Upland deforestation triggered an ecosystem state-shift in a kettle peatland

Authors


Correspondence author. E-mail: awi207@lehigh.edu

Summary

1. European settlement of eastern North America resulted in dramatic changes to ecosystems, although the dynamics and underlying causes of these changes are not always obvious. For example, land clearance likely exposed soils to increased wind erosion, potentially impacting downwind ecosystems indirectly through subsequent enhancement of dust deposition. We hypothesized that otherwise undisturbed wetlands were altered through this indirect disturbance mechanism, increasing nutrient availability and initiating a cascade of ecosystem-level changes.

2. We tested this hypothesis in a floating kettle peatland located in north-western Pennsylvania, USA, using an interdisciplinary approach. A series of peat cores were collected along a transect oriented parallel to the dominant wind direction. Palaeoecological techniques were used to identify signatures of upland deforestation and mineral matter deposition within the peat profiles. Elemental analyses were used to reconstruct historic availability of major macronutrients, plant community dynamics were reconstructed using plant macrofossils and tree rings (Pinus strobus) and testate amoebae were used as a proxy for microbial community dynamics.

3. Strong correlations between the concentration of ragweed (Ambrosia) pollen and fine-grained mineral matter linked upland deforestation to enhanced dust deposition on the peatland surface. Elemental analyses indicated that nitrogen, phosphorus and potassium concentrations increased coincident with dust deposition. Plant communities shifted from Sphagnum dominance to vascular-plant dominance coincident with enhanced dust deposition, including increased recruitment of P. strobus onto the peatland. Testate amoeba communities shifted towards those adapted to highly variable microenvironmental conditions and likely reflect broader changes in microbial communities.

4.Synthesis. Upland deforestation by European settlers triggered a cascade of ecological changes on a nutrient-poor peatland by enhancing dust deposition and nutrient delivery on the surface. These results demonstrate that indirect, unintended and often overlooked human disturbances can lead to dramatic structural and functional alterations of carbon-rich wetland ecosystems, highlighting the potential vulnerability of these systems in human-dominated landscapes.

Introduction

European colonization of eastern North America resulted in a profound and rapid transformation of natural ecosystems (Foster et al. 2004). While many direct consequences of land-cover conversion by European settlers, including altered nitrogen (N) cycling in forests (Goodale & Aber 2001) and dramatic changes in the physical and biological functioning of stream networks (Allan 2004; Walter & Merrits 2008) are well documented, comparatively few studies have examined the effects of historic upland land-use changes on the structure and function of adjacent wetlands. Understanding human legacies in these carbon-rich ecosystems is critically important to predicting how future climatic and land-use changes could alter carbon (C) cycling through this sizable terrestrial pool (Buffam et al. 2011). Although the legacies of past human disturbances are rarely recognized in modern wetland systems, several retrospective studies have documented pre-settlement conditions which were markedly different from modern wetland structure and speculated that these changes resulted from altered hydrology and/or increased overland sediment delivery associated with upland deforestation (Warner, Kubiw & Hanf 1989; Lamentowicz, Tobolski & Mitchell 2007; Rippke, Distler & Farrell 2010). However, detailed mechanisms are poorly understood, and other factors may have also changed in response to upland land clearance.

Increased aerial deposition of atmospheric dust associated with land clearance may represent a significant disturbance of wetlands, particularly nutrient-poor systems. For example, dust deposition in remote lakes of the western United States increased fivefold with the onset of extensive agricultural activities in the region (Neff et al. 2008). Ecologists have only recently begun to appreciate the potential consequences of dust deposition on natural systems (Okin et al. 2004; Hughes et al. 2008), yet concerns over the increasing intensity of human land-use, and anticipated climate changes of the coming century, highlight the critical importance of understanding this potential driver of ecosystem change (Field et al. 2010). In wetlands, and especially in nutrient-poor peatlands, increased dust deposition could initially go unnoticed; however, cumulative dust inputs may eventually lead to significant ecological changes by increasing nutrient availability (Redfield 1998), altering the rates of production and decomposition and changing the outcome of competitive dynamics.

Peat deposits represent natural environmental archives, recording both local changes in the peatland biota, as well as changes in the adjacent uplands through the preservation of pollen and other wind-blown particulate matter. For example, peat profiles have been used to track long-term trends in aerial deposition of heavy metals and to determine the relative contributions of natural and anthropogenic sources of pollutants (Shotyk et al. 1998; Givelet, Roos-Barraclough & Shotyk 2003; De Vleeschouwer et al. 2009; Farmer et al. 2009). Because these systems rely primarily on atmospheric inputs, they are naturally nutrient poor (Bridgham et al. 1996), and increasing anthropogenic N deposition during the past several decades has led to concerns over the potential effects of eutrophication on their structure and biogeochemistry (Bragazza et al. 2006).

Peatlands that form in small kettle depressions, which are common in previously glaciated portions of temperate North America, usually have discrete boundaries and generally maintain some connectively with groundwater (Kratz & Medland 1989; Bridgham et al. 1996). Many kettle peatlands consist of floating mats of live and decomposing vegetation, which rise and fall with fluctuations in basin water levels (Buell & Buell 1941; Whittington et al. 2007; Ireland & Booth 2011). As a result, the position of the water-table in floating kettle peatlands remains relatively constant throughout the year, making these systems less climatically sensitive than grounded bogs with domed centres (Booth 2010). Despite their reduced climatic sensitivity relative to some other peatlands, hydrological studies have demonstrated that groundwater flow paths in floating kettle peatlands are predominantly outward, making them functionally ombrotrophic and favouring Sphagnum-dominated plant communities comparable to those found on nutrient-poor raised bogs (Hemond 1980; Mouser et al. 2005; Rydin & Jeglum 2006). Thus, floating kettle peatlands represent a nearly ideal setting to assess the impacts of atmospheric dust deposition while minimizing the confounding effects of water-level fluctuations driven by climatic variability.

In this study, we assessed the historic response of a floating kettle peatland in north-western Pennsylvania, USA, to deforestation of the surrounding uplands by European settlers. We hypothesized that enhanced dust deposition associated with upland deforestation caused marked and long-lasting changes in nutrient availability that subsequently altered plant communities, microbial communities and long-term peat accumulation and decomposition processes. To test these hypotheses, we reconstructed the recent history of peatland development by collecting three sediment cores along a transect oriented generally parallel to the dominant wind direction. We used a suite of palaeoecological techniques to detect relative changes in forest cover on the surrounding uplands, quantify mineral matter inputs to the peatland, estimate past nutrient availability, reconstruct the dynamics of soil biota and plant communities and determine the timing of white pine (Pinus strobus) recruitment on the peatland surface.

Materials and methods

Study Region and Site Description

This study was conducted in Erie County, Pennsylvania, USA, which occupies c. 400 000 ha of the previously glaciated Allegheny Plateau (Fig. 1). In Erie County, the earliest land grants to European settlers occurred in the late 1700s (Whitney & DeCant 2003). Homesteaders found the soils and climate to be suitable for livestock production, and settlement commenced quickly after about 1850 (Whitney & DeCant 2003). Land clearance peaked between 1900 and 1910 when 70–80% of the county (∼300 000 ha) was deforested (Whitney & DeCant 2003). The regional climate is humid and continental with dominant winds blowing from the west-south-west (Fig. 1). From 1895 to 2008, the mean January and July air temperatures were −4.3 and 20.6 °C, respectively, and annual precipitation averaged 1040 mm (US Climate Division 36-10, NOAA, Earth System Research Laboratory, http://www.esrl.noaa.gov/psd/data/timeseries/). Over the past century, moisture availability in the region has been increasing, as indicated by a significant positive linear trend in monthly mean Palmer Drought Severity Index (R2 = 0.45, a = 0.004, < 0.0001).

Figure 1.

 Regional orientation map depicting the location of Titus Bog in Erie County, Pennsylvania, USA, and a detail map showing coring locations. The dominant wind direction of 230° was calculated by the Office of the Pennsylvania State Climatologist using 72 559 hourly observations collected between 1991 and 2005 at Venango Regional Airport (FKL) in Franklin, Pennsylvania (http://climate.met.psu.edu/www_prod/features/wind_roses/). FKL is c. 60 km south of Titus Bog. Note that the arrow is pointing in the leeward direction. Peat coring locations are depicted overlaying a 0.30-m resolution aerial photograph collected in April 2005. Tree cover on the peat mat is almost exclusively P. strobus, and the stem density is substantially greater along the northern, western and southern margins than in the centre or along the eastern margin. Note that cores A, B and C, were referred to as cores 4, 5 and 6, respectively, in Ireland & Booth (2011).

Within this landscape, we studied Titus Bog (41°57′07″ N, 79°45′30″ W), a peatland complex situated within an c. 16-ha glacial kettle (Fig. 1). Glacial deposits comprising the northern rim of the kettle serve as the drainage divide between two small watersheds (∼1800 ha each), but stream channels do not flow into or out of the basin. The surface of the peatland is about 20 m below the elevation of the surrounding uplands. The peatland complex is composed of a shrub-dominated swamp around the periphery and a centrally located floating peatland, which has had essentially the same geometry for the last several hundred years (Ireland & Booth 2011). Surface waters on the peatland are acidic (pH 3.24–5.04). Plant communities are generally characterized by Sphagnum mosses, but peatland shrubs, especially leather leaf (Chamaedaphne calyculata), sedges (Carex spp.) and club mosses (Lycopodiella spp.) are common, particularly in the central portion of the peatland. P. strobus occurs at high density (∼500 stems ha−1) around the western, northern and southern margins of the floating peatland and at low density (∼100 stems ha−1) in the interior (Fig. 1). There is no obvious evidence to suggest historic logging of the P. strobus population on the peatland and selective logging within this wetland complex seems unlikely given the floating character of the peatland and the extensive, shrub-dominated swamp around the perimeter. Examination of records archived by the Erie County Recorder of Deeds for an area of c. 230 ha surrounding Titus Bog indicated that local land-use history paralleled regional patterns (cf. Ireland, Oswald & Foster 2011). The earliest land grant that included Titus Bog occurred in 1787, intensive land clearance and agriculture commenced around 1852 and the peatland was purchased in 1967 by the Presque Isle Audubon Society and the Botanical Society of Western Pennsylvania for conservation purposes. Currently, Titus Bog is bordered by young second-growth hardwood forest to the north and north-west and active agricultural fields elsewhere (Fig. 1).

Field Methods

In 2008, three peat cores (Fig. 1) were collected along a south-west to north-east transect, generally parallel with the dominant wind direction, using a 10-cm diameter tripod-mounted piston corer. All cores were collected from low-lying microtopographic features, where average water-table depths (e.g. acrotelm–catotelm boundary) are near the surface year round (Ireland & Booth 2011; Sullivan & Booth 2011). Although all cores were collected from Sphagnum-occupied areas, the vascular plant cover was variable between coring sites; a P. strobus canopy covered coring site A, coring site B was in a open area where Lycopodiella species were common and a vascular plant canopy was lacking, and coring site C was within a relatively dense stand of C. calyculata. The stratigraphy of each sediment core was described in the field prior to being wrapped in plastic and aluminium foil and secured in rigid polyvinyl chloride shells for transport to cold storage. To assess general temporal patterns of P. strobus recruitment on the peatland, 96 individual trees were sampled with a 25-cm increment borer. Sampled trees were cored within 30 cm of the base and were generally selected to represent the size classes and spatial densities present on the peatland, but sampling was intentionally skewed towards mature bark characteristics and large size in an effort to include the oldest individuals.

Physical Properties of Peat Cores

In the laboratory, each peat core was cut into contiguous 1-cm slices. Because the goal of this study was to reconstruct relatively recent peatland development, and the basal peat ages for these cores were determined in a previous study (Ireland & Booth 2011); analyses and results described in this paper were restricted to the upper 30 cm of each core. In each core, dry bulk density (g cm−3) was estimated at 1-cm resolution by weighing volumetric (1 cm3) samples after drying at 80 °C for a period of 14 h. These samples were subsequently reweighed after 2 h of combustion at 550 °C, with per cent weight loss on ignition (LOI) taken as an estimate of the proportion of the sample composed of organic matter (Heiri, Lotter & Lemcke 2001; Chambers, Beilman & Yu 2011). LOI and dry bulk density were used to estimate the organic matter concentration of each sample by expressing dry bulk density in units of mg cm−3 and multiplying by the proportional weight lost during LOI analysis. Similarly, the concentration of inorganic matter (referred to hereafter as ‘mineral’) was estimated using the proportional weight remaining after LOI analysis. To examine the size class and semi-quantitatively estimate elemental composition of mineral matter contained in peat deposits, combusted LOI residue was retained at 5-cm intervals between 5 and 25 cm depth in each core, gently homogenized and examined using a Hitachi TM-1000 tabletop scanning electron microscope (SEM) equipped for energy-dispersive X-ray analysis. In each sample of homogenized LOI residue, three randomly placed 1-mm2 areas were examined at 150× magnification, an image was captured, and elemental abundances (% weight) were estimated through X-ray backscatter.

Carbon, Nitrogen and Phosphorus Concentrations

Volumetric samples of bulk peat (1 cm3) were collected at 1-cm resolution in core A and at 2-cm resolution in cores B and C. Samples were dried at 80 °C for 14 h, manually powdered using a mortar and pestle and homogenized immediately prior to sub-sampling by weight (2.7–4.3 mg for C and N; 5.2–9.9 mg for phosphorus [P]). The University of California-Davis Stable Isotope Facility performed C and N analyses using a PDZ Europa ANCA-GSL elemental analyzer. Standard procedures were used to extract particulate P and to measure optical density using a spectrophotometer at 885 nm (Solórzano & Sharp 1980), with laboratory standards included in each run. For each sample in which nutrient analyses were performed, measured C, N and P contents (μg g−1) were converted to units of mg mg−1 and multiplied by organic matter concentration (mg cm−3) to estimate C, N and P concentrations (mg nutrient cm−3) within the organic matter fraction of each sampled depth. Ratios of C : N and N : P were calculated from elemental concentrations and explored as proxies for peat decomposition (Kuhry & Vitt 1996) and historic community-to-ecosystem level nutrient limitations, respectively. Although speculative, previous work has suggested that an N : P ratio of 15 : 1 could potentially represent a critical threshold between P-limitation (>15 : 1) and N-limitation (<15 : 1) in both bryophyte-dominated and vascular plant–dominated wetland systems (Koerselman & Meuleman 1996; Walbridge & Navaratnam 2006).

Macrofossil Analysis

A semi-quantitative method of macrofossil analysis was adapted from techniques presented in Yu et al. (2003) and used to characterize stratigraphic changes in botanical composition at 1-cm resolution in core A and at 2-cm resolution in cores B and C. Volumetric samples (2 cm3) were disaggregated by gently wet sieving for c. 30 s, and all material >125 μm was retained. Captured material was placed into a gridded plastic dish and dispersed in c. 30 mL of water. The mixture was scanned under a standard dissecting microscope, and botanical constituents were classified as one of the six broad plant functional types or as unidentifiable, highly decomposed plant matter. Identifications were aided by the use of Mauquoy & van Geel (2007), but no attempt was made to classify botanical remains to the lowest possible taxonomic resolution. Because the large number of analyses performed in this study restricted the volumetric macrofossil sample size, concentrations were not calculated and data were only expressed on a percentage basis. The percentage of unidentifiable plant matter was interpreted as an estimate of peat decomposition (Cai & Yu 2011).

Microfossil Analysis

Standard procedures (Booth, Lamentowicz & Charman 2010) were used to isolate sub-fossil pollen grains and testate amoebae from volumetric peat samples (1 cm3) collected at 1-cm resolution in core A and at 1- to 2-cm resolution in cores B and C. Samples were boiled in deionized water for 10 min and wet-sieved through nested screens to retain material between 15 and 355 μm in diameter. Concentrated material was stained with Safranine dye and dispersed in glycerol. Microfossils were scanned at 400× magnification under a light microscope along non-overlapping transects. A known concentration of exotic Lycopodium spores was added to each sample and tallied along with pollen grains to enable estimation of pollen concentrations (Maher 1981) in units of (grains cm−3). Ragweed (Ambrosia spp.) pollen grains were used as a temporal marker for European settlement and land clearance (Turetsky, Manning & Wieder 2004). In cores B and C, pollen analysis was increased from 2-cm to 1-cm resolution across the Ambrosia rise to delineate the settlement horizon with the greatest possible accuracy.

Testate amoebae are a polyphyletic group of protozoa that produce morphologically distinct and decay-resistant shells. These organisms are commonly used as indicators of environmental change in peatland ecosystems, especially hydrology (Mitchell, Charman & Warner 2008a). Testate amoeba communities not only respond to changes in average hydrological conditions, but also other environmental factors, including the magnitude of microenvironmental variability at the surface of a peatland during the growing season, a factor that is influenced by type and density of vegetation cover (Sullivan & Booth 2011). Furthermore, testate amoebae are among the largest members of soil microbial communities and feed on bacteria, detritus and fungi, making their community composition sensitive to changing food sources (Gilbert et al. 2003; Jassey et al. 2011). Thus, testate amoebae can be used as an indicator for broader microbial community change. To assess changes in testate amoeba community composition through time, tests were identified and tallied until a minimum of 50 specimens were encountered in each sample. Although this total count is fewer than many studies employ, it is sufficient to identify major community shifts (Payne & Mitchell 2009; Forcino 2012), which was the goal of this study. Taxonomy followed Charman, Hendon & Woodland (2000) and Booth (2008). Habrotrocha angusticollis, a rotifer commonly associated with testate amoebae, was included in the counts and subsequent analyses. Five siliceous taxa (Corthion–Trinema, Euglypha rotunda, Euglypha tuberculata, Placocista spinosa and Tracheuglypha dentata types) were encountered in the uppermost samples of each core, but these were excluded from count totals and data analysis because they do not preserve well in oligotrophic peatlands (Mitchell, Payne & Lamentowicz 2008b). In samples containing these taxa, total counts were increased to ensure that the 50-specimen threshold was met after their removal.

Age-Depth Modeling

An age-depth model was constructed for each peat core using a series of accelerator mass spectrometry radiocarbon dates of plant macrofossils, the depth of the initial increase in mineral matter concentration, the depth of the maximum in Ambrosia pollen concentration and the age of the surface (Fig. 2). Descriptions of methods used in radiocarbon dating as well as laboratory identifiers and results for each date were presented in Ireland & Booth (2011). Radiocarbon dates were calibrated and age-depth models were constructed using software written in R (CLAM; Blaauw 2010). Within the age-depth models, the initial increase in mineral matter concentration and maximum in Ambrosia pollen concentration were assigned ages of 1850 ± 10 and 1900 ± 10, respectively, based on a regional land-use history constrained by the analysis of historical documents (Whitney & DeCant 2003). A smooth spline was used to model the data, and a Monte Carlo approach with 1000 iterations was used to estimate the 95% confidence interval for all age assignments. Peat accumulation rates (cm year−1) were calculated at each 1-cm depth interval in each core.

Figure 2.

 Age-depth models for peat cores A, B and C (Fig. 1). For each core, black error bars represent the full 2σ range of calibrated radiocarbon dates, and triangles represent the depths and ages of stratigraphic markers explained in text. Note that complete laboratory descriptions of all accelerator mass spectrometry radiocarbon dates are published in Ireland & Booth (2011). In each core, the solid black line indicates the age assignment for each depth and the grey-shaded zone represents the 95% confidence interval for the age-depth model. Horizontal dotted lines represent the maximum depth of analyses performed in this study. Note the apparent reductions in peat accumulation rates after European settlement.

Analysis of Pinus strobus Recruitment on Titus Bog

In the laboratory, tree cores were mounted on wooden blocks and sanded with progressively finer sandpaper until annual growth increments were visible under a standard dissecting microscope. Of the 96 original samples, two were removed due to rot and nine were removed because the core did not capture near-pith curvature, reducing the sample size to 85. Annual growth increments were counted using a Velmex tree ring system and the number of growth increments was used to estimate the year of recruitment for each stem. In cores that captured the pith imperfectly (n = 60), pith locators were used to estimate the number of missing rings (mean = 3; max = 9) following Applequist (1958). To account for uncertainty in age estimates, data were binned by decade and only the general pattern was interpreted.

Data Analysis and Synthesis

Within each core, the degree of association between Ambrosia pollen and mineral matter concentrations was determined by calculating Pearson’s correlation coefficient, excluding depths where pollen analysis was not performed. The software package PAST version 1.78 (Hammer, Harper & Ryan 2001) was used to summarize changes through time in multivariate X-ray backscatter, macrofossil and testate amoeba data by performing principle components analyses (PCAs) of the variance–covariance matrix for untransformed percentages. Due to small samples sizes in each core (n = 5), X-ray backscatter data were compiled from all three cores for PCA (n = 15). Separate PCAs were performed for each core with macrofossil and testate amoeba data, where sample sizes were substantially larger and variability among the cores was more pronounced. In the testate amoeba dataset, only taxa that exceeded 10% representation in at least one sample were included in the analysis. All data were plotted versus age, interpreted within the context of regional land-use history (Whitney & DeCant 2003), and interpretations were synthesized into a general conceptual model describing hypothesized processes and feedbacks.

Results

Ambrosia Pollen and Mineral Matter Concentrations

Prominent increases in concentration of both Ambrosia pollen and mineral matter were observed in each core (Fig. 3) and interpreted to represent deforestation of the surrounding uplands by European settlers (Turetsky, Manning & Wieder 2004). Pre-settlement Ambrosia pollen concentrations ranged from 0 to about 4000 grains cm−3, while peak concentrations ranged from about 17 000 to about 46 000 grains cm−3. This pattern was mirrored by mineral matter concentrations that ranged from 0.8 to 4.8 mg cm−3 during pre-settlement time and from 14.8 to 25.5 mg cm−3 during the inferred time of peak deposition. All cores displayed statistically significant correlations between Ambrosia pollen and mineral matter concentrations (Fig. 3). SEM observations of LOI residue documented only the presence of very fine-grained particles (generally <25 μm) within the mineral fraction of the peat deposits. Within each core, the initial increase in mineral concentration was interpreted as representing the onset of land clearance (∼1850) and the peak Ambrosia pollen concentration was interpreted as representing ∼1900, when regional deforestation was greatest (cf. Hölzer & Hölzer 1998).

Figure 3.

 Comparison of Ambrosia pollen, mineral matter, organic matter in bulk peat, C, N and P in the organic matter fraction and the elemental composition of the mineral matter fraction in all peat cores through time. Letters correspond to core locations presented in Fig. 1. Grey bands begin at 1850 and end at 1910, corresponding to regional settlement and maximum deforestation, respectively (Whitney & DeCant 2003). Note the statistically significant correlations between the concentrations of Ambrosia pollen and mineral matter as well as the marked shifts in concentrations of N and P and principle components analyse (PCA) axis 1 scores for X-ray backscatter data collected from the mineral fraction of the peat samples. Correlations between individual elements and PCA axis 1 scores were as follows: Si (−0.98), Al (−0.89), K (−0.85), Fe (−0.21), Ca (0.99), Mg (0.77) and S (0.64).

Peat Accumulation Rates and Elemental Concentrations

Age-depth models suggest that after the initial increase in mineral matter concentration around 1850, average peat accumulation rates decreased by factors of 1.6, 4.2 and 4.4 in cores A, B and C, respectively (Fig. 2). In core A, average carbon concentration increased by a factor of 1.9 after peak mineral concentration (Fig. 3). However, this pattern did not occur in cores B and C where average carbon concentrations remained relatively constant across the peaks in mineral matter concentration (Fig. 3). After peak mineral matter concentrations, N concentrations increased by factors of 3.9, 1.5 and 1.5, and P concentrations increased by factors of 6.7, 1.9 and 1.9 in cores A, B and C, respectively (Fig. 3). Energy-dispersive X-ray analysis indicated that the mineral fraction of peat deposits above the peak mineral concentrations was dominated by silicon (Si), aluminium (Al), potassium (K) and iron (Fe) while samples below this level were dominated by calcium (Ca), sulphur (S) and magnesium (Mg). The first principle component in this analysis explained 92.8% of the variability and closely tracked mineral matter concentrations (Fig. 3).

Macrofossils

In all cores, Sphagnum mosses were replaced by vascular plants coincident with the inferred timing of European settlement and maximum deposition of mineral matter on the peatland surface (Fig. 4). Shrubs, mostly C. calyculata, increased in abundance towards the top of each core. Pinus strobus needles and bud scales became abundant in the upper portions cores A and B. Unidentifiable plant material increased at the transition from Sphagnum- to vascular-plant dominance in all cores (Fig. 4).

Figure 4.

 Changes in macrofossil assemblages through time for cores A, B and C. Grey bands begin at 1850 and end at 1910, corresponding to regional settlement and maximum deforestation, respectively (Whitney & DeCant 2003). Note that major changes began around the inferred time of European settlement, including marked declines in Sphagnum, increases in vascular plants and increased representation of highly decomposed plant remains. Pinus strobus includes needles and bud scales.

Testate Amoebae

In total, 38 taxa were encountered in the analysis of testate amoebae, including the rotifer Habrotrocha augusticollis and the five siliceous testate amoebae that were excluded from the dataset. Of the remaining 33 taxa, 14 exceeded the representation threshold (10% in at least one sample) and were included in subsequent analyses (Fig. 5). Mirroring all other datasets, testate amoeba communities exhibited shifts in composition coincident with the inferred timing of European settlement and maximum deposition of mineral matter on the peatland surface (Fig. 5). Pre-settlement samples were generally dominated by Habrotrocha augusticollis (rotifer), Hyalosphenia papilo, Hyalosphenia elegans and Centropyxis aculeata types, while post-settlement samples tended to be dominated by Difflugia pulex, Difflugia pristis, Hyalosphenia subflava and Cyclopyxis arcelloides types. However, each core displayed slightly different pre- and post-settlement communities.

Figure 5.

 Testate amoeba data for cores A, B and C plotted through time. Grey bands begin at 1850 and end at 1910, corresponding to regional settlement and maximum deforestation, respectively (Whitney & DeCant 2003). Note that changes in community composition began in all cores coincident with European settlement, paralleling all other datasets.

Pinus strobus Recruitment on the Peatland

The oldest sampled P. strobus established on the peatland around 1902 and the youngest established around 1987 (Fig. 6). The median-estimated establishment date was 1952, and the inter-quartile range was 1925–72. Recruitment data agree with macrofossil data, which documented a rarity of P. strobus needles and bud scales before 1900 (Fig. 4).

Figure 6.

 Data synthesis comparing regional deforestation curve (Whitney & DeCant 2003) with mineral matter concentrations, C : N ratios, N : P ratios and principle components analyse (PCA) axis 1 scores for testate amoeba and macrofossil data from all peat cores as well as tree ring data documenting Pinus strobus recruitment on the peatland. Letters correspond to core locations presented in Fig. 1, and grey bands begin at 1850 and end at 1910, corresponding to regional settlement and maximum deforestation, respectively (Whitney & DeCant 2003). Note that all datasets in all cores display marked changes coincident with deforestation. PCA axis 1 accounts for 68.5%, 40.2% and 35.5% of the variability in testate amoeba data in cores A, B and C, respectively. In the case of macrofossil data, PCA axis 1 explains 83.5%, 60.7% and 61.9% of the variability in cores A, B and C, respectively. Pinus strobus recruitment patterns suggest that trees were rare on the peatland prior to European settlement, but that recruitment increased markedly following peak dust deposition on the peatland surface. Note that C : N ratios generally track macrofossil PCA axis 1 scores. Although speculative, the observed changes in N : P ratios may be indicative of a transition from P-limitation to N-limitation after dust deposition (Koerselman & Meuleman 1996; Güsewell & Koerselman 2002; Walbridge & Navaratnam 2006).

Discussion

An Anthropogenic, Nutrient-Mediated Ecosystem State-Shift

Collectively, these datasets suggest that upland deforestation by European settlers triggered an abrupt, nutrient-mediated ecosystem state-shift on Titus Bog (Fig. 6). Mineral deposition was linked to land clearance by strong positive correlations between the concentrations of mineral matter and Ambrosia pollen in all peat cores (Fig. 3), a pattern that has been documented before in comparable peatland systems (Hölzer & Hölzer 1998; Martínez Cortizas et al. 2005; Lomas-Clarke & Barber 2007; Hughes et al. 2008). Mineral matter deposited on Titus Bog was most likely derived from nearby upland soils and transported by aeolian processes. Evidence for an aeolian origin of mineral matter in these peat cores includes (i) the lack of stream channels flowing into Titus Bog, (ii) the lack of large slopes on the adjacent uplands that could have facilitated substantial sheet flow into the wetland complex, (iii) the presence of c. 30 m of shrub swamp between the centrally located peatland and surrounding uplands, and (iv) the small particle size of the mineral matter itself (cf. Gorham & Tilton 1978; Santelmann & Gorham 1988). Furthermore, X-ray backscatter data from all three cores indicated that Si, which has long been used as a tracer of dust deposition on peatlands (Chapman 1964; Hölzer & Hölzer 1998; Martínez Cortizas et al. 2005; Lomas-Clarke & Barber 2007; Hughes et al. 2008), was the most abundant element in LOI residue collected within the mineral matter peak and that Si is more abundant above the depth of maximum mineral concentration than below (Fig. 3). Finally, core A, which was collected along the windward margin of the peatland, recorded more dramatic shifts in all measured variables than core B or C, further suggesting that windblown material likely caused nutrient enrichment of the ecosystem.

In all cores, peat deposits above the depth of peak mineral matter and Ambrosia pollen were enriched in N, P and K relative to peat deposits below this level (Fig. 3), consistent with the hypothesis of dust fertilization. P enrichment was especially pronounced, particularly in core A (Fig. 3). Modern studies have shown that windblown dust particles can be composed of up to 0.2% P by weight (Redfield 1998) and often represent the most important source of P replenishment to peatlands (Le Roux, Laverrret & Shotyk 2006). This is especially true in highly disturbed agricultural settings where dry P deposition can reach 50–100 mg m−2 year−1, two to three times greater than less disturbed, forested regions (Redfield 1998).

The post-disturbance shift from Sphagnum-dominated to vascular-plant-dominated plant communities is also consistent with greater nutrient availability on the peatland surface (Rydin & Jeglum 2006). Similarly, expansion of P. strobus on the peatland is consistent with fertilization, especially in terms of P enrichment, as trees can be strongly P-limited in peatland systems and can respond quickly to increased P availability (Rydin & Jeglum 2006). Other studies of peat profiles have documented correlations among palynological indicators of upland land-use change, geochemical properties of the peat deposits and macrofossil records of vegetation shifts. For example, Hughes et al. (2008) demonstrated strong linkages between palynological indicators of pastoral disturbance and marked declines in a particular species of Sphagnum moss on raised bogs in the British Isles and suggested that this species (Sphagnum austinii) was sensitive to aerial deposition of dust particles containing nutrients and/or charcoal.

Testate amoeba communities also underwent marked structural changes coincident with mineral matter deposition, nutrient enrichment and plant community shifts, especially in core A (Figs 5 and 6). Recorded community shifts were not strictly consistent with directional changes towards wetter or drier conditions on the peatland surface, which is not surprising as such directional shifts would not be expected to occur on a hydrologically stable floating peatland (Booth 2010). Rather, data suggest that coincident with the change in plant communities, testate amoeba communities shifted towards those more tolerant of high-magnitude variability in micrometeorological conditions, such as those characterized by high abundances of D. pulex and H. subflava (Sullivan & Booth 2011). Shifts from densely growing Sphagnum mosses to relatively less dense vascular plants would have enhanced micrometeorological variability at the peatland surface, affecting the upper few centimetres where testate amoebae live. Furthermore, food sources for testate amoebae may have changed, due to shifts in the composition of microbial communities, and this may also have contributed to changes in testate amoeba community composition (Jassey et al. 2011). Variation in testate amoeba communities among cores likely resulted from relatively small dissimilarities in the microtopography of coring locations. These differences highlight that core samples generally record very local changes, illustrating the importance of collecting and analysing multiple cores when attempting to reconstruct whole-ecosystem dynamics.

Together, changes in the mineral matter concentrations (Fig. 3) and proportions of highly decomposed plant material (Fig. 4) suggest that nutrient enrichment stimulated microbial decomposition. Reduced peat accumulation rates in the upper portions of these peat cores (Fig. 2) are also suggestive of enhanced decomposition. A post-disturbance increase in rates of peat decomposition is generally consistent with increased P availability, as microbial respiration can be P-limited in nutrient-poor peatlands (Amador & Jones 1993). It is also possible that the expansion of vascular plants yielded less recalcitrant litter than the preceding Sphagnum-dominated peatland community, facilitating relatively high rates of decomposition and rapid recycling of resources (Verhoeven, Maltby & Schmitz 1990; Aerts, Verhoeven & Whigham 1999) (Fig. 7).

Figure 7.

 Conceptual model summarizing the proposed dynamics that were triggered by deforestation of the surrounding uplands. Boxes and text in black are underpinned by data collected in this study, whereas those in grey represent hypotheses. Data indicate that human deforestation led to dust deposition on the Sphagnum-dominated, nutrient-poor surface of Titus Bog. Elemental data strongly suggest that dust deposition transported nutrients (especially, N, P and K) to the system, and this fertilization likely led to changes in plant communities. Changes in testate amoeba communities were likely caused by some combination of higher microenvironmental variability associated with the more open vegetation canopy, as well as changing food sources caused by broader changes in microbial communities. Macrofossil data documented more decomposition within the post-settlement vascular plant communities than within the pre-settlement Sphagnum-dominated communities. Vascular plant communities may have produced less recalcitrant litter, sustaining high rates of decomposition and rapid nutrient cycling (Verhoeven, Maltby & Schmitz 1990; Aerts, Verhoeven & Whigham 1999) and providing a feedback mechanism to maintain the new ecosystem state.

While all data indicate that aerial dust deposition fertilized Titus Bog, inferring past nutrient limitation and identifying which specific nutrients led to the ecological changes is challenging. Primary productivity in oligotrophic peatlands is commonly thought to be N-limited (e.g. Bragazza et al. 2006), although others have suggested P-limitation (Walbridge & Navaratnam 2006), potentially in conjunction with K-limitation (Güsewell & Koerselman 2002). Recent work has suggested that many ecosystems, including freshwater wetlands, could in fact be co-limited by N and P, with the addition of both nutrients resulting in strong synergistic effects (Elser et al. 2007).

Nutrient ratios are frequently used to explore the nature of nutrient limitation and ecological processes. In all cores at Titus Bog, C : N ratios were generally consistent with average values reported for bulk peat (Limpens, Heijmans & Berendse 2006) and tracked botanical changes recorded in macrofossil data (Fig. 6). Observed negative excursions in C : N ratios towards the surface were inconsistent with expectations for a decomposition-driven pattern (Kuhry & Vitt 1996) and more likely reflected changes in source material (Dorrepaal et al. 2005). In all cores, N : P ratios exhibited negative shifts coincident with maximum mineral matter concentrations (Fig. 6). The direction of these shifts suggested that Titus Bog may have tended towards P-limitation prior to European settlement (>15 : 1) and that after dust deposition the system may have become more N-limited (<15 : 1) (Walbridge & Navaratnam 2006). More research would be required to systematically test this hypothesis; however, data suggest N, P and K were transported to the surface of Titus Bog, making it likely that increased nutrient availability triggered the dramatic ecological changes, even if the nature of the preceding nutrient limitation is uncertain.

Conclusions and Conservation Implications

Extensive efforts have gone towards predicting responses of C-rich peatland ecosystems to large-scale increases in anthropogenic nutrient deposition (e.g. Bragazza et al. 2006). Results of this study suggest that peatlands in agricultural landscapes may be vulnerable to nutrient enrichment through non-point-source dust deposition, with considerable community-to-ecosystem level consequences. At Titus Bog, dust deposition was associated with reduced peat accumulation rates (Fig. 2), nutrient enrichment (Fig. 3), shifts from Sphagnum moss to vascular plant communities (Figs 4 and 6) and shifts in testate amoeba assemblages (Fig. 5) that likely reflected broader changes in microbial communities. Although speculative, N : P ratios suggest that prior to European settlement, the peatland tended towards P-limitation and that dust deposition associated with land clearance pushed the system towards N-limitation (Fig. 6). Although the mineral fraction of peat deposits never exceeded 20% of total weight, the nutrient inputs were apparently large enough to exceed a critical threshold for this oligotrophic system, leading to a cascade of ecological changes (Scheffer et al. 2001). These results suggest that Titus Bog had limited capacity to resist nutrient-driven changes.

Sphagnum mosses are relatively abundant on the modern peatland surface, suggesting some recovery during the past century and obscuring the dramatic shifts that occurred in the recent past. However, the legacy of dust deposition likely still remains, as vascular plants are substantially more common today than they were prior to the transient disturbance event. Furthermore, the strong spatial patterning of P. strobus (Fig. 1) is apparently unrelated to the developmental history of the peatland (Ireland & Booth 2011), but more likely is related to land-use history in the adjacent upland (cf. Houlahan et al. 2006) and possibly the influence of the dominant westerly surface winds in depositing more dust along the western margins (Santelmann & Gorham 1988). Our results underscore the benefits of including a palaeoecological perspective in developing strategies for the conservation, management and restoration of C-rich peatlands (Vasander et al. 2003; Williams 2011), reaffirm the value of upland buffers along wetland margins and highlight the potential sensitivity of these systems to past and future changes in usage of the surrounding landscape.

Acknowledgements

This project was supported by grants from the Society of Wetland Scientists and the Lawrence Henry Gipson Institute for 18th Century Studies (Lehigh University). Julie Loisel, Maura Sullivan and William Ireland assisted in the field. Travis Andrews performed preliminary testate amoeba counts, which helped to move this project forward. Donald Morris provided laboratory space, reagents and guidance for phosphorus extractions, and Christopher Dempsey assisted in the analysis. Molly O’Neil processed peat samples for microfossil analysis. Earlier versions of this manuscript benefited greatly from critiques by Zicheng Yu, Frank Pazzaglia, Daniel Charman, Edward Mitchell and an anonymous reviewer.

Ancillary