1 ALTERNATIVE STABLE STATES AN D SELF-ORGANIZED PATTERNING: EVERGLADES RIDGE AND SLOUGH MOSAIC By DANIELLE WATTS A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2008
2 2008 Danielle Watts
3 To my friends, my husband, and the Everglades.
4 ACKNOWLEDGMENTS This thes is represents much more than the research on the pages; it is the culmination of the encouragement of a large number of people to follow my heart and to find a vocation that is also an avocation. As a result I am doing what I love, and would rather be nowhere else, doing nothing else, than what I am doing right now. There are a number of people whom I must acknowledge; they made this thesis happen either via their expertise or their help in my keeping my sanity. Among the former, Sanjay Lamsal provided invaluable help with all things spatial, and Jim Heffernan was a major part of the theoretical development of this project. As for my sanity, I would ha ve been lost without Elizabeth Deimeke, Danielle King, Dina Liebowitz Lauren Long, and Justin Vogel. Not only did they help make graduate school fun, they also act ed as field technicians when I was at my most desperate. Among others who helped make fieldwork happen are Jason Evans and Laura Schreeg. Thank you for jumping in and lending a hand--I couldnt have done this without you. My husband, Adam Watts, was amazing through out, acting as sometime editor, field hand, nursemaid, and best friend. I would like to express my gratitude to my committee members, Mark Clark and Todd Osborne, for their mentoring and advice throughout this project. Finally, I am indebted to my advisor, Matthew Cohen, for his unfailing support and encouragement to stretch my intellect. Dr. Cohen s willingness to take a chance on the hair-brained schemes of a green graduate student is the buildin g block this thesis is formed upon.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........6 LIST OF FIGURES.........................................................................................................................7 ABSTRACT.....................................................................................................................................8 CHAP TER 1 LITERATURE SYNTHESIS.................................................................................................10 Introduction................................................................................................................... ..........10 The Landscape........................................................................................................................12 Hydrologic Changes............................................................................................................. ..13 Plant Communities.............................................................................................................. ....15 The Processes..........................................................................................................................18 Research Objectives............................................................................................................ ....23 Objective 1.......................................................................................................................23 Objective 2.......................................................................................................................23 2 EVIDENCE FOR PERSISTANCE AND LOSS OF ALTERNATIVE STABLE STATES IN THE RIDGE SLOUGH MOSAIC OF THE EVERGLADES ...........................30 Introduction................................................................................................................... ..........30 Methods..................................................................................................................................33 Study Area.......................................................................................................................33 Sampling..........................................................................................................................34 Data Analysis...................................................................................................................35 Results.....................................................................................................................................38 Discussion...............................................................................................................................39 3 SUMMARY AND CONCLUSIONS.....................................................................................57 LIST OF REFERENCES...............................................................................................................63 BIOGRAPHICAL SKETCH.........................................................................................................69
6 LIST OF TABLES Table page 1-1 Net aboveground primary productivity estimates in the Everglades................................. 29 2-1 Summary of regressions for slough wate r d epths as a function of latitude....................... 47 2-2 Summary of regressions for selected areas fo r ridge water depths. ................................... 48 2-3 Kurtosis of the water depth distributions for each sampling unit...................................... 50 2-4 Results of a-priori defined community water depth analysis ............................................. 52 2-5 Results of spatial analyses.................................................................................................56 3-1 Hypotheses of mechanisms gove rning ridge-slough m aintenance.................................... 62
7 LIST OF FIGURES Figure page 1-1 Aerial views of the ridge-slough mosaic........................................................................... 25 1-2 Vegetative communities of the ri dge-slough portion of the Everglad es............................ 26 1-3 Feedbacks to the multiple steady states of the ridge and slough portion of the Everglades ..........................................................................................................................27 1-4 Point model for theoretical relationship of the carbon balance in ridges and sloughs ......28 2-1 Conceptual model for autogenic feed backs m aintaining stable alternative ecosystems..................................................................................................................... ....44 2-2 Bifurcation model for alternative stab le states with hypot hetical water depth distributions super-im posed............................................................................................... 45 2-3 Map of South Florida with landscape sampling blocks..................................................... 46 2-5 Average water depths of each comm unity of the landscape sam pling units..................... 51 2-6 Relative incidence of vegetation communities by sampling unit...................................... 53 2-7 Correlograms of water dept hs with lag distances of h=20m ............................................. 54 2-8 Metrics of spatial pattern for sampling units..................................................................... 55
8 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science ALTERNATIVE STABLE STATES AN D SELF-ORGANIZED PATTERNING: EVERGLADES RIDGE AND SLOUGH MOSAIC By Danielle Watts August 2008 Chair: Matthew Cohen Major: Interdisciplinary Ecology The Everglades in South Florida is a large subtropical peat wetland with important hydrologic, ecologic, and cultura l values. The ridge and slough mosaic is a major landscape component of the Everglades, characteri zed by elevated ridges of sawgrass ( Cladium jamaicense ) interspersed among deeper water sloughs comprised of floating and emergent species as well as calcareous periphyton. Aut ogenic feedbacks among depth and duration of inundation, plant community composition, and net pe at accretion are hypothesized to create and stabilize these multiple ecosystem equilibria. While significant research has focused on processes leading to these potential stable states, few studies have formally examined the hypothesis that ridge and slough represent alterna tive stable states, and none has examined the resilience of alternative stability domains to hydrologic change. In this study, water depth (a proxy for soil elevation) was measured along a gradient of hydrologic impairment in the ridge-slough region of the central Everglades to evaluate predictions that follow from the multiple stable state hypothesis: 1) soil elevations s how strong fidelity to community type with little overlap in elevation distributions in ar eas with minimum hydrologic im pacts; 2) increasing hydrologic modification increases overlap between eleva tion distributions; 3) hydrologic modification increases the variance of elevations within comm unities; 4) kurtosis of the joint soil elevation
9 distribution increases with in creasing hydrologic modification; and 5) spatial anisotropy (diagnostic of landscape orientati on with flow), and spatial stru cture (diagnostic of landscape scale self-organization) declin e with hydrologic modification. Re sults accord with predictions; properties of bimodality and spatial structure ar e strong in sites with minimal hydrologic impact and change significantly with hydrologic modification. These non-linear ecosystem processes and feedbacks need to be incorporated into m odeling and restoration plan ning in the Everglades, as they may act as constr aints on restoration goals.
10 CHAPTER 1 LITERATURE SYNTHESIS Introduction Alternative stable states have been desc ribed for numerous ecosystems in the recent literature (Scheffer and Carpenter, 2003; S uding et al., 2004; Schrde r et al., 2005). These ecosystems exhibit resistance to changes over some range of an ecosystem driver via homeostatic feedbacks; however, th resholds at the edge of that ra nge lead to catastrophic regime shifts to some new self-reinforcing ecosystem st ate (Scheffer et al., 2001 ). A key component of these shifts is hysteresis (path dependency) wher ein the threshold at whic h regime shifts occur differ in opposing directions of change. Regime shif ts between alternative stable states have been expressed at a population level (e.g., shifts between high biomass and low biomass in the presence of a predator) and at an ecosystem le vel (e.g., nutrient loading driving a lake system from macrophyte dominance to algal dominance) (s ee review of concepts in Beisner et al., 2003). Alternative stable state theory has more recentl y been extended in la ndscapes with regular patterning, wherein feedback mechanisms from a plant community engineer ecosystem attributes at the local scale to their favor (e.g., changing nutrient content, in filtration rate, soil elevation) rendering more distal locations disfavorable to them; the re sult of such scale-dependent autogenic reinforcement is la ndscape patterning (Ludwig et al ., 1999; Rietkerk et al., 2004a; Rietkerk et al., 2004b; Rietkerk and van de Koppel, 2008). Among these feedbacks in patterned landscapes are a redistribution of water in arid systems; competition for limiting resources in savannas; and ponding of water and advective transport of nutrient by differences in transpiration rates in peatlands (Rietkerk et al., 2004a).
11 The ridge-slough mosaic, the dom inant landscape feature of the central peat-dominated Everglades, exhibits corrugated soil elevati on patterning suggestive of those landscapes increasingly understood as being st ructured by alternative stable states. Ridges, which occur at higher soil elevations, are dominated by sawgrass ( Cladium jamaicense ). Interspersed among the ridges are deeper water sloughs, comprised of floa ting-leaved and emergent species as well as extensive periphyton. This patterning has been in existence for the last 2,700 years, and the centers of the ridges and sloughs have been invariant for the la st 1000 years (Bernhardt et al., 2004). However, whether the ridge and slough mosaic represent alternative st able states in the Everglades, with multiple state equilibria, is unknown. For patterned systems phase shifts can occur at the landscape, from a self-organized patchy state to a homogenous state (Rietkerk et al., 200 4a). The model for shif ts between homogeneity and self-organized patchiness invokes a region of global bistability (Rietkerk et al., 2004a) along a gradient of environmental conditions wher ein alternative ecosystem states are stable. Outside of the region, only one state is stable. Th e implications of this model for the Everglades are outlined by Wu et al. (2006) who described catastrophic shifts from the ridge-slough pattern to a nearly homogenous state driven by landscape hydrologic modifications. A key line of evidence for assessing whether a lternative stable states describe ridges and sloughs is peat elevation differences between th e two ecosystems. More specifically, transition zones between the two ecosystems are very dis tinct and narrow in th e least hydrologicallyaltered parts of the landscape. An analysis of the distributions of elevations should indicate discreteness between the two ecosystems. A histogram of el evations should center around two distinct distributions (ridges a nd sloughs), suggesting that localiz ed feedbacks maintain discrete ecosystem states. Moreover, soil elevation distributions should allow us to evaluate the effect of
12 hydrologic modification on the stab ility of the ridge and slou gh mosaic. Should ridges and sloughs appear to be alternative stable states, then it is important to incorporate ecosystem thresholds and feedbacks to any model and restor ation plan in the Everglades, as these may act as constraints on restoration goals (Suding et al., 2004). Evidence c ontrary to this model would be in a spatially stochastic structure of peat elevations. That is, if there are not multiple ecosystem equilibria, elevation differences betw een ridges and sloughs should be random due to non-directional ecosystem development pathways. A lack of bimodal distribution would indicate that the system is in a constant, non-linear transitional state betw een ridges and sloughs. In this chapter, I outline what is known about pattern formation and decline in the Everglades ridge-slough landscape, and discu ss the relative merits of various proposed mechanism for that phenomenon. In Chapter 2, I eval uate evidence for alternative stable states, and examine the effects of hydrologic modifica tion on alternative state stability. Chapter 3 discusses the implications of th is study, and suggests future dire ctions for research on the ridgeslough pattern in the central Everglades, emphasi zing inference about a lternative hypothetical mechanisms for ridge-slough maintenance. The Landscape Much of Florida, south of Lake Okeechobee, is a single large drainage basin that includes the Everglades, the Big Cypress basin, Shar k Slough, and the salt marshes and mangrove swamps of the coastal zone. Historically its hyd rology was dominated by seasonal rainfall, with great sheets of shallow water moving slowly south over a uni quely flat landscape (vertical gradients of 3 cm per 1 km) and draining into Florida Bay. The Everglades is the largest subtropical wetland complex in the US, a nd was once comprised of 500,000 ha (5,000km2) of sawgrass marshes, wet prairies, slough aquatic communities, and tree islands (Loveless, 1959). These vegetative communities are associated with specific hydrol ogy and water depths
13 (Loveless, 1959; Newman et al., 1996; Busch et al., 1998). The central Everglades is a peatdominated wetland underlain by limestone bedrock which form ed wetland environments approximately 5000-4500 YBP (Gleason and Stone, 1997). Gradients in soil depth and plant productivity occur with varia tions in topography and hydrology (Craft and Richardson, 1993b; Gleason and Stone, 1997). The Everglades has uniquely low topographic relief; increases in soil elevation as little as 10 cm can result in a 45% reduction in mean water depth and a 20% reduction in inundation frequency in some areas (David, 1996). The central portion of the Ev erglades is largely a mosaic of ridges and sloughs, where higher elevation peat ridges dominated by sawgrass or tree island vegetation are oriented parallel to water flow (Figure 1-1), interspersed am ong lower elevation sloughs dominated mainly by either short hydroperiod marsh vegetation or fl oating aquatic vegetation (Ross et al., 2006). The wet season occurs from May through October w ith the dry season occurring November through April (Duever et al., 1997). The precipitation in the wet season is primarily from localized thunderstorms and is therefore erratic in distribution both temporally an d spatially. The plant communities comprising the historic landscape ar e adapted to oligotrophic conditions (i.e., surface waters with low concentrations of dissolved minerals and nutrients). Hydrologic Changes Since the turn of the 20th century, large sections of the Ev erglades have been drained via canals (Light and Dineen, 1997), converted to ag riculture (Snyder and Davidson, 1997), enriched with phosphorus (Davis and Ogden, 1997), and unnaturally flooded by the construction of impoundments (Kushlan, 1990). Water management projects in the Everglades culminated in the development of 1000 miles of canals, 720 mile s of levees, 16 pumping stations, and approximately 200 control structures. The new hydrologic system redirects vast quantities of water to the coasts, resulting in 70% less water reaching Everglades National Park and Florida
14 Bay (Perry, 2004). The historic Ev erglades is now partitioned in to the Everglades Agricultural Area (EAA), Water Conservation Areas (WCA) 1, 2, and 3, and the Everglades National Park (ENP). While the compartmentalization of the Ever glades has helped to preserve the marshes by restricting urban and agriculture development, compartmentalization has also had negative consequences on marsh biota as the water flow and hydroperiod have been altered (Figure 1-1). Ongoing conflicts over water resources between human users and ecosystem requirements led, in part, to the Central and So uthern Florida Project Comprehensive Review Study (C & SF project), authorized by US Congre ss in 1992 to study the feasibility of modifying water-control structures and ope rations to restore the south Fl orida ecosystem and provide for other water-related needs of the region (Perry, 2004). The C & SF project reported a general decline in the health of the Everglades, including a 90-95% loss of wading birds; species loss (there are 68 federally liste d species); deteriorated water quality via eutrophication and contaminants; habitat degradation and loss; and a decline in fisheries in estuaries and Florida Bay (Perry, 2004), which led to the creation of the Comprehensive Everglades Restoration Plan (CERP) as a framework for implementing hydrologi c restoration. The main goals were improved water quality and restoration of th e seasonal water flows to ENP. Among the restoration goals were distribu tion of 80% of available water to support ecosystem needs, 20% to urban use, and water delivery with a more na tural hydropattern (Perry, 2004). More than 240 miles of levees and canals ar e scheduled to be removed, including most of the Miami canal in WCA 3, Tamiami Trail (US-41) will be rebuilt with br idges and culverts, and the levee separating the Big Cypress National Pr eserve from ENP will be removed (Perry, 2004). Other goals include the removal of contaminan ts (i.e., pesticides, biological oxygen demand (BOD), bacteria, and suspended solids) and phosphorus.
15 While anthropogenic nutrient inputs ha ve strong effects on the plant and algal communities in the Everglades in areas adjacent to canal structures, nutrient loads are largely attenuated for ecosystems in the interior of WCA 3A (Childers et al., 2003). As such, this is an ideal area for understanding the role of hydrology in shaping the Everglades, decoupled from nutrient loading effects. WCA 3A is leveed on three sides. Tamiami Trail, constructed between 1915-1928 to connect Miami on the east coast with Naples on the west coast, delineates the southern boundary between WCA 3A and ENP. While the road contains numerous culverts and bridges designed to allow flow from north to so uth, a levee (Levee 29) still regulates flow, which has increased water depth on the WCA 3A side and, in turn, disrupted the seasonal hydrologic pattern (Childers et al., 2003), reduced vegeta tive diversity (David, 1996), and increased the proportion of slough area to ridge area (Wu et al ., 2006). The northern area of WCA 3A has an inundation period of 32-61% of the year (mean water depth is 10-18cm); in contrast, the impounded southern part of 3A is much wetter, with an inundati on period of 96% of the year (mean water depth > 60cm) (David, 1996). Reductions in spatial heterogene ity associated with hydrologic modification (Wu et al. 2006) alter the size and diversity of habitats, with important negative consequences for many wildlif e populations (Davis and Ogden, 1997). Plant Communities Herbaceous communities in the Everglades are generally grouped according to dominance by one of the following: 1) Cladium jamaicense Crantz, either in nearly monospecific stands or sparsely in marl prairies; 2) Eleocharis cellulosa in wet prairie habitats usually combined with Sagittaria spp and Rhyncospora spp; 3) Panicum spp and Paspalidium spp, also in wet prairies; or 4) Utricularia spp and Bacopa spp, generally found in sloughs (Busch et al., 1998) (Figure 1-2). Other species, such as Nymphaea odorata are associated with deeper water habitats and frequently are found in sloughs. Ther e is a gradient of increasing water depth and
16 above water decreasing habitat co mplexity as one moves from C. jamaicense stands toward wet prairies and thence to slo ughs (Jordan et al., 1997). C. jamaicense is a rhizomatous perennial sedge adapted to oligotrophic conditions (Steward and Ornes, 1975; Lorenzen et al., 2001) which may help explain its dominance in the Everglades. It reproduces primarily vegetativ ely, chiefly through rhiz ome production, but also through witchs brooms, an unusual vegetative prolif eration on the reproductive organs (Miao et al., 1998). Physiological constraints, including lo w seed germination rates (Lorenzen et al. 2000) and growth responses to oxygen limitation (Lorenzen et al. 2001) restrict the spatial extent of C. jamaicense in this landscape. One physiological attribute precluding C. jamaicense from deeper water habitats may be its root response to reducing conditions. Chabbi et al. (2000) found the r oot response to plant oxygen demand was limited. A detailed examination of radial oxygen loss to root tips showed a mechanism for developing barriers to leakage al ong root axes (Chabbi et al. 2000). However, increased alcohol dehydrogenase activ ity and ethanol concentrations in C. jamaicense suggest that this mechanism is not sufficient to pr event oxygen deficiencies under extended inundated conditions. Chabbi et al. ( 2000) additionally found the deve lopment of aerenchyma and subsequent increases in root porosity are also limited in C. jamaicense Further compounding this limitation, oxidation of the rhizosphere after flooding is re duced, which may limit mineralization and availability of nutrients in the root zone. These physiological limitations may be key to understand ing the presence of C. jamaicense along higher elevation ridges and its absence in deeper water sloughs; moreover, re productive constraints may be important to understanding the time domain of sawgrass encroachment into sloughs with hydrologic modification.
17 Sloughs are a few inches to one two feet below the elevation of adjacent ridges (Loveless, 1959). The dominant vegetation is Nymphaea odorata and Nuphar advena as well as sparse, stunted C. jamaicense and Utricularia spp (Loveless, 1959). Utricularia is abundant in many slough areas and may play an important role in the formation of organic soils in sloughs (Loveless, 1959). Lewis (2006) found a higher residual fiber content in Utricularia than any other slough species, though values we re substantially lower than for C. jamaicense. Periphyton is an important component of open wa ter habitats in the Everglades and varies both temporally and spatially in terms of bi omass, productivity, and species richness and diversity in relation to macrophyte abundance an d nutrients (Gleason and Spackman Jr., 1974; Wood and Maynard, 1974; McCormick et al ., 1998; McCormick and Stevenson, 1998). Periphyton plays an important role in so il formation, as calcium carbonate (CaCO3) can represent more than 90% of the weight of periphyton in some Everglades communities (Browder et al., 1982). High pH conditions cause CaCO3 to precipitate, which can yi eld material for sediment building. Where C. jamaicense is abundant, periphyton is sparse (7-52g AFDM m-2) and less productive (0.25-0.70 g C m-2 d-1) than in sloughs (100-1600g AFDM m-2; 1.75-11.49g C m-2 d-1) (McCormick et al., 1998). Interestingly, unlike w ith macrophytes, water depth does not appear to be a factor influencing biomass and speci es composition of periphyton (Wood and Maynard, 1974; Browder et al., 1982; McCo rmick et al., 1998). However, phosphorus loads do regulate periphyton production and respiration rates (Iwan iec et al., 2006). Even small changes in phosphorus, from less than 7 to 10-20 g L-1 can change the periphyton community to one dominated by filamentous green algae (McC ormick et al., 1998). Periphyton acts as a phosphorus sink, and therefore has a role in preserving the oligotr ophic nature of the Everglades (McCormick et al., 1998; Vaithiyanathan and Richardson, 1998). Clear differences in the
18 characteristics and life history of the vegetation between ridges, sloughs and wet prairies sets the stage for associated and cri tically important differences in ecological processes. The Processes Proposed explanations for the development of ridges and sloughs have included the dislodging of patches of water lily marsh peat under deep water conditions (Gleason and Stone, 1997); flocculent transport and sediment deposition resulting in the creation of ridges (Jorczak, 2006); high-energy events depositing floc and or ganic matter on the interior of ridges, thus depositing greater amounts of nutrients on ridges (Leonard et al., 2006); or high flow volumes with a scouring action on sloughs, maintaining a distinct elevation difference between ridges and sloughs (Ogden, 2005). The velocity of water and its action on elevation does not appear to be an important factor, as it does not vary among ridges and slou ghs (Leonard et al., 2006); however, there is seasonal variation of the velocity of water be tween adjacent ridges and sloughs, an observation that Jorczak (2006) links to water levels. The sa me study showed an increase in total suspended solids in the water column moving from north to south from northern WCA3A to Everglades National Park, and slightly higher suspended soli ds in sloughs compared to ridges. Another study similarly found accumulation rates of particulates to be lower on ridges than in sloughs (Leonard et al., 2006). Accumulation rate s are generally low in Shark River Slough (mean rate of particulate accumulation in Oct. 2003 of 44.8 gdw d-1 m-2 and 122.4 gdw d-1 m-2 for ridges and sloughs, respectively). As such, sediment deposition from the water column as the mechanism of ridge formation and maintenance is unlikely. It is possible that physical structure of sawgrass on a ridge may act as a baffle, allowing fine organi c matter and nutrients to settle out of the water column, and thereby increase the nutrient status of the ridge. If so, this could help explain higher
19 productivity on ridges. Unfortunately, an analysis of the mass and characteristics of particulates deposited on ridges has not yet been done. Another mechanism proposed for maintaining multiple stable equilibria focuses on autogenic feedbacks between soil elevation, plant productivity and peat oxida tion (Givnish et al., 2007; Larsen et al., 2007). As Odum (1971) elegan tly stated Ecosystems are capable of selfmaintenance and self-regulation as are thei r component populations and organisms. The processes of maintenance and self-regulation of the Everglades center around hydrological effects on community composition, decomposition and productivity. As discussed above, water depth has an influence on the species present at any particular location, and is an important control on ecosystem productivity via both changes in composition and inundation stress. Decomposition of plant matter is also influenced by soil elevation by affecting both the diffusion of oxygen through the water columne to the sedime nt and the probability that a particular location will be exposed to the air with natural variation in water depth. The combined processes of plant (and litter) production and subsequent mineralization control peat accretion; changes in water depth that occur ins response to changes in peat accretion act as a feedback mechanism controlling plant community composition (Figure 1-3). Despite strongly differential ra tes of primary productivity, ridges and sloughs must accrete peat at roughly the same rate in order for the la ndscape pattern to be pe rsistent. To compensate for large differences in productivity, carbon mineralization from the soil and water column (as CO2 and CH4), must be higher in ridges. Oxidation di fferences are likely to be most pronounced during seasonal low water, when ridge sites have a higher probability of exposure and therefore aerobic oxidation. This mechanism is supported by ot her studies showing that water depth has a strong influence on CO2 evolution in the Everglades (D ebusk and Reddy, 2003; Jorczak, 2006).
20 Time integrated ecosystem carbon accretion rate s at equilibrium must be equal between ridges and sloughs (Fig 1-4a); were this not tr ue, elevation differences between ridge and slough would continue to grow unchecked by feedbacks such as accelerated re spiration. Moreover, if ridges and sloughs represent alternative stable states any location that is a ccreting peat at rates different from the equilibrium st ate would be unstable; that is, ecosystem feedbacks in the form of changes in production and/or respiration would force the carbon balance back to the equilibrium state. For example, a particularly shallow location would experience accelerated soil respiration such that soil elevations would decline and the system would move back to equilibrium. Similarly, oxidation rates at deep wa ter sites would force the system to accrete soil, moving that location towards the slough equili brium. What happens in the systems falling between equilibrium states may be more vari ed, where the elevation may be either rising, moving towards ridge systems, or lowering, moving towards the slough system; moreover, community composition shifts may lead to catast rophic transitions betwee n states. The keystone driver for these feedback contro ls on the carbon balance is h ydrology (water depth); as such, when hydrology is altered, shifts between ridges and sloughs are expected to ensue. The transition from ridge to slough is expected to be strongly influenced by competition between sawgrass and slough specie s (Fig 1-4b). The directiona lity of change should be indicated by the carbon balance: If the net carbon accretion is highe r than the landscape equilibrium rate, then the system is moving towards a ridge. Alternatively, if the net carbon accretion is less than equlibrium, then the system is moving towards a slough. Understanding peat formation, then, is necessary to understand the variations in the topography found in the ridge and slough mosaic.
21 The characteristics of peat formed in ri dges and sloughs differ in many ways. Vegetation on ridges has greater residual fi ber content and a higher C:N ratio (Lewis, 2005). Many slough species require buoyancy in leaf structures and therefor e lack the carbon structures necessary in emergent plants such as C. jamaicense (Lewis, 2005). While hydroperi od has an effect on litter decomposition (Lewis, 2005), it may be the recalcitr ant nature of C. jamaicense dictating the litter decomposition ra te on ridges. However, when peat is exposed to oxygen, it decomposes at a much higher rate. Peat has been esti mated to accrete at rates of 2.8-3.2 mm yr-1 at reduced hydroperiod areas, 1.6-2.0 mm yr-1 at higher hydroperiod areas, and as fast as 4.0-5.67 mm yr-1 in phosphorus enriched areas (C raft and Richardson, 1993a, b). Another factor essential to understanding peat formation is in the recalcitrance of the peat. Jorczak (2006) found no diffe rence in carbon dioxide (CO2) emission from ridge soils versus slough soils. This finding suggests that although vegetative matte r decomposes more readily in sloughs, the recalcitrance of the soils is similar once the litter becomes soil. Jorczak (2006) also reported a threefold in crease in soil respir ation when the water table fell below the soil surface. Other studies have shown a similar increase in CO2 flux when water tables are decreased. Debusk and Reddy (2003) reported an increasing flux of CO2 in Everglades peat profiles as water levels were deceased incrementally between 0 and -15 cm. However, while studies have correlated methane (CH4) production to water table de pth, temperature, and pH fluctuations, a thorough understanding of meth ane production has thus far proven elusive (Bachoon and Jones, 1992; Drake et al., 1996; Debusk and Reddy, 2003; Chauhan et al., 2004; Jorczak, 2006): CH4 may be oxidized through the rhizosphere lost to methanogenesis, or fluxed through vegetative matter. The flux through vegetative matter is a particularly important factor,
22 as increases in C. jamaicense biomass in a marl prairie were closely correlated with rates of methane emission for the ecosystem (Whiting et al., 1991). Sawgrass communities are highly productive, with net aboveground primary productivity (NAPP) values ranging from 300 to 5656 gdw m-2 yr-1 (Table 1-1), and are consistently higher than either slough or wet prairie communities. Increasing annual water depths and hydroperiod reduce the productivity of sawgrass-dominated communities (Childers et al., 2006), which may help explain the transition of ridges to sl oughs in areas affected by impounded water. Productivity of C. jamaicense also varies concurrently with phosphorus (Daoust and Childers, 1999). While ridges are dominated by C. jamaicense, slough vegetation is mu ch more variable, and therefore the productivity of the ecosystem is expected to be much more variable (Table 11). This is particularly true as sloughs may be dominated by sedges, N. odorata, Eleocharis spp, periphyton, or change over the course of a y ear in plant dominance (Daoust and Childers, 1999), with reciprocal changes in ecosystem productivity. This study has two objectives: First, test hypotheses drawn from alte rnative stable state theory in the Everglades ridge -slough region. Second, quantify the effects of altered hydrologic regimes on the properties of the bi-stable ridge -slough landscape. To accomplish these tasks, I examined the distribution of so il elevations across la rge landscape blocks (2x4 km) spanning a gradient of hydrologic impairment in the central Everglades. The hypothe sis that ridges and sloughs represent alternative stable states leads to the predictions that: 1) soil elevations show strong fidelity to community type with little overlap in probability dens ity functions (pdf) of elevation in minimally hydrologically impacted areas; 2) increasing hydrologic modification decreases pdf separability; 3) hydrologic modificati on increases the variance of elevations within communities; 4) kurtosis of the joint soil elev ation distribution decreases with increasing
23 hydrologic modification; and 5) sp atial anisotropy declines w ith hydrologic modification. The results born out of this project, presented in Chapter 2, have a third intentionto develop simple, diagnostic measures of ecosystem health in the Everglades. Research Objectives Objective 1 Determ ine the extent to which ridges and slough s are discrete ecosystems via an analysis of the distributions of elevations. Hypothesis 1 : The ridge-slough landscape is an expr ession of alternative stable states. Prediction 1-1 : There is a bimodal distribution of elevation heights. Rationale: If the system is comprised of a series of gradients leading from ridges to sloughs, then water depths are clustered around a mean, which is the intermediary between ridges and sloughs. Thus, bimodality should indica te the discreteness of the boundary between ridges and sloughs. Prediction 1-2 : Vegetative communities exhibit a fi delity to discrete soil elevations. Prediction 1-3 : Autocorrelation is high in near-point neighbors, decreasin g with distances. Rationale: It has been suggested that scale-dependent feedbacks (locally positive, distally neutral or negative) lead to landscape patterning (Rietkerk and van de Koppel, 2008). These feedbacks would be evidenced by initially high autocorrel ation followed by decreasing correlation at each site. Objective 2 Determ ine the extent to which hydrologic modification has altered the underlying characteristics of ridge-slough patterning. Hypothesis 2 : Recent hydrologic modifications have le d to a loss of the characteristics of the patterning of the ridge-slough landscape. Prediction 2-1: Probability density function separability decreases at sites with hydrologic modification. Prediction 2-2: The occurrences of vegetative communities shifts under hydrologic modification.
24 Prediction 2-3: The variance of elevations within each community increases in areas with hydrologic modification. Rationale: As each community nears its wate r depth threshold, increases in transitional areas are observed. Prediction 2-4: Spatial anisotropy is lost under hydrological modification. Rationale: Given that the conserved portions of the ridge-slough mosa ic are anisotropic, and have long been long thought to be related to water flow, then hydrologic alteration (drainage and thus reduced flow rates; im poundment) should reduce anisotropy. Prediction 2-5: Spatial structure is lost under hydrologic modification. Rationale: While there is a ce rtain amount of stochasticity to soil characte ristics, if predictions 1-1 and 1-2 are correct, that is, vegetative communities have a strong fidelity to nonoverlapping peat elevations, then we can assume there is a high amount of non-stochastic spatial structure. The processes leading to this spatia l structure (community dyna mics, respiration, etc), however, can be assumed to be functioning diffe rently under hydrologic modification, increasing the stochastic nature of any patterning present.
25 Figure 1-1. Aerial views of the ridge-sl ough mosaic. Recent (last 100 year) hydrologic modifications have resulted in a genera l loss of the historic patterning of the landscape. WCA 3B has significantly reduced flow volumes and rates, whereas the southern portion of WCA 3A-S has impound ment of water. The central portion of WCA 3A-S is considered to have the mo st well conserved hydr ologic conditions, and thus the most well conserved ridge-slough pa tterning. The areas show n are sections of the landscape blocks sampled, as described in Chapter 2.
26 Figure 1-2. Vegetative communities of the ridge-slough portion of th e Everglades (excluding tree islands): (a) sawgrass ridge; (b) wet pr airie, generally dominated by either Panicum spp or Eleocharis spp; and (c) sloughs, generally dominated by Utricularia spp and N. odorata (picture (c) courte sy of Tyler Jones).
27 Figure 1-3. Feedbacks to the multiple steady states of the ridge and slough portion of the Everglades. Interactions between hydrol ogic conditions and pr oductivity/respiration (a) lead to the maintenance of either a ridge or a slough (b), with differing communities and peat elevatio ns. Large changes to the hydrologic conditions could, however, cause a shift from one ecosystem type to another.
28 Figure 1-4. Point model for theoretical relations hip of the carbon balance in ridges and sloughs. There is a range of water depths ideal for ridges and sloughs, where the ecosystem carbon balance at any point will move to wards the equilibrium state over any time period (a). When the water depth is at some intermediate level for an extended period of time, the effect of competition between the ridge and slough communities is seen (b), where sawgrass reaches a threshold le vel whereby it can no longer compete with the more opportunistic species found in sloughs, causing a community and thereby ecosystem shift.
29 Table 1-1. Net aboveground primary productivity estimates in the Everglades. Community Location g DW m-2 yr-1 Citation Sawgrass 2991 (Daoust and Childers, 1998) Sawgrass 802-3035 (Davis, 1989) Sawgrass Northern ENP 300-850 (Daoust and Childers, 1999) Wet Prairie Northern ENP 30-135 (Daoust and Childers, 1999) Wet Prairie 409 (Daoust and Childers, 1998) Wet Prairie Eastern Big Cypress 43.8-317.55 (Porter Jr., 1967) R/S Sawgrass Shark River Slough 166-800* (Ewe et al., 2006) Tall Ridge Sawgrass North 3A Central 3A North ENP Central ENP 4295.2 3667.7 5656.2.8 3765.6.2 (Lewis, 2005) Short Ridge Sawgrass North 3A North ENP Central ENP 2486.7.9 1094.6 13.4 3487.4.6 (Lewis, 2005) Slough North 3A Central 3A North ENP Central ENP 1683.5 .1 844.9 406.6 663.8 213.7 1515.7 .7 (Lewis, 2005) *Converted from g C m-2 yr-1 assuming 48% C content in sawgrass vegetation.
30 CHAPTER 2 EVIDENCE FOR PERSISTANCE AND LOSS OF ALTERNATIVE STABLE STATES IN THE RIDGE SLOUGH MOSAIC OF THE EVERGLADES Introduction Alternative stable state theory suggests that within ecosystems with the same exogenous drivers, two or more distinct stable states can ex ist. Generally, these stable states exhibit shifts between alternative basins of attraction in community co mposition (Holling, 1973) that are internally maintained via positive feedbacks between communities and ecosystem variables. These systems undergo catastrophic shifts between system states as ecosystem driver variables change (Scheffer and Carpenter, 2003; Sudi ng et al., 2004; Schrder et al., 2005), and are characterized by path-dependent hysteretic behavi or. That is, changes in ecosystem state lag changes in an ecosystem driver because of positi ve feedbacks that resist change; resistance in both directions results in transiti on thresholds that differ with the direction of the shift (Scheffer et al., 2001). A number of ecosystems exhibit regime shifts characteristic of alte rnative stable states (reviewed in Didham et al., 2005; Schrder et al., 2005). Systems that manifest multiple stable states in space rather than time (i.e., patterned landscapes) ha ve been the subject of recent interest (reviewed in Rietkerk et al., 2004a). These systems fo rm landscape pattern from local and regional feedback mechanisms wherein orga nisms (principally plants) change the nutrient content, water status, and/or soil elevation with reciprocal implic ations in ecosystem structure, composition and function (Ludwig et al., 1999; Rietkerk et al., 2004a; Rietkerk et al., 2004b). These scale dependent feedbacks (positive feedb acks locally, negative feedbacks at distance) lead to patterned structure in the landscap e (Rietkerk and van de Koppel, 2008). Among the feedbacks that create these patterns are a redistri bution of water in arid systems; competition for limiting resources in savannas; and ponding of water and convective tr ansport of nutrient by
31 differences in transpiration rates in peatlands (Rietkerk et al., 2004a). Regime shifts in these systems occur not only as local sh ifts between alternative ecosyst em states, but also landscapelevel shifts from self-organized patchy states to homogenous states as conditions change (Rietkerk et al., 2004a); that is, under some environmental conditions, only one state is stable, and landscape patterning is lost. The Florida Ever glades provides an excellent model system for testing predictions of alternativ e stable state theory, as well as increasing our understanding of the mechanisms underlying ecosystem patterning. The central portion of the Ever glades is comprised of the patterned ridge-slough mosaic. The system was historically a flow-through, oligotrophic system which has since been compartmentalized into water conservation areas (WCA) with distinct hyd rologic regimes (SCT, 2003) and, in areas, enriched with phosphorus. Ridges are dominated by sawgrass ( Cladium jamaicense ), and are more productive systems with high peat accretion potential (Loveless, 1959; Craft and Richardson, 1993a; Childers et al., 2003; SCT, 2003). Sloughs are commonly dominated by Nymphea odorata Nymphoides aquatica and Utricularia species, although they may be dominated by various graminoid and rush sp ecies in an assemblage frequently referred to as wet prairies (Loveless, 1959; Jordan et al., 1997; Childers et al., 2003). Sloughs are generally lower in elevation than ridges, with characteri stic vegetation that produc es more labile litter leading to lower peat accret ion potential (Vaithiyanathan and Richardson, 1998; Lewis, 2005; Jorczak, 2006). In areas where historic conditions pe rsist, vegetation patterning is highly regular, and oriented with landscape water flow; the soil elevation differences between communities have led to the landscape being referred to as corrugated (Baldw in and Hawker, 1915; Loveless, 1959; Sklar et al., 2004). Recent loss of this spat ial vegetative patterning in some areas has been identified as a major concern for restorati on activities (SCT, 2003; Ogden, 2005). Wu et al.
32 (2006) have suggested that loss of spatial pattern towards a ho mogenous landscape represents a catastrophic shift sensu Rietkerk (2004a), and have deve loped spatial indices for pattern degradation. This vegetative pa ttern loss has not been correlat ed to soil elevation changes; elevation is inferred from vegetative cover. Recent studies have established that soil elevation in this region (excluding tree islands) is dec oupled from the underlying bedrock (SCT, 2003; Givnish et al., 2007), suggesting th at some other process, presum ably autogenic feedback from the vegetative communities, has led to spatial patterns of peat elevations across the landscape. Further, this implies that the loss of spatial patt erning is driven by altera tions in these autogenic feedbacks. Our conceptual model for patterning in th e Everglades ridge-slough proposes autogenic feedback mechanisms in carbon budgets maintain stable landscape patterning. Feedbacks among depth and duration of inundation, plant commun ity composition, net primary production, and peat accretion and decomposition are hypothesized to create and ma intain multiple peat accretion equilibria (high production, high re spiration ridges; low producti on, low respiration sloughs). To maintain landscape patterning, long-term accretion ra tes in ridges and sloughs must be similar, despite strongly differential primary productivity, presumably because of a dynamic balance between production and respiration rates (Figur e 2-1). Alterations in hydrologic conditions, which are emblematic of human management of th e Everglades, are expected lead to changes in the carbon budget and disequilibrium in peat accretion and ultimately to shifts between ecosystem states. Recent literatur e describes elevation differences between intact ridges and sloughs and the loss of these differences with hydrologic alteration (G ivnish et al., 2007). However, efforts to quantify elevation differences in response to known hydrologic changes are not sufficiently developed to be diagnostic of regime shifts (from ridge-slough to a flattened
33 landscape), nor can we predict the hydrologic co nditions under which transitions between ridge and slough states will occur. Finally, quantitativ e information linking elevation differences to changes in the plant communities is limited, which represents a critical knowledge gap since this relationship is hypothesized to be the foundation of bi-modality maintenance in the ridge-slough mosaic. Here we consider the use of p eat elevation distributions to in dicate autogenic state stability and examine how elevation distribution propert ies change along a gradient of hydrologic impairment. Using a bifurcation point model for alte rnative stable states (adapted from Sheffer et al., 2001, Figure 2-2), we developed a series of pr edictions emerging from the hypothesis that the ridge-slough landscape represents alternative stable states. First, distinct bimodality in peat elevations should be evident in what is c onsidered the most well conserved hydrologic conditions in the ridge-slough mosaic. Second, thes e peat elevation distributions should show significant shifts away from bimodality with hydr ologic alteration, with co rresponding shifts in plant community prevalence and fide lity to soil elevations. Conse quently, self-organizing spatial macro-structure in the landscape is lost with hydrologic altera tion. If true, we predict that statistical metrics of pattern (s patial anisotropy and relative spatial struct ure) in the peat topography evident in conserved areas will be lost with hydrologic modification. While the metric of interest is peat elevation, we use water depth as a local el evation proxy throughout. Methods Study Area For at least the last 2,700 years, m ost of the central Everglades has consisted of the ridgeslough mosaic, with the centers of ridges and sloughs persistent in their present configuration for at least 1,000 years (Bernhardt et al., 2004). Presently, the ridge-slough region is restricted to Water Conservation Area (WCA) 3 and ENP (SCT 2003), with the larges t area of peat-based
34 ridge-slough mosaic located in WC A 3. This change is attributed, at least in part, to hydrologic changes in the 20th century (Light and Dineen, 1997) asso ciated with the compartmentalization of the Everglades. Recent (100 years) hydrol ogic changes has been linked with sawgrass expansion, producing or exploiti ng topographic flattening (Bernhard t et al., 2004). The modern hydroperiod of WCA 3 has been modified for use as a shallow storage reservoir (Walters et al., 1992), and the presence of road co rridors orthogonal to flow has led to hydroperiod shortening in the north, and impoundment in th e south. These landscape modifi cations to regional hydrology in WCA3A comprise a hydrologic gradient, with the central area considered well conserved, and dry and impounded end-members in the north and south (District, 1992). Seven 2x4 km landscape-sampling blocks we re located throughout WCA 3, oriented along flow lines (Figure 2-3). Six of these landsca pe blocks were located in hydrologic partitions that represent the modern hydrologic conditions in WCA3A (drained, conserved, and impounded conditions), as well as a dry, low-flow end member located in WCA 3B. Sampling Sa mpling in each unit was done over a single day between September and December 2007. Since our metric of interest was relative el evation differences between ridges and sloughs, benchmarked water depths are not essential to the analysis. As such, water depths are not comparable among sampling units due to change s in hydrologic conditions over the sampling period; all analyses were performed within samp ling units. Up to 30 randomly located clusters were placed in each sampling unit. Clusters cons isted of measurements of water depth at the center point and points at 5 m and 25 m away from center in the or dinal directions (north, south, east and west). Water depth data act as surroga tes for peat elevation, and are analyzed with respect to distributions of depths. All plant speci es were recorded at each location, and the cover of the dominant species noted.
35 Data Analysis To avoid pseudoreplication resulting from shor t-range spatial autocorrelation (Legendre, 1993), the 5-m data points within each cluster were removed from distribution and vegetation analyses. The full data set was included for all spatial analyses. To test for bimodal distribution of water depth, the Bayes information criterion (BIC), a model comparer, was used to determine whethe r a single normal or a bimodal (two normal) distribution better fits the joint probability density function. The BIC penalizes additional parameters more strongly than does the Akaike information criterion (AIC ), and thus at large sample sizes is a more conservative comp arison of models (Burnham and Anderson, 2004). Water depth frequency in each sampling unit was fit to single normal (1) and mixed normal (2) distributions: Ps = N (i, i) (1) Pm = q N (1, 1) + (1 q ) N (2, 2) (2) where q is a variable representing the probability of falling within the first of the two normal distributions, and N is a normal distribution with mean i and standard deviation i. To understand the separation of the joint probability f unction across the hydrologic gradient, a measurement of the joint probability function kur tosis (increasing density of points in the shoulders of the probability density function), th e fourth moment of the pdf, was modeled as: N(-1) --4 i (xi )4 3 (3) where N is the length of ( x ). Negative kurtosis indicates a high er density of observations in the shoulder of the distribution than expected with a standard norm al; extreme kurtosis indicates a bi-modal density function, and as such serves as a simple diagnostic of joint probability function separation.
36 A priori vegetative classificat ions were used to separate wa ter depth data into separate distributions to test for significant differe nces in water depths between communities. C. jamaicense dominance led to a label of ridge, N. odorata or Utricularia spp. dominance was labeled slough, and areas dominated by graminoids or sedges other than C. jamaincense were labeled wet prairie. All data poi nts not falling into one of these cl assifications were not used for vegetative analyses. The resulting community-speci fic mean water depth values were compared within each sampling unit using a Students t-te st. Linear regression between latitude and water depth within each landscape block was used to determine if the assumption of within-unit hydrologic uniformity is valid. Because ridge s and sloughs each exhibit separate normal distributions, we regressed only the slough water depths (Table 21). The assumption of uniform hydrologic conditions within each landscape block was validated in most units, which exhibit non-significant slopes of slough water depth with latitude. However, the Drained and Conserved 1 landscape blocks both have sign ificant negative slopes, sugges ting hydrologic conditions are not uniform across their 4-km lengt h and, specifically, that water gets deeper at the southern end of the unit. Ridge water depths also exhibit a significant negativ e slope for the Drained landscape block (Table 2-2), but exhibited a significant pos itive slope in the Conserved 1 landscape block, suggesting that the ridge-slough el evation difference in that sampling unit increases dramatically with distance south. To test our conceptual model for scaledependent feedbacks, patterns of spatial autocorrelation were determined within each landscape block (Rietkerk and van de Koppel, 2008) using GS+ software (Gamma Design Software, Plai nwell, MI USA). As the landscape is expected to show declining corre lation as distances increase, a measure of autocorrelation was done first in the near-neighbors, us ing 5m lag classes, and then at larger distances, using 20m lag
37 classes. Autocorrelation was used to find repeatin g patterns over space. A covariance matrix was created, where the covariance for an interval distance (class h ) is cov( h ) xixi h xi xi h Nh (4) where Nh is the total number of paired samples for the lag class h, xi is the measured value of point i xi+h is the measured vale of point i + h, xi is the mean of all xi for lag class h, and xi h is the mean of all xi+h. Autocorrelation is then defined as C ( h ) cov( h )ii h (5) where i is the standard deviation of the measured values. The function C ( h) gives a value in the range [-1,1], where 1 indicates perfect correla tion and -1 indicates pe rfect anti-correlation. Spatial analyses were performed using ArcGIS v. 9.2 (ESRI, Redlands, CA USA). Spatial characteristics of variables are modeled by the semi-variogram function ( h). A semivariogram describes the degree of spatial dependen ce between observations as a function of distance (lag). The semi-variogram then has proper ties of a sill, which is a measure of the lag variance; a range, which is the distance at whic h the semi-variogram reaches the sill; and a nugget effect, which is the sum of microstructure and measurement error. A ratio of the nugget to sill gives a value of Q which is a measure of the spatial dependence of soil properties (Cambardella et al., 1994). Strong spatial struct ure is represented by values greater than 75%. Values 2575% suggest moderate spatial structures, and valu es less than 25% indicate low spatial structures. Spatial anisotropy was determined by ex amining the range of autocorrelation in orthogonal directions. Anisotropy fact or is calculated as the ratio of the major range (direction of
38 maximum autocorrelation) to the minor range (autocorrelation range orthogonal to the major axis). Values of 1 indicate isotropic semivari ance; the expected condition for a conserved ridge slough landscape is strong anisotropy. As our main interest was evidence for highly localized anisotropy, we restrained the analysis to 100m. Results In conserved sites, the water depth density f unction exhibited distinct bimodality (Figure 2-4). This bimodal signature is lost with bot h increasing drainage a nd increasing impoundment. Kurtosis follows a similar trend (Table 2-3), in which drained conditions have positive values, conserved conditions have negati ve values (i.e., density functi ons in these areas are highly kurtotic) and impounded conditions exhibit a shift back towards positive values. When probability density functions are ev aluated on a community basis, significant differences are observed between mean water de pths in ridge and sl ough communities (Figure 25). However, the separation of the means, inferred from the t value, decreases in both directions of hydrologic impairment compared with conser ved conditions, indicatin g greater overlap in water depths between communities (Table 2-4). Moreover, the variance in water depths in both vegetative communities clearly increases with impoundment, and may decrease under drained conditions. The distribution of vegetative community type s varied with position within the hydrologic gradient. A substantially large fraction of sites were ridges in the drained/low flow site (WCA 3B) than at other sites (Figure 2-6). Ridge incidence decreases systematically with impoundment, while sloughs increase in incidence. When slough sites are partitioned into wet prairies (dominated by emergent vegetation) and deep sl oughs (submerged and floating-leaved aquatics), a clear decline in the incidence of wet prai ries is observed from conserved through impounded conditions (Fig. 2-6).
39 Near-distance autocorrelation for the well-conserved sites and the first transition site was highest among all the sites (bet ween 0.85 and 1), with autocorr elation the lowest among the drained sites (between 0.5 and 0.6) (Figure 2-7). With larger ranges and lag distance, a periodicity is observable at all sites, with au tocorrelation varying between strongly positive and negative correlation (Figure 2-7). In all cases, positive and negative correlation reduces with increased distance, however, so the auto correlation becomes increasingly weak. Relative spatial structure (Q) was high (Q>75) for conserve d and impounded sites, but low (<40) for both of the drained sites (Figure 28). Anisotropy, however, was much lower for both the impounded sites (Transition 1, 2, and Impounde d), and high (>1.5) for both conserved and drained sites. Values for the anisotropy fact or ranged from 1.1 in the Impounded block (effectively isotropic) to 2.2 (strongly anisotropic) in the Conserved 2 bock. Discussion The evolution of patterning in the Everglades has been variou sly attributed to sediment transport (Leonard et al., 2006), differential flow volumes (Sklar 2005), a nd the redistribution of limiting nutrients, particularly phosphorus (Ross et al., 2006; Givnish et al., 2007; Rietkerk and van de Koppel, 2008). We propose a different model, in which peat accretion processes, and autogenic feedbacks therein, coupled with hydrology regulate la ndscape patterning. While there have been a number of studies concluding that anthropogenic hydrologic modification has resulted in community shifts in composition an d extent (reviewed in Ogden, 2005), there have been few comprehensive studies on correspondin g changes in the topog raphy in the ridge-slough landscape (although see Givnish et al ., 2007). As our results are high ly suggestive that ridges and sloughs are a synthesis not only of vegetative community, but also of autogenic processes leading to peat elevation differences, from here on ridge and slough will represent these combined ecosystem variables.
40 Alternative stable states in patterned lands capes are predicted to exist within a discrete range of ecosystem drivers (Rietkerk et al., 2004a). Outside of that region of bistability, environmental conditions are such that feedb acks cannot maintain one or the other of the ecosystem types, and the system trends toward s spatial homogeneity. Alternative stable states, then, exist not only as phase shifts from one ecosystem to another, but also as the landscape possibility of self organized patchy vs homogeno us states. Our data support the hypothesis that changes in environmental drivers (i.e., hydrol ogic modification) lead to conditions where landscape bimodality is removed. These results ar e strongly consistent with predictions that emerge from the alterna tive stable state theory. That the system is made of two phases (ridge and slough) with sharp boundaries is evidenced by the bimodality in conserved and transitionary landscape blocks. Patterning comprised of gradual transitions would exhibit unimodal distributions in water depths. The loss of bimodality with hydrologic modification suggests that vegetative patter ning in the drained and impounded sites is a residual signature; the shar p boundaries between ridg es and sloughs have disappeared altogether, and it can be surmised that vegetative differences will ultimately disappear as well. Soil elevation, then, is the keystone variable in th is system. Studies of vegetation change are forced to infer soil eleva tion changes from vegetation, but with key model failures. For example, Wu et al. (2006) infer from aerial photography that northern 3A (near our Drained block) and southern 3A (near our Imp ounded block) exhibits intact (or nearly so) ridge-slough patterning. Our data, however, s how that vegetative changes lag behind soil elevation parameters. Increased water depth variance within each ve getative community was expected to occur with hydrologic modification in both directi ons. Evidence supports th is prediction with
41 impoundment but not with drainage, which was at first contra ry to our expectations. The rationale for the original expectation was that vegetative communities would exhibit hysteretic behavior in response to hydrologic change. Thus we would expect to see species diagnostic of a community (e.g., C. jamaicense ) over a wider range of soil el evation conditions when the landscape is in transition. We attribute the obs ervation of increased va riance behavior with impoundment, but not with drainage, to ecophysio logical attributes of sawgrass. Despite sawgrass physiological stress and decreased growth with increas ing inundation (Newman et al., 1996), which limits its spatial expansion excep t during periods of drought (Brewer, 1996; Pezeshki et al., 1996; Weisner a nd Miao, 2004), sawgrass appears to persist in deep water via rhizome extension upwards in the water column (presumably overcoming redox induced stress). Slough vegetation, in contrast, is easily out competed, and res ponds rapidly to both interannual variation, as well as long-t erm hydrologic change associat ed with both drainage and impoundment (David, 1996; Busch et al., 1998). The differences in vegetative response dynamics to hydrologic change (slow loss of sawgrass with impoundment, rapid loss of slough vegetation with drainage) results in increasing presence of em ergent vegetation under drained conditions, but asymmetrical rate s of change with impoundment. Strong positive autocorrelation was observed at short-range in all landscape blocks, with clear declines with distance. More importantly for understandi ng landscape patterning, negative autocorrelations are observed at moderate distance and periodicity is evident at the block scale for sites with conserved pattern (Figure 2-7). This is consistent with expectations from selforganizing patterned landscapes. As communities e xhibit fidelity to restricted ranges for water depths, autocorrelation would be expected to be high over distan ces coinciding with the spatial wavelength of ridge and slough patterning.
42 Rietkerk and van de Koppel (2008) postulate that short-scale positive feedbacks explain the sharpness of ecosystem boundaries, while la rge-scale negative feedbacks determine the pattern of self-organized patchy systems. They further propose that the density of the organism engineer (in this case sawgra ss) in a landscape determines the strength of scale-dependent feedbacks. This scale-dependency is exhibited by the periodicity and redu ction in autocorrelation of water depths in our study. Further, the cata strophic shift from patterned to homogenous states is evidenced by the concurrent loss of peat bi modality, spatial structure, and the loss of anisotropy at the hydrologi cally modified sites. Hydrologic changes over the last 50 years in WCA 3, considered widely to be the most well conserved ridge-slough mosaic, are driving the system away from the conditions that maintain patterned ridge-slough landscapes. Drai ned sites exhibited dramatic leveling of the landscape, decreased incidence of slough ecosystems (replaced by wet prairies), and decreasing spatial structure, suggesting that the ridge-slough mosaic is cha nging towards a sawgrass prairie. The residual signature of bi-modal vegetative communities under drained conditions is likely an artifact of sawgrass autecology; sawgrass expa nds primarily via vegetative growth (Brewer, 1996; Daoust and Childers, 1999; Lorenzen et al., 2000; Lorenzen et al., 2001) and thus is not an effective colonizer of disturbed areas. With in creasing drainage leadi ng to the succession of slough to wet prairie, and a complete loss of peat height bimodality (observed at the drained endmember block), it seems likely that the wet prairie state is transitional, and will succeed to sawgrass given time required for vegetative prop agation. That is, transition from slough to emergent vegetation suggests that these locations are out of equilibrium; loss of soil elevation differences suggests that future plant commun ity composition will be determined by competitive interactions, with the expectation that sawgrass will eventually dominate. Further, it appears that
43 the responses of landscape stru cturing to drainage and impound ment differ. Spatial structure (diagnostic of landscape self-organiz ation) is lost in the drained areas, suggesting that sawgrass expansion and slough conversion to wet-prairie have flattened the landscape. In contrast, impoundment does not lead to a loss of spatial structure (that is, autocorrelation remains strong), but does appear to lose anisotropy, or direc tionality. The mechanism maintaining anisotropy under drained conditions is unknown, though maintenance of flow velocities despite changes in hydroperiod are likely to be a factor. Among the long-term goals of the Comprehe nsive Everglades Restoration Plan (CERP) is the restoration of the depth, duration, timing and velocity of water across the landscape (SCT, 2003; Perry, 2004). Since much of the fish, insect, and plant diversity occurs in sloughs, there is a clear biological imperative to understand how to maintain the ridge-slough pattern (Davis and Ogden, 1997). Our results show that simple statistical descriptions of wa ter depth distributions, including anisotropy, spa tial structure, and bimodality, can ac t as diagnostic measures of ridgeslough health. Further, these measur es are sensitive and specific to the magnitude and direction of hydrologic modification and may therefore be useful for effec tive ecosystem monitoring. We can also infer from our data that restoring hi storic water flows to WCA 3 will not necessarily quickly restore the original ridge -slough conditions. There appear to be lags in the shifts from intact ridge-slough to a degraded pa ttern; if these lags ex ist in reverse then we can expect a long ecosystem recovery. Where the landscape has lo st patterning altogeth er, recovery can be expected to take even longer, constraining rest oration goals (Zedler, 2000; Suding et al., 2004).
44 Figure 2-1. Conceptual model for autogenic feedbacks maintaining stable alternative ecosystems. Peat accretion is a net ecosystem variable, synthesizing the ecosystem processes of productivity and respiration. At the locali zed level, both ridges and sloughs have a hydrologic condition (water depth, inundation du ration, velocity, etc) at which either a ridge or a slough is the likely community to be found. Changes in the hydrologic condition may alter the peat ac cretion rates, but the aut ogenic feedbacks within the communities will drive the system back to an equilibrium point. However, the system can be driven to whole-scale shifts if hydrologic alterations are of a large enough magnitude within a parameter space that would be the region of transition.
45 Figure 2-2. Bifurcation model for alternative stable states with hypothetical water depth distributions super-imposed. Under natu ral hydrologic conditions, the ridge-slough pattern is the most stable state. Howeve r, a movement away from those hydrologic conditions leads to a uniform, unimodal state of either all ridge (s awgrass prairie) or all slough.
46 Figure 2-3. Map of South Florida with landscape sampling blocks. Insert: Locations of nested, randomly-located sites with in the sampling blocks.
47 Table 2-1. Summary of regressions for slough water depths as a function of latitude. Site designation Intercept Slope R2 Drained/Low flow 5.4e+02 -1.2e-03 0.06 Drained 5.3e+02 -2.0e-03 *** 0.16 Conserved 1 3.6e+02 -1.4-03 *** 0.17 Conserved 2 -7.6e+01 5.6e-04 -0.02 Transition 1 1.7e+02 -5.2e-04 -0.01 Transition 2 -2.6e+021.5e-030.02 Impounded -4.4e+01 3.9e-04 -0.01 Regression summary for Slough water depth where water depth = a + b (latitude) Significance codes: 0 *** 0.001 ** 0.01 0.05
48 Table 2-2. Summary of regressions for selected areas for ridge water depths. Site designation Intercept Slope R2 Drained/Low flow 1.9e+02-7.6e-04*0.01 Drained 5.5e+02 -2.1e-03 *** 0.17 Conserved 1 -5.9e+02 2.6e-03 ** 0.12 Conserved 2 545.7 0.00.08 Transition 1 -3e+021.2e-030.17 Transition 2 797.5 0.0* 0.08 Impounded -2.1e+02 1.1e-03 0.52 Regression summary for Ridge water depths where water depth = a + b (latitude). Significance codes: 0 *** 0.001 ** 0.01 0.05
49 Figure 2-4. Water depths (local proxy for peat elevation) within each landscape block are shown as histograms. The lines indicate the probabi lity density function of the best model indicated by the BIC score. The greater BIC fit (indicated in bold) indicates whether a single normal distribution (Ps) or a mixture of two normal distributions ( Pm) best describes the data. Best-fit parameters are shown, where q is the weight of the first distribution when a mixed di stribution has the better f it. Note: NA=not applicable.
50 Table 2-3. Kurtosis of the water depth distributions for each sampling unit. Site designation Kurtosis Drained/Low flow 0.24 Drained 0.32 Conserved 1 -1.37 Conserved 2 -1.23 Transition 1 -1.34 Transition 2 -1.17 Impounded 0.10
51 0 10 20 30 40 50 60 70 80 Drained/Low Flow DrainedConserved 1Conserved 2Transition 1Transition 2ImpoundedAvg Water Depth (cm) Ridge Slough/Wet Prairie Figure 2-5. Average water depths of each commun ity of the landscape samp ling units. Error bars indicate standard deviation (summary of results are found in Table 2-3).
52 Table 2-4. Results of a-priori defi ned community water depth analysis Site designation Community t-value p-value mean (cm) var (cm2) -3.98 0.004 Ridge 32.73 21.20 Drained/Low flow Slough/Wet Prairie 39.38 20.98 -4.92 3.03E-06 Ridge 20.88 28.02 Drained Wet Prairie 25.32 21.34 -17.33 < 2.2E-16 Ridge 20.94 41.62 Conserved 1 Slough 37.09 14.71 -14.21 < 2.2E-16 Ridge 28.73 46.51 Conserved 2 Slough 48.43 41.00 -17.52 < 2.2E-16 Ridge 35.47 60.55 Transition 1 Slough 61.03 54.89 -6.82 1.41E-09 Ridge 40.65 153.13 Transition 2 Slough 57.88 143.25 -7.01 1.26E-08 Ridge 21.07 118.51 Impounded Slough 36.94 72.43
53 0% 20% 40% 60% 80% 100%Drained/Low Flow DrainedConserved 1Conserved 2Transition 1Transition 2ImpoundedSampling Location Emergent Wet Prairie Deep Water Slough Sawgrass Ridge Figure 2-6. Relative incidence of vegeta tion communities by sampling unit. A priori classifications were designated by commun ity dominance, i.e., sawgrass ridges are dominated by C. jamaicense, sloughs are dominated by either N. odorata or Utricularia spp, and wet prairies are dominated by graminoids or sedges, excluding C. jamaicense
54 Figure 2-7. Correlograms of water depths with lag distances of h=20m, demonstrating that water depths (cm) are significantly positively co rrelated at short distances (0-50m), but either uncorrelated or negatively correlated at larger distances (around 100m for all sites). Autocorrelation periodicity, most evident in the conserved and transitional sites is suggestive of the wavelength of lands cape pattern. Black lines indicate 95% confindence limits of the cross correlation values based on the number of pairs within the initial lag class h. The distances along the x-axis are the average distances among all pairs within each lag class.
55 Figure 2-8. Metrics of spatial pattern for sampling units. Relative structure (Q) indicates the percentage of spatial semi-variance explained by the model (i.e., non-random spatial variability) while the anisotropy factor is the ratio of the ranges in the major and minor directions (1.0 where spatial pattern is isotropic). Summary of analyses is found in Table 2-5.
56 Table 2-5. Results of spatial analyses. Major range Minor range Anisotropy factor Direction Nugget Sill Q Drained/Low flow 49.6 28.11.83509.7 15.1 35.8 Drained 53.0 37.01.435611.0 16.2 32.1 Conserved 1 52.6 28.11.93604.8 24.5 80.4 Conserved 2 125.0 58.02.231512.3 67.9 81.9 Transition 1 52.5 42.91.23316.5 97.0 83.0 Transition 2 89.3 72.41.229516.2 189.2 91.4 Impounded 51.6 44.91.14614.2 89.1 84.1 The major and minor range, nugget, and sill are values interpreted from the semivariogram generated for each landscape block. All analyses were done with in 100m. The anisotropy factor is ratio of the major range to the minor range. Values of 1 indicate isotropic semivariance; the expected condition for a conserved ridge slough landscape is strong anisotropy. Q is then calculated as the partial sill/(partial sill + nugge t) 100, which is a measure of the spatial variation not explained by error. Values greater than 75 indicated a strong spatial structure to the variable; values less than 75 indicate weak to moderate spatial structure
57 CHAPTER 3 SUMMARY AND CONCLUSIONS The descriptive inform ation in Chapters 1 a nd 2 provide evidence of alternative stable states, but lead to the key question of the mechanisms that maintain the sharp boundaries between ridges and sloughs. The use of alternative stable state and ecosystem pattern theories incorporates community dynamics and ecosystem a nd landscape level processes together to help explain patterns observed in the ridge-slough mosaic of the central Everglades. Chapter 2 examined evidence for bimodality in peat elevat ions and clearly demonstrated two distinct ecological states, with tightly bounded transition al zones. These states, ridge and slough, are defined based on two linked factors: water depths and vegetative communities. Moreover, the patterning of this landscape is highly struct ured and anisotropic unde r historic hydrologic regimes. All of these variables are altered with hydrologic modification, although not always in the same direction (i.e., loss of anisotropy with impoundment but a pparently not with drainage). The loss of peat elevation bimodality occurs with both drainage and impoundment, however, suggesting that bimodality is a ke y indicator of landscape stability. Bimodality in keystone ecosystem variables (e.g., phytoplankton abundance in shallow lakes, soil elevation in pattern ed peatlands, tree density in semi-arid grasslands, macrophyte density in desert streams) has been used as evid ence of alternative stable states. The presence of bimodality suggests regime shifts between alterna tive states; intermediate states are unstable, transitory by definition, and therefore not widely observed Examples of variables exhibiting bimodality are clorophyll a concentrations in lakes (Bayle y and Prather, 2003); mating systems in plants (Vogler and Kalisz, 2001); vegetative f eatures of hydrologically influenced calcareous dunes in the Netherlands (Adema et al., 2002); an d a number of variables in marine systems (reviewed in Petraitis and Dudgeon, 2004). While bi modality is in concordance with predictions
58 emerging from alternative stable state theory other explanations supporting predictions of bimodality are still possible and further eviden ce, preferably experimental, is necessary (Schrder et al., 2005). That is, despite strong ev idence of bimodality in the ridge-slough mosaic, we still lack mechanistic understa nding of scale-dependent feedb acks that lead to alternative stable states, which constrains our ability to restore and maintain the landscape. A number of hypothesized mechanisms have been suggested for ridge-slough maintenance (Table 3-1). These alternativ e hypotheses for ridge-slough crea tion and maintenance lead to predictions which future work will empirically evalua te. It is plausible that two or more of these processes are occurring simultaneously. Future work should involve no t only evaluating these mechanisms, but also attempting to determine their relative importance. The first hypothesis and prediction outlines peat accretion equilibrium. The conceptual model as presented (Fig. 1-4) posits that the car bon budget, and the autogenic feedbacks therein, are the key unknowns in this system. Specifically a theoretical point model for peat accretion suggests that there is a landscap e-scale equilibrium accretion stat e. This mean accretion rate across the landscape can arise along two separate self-organizing pathways: one high production and high respiration (ridge), and the other low production and lo w respiration (slough). These two pathways, then, are the multip le state equilbria of intere st. Under historic hydrologic conditions, locations with elevati ons different from an equilibrium level are driven towards one of those equilibria by changes in peat accretion rates. For example, shallow water conditions will generally increase produc tivity on ridges in response to increased oxygen availability. However, oxygen availability also increases respiration rates, canceling out or surpassing the incremental increases in productivity, leading to peat oxidati on until the soil elevation is at the equilibrium. Similarly, sloughs at lower elevation than the equilibrium slough elevation will tend to
59 accumulate peat more rapidly due to reduced exposure probability. Peat accretion will raise the local elevation until it is again at the equilibrium level. The second hypothesis, related to peat accre tion equilibrium, invokes a scale-dependent negative feedback. To regulate the point-based equilibria at a landscape level, the hydraulic requirements of moving necessary flow volumes off the landscape will act as the negative feedback. Specifically, as peat elevation rises locally, water de pths in the surrounding landscape increase to accommodate a fixed flow. The simp licity of this process is attractive as the mechanism is sufficient to explain the orient ation and prevalence of ridge and slough system, and requires no invocation of sediment tr ansport in the face of low velocities. Since peat accretion rates are an emergent property of the lo cal scale carbon budget, hypotheses about the mechanisms that create and maintain alternative stable states involve interactions of productivity, respiration (deco mposition), community composition shifts (and associated changes in carbon i nputs) and hydrology. Demonstrat ing peat accretion equilibria would provide mechanistic evidence for the existen ce of alternative stable states in the ridgeslough region, and aid in understanding the hydrol ogic requirements for landscape maintenance. The hydrologic gradient present in WCA 3A allows for an excellent natural experiment for these hypotheses. That is, directionality at the la ndscape-scale can be inferred from carbon budgets spanning the drained areas to the impounded areas. Further, modeling the negative feedback would give us the hydrologic regime within whic h patterning is stable. This information would be vital to management and restoration schemes, wh ere the goal is to maintain the historic ridgeslough patterning. Other mechanisms that have been postulated to lead to the maintenance of ridges and sloughs revolve primarily on water flow and the move ment of particles. Th e literature primarily
60 focuses on erosive and depositional mechanisms (see review of hypotheses in Larsen et al., 2007). While a great deal of work has been done on modeling potential cont ributions of either scour or depositional processes (Leonard et al., 2006; Ross et al., 2006; Larsen et al., 2007), field observations of both processes are lacking. The plausibility of m ovement of any material other than that which is neutrally buoyant is very low; observed flow rates are generally less than 2 cm s-1 (Leonard et al., 2006), far below entrainment ve locities for even soil floc particles. This makes the third hypothesis (Table 3-1) difficult to find credible. However, legacy signatures of processes that accelerate or reduce accretion rates (i.e., via de position of exogenous material or nutrient subsidies, the final two hypotheses) shou ld be observed in the soils. Sediment deposition of calcitic material derived from sloughs and de posited at ridge edges would result in higher concentrations of Ca in ridge soils. Similarly, deposition of nutrient-r ich organic particles on ridge edges would lead to localized enrichme nt, allowing increased sawgrass production despite anoxic stress at ridge edges. Nutrient stoichiometr y in peat soils at the ed ges and centers of ridge and slough ecosystems will be indicative of this mechanism. It is evident that our understanding of the pro cesses that creates and maintains patterning in the central Everglades is incomplete. Regardless of the mechanisms that develop and maintain this region, consideration of the scale of their action and interac tion is an important unknown. As Rietkerk and van de Koppel (2008) propose--and this study demonstrates--there are multiple scales at which processes act, and their relati ve importance differs accordingly. Failure to incorporate scale into studies of patterned landscapes could lead to misinterpretation of results, and ultimately to poor management or restoration plans (Zedler, 2000). The variables measured in this study respond to hydrologic modification, and are sensitive to the direction and magnitude of that modifica tion. Simple statistical descriptions of the
61 distributions of elevation, particularly anis otropy and bimodality, therefore provide diagnostic measures of ridge-slough health. Effective (i.e ., sensitive and specific) diagnosis of ecosystem change is increasingly important, as managers of CERP have focused on the Field of Dreams method of restoration (Hilderbrand et al., 2005). That is, there is a tacit expectation that restoring historic water regimes will restore the ridge -slough mosaic (SCT, 2003; Perry, 2004). Given evidence for alternative stables states and the time domain of the keystone variable that responds to bi-modal ecological conditions (i.e., peat accr etion), there is a high likelihood that ecological restoration will lag hydrologic restoration by a lo ng time, particularly where landscape flattening has occurred. Providing realistic expectations for the restoration timeline will require a welldeveloped understanding of th e processes that create and ma intain the desired ecological condition.
62 Table 3-1. Hypotheses of mechanisms governing ridge-slough maintenance. Hypothesis Source Predictions Feedbacks of community carbon budgets lead to multiple system equilibria. This study Peat accretion rates are, on average, the same between ridges and sloughs. For areas not falling within the peat accretion equilibrium, there is directionality to the carbon budget, forcing the system back to the equilibrium. Locally positive feedbacks (productivity and respiration) and landscape negative feedbacks (hydrology) leads to regular pattern formation (This study, Rietkerk and van de Koppel, 2008) Associated with autogenic hypothesis; a parameter space exists in hydrology (water depth, duration, pattern) wherein a patterned landscape is more stable than a homogenous landscape. Sloughs are formed via erosive processes (scour). (SCT, 2003) Flow velocities in sloughs are lower than ridges, which lead to erosion and lowered peat elevation. Sediment transport in the form of neutrally buoyant, inorganic material preferentially settles on ridges. (Leonard et al., 2006; Larsen et al., 2007) Soil calcium concentrations are higher at ridge edges than either ridge or slough centers. Sediment transport in the form of neutrally buoyant, nutrient-rich material enriches ridge edges preferentially to ridge centers, leading to increased sawgrass productivity. (Larsen et al., 2007) Nutrient stoichiometry at ridge edges is an intermediate between ridges and sloughs (e.g., lower C:N on ridges, higher C:N in sloughs).
63 LIST OF REFERENCES Ade ma, E.B., Grootjans, A.P., Petersen, J., Grijpst ra, J., 2002. Alternative stable states in a wet calcareous dune slack in The Netherlands. Journal of Vegetation Science 13, 107-114. Bachoon, D., Jones, R.D., 1992. Potential rates of methanogenesis in sawgrass marshes with peat and marl soils in the Everglades. So il Biology and Biochemistry 24, 21-27. Baldwin, M., Hawker, H.W., 1915. Soil survey of the Fort Lauderdale Area, Florida. U.S. Department of Agriculture, pp. 751-798. Bayley, S.E., Prather, C.M., 2003. Do wetland lakes exhibit alternative st able states? Submersed aquatic vegetation and chlorophyll in western boreal shallow lakes. Limnology and Oceanography 48, 2335-2345. Beisner, B.E., Haydon, D.T., Cuddington, K., 2003. Alternative stable states in ecology. Front. Ecol. Environ. 1, 376-382. Bernhardt, C.E., Willard, D.A., Marot, M., Holmes, C.W., 2004. Anthropogenic and natural variation in ridge and slough pollen assemb lages. U.S. Geological Survey, p. 47p. Brewer, J.S., 1996. Site differences in the cl one structure of an emergent sedge, Cladium jamaicense. Aquatic Botany 55, 79-91. Browder, J.A., Cottrell, D., Brown, M., Newman, M., Edwards, R., Yuska, J., Browder, M., Krakoski, J., 1982. Biomass and primary pr oduction of microphytes and macrophytes in periphyton habitats of the southe rn Everglades. In: Center, S.F.R. (Ed.). Everglades National Park, p. 49. Burnham, K.P., Anderson, D.R., 2004. Multimode l inference: Understanding AIC and BIC in model selection. Sociological Me thods and Research 33, 261-304. Busch, D.E., Loftus, W.F., Oron L. Bass, J., 1998. Long-term hydrologic effects on marsh plant community structure in the southe rn Everglades. Wetlands 18, 230-241. Cambardella, C.A., Moorman, T.B., Parkin, T.B., Karlen, D.L., Novak, J.M., Turco, R.F., Konopka, A.E., 1994. Field-scale variability of soil prop erties in central Iowa soils. Soil Sci. Soc. Am. J. 58, 1501-1511. Chauhan, A., Ogram, A., Reddy, K.R., 2004. S yntrophic-methanogenic associations along a nutrient gradient in the Florida Everglades. Applied and Environmental Microbiology 70, 34753483. Childers, D.L., Doren, R.F., Jones, R., Noe, G.B., Rugge, M., Scinto, L.J., 2003. Decadal change in vegetation and soil phosphorus pattern acro ss the Everglades landscape. Journal of Environmental Quality 32, 344-362.
64 Childers, D.L., Iwaniec, D., Rondeau, D.N., Rubio, G., Verdon, E., Madden, C.J., 2006. Responses of sawgrass and spikerush to variation in hydrologic drivers an d salinity in Southern Everglades marshes. Hy drobiologia 569, 273-292. Craft, C.B., Richardson, C.J., 1993a. Peat accretion and N, P, and organic C accumulation in nutrient-enriched and unenriched Everglades peatlands. Ecological Applications 3, 446-458. Craft, C.B., Richardson, C.J., 1993b. Peat accretion and phosphorus accumulation along a eutrophication gradient in the northern Everglades. Biogeochemistry 22, 133-156. Daoust, R.J., Childers, D.L., 1998. Quantif ying aboveground biomass and estimating net aboveground primary production for wetland macr ophytes using a non-destructive phenometric technique. Journal of Aquatic Botany 62, 115-133. Daoust, R.J., Childers, D.L., 1999. Controls on emergent macrophyte composition, abundance, and productivity in freshwater Everglades wetland communities. Wetlands 19, 262-275. David, P.G., 1996. Changes in plant communities relative to hydrologic cond itions in the Florida Everglades. Wetlands 16, 15-23. Davis, S.M., 1989. Sawgrass and cattail producti on in relation to nutrient supply in the Everglades. In: Sharitz, R.R., Gibbons, J.W. (Eds .), Freshwater Wetlands and Wildlife. US DOE, Office of Scientific and T echnical Information, Oak Ridge, TN, USA, pp. 325-341. Davis, S.M., Ogden, J.C., 1997. Toward ecosystem restoration. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: The Ecosystem and Its Rest oration. St. Lucie Press, Boca Raton, pp. 769796. Debusk, W.F., Reddy, K.R., 2003. Nutrient and hydrology effects on so il respiration in a northern Everglades marsh. Journal of Environmental Quality 32, 702-710. Didham, R.K., Watts, C.H., Norton, D.A., 2005. Are systems with strong underlying abiotic regimes more likely to exhibit alterna tive stable states? Oikos 110, 409-416. District, S.F.W.M., 1992. Surface water improvement and management plan for the Everglades: Supporting information document. SFWMD, West Palm Beach, FL. Drake, H.L., Aumen, N.G., Kuhner, C., Wagne r, C., Griehammer, A., Schmittroth, M., 1996. Anaerobic microflora of Everglades sediments: effects of nutrients on population profiles and activities. Applied and Environmental Microbiology 62, 486-493. Duever, M.J., Meeder, F.F., Meeder, L.C., McColl on, J.M., 1997. The climate of South Florida and its role in shaping the Everglades ecosy stem. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: The Ecosystem and Its Restorati on. St. Lucie Press, Boca Raton, pp. 225-248. Ewe, S.M.L., Gaiser, E.E., Childers, D.L., Iw aniec, D., Rivera-Monroy, V.H., Twilley, R.R., 2006. Spatial and temporal patterns of above ground net primary productivity (ANPP) along two freshwater-estuarine transect s in the Florida Coastal Ever glades. Hydrobiologia 569, 459-474.
65 Givnish, T.J., Volin, J.C., Owen, D., Volin, V. C., Muss, J.D., Glaser, P.H., 2007. Vegetation differentiation in the patterned landscape of the central Evergl ades: importance of local and landscape drivers. Global Ec ology and Biogeography, 1-19. Gleason, P.J., Spackman Jr., W., 1974. Calcareo us periphyton and water chemistry in the Everglades. In: Gleason, P.J. (Ed.), Environments of South Florida: Past and Present. Miami Geological Society, Coral Gables, pp. 146-181. Gleason, P.J., Stone, P., 1997. Age, origin, and la ndscape evolution of the Everglades peatland. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades : The Ecosystem and Its Restoration. St. Lucie Press, Boca Raton, pp. 149-197. Hilderbrand, R.H., Watts, A.C., Randle, A.M., 2005. The myths of restoration ecology. Ecol. Soc. 10, 19 (online). Holling, C.S., 1973. Resilience and stability of ecological systems. A nnual Review of Ecology and Systematics 4, 1-23. Iwaniec, D.M., Childers, D.L., Rondeau, D.N., Madden, C.J., 2006. Effects of hydrologic and water quality drivers on periphyton dynamics in the southern Everglades. Hydrobiologia 569, 223-235. Jorczak, E., 2006. Influence of hydrology on Evergl ades ridge and slough soil topography. Soil and Water Science. University of Florida, Gainesville, p. 63. Jordan, F., Jelks, H.L., Kitchens, W.M., 1997. Habitat structure and plant community composition in a northern Everglades wetland landscape. Wetlands 17, 275-283. Kushlan, J.A., 1990. Freshwater Marshes. In: My ers, R.L., Ewel, J.J. (Eds.), Ecosystems of Florida. University of Central Florida Press, Orlando, pp. 324-362. Larsen, L.G., Harvey, J.W., Crimaldi, J.P., 2007. A delicate balance: eco hydrological feedbacks governing landscape morphology in a lotic p eatland. Ecological Monographs 77, 591-614. Legendre, P., 1993. Spatial Autocorrelation: Trouble or New Paradi gm? Ecology 74, 1659-1673. Leonard, L., Croft, A., Childers, D.L., Mitchell-B ruker, S., Solo-Gabriele, H., Ross, M.S., 2006. Characteristics of surface-water flows in the ridge and slough landscape of Everglades national Park: implications for particulate transport. Hydrobiologia 569, 5-22. Lewis, C.G., 2005. Linkages among vegetative substrate quality, biomass production, and decomposition in maintaining Everglades ridg e and slough vegetative communities. Soil and Water Science. Florida, Gainesville, p. 58. Light, S.S., Dineen, J.W., 1997. Water control in the Everglades: a historical perspective. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: The Ecosystem and Its Restoration. St. Lucie Press, Boca Raton, pp. 47-84.
66 Lorenzen, B., Brix, H., McKee, K.L., Mendelssohn, I.A., Miao, S., 2000. Seed germination of two Everglades species, Cladium jamaicense and Typha domingensis Aquatic Botany 66, 169180. Lorenzen, B., Brix, H., Mendelssohn, I.A., McKee, K.L., Miao, S., 2001. Growth, biomass allocation and nutrient use efficiency in Cladium jamaicense and Typha domingensis as affected by phosphorus and oxygen availability. Aquatic Botany 70, 117-133. Loveless, C.M., 1959. A study of the vegetation in the Florida Everglades. Ecology 40, 9. Ludwig, J.A., Tongway, D.J., Marsden, S.G., 1999. Stripes, stands, or stipples: modelling the influence of three landscae banding patterns on re source capture and productivity in semi-arid woodlands, Australia. Catena 37, 257-273. McCormick, P.V., Shuford III, R.B.E., Backus J.G., Kennedy, W.C., 1998. Spatial and seasonal patterns of periphyton biomass and productivity in the northern Everglades, Florida USA. Hydrobiologia 362, 185-208. McCormick, P.V., Stevenson, R.J., 1998. Periphy ton as a tool for ecological assessment and management in the Florida Everglades. Journal of Phycology 34, 726-733. Miao, S., Kong, L., Lorenzen, B., Johnson, R.R., 1998. Versatile modes of propagation in Cladium jamaicense in the Florida Everglades. Annals of Botany 82, 285-290. Newman, S., Grace, J.B., Koebel, J.W., 1996. E ffects of nutrients and hydroperiod on Typha, Cladium, and Eleocharis: Implications for Evergl ades restoration. Ecological Applications 6, 774-783. Odum, E.P., 1971. Fundamentals of Ecology. W. B. Saunders Company, Phladelphias, PA. Ogden, J.C., 2005. Everglades ridge and slough conceptual ecological model. Wetlands 25, 810820. Perry, W., 2004. Elements of South Florida's Comprehensive Everglad es Restoration Plan. Ecotoxicology 13, 185-193. Petraitis, P.S., Dudgeon, S.R., 2004. Detection of a lternative stable states in marine communities. Journal of Experimental Mari ne Biology and Ecology 300, 343-371. Pezeshki, S.R., DeLaune, R.D., Kludze, H.K., Choi, H.S., 1996. Photosynthetic and growth responses of cattail ( Typha domingensis ) and sawgrass (Cladium jamaicense) to soil redox conditions. Aquatic Botany 54, 25-35. Porter Jr., C.L., 1967. Composition and productiv ity of a subtropical prairie. Ecology 48, 937942. Rietkerk, M., Dekker, S.C., de Ruiter, P.C., va n de Koppel, J., 2004a. Self-organized patchiness and catastrophic shifts in ecosystems. Science 305, 1926-1929.
67 Rietkerk, M., Dekker, S.C., Wassen, M.J., Ve rkroost, A.W.M., Bierkens, M.F.P., 2004b. A putative mechanism for bog patterning. The American Naturalist 163, 699-708. Rietkerk, M., van de Koppel, J., 2008. Regular pattern formation in real ecosystems. Trends in Ecology and Evolution 23, 169-175. Ross, M.S., Mitchell-Bruker, S., Sah, J.P., Stotho ff, S., Ruiz, P.L., Reed, D.L., Jayachandran, K., Coultas, C.L., 2006. Interaction of hydrology a nd nutrient limitation in the Ridge and Slough landscape of the southern Everglades. Hydrobiologia 569, 23. Scheffer, M., Carpenter, S., 2003. Catastrophic re gime shifts in ecosystems: linking theory to observation. Trends in Ecol ogy and Evolution 18, 648-656. Scheffer, M., Carpenter, S., Foley, J.A., Folk e, C., Walker, B., 2001. Catastrophic shifts in ecosystems. Nature 413, 591-596. Schrder, A., Persson, L., De Roos, A.e.M., 2005. Direct experimental evidence for alternative stable states: a review. Oikos 110, 3-19. SCT, 2003. The role of flow in the Everglades ridge and slough landscape. Science Coordinating Team, South Florida Ecosystem Restorat ion Working Group, Miami, FL, USA. Sklar, F.H., Coronado-Molina, C.A., Gras, A., Ru tchey, K., Gawlik, D.E., Crozier, G., Bauman, L., Hagerthy, S., Shuford, R., Leeds, J.A., Wu, Y., Madden, C.J., Garrett, P.B., Nungesser, M., Korvela, M., McVoy, C., 2004. Ecological e ffects of hydrology. Everglades Consolidated Report. South Florida Water Management Di strict, West Palm Beach, FL, pp. 6-58. Snyder, G.H., Davidson, J.M., 1997. Everglades agri culture: past, present, and future. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: The Ecosys tem and Its Restoration. St. Lucie Press, Boca Raton, pp. 85-116. Steward, K.K., Ornes, W.H., 1975. The auteco logy of sawgrass in the Florida Everglades. Ecology 56, 162-171. Suding, K.N., Gross, K.L., Houseman, G.R., 2004. Alternative states and positive feedbacks in restoration ecology. Trends in Ecology and Evolution 19, 46-53. Vaithiyanathan, P., Richardson, C.J., 1998. Bi ogeochemical characteristics of Everglades sloughs. Journal of Environmental Quality 27, 1439-1450. Vogler, D.W., Kalisz, S., 2001. Sex among the flower s: The distribution of plant mating systems. Evolution 55, 202-204. Walters, C., Gunderson, L., Holling, C.S., 1992. E xperimental policies for water management in the Everglades. Ecological Applications 2, 189-202.
68 Weisner, S.E.B., Miao, S., 2004. Use of morphological variability in Cladium jamaicense and Typha domingensis to understand vegetation changes in an Everglades marsh. Aquatic Botany 78, 319-335. Whiting, G.J., Chanton, J.P., Barlett, D.S., Happell, J.D., 1991. Relationships between CH4 emission, biomass, and CO2 exchange in a su btropical grassland. Journal of Geophysical Research 96, 13067-13071. Wood, F., Maynard, N.G., 1974. Ecology of the mi cro-algae of the Flor ida Everglades. In: Gleason, P.J. (Ed.), Environments of South Florid a: Pest and Present. Mi ami Geological Society, Coral Gables, pp. 123-145. Wu, Y., Wang, N., Rutchey, K., 2006. An analys is of spatial complexity of ridge and slough patterns in the Everglades ecosystem Ecological Complexity 3, 183-192. Zedler, J.B., 2000. Progress in wetland restoratio n ecology. Trends in Ecology and Evolution 15, 402-407.
69 BIOGRAPHICAL SKETCH Danielle Watts was born in 1979, in Oklahom a City, Oklahoma, the only child of an only child. Danielle was raised in the Florida Keys by her mother and grandmother, and in this environment lay the genesis of her later interests. After graduating from Key West High School in 1997, Danielle continued her education, graduating with an Associates in Arts from th e Florida Keys Community College in 1999. After effectively throwing a dart at a map of Florid a, Danielle moved on to gaining a degree in Wildlife Ecology and Conservation at the University of Florida in Gainesville, FL. It was during this time that she realized that her real intere sts lay in ecosystem development, and performed an independent study for Dr. Tim Martin on the ecophysiology of two pineland species, Ilex glabra and Serenoa repens A years experience working for Drs. Ted Schuur and Michelle Mack after her graduation helped further refine her interests in plant-soil interactions. It was at this time, however, that Danielle chose to realize a life -long dream of living in Africa. Early in 2003, Danielle fell in love a nd married Adam Watts. Together, the two joined Peace Corps, and spent 2004-2006 working as Agro forestry Extension Agents in the Fouta Djallon of the Republic of Guinea. It was at the to p of a plateau, during a particularly brutal subSaharan day that Danielle realized that her true environmental pa ssion laid in low-lying, tropical wetlands. Danielle was then accepted into the Interdisciplinary Ecology program, returning to the University of Florida. Chasing after opportuni ties to work in the Florida Everglades has culminated Danielles ecological, intellectual, and emotional passions. She continues to explore her interests concerning proce sses influencing ecosystem development, expanding on her masters research and continuing on for her PhD.