay rates, and nutrient dynamics promoting the differential processes that lead to lower C loss and net P accumulation in tree islands and loss of C and P in the marsh (Brinson et al. 1981; Webster and Benfield 1986; Baker et al. 2001; Middleton and McKee 2001, Table 2).
Hydrologic modifications alter mechanisms promoting landscape differentiation
In oligotrophic wetland environments, the decomposition of low nutrient quality leaf litter typically takes place in a soil and water environment of similarly low nutrient status. In general, studies of litter decomposition in oligotrophic Everglades wetlands (soil TP: 100–400 μg P g−1; water TP: 0.1–0.2 μM l−1; Noe et al. 2001) have shown net N immobilization and either mineralization or no change in P content (Newman et al. 2001; Debusk and Reddy 2005; Corstanje et al. 2005; Rubio and Childers 2006). However, when P concentrations of surface water or soil exceed oligotrophic levels, either as a function of natural or human-induced P sources, net immobilization is the most consistent result (Newman et al. 2001; Davis et al. 2003; Debusk and Reddy 2005; Corstanje et al. 2005; Rubio and Childers 2006). Thus, as tree islands have moderate (above oligotrophic levels) soil and surface water P concentrations, P immobilization would be expected. Whereas in the marsh, we found net P mineralization under conditions where inflowing water was of low nutrient status (Figs. 4 and 5). Yet, with greater than oligotrophic levels of inflowing P, net P immobilization becomes independent of ecosystem type leading to similar net accumulation of P (Fig. 5, Table 2). This can be conceptualized with two contrasting models depicting the net response of leaf litter decomposition in tree islands and adjacent marsh in conditions of low and high P in inflowing surface water (Fig. 6, scenarios A and B). Thus, under unimpacted conditions, positive and negative feedbacks occurring in contrasting environments via decomposition processes promote the maintenance of heterogeneous landscape features in this system. Whereas, with P loading associated with hydrologic interventions, at least in the short term, there are similarly positive feedbacks (unidirectional) that emerge in contrasting environments that degrade this mechanism for landscape differentiation (Fig. 6). Strong linear relationships between hydrologic factors and decomposition values further elucidated these patterns.
This study suggested that both water level and the supply of P by inflowing water were important factors regulating tree island P immobilization via litter decomposition (Table 3). Peatland tree islands of this area have soils with greater P concentrations than surrounding marshes, but are dominated by tree species delivering senesced leaves of very low P content and high C content (N:P and C:P ratios typically 100 and 8,500, respectively; Troxler Gann et al. 2005). This constant influx of low quality litter coupled with slightly acidic soil and surface water conditions promote a substrate for accumulation of microbial biomass P and CaPO4 precipitation (Keuhn et al. 2000; Noe et al. 2001). If plant species maintain delivery of low quality litter (high N:P, high C:P), despite increased surface water loads of P, tree islands will likely accumulate P. However, under circumstances in which hydrologic modifications alter litter nutrient quality or species composition favoring plants with higher quality litter, C and P mineralization of leaf litter in peatland tree islands is a potential outcome, i.e., convergence of ecosystem types (Fig. 6, scenario C).
The marsh ecosystem, on the other hand, appeared to be more sensitive to flow rate directly (Table 3). While nutrient loads into the C-111 would co-vary with flow rates, nutrient loads were not similarly correlated with decomposition values after 64 weeks (Table 3). Thus, advective sheet flow with even a minor change in P concentrations of surface water appeared to influence N and P dynamics and P accumulation by decomposing leaf litter within the marsh ecosystem. This coincided with positive relationships with mass loss and decay rates, and negative relationships with absolute C mass, suggesting a mechanism that promotes maintenance of mineral soils in the marsh landscape. Furthermore, considering that mass loss was highest where flow rates were highest (in the marsh of the Flow treatment level; 55.7 ± 4.1%) as compared with mass loss in the marsh of Wall and No Flow levels (38.9 ± 4.17 and 38.1 ± 4.1%) suggests an additive factor influencing C and nutrient remineralization in the marsh system. It is interesting to note that given a situation where light, nutrient and temperature-limited conditions are removed (marsh, flow treatment), flow rates promote the highest mass loss and decay rates, as has been found in some stream studies (Lepori et al. 2005; Ardon et al. 2006).
The similar P concentration and accumulation responses that occurred in contrasting wetland ecosystems of the Flow treatment level strongly suggests that canal influences modify mechanisms for landscape differentiation in this study. Furthermore, small differences in P loading can determine whether tree island and marsh ecosystems have differential P dynamics or similarly accumulate P. Our study suggests another mechanism by which reduced landscape heterogeneity may occur in hydrologically-altered wetland landscapes, at least in the short term. With long term P loading, and changes in tree island canopy structure or leaf quality, convergence of ecosystem types, and loss of landscape heterogeneity, is a possible result, and may signal an altered tree island ecosystem state.
We would like to thank the Wetland Ecosystems Ecology group for field and lab support, as well as critical reviews that greatly improved this manuscript. Critical to the accomplishment of this work were Damon Rondeau and undergraduate research assistants Josh Mahoney and Simone Normile. This research was partially supported by the South Florida Water Management District under several sequential contracts, and by the National Science Foundation through the Florida Coastal Everglades Long-Term Ecological Research Program (DEB-9901514).