Wetland microtopography is a visually striking feature, but also critically
influences biogeochemical processes at both the scale of its observation
(10
Microtopography, or the small-scale structured variation (10
Wetland microtopography changes the spatial distribution of relative water levels, affecting vegetative composition and growth, which, in turn, may reinforce microtopographic development. For example, seedlings often fare better on elevated microtopographic features such as downed woody debris or tree-fall mounds (Huenneke and Sharitz, 1990). The resulting increased vegetation root growth and associated organic matter inputs on such features may subsequently support hummock expansion. In this way, vegetation may reinforce and maintain its own hummock microtopography (and thus preferred environmental conditions). Growing research across different ecosystems suggests that such reinforcing processes, or feedback loops, may be common between biota and their environment, and may result in characteristic, self-organized patch features (Rietkerk and Van de Koppel, 2008; Bertolini et al., 2019). By quantifying the structure and patterning of these features, we may therefore make process-based inferences about latent feedback mechanisms (Turner, 2005; Quintero and Cohen, 2019).
Spatial patterning of landscape patches has been observed in many systems, such as the striping of vegetated patches in arid settings or maze-like patterns in mussel beds (Rietkerk and Van de Koppel, 2008), where researchers have inferred responsible feedback mechanisms (as opposed to random processes) using a suite of diagnostic indicators. There is a large body of literature where such measurements are used to identify patterned systems and to infer their latent feedbacks (see Pascual et al., 2002; Pascual and Guichard, 2005; Kéfi et al., 2011, 2014; Quinton and Cohen, 2019 and references therein). We suggest that these diagnostic indicators are extensible to the analysis of wetland microtopography, thereby allowing us to assess mechanisms that maintain and reinforce patterns of hummock patches. Here, we focus on three common methods of inference. First, multimodal distributions in environmental variables, such as vegetation composition, soil texture, and, in our case, elevation (and see Rietkerk et al., 2004; Eppinga et al., 2008; Watts et al., 2010), indicate positive feedbacks to patch growth, where local patch conditions promote further patch expansion (Scheffer and Carpenter, 2003; Pugnaire et al., 1996). Second, the presence of characteristic patch sizes implies that limits to patch growth operate at local scales as opposed to system scales (Manor and Shnerb, 2008; von Hardenberg et al., 2010). Limited patch growth results in a distinct absence of large patches, and, thus, a truncation of the size distribution (Kéfi et al., 2014; Watts et al., 2014). Third, regular spatial patterning of patches (Rietkerk et al., 2004), or spatial overdispersion of patches (i.e., uniformity of patch spacing is greater than expected by chance), implies a coupling of both local-scale positive feedbacks to patch growth and local-scale negative feedbacks to patch expansion (Watts et al., 2014; Quinton and Cohen, 2019). Here, we extend this inferential theoretical framework to characterize patterning and infer the genesis and persistence of wetland microtopography.
Our conceptual model of wetland microtopographic development posits elevation–plant productivity feedbacks that result in elevation bimodality, characteristic patch sizes, and patch overdispersion (Fig. 1). We suggest that many mechanisms may initiate microtopographic development, including direct actions from biota (e.g., burrowing or mounding), indirect actions from biota (e.g., tree falls or preferential litter accumulation), and abiotic events that redistribute soils and sediment (e.g., extreme weather events). However, regardless of the initiation mechanism, we hypothesize that elevated microsites provide relief from hydrologically induced anaerobic conditions, promoting plant establishment and growth, evapoconcentration of nutrients (Eppinga et al., 2009), increased organic matter accumulation and subsequent soil elevation (Harris et al., 2019), and so on (top, solid loop on the right-hand side of Fig. 1). These positive feedbacks ultimately induce soil elevation bimodality, where microtopographic features belong to either a stable hummock or stable hollow elevation state (Rietkerk et al., 2004, Eppinga et al., 2008; Watts et al., 2010). Negative feedbacks eventually limit this growth; otherwise, hummocks would have no vertical or lateral limit. Vertical negative feedbacks may result from increased decomposition as hummocks grow vertically and their soils become more aerobic (Minick et al., 2019a, b; bottom, dashed loop on the right-hand side of Fig. 1). Lateral negative feedbacks may result from canopy competition for light among trees located on hummocks, or from competition for nutrients among hummocks (Rietkerk et al., 2004; Schröder et al., 2005; Eppinga et al., 2009), leading to spatial overdispersion and common patch sizes. Finally, we predict that the strength of these feedback loops that grow and maintain hummocks will likely increase with wetter conditions (blue shading in Fig. 1). In contrast, hummock–hollow terrain and patterns may be less evident at drier sites where soils are nearly always unsaturated and aerobic, weakening the elevation–productivity feedback (Miao et al., 2013; Miao et al., 2017). In a companion study we found support for this overall model, where we observed vegetation and soil chemistry associations with hummock structures, indicative of elevation–productivity feedbacks, and that these associations were greatest at the wettest sites (Diamond et al., 2019). Here, we add to that work by assessing the structure and pattern of hummock features and the extent to which they are influenced by the hydrologic regime.
Conceptual model for autogenic hummock maintenance in wetlands. Incipient mechanisms create small-scale variation in soil elevation that is amplified by autogenic feedbacks, which grow and maintain elevated hummock structures. Solid lines indicate positive feedback loops, and dashed lines indicate negative feedback loops. Font in italics refer to feedback processes hypothesized to only affect the lateral hummock extent (thus the hummock area), whereas standard font indicates mechanisms that affect both the vertical and lateral hummock extent. Processes in blue indicate that these mechanisms are influenced by hydrology. Soil mass refers to the amount of (organic) soil in a hummock, which can include roots, leaves, and decaying organic matter.
In this study, we evaluated wetland soil elevations, hummock spacing, and
hummock sizes and their associations with hydrologic regimes in black ash
( elevation will exhibit a bimodal distribution, but the degree of bimodality and the overall variability in elevation will be greater in wetter sites than drier sites; surface topography will not reflect subsurface mineral topography, but will instead be representative of self-organizing processes at the soil surface; hummock heights will be positively correlated with water levels at site and within-site scales; hummock patches will exhibit spatial overdispersion, which will be more evident at wetter sites; cumulative distributions of hummock areas (and perimeters and volumes) will correspond to a family of truncated distributions (e.g., exponential or lognormal), indicating a characteristic patch size, with wetter sites exhibiting more large (with respect to area) hummocks than drier sites.
Site information for 10 black ash study wetlands.
To test our hypotheses, we investigated 10 black ash wetlands of varying
sizes and hydrogeomorphic landscape positions in northern Minnesota, USA
(Fig. 2; Table 1). Thousands of meters of sedimentary rocks overlay an
Archean granite bedrock geology in this region. Study sites are located on a
glacial moraine landscape (400–430 m a.s.l.) that is flat to gently rolling, with the black ash wetlands found in lower landscape positions that commonly grade into aspen- or pine-dominated upland forests. The climate is
continental, with a mean annual precipitation of 700 mm and a mean growing
season (May–October) temperature of 14.3
Map of black ash wetland sites. Sites are colored by their mean organic horizon depth. Imagery provided by © Google Maps 2019.
As part of a larger effort to understand and characterize black ash wetlands
(D'Amato et al., 2018), we categorized and grouped each wetland by its
hydrogeomorphic characteristics as follows: (1) depression sites (“D”,
Overstory vegetation at the 10 sites is dominated by black ash, with tree
densities ranging from 650 stems ha
Soils in black ash wetlands in this region tend to be Histosols characterized by deep mucky peats underlain by silty clay mineral horizons, although there were clear differences among site groups (NRCS, 2019). Depression sites were commonly associated with Terric Haplosaprists of the poorly drained Cathro or Rifle series with O horizons approximately 30–150 cm deep (Table 1). Lowland sites were associated with lowland Histic Inceptisols of the Wildwood series, which consist of deep, poorly drained mineral soils with a thin O horizon (
To characterize the microtopography of our sites, we conducted a terrestrial
laser scanning (TLS) campaign from 20 to 24 October 2017. We chose this period to ensure high-quality TLS acquisitions, as it coincided with the time of least vegetative cover and the least likelihood for inundated conditions. During scanning, leaves from all deciduous canopy trees had fallen and grasses had largely senesced. Standing water was present at portions of three of the sites and was typically dispersed across the site in small pools (ca. 0.5–2 m
To validate the TLS surface model products, we installed sixty 2.54 cm radius spheres on fiberglass stakes exactly 1.2 m above ground surface at each site. Using the validation locations, we could easily calculate the exact surface elevation (i.e., 1.2 m below a scanned sphere) of 60 points in space. We installed 39 (13 at each plot) validation spheres at points according to a random walk sampling design, and placed 21 (7 at each plot) validation spheres on distinctive hummock–hollow transitions. We placed the 1.2 m tall validation spheres approximately plumb to reduce errors due to horizontal misalignment.
We processed the point clouds generated from the TLS sampling campaign to generate two products: (1) site-level 1 cm resolution ground surface models, and (2) site-level delineations of hummocks and hollows. The details and validation of this method are described completely in Stovall et al. (2019), but a brief summary is provided here.
For each site, we first filtered the site-level point-clouds in the CloudCompare software (Othmani et al., 2011) and created an initial surface model with the absolute minima in a moving 0.5 cm grid. We removed tree trunks from this initial surface model using a slope analysis and implemented a final outlier removal filter to ensure all points above ground level were excluded. Our final site-level surface models meshed the remaining slope-filtered point cloud using a local minima approach at a 1 cm resolution. We validated this final 1 cm surface model using the 60 validation spheres per site.
Before we analyzed surface models from each site, we first detrended sites
that exhibited site-scale elevation gradients (e.g., 0.02 cm m
We classified the final surface model into two elevation categories: hummocks and hollows. We first classified hollows using a combination of normalized elevation and slope thresholds; hollows have less than average elevation and less than average slope. This combined elevation and slope approach avoided confounding hollows with the tops of hummocks as the tops of hummocks are typically flat or shallow sloped. We removed hollows and used the remaining area as our domain of potential hummocks.
Within the potential hummock domain, we segmented hummocks into individual features using a novel approach – TopoSeg (Stovall et al., 2019) – and thereby created a hummock-level surface model for each site. We first used the local maximum (Roussel and Auty, 2018) of a moving window to identify potential microtopographic structures for segmentation. The local maximum served as the “seed point” from which we then applied a modified watershed delineation approach (Pau et al., 2010). The watershed delineation inverts convex topographic features and finds the edge of the “watershed”, which in our case are hummock edges. The defined boundary was used to clip and segment hummock features into individual hummock surface models.
For each delineated hummock within each site, we calculated the perimeter length, total area, volume, and height distributions relative to both local hollow datum and to a site-level datum. To calculate area, we summed the total number of points in each hummock raster multiplied by the model resolution (1 cm
To validate the hummock delineation, we compared manually delineated and automatically delineated hummock size distributions at one depression site (D2) and one transition site (T1), both with clearly defined hummock features. We omitted using a lowland site for validation because none of these sites had obvious hummock features that we could manually delineate with confidence. We manually delineated hummocks for the D2 and T1 sites with a qualitative visual analysis of raw TLS scans using the clipping tool in CloudCompare (2018). Stovall et al. (2019) found no significant differences between the manual and automatically segmented hummock distributions, and feature geometry had an RMSE of less than approximately 20 %.
After the automatic delineation procedure and subsequent validation, we
performed a data cleaning procedure by manually inspecting outputs in the
CloudCompare software. We eliminated clear hummock mischaracterization that
was especially prevalent at the edges of sites, where point densities were
low. We also excluded downed woody debris from further hummock analysis
because, although these features may serve as nucleation points for future
hummocks, they are not traditionally considered hummocks and their distribution does not relate to our broad hypotheses. Finally, we excluded
delineated hummocks that were less than 0.1 m
Validation of surface models using the validation spheres indicated that
surface models were precise (RMSE of
Hummocks delineated from our algorithm were generally consistent in
distribution and dimension with manually delineated hummocks. However, the
automatic delineation located hundreds of small (
To address our hypothesis that hydrology is a controlling variable of
microtopographic expression in black ash wetlands, we instrumented all 10 sites to continuously monitor water level dynamics and precipitation. Three sites (L1, L2, and L3; Slesak et al., 2014) were instrumented in 2011 and seven in June 2016 following the same protocols. At each site, we placed a fully slotted observation well (schedule 40 PVC, 5 cm diameter, 0.025 cm wide slots) at approximately the lowest elevation; at the flatter L sites, wells were placed at the approximate geographic center of each site. The ground surface at the well served as each site's datum (i.e., elevation
To quantify the control that underlying mineral layer microtopography has on surface microtopography, we conducted synoptic measurements of mineral layer depth and thus organic soil thickness at each site. Within each of the 10 m diameter plots used for TLS at each site, we took 13 measurements (co-located with the randomly established validation spheres) of depth-to-mineral-layer using a steel 1.2 m rod. At each point the steel rod was gently pushed into the soil with consistent pressure until resistance was met and the depth to resistance was recorded (resolution of 1 cm) as the “depth-to-mineral-layer”. We then associated each of these depth-to-mineral-layer measurements with a soil elevation based on TLS data and the site-level datum (i.e., elevation at the base of each site's well).
We calculated simple hydrologic metrics based on the 3 years (2016–2018) of water level data for each site. For each site, we calculated the mean and variance of water level elevation relative to ground surface at the well, where negative values represent belowground water levels and positive values indicate inundation. We also calculated the average hydroperiod of each site by counting the number of days that the mean daily water level was above the soil surface at the well each year, and averaging across years.
Our first line of inquiry was to evaluate the general spatial distribution
of elevation at each site. We first calculated site-level omnidirectional
and directional (0, 45, 90, and 135
Our second line of inquiry was to evaluate the degree of elevation bimodality in these systems, which is indicative of a positive feedback between hummock growth and hummock height (Eppinga et al., 2008). Based on the classification into hummock or hollow from our delineation algorithm, we plotted site-level detrended elevation distributions for hummocks and hollows and determined a best-fit Gaussian mixture model with Bayesian information criteria (BIC) using the “mclust” package (Scrucca et al., 2016) in R (R Core Team, 2018), which uses an expectation-maximization algorithm. Mixture models were allowed to have either equal or unequal variance, and were constrained to a comparison of bimodal versus a unimodal mixture distribution.
We assessed the importance of mineral layer microtopography on soil surface
microtopography by comparing the depth-to-mineral-layer measurements with the soil surface elevation TLS measurements. We first calculated the elevation of the mineral layer relative to each site-level datum by subtracting the depth-to-mineral-layer measurement from its co-located soil elevation measurement estimated from the TLS campaign. We then plotted the depth-to-mineral-layer measurement (hereafter referred to as “organic soil
thickness”) as a function of this mineral layer elevation, noting which
points were on hummocks or hollows as determined from the TLS delineation
algorithm. We fit linear models to these points and compared the regression
slopes to the expected slopes from (1) a scenario where surface microtopography is simply a reflection of subsurface microtopography (slope of 0, or constant organic soil thickness), and (2) a scenario of flat soil surface where organic soil thickness negatively corresponds to varying mineral layer elevation (slope of
To test our hypothesis that hydrology is a broad, site-level control on hummock height, we first regressed site mean hummock height against site mean daily water level. We also conducted a within-site regression of individual hummock heights against their local mean daily water level. To do so, we first calculated a local relative mean water level for each delineated hummock location by subtracting the elevation minimum of the hummock (i.e., the elevation at the base of the hummock) from the site-level mean water level elevation. This calculation assumes that the water level is flat across the site, which is likely valid for the high permeability organic soils at each site, low slopes (
To test whether there was regular spatial patterning of hummocks at each
site, we compared the observed distribution of hummocks against a theoretical distribution of hummocks subject to complete spatial randomness (CSR) with the R package “spatstat” (Baddeley et al., 2015). We first extracted the centroids and areas of the hummocks using TopoSeg (Stovall et al., 2019) and created a marked point pattern of the data. Using this point pattern, we
conducted a nearest-neighbor analysis (Diggle, 2002), which evaluates the
degree of dispersion in a spatial point process (i.e., how far apart on average hummocks are from each other). If hummocks are on average further
apart (using the mean nearest-neighbor distance,
To test the prediction that hummock sizes are constrained by patch-scale
negative feedbacks, we plotted site-level rank-frequency curves (inverse
cumulative distribution functions) for hummock perimeter, area, and volume.
These curves trace the cumulative probability of a hummock dimension
(perimeter, area, or volume) being greater than or equal to a certain value
(
Hydrology varied across sites, but largely corresponded to hydrogeomorphic
categories (Table 2). Depressions sites were the wettest sites (mean daily water level of
Daily water level summary statistics for black ash study wetlands.
Semivariograms demonstrated much more pronounced elevation variability at
depression and transition sites than at lowland sites (Fig. 4). In general, lowland sites reached overall site elevation variance (sills, horizontal dashed lines) within 5 m, but best-fit ranges (dotted vertical lines in Fig. 4) were less than 1 m. In contrast, best-fit semivariogram ranges for
depression and transition sites were several times greater. Therefore,
depression and transitions sites have much larger ranges of spatial
autocorrelation for elevation than lowland sites. Semivariograms were all best fit with Matérn models with Stein parameterizations, and nugget
effects were extremely small in all cases (average
Omnidirectional semivariograms for site elevations by hydrogeomorphic category (D refers to depression, L refers to lowland, and T refers to transition). Sites are colored according to their number within their hydrogeomorphic category. Dotted vertical lines indicate best-fit ranges, and horizontal dashed lines indicate best-fit partial sills (sill – nugget).
We observed bimodal elevation distributions at every site, with hummocks
clearly belonging to a distinct elevation class separate from hollows (Fig. 5). Bimodal mixture models of two normal distributions were always a
better fit to the data than unimodal models based on BIC values. Differences
in mean elevations between these two classes ranged from 12 cm at the
lowland sites to 20 cm at depression sites, and hummock elevations were more
variable than hollow elevations across sites. Across sites, 27 %
Relative elevation probability densities for each site, colored by
hummock and hollow. The text indicates the difference in mean elevation (
Across sites, organic soil thickness varied and was greatest at the lowest
mineral layer elevations, indicating that surface microtopography is not simply a reflection of subsurface mineral layer topography with constant
overlying organic thickness (as illustrated with by the dotted “subsurface reflection” line in Fig. 6). In contrast, at most sites, except for D1 and L2, there was a strong negative linear relationship between soil thickness and mineral layer elevation, with five sites exhibiting slopes near
Organic soil thickness (measured as depth to resistance) as a function of mineral layer elevation. Points are filled by their microsite.
The dashed (
We observed a significant (
Hummock height as a function of mean water level.
Within sites, we also observed clear positive relationships between individual hummock heights and their local mean daily water level (Fig. 7b). At all but two of the sites (D4 and L1), individual hummock
heights within a site were significantly (
All sites characterized as depressions or transitions exhibited a significant
(
Hummock nearest-neighbor distance distributions across sites. Bars are scaled density histograms overlaid with best-fit normal distributions (red lines). The text indicates the mean nearest-neighbor distance (
Hummock dimensions (perimeter, area, and volume) were strongly lognormally
distributed across sites (Fig. 9), although exponential models were typically only slightly worse fits. For each hummock dimension, site fits were similar within site hydrogeomorphic categories, but drier lowland site distributions were clearly different from wetter depression and transition site distributions, which were more similar (Fig. 9). Lowland sites had significantly lower (
Inverse cumulative distributions of hummock dimensions (perimeter, area, and volume) across sites (points), split by hummock dimension and site type. The
We tested our hypothesis that microtopography in black ash wetlands self-organizes in response to hydrologic drivers (Fig. 1) using an array of commonly used diagnostic tests from landscape ecology, including analyses of multimodal elevation distributions, spatial patterning, and patch size distributions. We further analyzed the influence of hydrology on these diagnostic measures and tested a potential null hypothesis that surface microtopography was simply a reflection of subsurface microtopography. Diagnostic test results of elevation bimodality, hummock spatial overdispersion, and truncated hummock areas along with clear hydrologic influence on microtopographic structure provide strong support for our hypothesis.
Bimodal soil elevation distributions at all sites suggest that the microsite separation into hummocks and hollows is a common attribute of black ash wetlands. Soil elevation bimodality was most evident at the wetter depression and transition sites, where hummocks were more numerous and occupied a higher fraction of the overall site area (15 %–20 %). Sharp boundaries between hummocks and hollows were not always observed in soil elevation probability densities (Fig. 5), which may be indicative of weak positive feedbacks between primary productivity and elevation (Rietkerk et al., 2004; Fig. 1). Conversely, modeling predictions indicate that if evapoconcentration feedbacks (i.e., that hummocks harvest nutrients from hollows through hydraulic gradients driven by hummock–hollow ET differences) are strong, boundaries between hummocks and hollows will be less sharp (Eppinga et al., 2009), possibly implicating hummock evapoconcentration as an additional feedback to hummock maintenance (Fig. 1). Greater levels of soil chloride in hummocks relative to hollows in these systems may be an additional layer of evidence for this mechanism (Diamond et al., 2019).
We also observed clear evidence of decoupling between surface microtopography and mineral layer microtopography at all of our sites. Hollows were best represented by a smooth surface model, with a relatively constant surface elevation despite variable underlying mineral soil elevation. Importantly, we also observed that regardless of underlying mineral layer, hummocks had greater soil thickness than hollows (Fig. 6). That is, irrespective of mineral layer microtopography, hummocks are maintained at local elevations that are higher than would be predicted for a smooth soil surface. Moreover, drier lowland (L) sites had less clear patterns in this regard than the wetter depression (D) or transition (T) sites, supporting our hypothesis for hydrology driven hummock development. We also note that some measurement locations had deeper organic soils than we could measure with our rod (particularly at our wettest sites) and that this is likely further evidence for our contention that hummocks are self-organized mounds on a smooth surface of organic soil, rather than an argument against it. Smoothing of soil surfaces relative to variability in underlying mineral layers or bedrock is observed in other wetland systems where soil creation is dominated by organic matter accumulation (e.g., the Everglades; Watts et al., 2014). This implies that deviations from these smooth organic soil surfaces are related to other surface-level processes, such as spatial variation in organic matter accumulation resulting from hypothesized elevation–productivity feedbacks.
Hummock heights relative to mean site-level water level were approximately 30 cm, aligning with field observations of relatively constant hummock height within sites. Generally consistent hummock height across sites in conjunction with clear bimodality in soil elevations supports the contention that hummocks and hollows are discrete, self-organized ecosystem states (sensu Watts et al., 2010). However, variability in site-level hummock heights – especially at depression and transition sites – may partially be attributable to hummocks in nonequilibrium states. From our feedback model (Fig. 1), it seems reasonable that within a site, some hummocks may be in growing states (e.g., increasing in height over time via the elevation–productivity positive feedback) and some may be in shrinking states if hydrologic conditions have recently become drier (e.g., decreasing in height via the elevation–respiration negative feedback), the combination of which may result in a distribution of hummock heights centered around an equilibrium hummock height. Future efforts could leverage time-series observations of hummock properties (e.g., area, height, and volume), but we note the likely decadal timescales required to detect hummock growth or shrinkage (Benscoter et al., 2005; Stribling et al., 2007).
Local hydrology exhibited clear control on hummock height, providing evidence for our hypothesis that hummocks are a biogeomorphic response to hydrologic stress in wetlands. We found support for this contention at both the site level and the hummock level. The tallest hummocks were consistently located at the wettest sites and in the wettest zones within sites. At the site scale, 85 % of the variance in the average hummock height could be explained by the mean water level alone. Within sites, the local mean water level explained 35 % of the variability in hummock height on average (Fig. 7); the prevalence of nonequilibrium hummock states may explain much of the additional variability. The considerable variation in the ability of local water levels to explain hummock height within sites (adjusted
We found clear support for our hypothesis that hummocks are non-randomly distributed in our wettest study sites. Hummocks exhibited spatial overdispersion at all sites, but this overdispersion was only significant at depression and transition sites (Fig. 8). Significant spatial overdispersion indicates regular hummock spacing in contrast to clustered distributions or completely random placement. Regular patterning of landscape elements is observed across climates, regions, and ecosystems (Rietkerk and Van de Koppel, 2008), and is indicative of negative feedbacks that limit patch expansion (Quinton and Cohen, 2019). Our results indicate similar patterning for forested wetland microtopography and, importantly, demonstrate the hydrologic controls on that patterning. Hydrology appears to be a common driver in regular pattern formation in wetlands (Heffernan et al., 2013) and drylands (Scanlon et al., 2007). Thus, water stress – both too much (Eppinga et al., 2009) and too little (Deblauwe et al., 2008; Scanlon et al., 2007) – appears to be an important regulator of patch distribution across the landscape.
We observed lognormal hummock size distributions, suggesting that some hummocks may attain very large areas (i.e., over 10 m
Characteristic hummock sizes in association with overdispersion in black ash wetlands suggest that hummocks are laterally limited in size by negative feedbacks on the scale of meters (Manor and Shnerb, 2008). We posit that there are two patch-scale negative feedbacks: (1) overstory competition for nutrients and (2) understory and overstory competition for light. Hummocks associated with black ash trees, which account for more than 85 % of measured hummocks, are likely limited in area by the radial growth of the trees' root systems. Evapoconcentration feedbacks bring nutrients to the tree roots, limiting the degree to which roots must search for them (Karban, 2008), and therefore limiting root lateral expansion. Indeed, evidence suggests that a majority of fine tree roots occur within hummocks in forested wetland systems (Jones et al., 1996, 2000). Moreover, finite nutrient pools may lead to development of similarly sized nutrient source basins for each hummock, further limiting lateral hummock expansion (Rietkerk et al., 2004; Eppinga et al., 2008). Black ash trees must also compete for light with other ash trees, but leaf area is typically low in these systems (Telander et al., 2015). Low LAI and observed crown shyness (sensu Long and Smith, 1992) in black ash wetlands may imply less competition among individuals than would be expected in mixed stands (Franco, 1986). Conversely, lower than expected canopy competition for light in the overstory may increase light availability for understory hummock species, and allow subsequent hummock expansion from the understory. Therefore, based on evidence and observations presented here and in Diamond et al. (2019), we suggest that a major difference between microtopography in forested versus non-forested wetland systems will be the size distributions and spacing of hummocks. In other forested systems, hummocks associated with trees will likely be limited in size, exhibiting characteristic sizes and spacing due to local negative feedbacks from the crown competition. In contrast, non-forested wetland hummocks may have a much wider distribution of size classes, where negative feedbacks to hummock expansion may be largely due to local nutrient competition effects (e.g., Eppinga et al., 2008).
In this work, we used common landscape ecology diagnostics to characterize
microtopographic patterns and infer the responsible reinforcing processes, including analyses of multimodal distributions of elevation, spatial patterns of hummock patches, and hummock size distributions. Other recent work has used nearly identical diagnostic measurements to infer self-organization of depressional wetland features (
The consequences of wetland microtopography are clear at small scales, but can also scale to influence site- and regional-scale processes. For example, microtopographic expression results in a drastic increase in surface area within wetlands. We conservatively estimate an average of 22 % and up to a 42 % relative increase in surface area due to the presence of hummocks (i.e., additional surface area provided by the sides of hummocks; Table 3). These estimates comport with studies in tussock meadows, where tussocks (ca. 20 cm tall) increased surface area by up to 40 % (Peach and Zedler, 2006). Furthermore, increases in the diversity of biogeochemical processes occurring at the individual hummock or hollow scale (Deng et al., 2014) likely aggregate to influence ecosystem functioning at large scales. For example, microtopographic niche expansion allows for local material and solute exchange between hummocks and hollows, creating coupled aerobic–anaerobic conditions with emergent outcomes for denitrification (Frei et al., 2012) and carbon emission (Bubier et al., 1995; Minick et al., 2019a, b).
Relative area increase by hummocks across sites.
While our results implicate hydrology as a major determinant of microtopographic structure and pattern, microtopography can reciprocally influence system-scale hydraulic properties. Results from our hummock property analysis indicate that hummock volume displacement may be a significant factor in water level dynamics of wetlands. Specific yield, which governs the water level response to hydrologic fluxes, is commonly assumed to be unity when wetlands are inundated. However, inclusion of microtopography may render this assumption invalid, with hummock volumes up to 30 % of site volumes (Table 4). These observations are supported in other studies of microtopographic effects of specific yield (Sumner, 2007; McLaughlin and Cohen, 2014; Dettmann and Bechtold, 2016). Therefore, while hydrology exerts clear control on the geometry of hummocks, hummocks may exert reciprocal control on hydrology by amplifying small hydrologic fluxes into large water level variations.
Hummock volume displacement ratios for all sites.
Last, black ash hummocks provide unique microsite conditions that support increased vegetation growth and diversity (Diamond et al., 2019), aligning with observations in other wetland systems (Bledsoe and Shear, 2000; Peach and Zedler, 2006; Økland et al., 2008). Accordingly, recent wetland restoration efforts have begun to use microtopography as a strategy to promote seedling success and long-term project viability (Larkin et al., 2006; Bannister et al., 2013; Lieffers et al., 2017). Specific to our focal system, there are increasing efforts to mitigate potential black ash loss due to the emerald ash borer and possible regime shifts to marsh-like states (Diamond et al., 2018). We posit that hummock presence and persistence may allow for future tree seedlings to survive wetting up periods following this ash loss (Slesak et al., 2014), and for consequent resilience of forested ecosystem states.
Overall, this study adds to the growing body of evidence that the structure and regular patterning of wetland microtopography is an autogenic response to hydrology. Although the imprint of biota on landscapes may be masked by the signature of larger-scale physical processes (Dietrich and Perron, 2006), we show clear evidence here for a microtopographic signature of life.
Code for analysis and figure creation is available at
The supplement related to this article is available online at:
JSD and DLM created the conceptual framework, questions, and hypotheses. AS and JSD developed the TLS procedure and carried out measurements and subsequent analysis/coding; JSD and RAS carried out hydrology measurements. JSD conducted all data analysis and wrote the paper. All co-authors contributed significantly to editing the paper.
The authors declare that they have no conflict of interest.
We gratefully acknowledge the field work and data collection assistance provided by Mitch Slater, Alan Toczydlowksi, and Hannah Friesen. The authors also acknowledge two anonymous reviewers and Victor Lieffers, whose comments and suggestions improved this paper.
This project was funded by the Minnesota Environmental and Natural Resources Trust Fund, the USDA Forest Service Northern Research Station, and the Minnesota Forest Resources Council. Additional funding was provided by the Virginia Tech Forest Resources and Environmental Conservation department, the Virginia Tech Institute for Critical Technology and Applied Science, and the Virginia Tech William J. Dann Fellowship. Jacob S. Diamond is supported by POI FEDER Loire no. 2017-EX001784, the Water Agency of Loire Catchment AELB, and the University of Tours.
This paper was edited by Sally Thompson and reviewed by two anonymous referees.