Wetlands (2013) 33:1139–1149 DOI 10.1007/s13157-013-0469-y
ARTICLE
Reconstructing Vegetation Response to Altered Hydrology and its Use for Restoration, Arthur R. Marshall Loxahatchee National Wildlife Refuge, Florida Christopher E. Bernhardt & Laura A. Brandt & Bryan Landacre & Marci E. Marot & Debra A. Willard Received: 30 January 2013 / Accepted: 30 July 2013 / Published online: 15 August 2013 # US Government 2013
Abstract We present reconstructed hydrologic and vegetation trends of the last three centuries across the Arthur R. Marshall Loxahatchee National Wildlife Refuge, Florida in order to understand the effects of 20th century water management. We analyzed pollen assemblages from cores at marsh sites along three transects to document vegetation and infer hydroperiod and water depth both before and after human alteration of Everglades hydrology. In the northern and central part of the Refuge, late Holocene water levels were higher and hydroperiods longer than the last 100 years. Post-1950 was a time of several different water management strategies. Pollen assemblages indicate drier conditions post-1950 in the northern and central parts of the Refuge, whereas sites in the southern Refuge are wetter and vegetation turnover is higher. Throughout the Refuge, Sagittaria pollen declines with the onset of water management, and may indicate a loss of greater variation in hydroperiods across years and water depths between seasons. Paleoecological evidence provides clear estimates of the vegetation response to hydrologic change under specific hydrologic regimes. Keywords Everglades . Hydrologic change . Pollen . Restoration Electronic supplementary material The online version of this article (doi:10.1007/s13157-013-0469-y) contains supplementary material, which is available to authorized users. C. E. Bernhardt (*) : B. Landacre : D. A. Willard U.S. Geological Survey, 926A National Center, Reston, VA 20192, USA e-mail:
[email protected] L. A. Brandt U.S. Fish and Wildlife Service, 3205 College Ave., Davie, FL 33314, USA M. E. Marot U.S. Geological Survey, 600 4th Street South, St. Petersburg, FL 33701, USA
Introduction Transformation of wetland ecosystems for agricultural and urban uses over the last four centuries has reduced United States wetland acreage by more than half, from ~89.4 million hectares (ha) in the early 17th century to 42.7 million ha in 1997 (Dahl and Allord 1996; Dahl 2000, 2006). Although substantial tracts of wetlands were drained for agriculture and navigation in the 19th century, engineering advances early in the 20th century resulted in some of the most substantial drainage and “reclamation” of wetlands nationwide. Prior to drainage efforts in the late 19th century, it is estimated that wetlands covered 8.2 million ha in the state of Florida (Dahl 2005), of which the Everglades comprised 1.2 million ha (Davis et al. 1994). By AD 2005, Florida wetlands covered only 4.6 million ha, and the Everglades were reduced to approximately 0.6 million ha (Dahl 2005; Lodge 2010). Since the 1970s, however, the rate of wetland loss has been reduced by 81 %. In addition, through a range of restoration measures, the condition of wetlands that have been altered has been improved. State and federal agencies are cooperating in design and implementation of the Comprehensive Everglades Restoration Plan (CERP) to restore and preserve the water resources of central and southern Florida including the Everglades (U.S. Army Corps of Engineers and South Florida Water Management District 1999). A primary part of CERP is to restore timing, distribution and volume of flow to wetland ecosystems that more closely match their pre-drainage conditions. The CERP incorporates a science-based adaptive assessment approach that includes development of restoration targets and subsequent documentation of wetland response to hydrologic and environmental changes on a range of time scales. Typically, hydrologic targets for the wetlands have been based on results from the Natural Systems Model (NSM), a model developed by the South Florida Water Management District to describe the hydrologic responses of the ecosystem lacking canals and other water management features (Fennema
1140
et al. 1994). Flow in the NSM is driven by rainfall and other hydrologic conditions measured between AD 1965 and AD 2000. This relatively short time frame does not capture multidecadal scale variability associated with climate phenomena such as the Atlantic Multidecadal Oscillation (AMO) and North Atlantic Oscillation (NAO) and therefore may not adequately portray natural hydrologic patterns in the unaltered system. We use paleoecological proxies to extend our understanding of hydrologic and vegetation conditions backward from the modern managed system to the centuries preceding hydrologic alteration of the system. Previous research (Cronin et al. 2002; Donders et al. 2005) highlighted the role played by multidecadal droughts in structuring key habitats within the greater Everglades ecosystem and the response of aquatic habitats to El Niño Southern Oscillation (ENSO) cycles. These multidecadal scale studies provide data on baseline levels of ecosystem and climate variability for comparison with recent and future changes caused by anthropogenic factors such as modification of land cover and water management. These and other studies (Willard et al. 2006b; Bernhardt and Willard 2009) indicate that changes in the last half of the 20th century greatly exceed those of the last few millennia. In addition they also indicate substantial spatial heterogeneity of responses; some areas are much wetter than they were historically and some are much drier. In this study we apply paleoecological techniques to reconstruct hydrologic and vegetational trends of the last three centuries across the Arthur R. Marshall Loxahatchee National Wildlife Refuge (hereafter the Refuge). We focus on the last 300 years to provide a sufficient temporal scale to identify multidecadal fluctuations in pollen assemblages before anthropogenic modification of the system. We present palynological results distributed throughout the Refuge to infer vegetation, hydroperiod, and water depth at discrete time intervals. Results from these sites document the spatial and temporal patterns of vegetation and hydrologic change within the Refuge both pre-drainage (prior to AD 1900) and post-drainage (after AD 1900). We compare these patterns to known watermanagement actions and discuss our findings in the context of long-term restoration and management goals for the Refuge.
Study Area Initial attempts to drain and reclaim the Everglades began in the AD 1880 s, but the first truly successful water diversion occurred in the first decades of the 20th century (Light and Dineen 1994; Lodge 2010). By AD 1930, the Hoover Dike around Lake Okeechobee was completed, four canals diverted water from the wetland to the Atlantic Ocean, and the Tamiami Trail was completed as a transportation link between Miami and Naples. The canals collectively drained >600,000 ha
Wetlands (2013) 33:1139–1149
annually, resulting in exposure of previously submerged peats and their subsidence due to physical compaction, burning, and oxidation (Sklar and van der Valk 2002). Further drainage and compartmentalization of the Everglades occurred after severe flooding in AD 1947–48. The “Central and Southern Florida Project for Flood Control and other Purposes” (or C&SF Project) was signed into law by Congress in AD 1948 to centrally regulate all phases of water distribution. Using an extensive system of levees, canals, and pump stations, this project included construction of three Water Conservation Areas (WCAs) to ensure adequate regional water supply during droughts and store excess water during wet years (Light and Dineen 1994). The Refuge encompasses the northernmost of three WCAs in the greater Everglades ecosystem (Fig. 1). In AD 1951, WCA-1 came under the control of the U.S. Fish and Wildlife Service, as such becoming the Refuge, in order to protect and manage the unique northern Everglades habitat (U.S. Fish and Wildlife Service 2000). Canal and levee construction began in the early AD 1950s; by AD 1961 the 57,212 ha WCA-1 was completely impounded, and the major pump station (S5A) and gates (S10s) were in place. Since July AD 1960, four different water regulation schedules have been in effect. The water regulation schedules are a rule curve that provides guidance on how to manage the hydrology in the Refuge at different times of the year and different water levels. The current schedule has been used since AD 1995. The three previous water regulation schedules operated from AD 1975–1995, AD 1969–1975, and AD 1960–1969. Water regulation schedules were changed when managers believed that the existing schedule was creating conditions that were either too dry (AD 1960–1969 and AD 1975–1995 schedules) or too wet (AD 1969–1975 schedule) (Brandt 2006). Before impoundment of the Refuge, water depth and flow patterns were driven by seasonal precipitation and overland sheet flow from Lake Okeechobee and the Kissimmee River watershed. Conditions in the Refuge fluctuated from dry (November–May) to wet (June–October), creating a dynamic flow through system with a mosaic of habitats including slough, wet prairie, sawgrass stands, and thousands of tree islands (Jordan et al. 1997). Sloughs, dominated by Nymphaea (waterlily), Nymphoides (floating heart), and Nuphar (spatterdock) (Gunderson 1994), occur in relatively deep water, produce little plant biomass, and are inundated for >10 months of the year. Sawgrass strands, dominated by tall stands of Cladium (sawgrass), occur on higher elevations (0.1–0.2 m higher than sloughs) in relatively shallower water, produce abundant plant biomass, and have shorter hydroperiods (6–9 months). Wet prairies, dominated by short sedges (i.e., Rhyncospora, Eleocharis) and short stands of Cladium, are intermediate to sloughs and sawgrass marshes in hydroperiod and biomass (Jordan et al. 1997) and occur on sites 0.01–0.2 m higher than sloughs (Brandt 2006). Tree
Wetlands (2013) 33:1139–1149 81°
a
80°
27 °
LAKE
b
OKEECHOBEE W es
th
Beach Cana l
WCA 1 Loxahatchee
R
ive
l na Ca
N ew
tP alm
ro bo lls Hi
Nor al i Can Miam
r
Ca
WCA 2A
l na
Fig. 1 a. Location of Arthur R. Marshall Loxahatchee National Wildlife Refuge Water Conservations Area 1 (WCA1), Everglades National Park, and primary canals. b. Location of cores collected in the Refuge in relation to ground surface elevation (in cm). Elevation data are from EDEN DEM website (http://sofia.usgs.gov/metadata/ sflwww/eden_em_oc11.html)
1141
Alligator Alley WCA 3A
2B
South New River Canal 26 °
Tamiami Trail
BISCAYNE BAY Shark River Slough
Taylor Slough
FLORIDA BAY
islands are found on the highest ground and are scattered throughout the Refuge, providing relatively dry habitats for many plant and animal species (Brandt et al. 2003). Tree islands in the Refuge are primarily either “pop-up” bayhead islands or strand islands. Pop-up islands are those formed from peat that becomes buoyant and breaks free, typically from a slough bottom, and floats becoming lodged on top of a sawgrass ridge (Brandt et al. 2000; Lodge 2010). Strand islands form on peat ridges between sloughs and are lower in elevation than pop-ups (Loveless 1959). Impoundment of the Refuge in the mid-20th century disrupted the previously dynamic flow patterns by eliminating connectivity of sheet flow. As a result, there now is a strong north to south gradient in water depth and hydroperiod, and areas in the southern Refuge are wetter than areas in the north.
Methodology Sediment cores were collected in AD 2004 and AD 2005 from marshes in three east–west transects across the Refuge to capture the last few centuries of sediment accumulation (see Fig. 1b). We collected sediment cores using a piston corer with a 10 cm diameter core barrel. Although peats in the Refuge are up to 4.5 m thick (Richardson et al. 1990), we collected only the upper 1–1.5 m of peat because the study focused on the last few centuries, which typically are contained in the upper meter of Loxahatchee peats. After core collection, we extruded sediment from the core barrel and sampled it at 1 cm increments for the upper 20 cm and at 2 cm increments at greater depths. We described sediment lithology as samples
ATLANTIC OCEAN 20 40 0 kilometers 25°
were extruded. We dried samples in a 50˚ C oven and subsampled sediments at the base of each core and at 20 cm increments above the base for radiocarbon dating. Bulk peats were dated using conventional radiocarbon dating because plant macrofossils (i.e. seeds) or identifiable material was not found. Peat in the Refuge is composed of both emergent and submergent plant material, as such radiocarbon dating of bulk peat could incorporate older carbon and yield older dates. However, bulk peat radiocarbon dates from both the Refuge and the greater Everglades have yielded robust age-depth models for paleoenvironmental reconstructions in other studies (Willard et al. 2006b; Bernhardt and Willard 2009). Age-depth models for the last century of deposition are based on 210Pb (lead-210), first occurrences of pollen of the exotic plant Casuarina (Australian pine), which was introduced to south Florida in the late 19th century (Langeland 1990) and appears in sediments by AD 1910 (Wingard et al. 2003) and bulk radiocarbon dates (Table 1). Lead-210 accumulation rates were calculated by fitting an exponential decay curve to the measured data using least squares optimization and making the assumptions of a constant initial excess 210Pb concentration (the CIC [constant initial concentration] model; Online Supplementary Figs. 1, 2, 3). Because the peat accumulating in the Everglades is formed by the detritus of local wetland vegetation, we used the CIC model because it assumes that deposition has been constant and not episodic. Age-depth models (Online Supplementary Figs 4, 5, 6) are constructed using the open-source the age-depth modeling software clam 2.1 (Blaauw 2010) using the statistical software R 2.15.0 (R Development Core Team 2012). Radiocarbon dates are calibrated in clam 2.1 using the IntCal09 calibration
1142
Wetlands (2013) 33:1139–1149
Table 1 Radiocarbon dates Core
Beta #
Depth in core (cm)
Latitude
Longitude
Material dated
δ 13C
Conventional C age
Error
14
Calibrated age plusa
Calibrated age minusa
Age AD
04-9-20-7
200693
20–22
26.5547
80.3599
peat
−24.3
180
70
422
0
1528–1950
04-9-20-7
200694
40–42
26.5547
80.3599
peat
−25.3
620
50
666
540
1284–1410
04-9-21-2
200697
20–24
26 28.236
80 22.59
peat
−25.3
330
60
504
289
1446–1661
04-9-21-2
200698
40–44
26 28.236
80 22.59
peat
−25.1
750
60
790
561
1160–1389
05-7-26-4
215151
20–22
26.4255
80.2965
peat
−25
120
50
280
0
1670–1950
05-7-26-4
215152
40–42
26.4255
80.2965
peat
−22.5
610
50
664
537
1286–1413
All dates were obtained on bulk peat samples and analyzed by Beta Analytic (Miami, FL, USA) Radiocarbon ages reported years before present where present is 1950 Upper and lower limits based on two sigma errors in calibration a
Ages are calibrated using Clam 2.1 (Blaauw 2010) against the IntCal09 calibration curve (Reimer et al. 2009)
curve (Reimer et al. 2009). Age-depth models are simple linear regression models, with age reversals excluded, between age calculations from 210 Pb, the first occurrence of Casuarina, and bulk 14C ages. Approximately 0.5–1.0 g of dry sediment was used for palynological analysis. Pollen and fern spores were concentrated from these samples using standard palynological techniques (Willard et al. 2001a, b; Traverse 2007). The samples were first acetolyzed (9 parts acetic anhydride : 1 part sulfuric acid) in a hot-water bath (100 °C) for 10 min, then neutralized and treated with 10 % KOH in a hot-water bath for 15 min. Neutralized samples were sieved with 10 μm and 200 μm sieves, and the 10–200 μm fraction was stained with Bismarck Brown, mixed with warm glycerin jelly, and mounted on microscope slides. Raw data for pollen samples are reposited at the US Geological Survey South Florida Information Access (SOFIA) site (http://sofia.usgs.gov). Pollen and spore identification (minimally 300 grains per sample) was based on reference collections of the United States Geological Survey (Reston, VA) and Willard et al. (2006a). Pollen diagrams are based on total pollen sum. In the Everglades, the combination of upland taxa and wetland taxa in the pollen sum is sensitive enough to distinguish between wetland communities only meters apart and record their response to changes in hydrology (Willard et al. 2001b; Bernhardt and Willard 2009). Our interpretations of past plant communities are based on the quantitative method of modern analogs (Overpeck et al. 1985). We calculated squared chord distance (SCD) between downcore pollen assemblages and a suite of surface samples (225) collected throughout southern Florida in the early AD 1960s and AD 1995–2002 (Willard et al. 2001b and this research) to define the similarity between each fossil and modern pollen assemblage. Internal comparison among surface samples from vegetation types indicates that samples with SCD values
