The pages of this journal are having a lively exchange of views on the idea of Potential Natural Vegetation (PNV), starting with a critique of the concept (Chiarucci et al. 2010). The discussion has been centred on examples and applications from Europe, and has not made full use of paleoecological records. Those records, particularly from relatively undisturbed settings, can reveal how natural vegetation actually behaves, and whether (and at what scales) the PNV concept might correspond to actual vegetation.
Two tenets underlie applications of PNV (Küchler 1964, 1967; Härdtle 1995). First, PNV must represent vegetation in the absence of human activity – hence the “natural” component. Second, PNV must represent vegetation in a mature state. This may be rephrased as the vegetation that would develop and persist on a site in the long-term absence of disturbance (e.g. fire) and ecologically significant climate change. Thus, PNV is analogous to ‘climax’ vegetation (Küchler 1964, 1967) insofar as it represents the expected stable endpoint of successional processes – i.e. “mature” vegetation. In that context, PNV bears strong similarity to the ‘typal community’ concept of Daubenmire (1952, 1966), whereby for forests, the community type would comprise the most shade-tolerant and long-lived tree species capable of occupying a site for given local conditions (slope, aspect, soils). Strict application of the ‘mature’ or ‘climax’ requirement often yields a substantial difference between ‘actual natural’ and “potential natural” vegetation. For example, Küchler (1964) and Grossman et al. (1998) mapped presumed climax types (Picea–Abies forest in the Rocky Mountains and mixed hardwood–Pinus elliottii forest in the southeastern Coastal Plain) in regions where fire-dependent forests (respectively, Pinus contorta forests and P. palustris forests) grow today.
As Küchler (1967) observed, North America seems ideally suited for PNV application, because of the dampened preindustrial human imprint and the continued existence of extensive natural vegetation in many regions to provide reference conditions. Although humans have occupied North America for at least the last 12 000 years, populations were relatively small and diffuse until ca. 1000 CE (Munoz et al. 2010), and in nearly all of North America north of central Mexico they had very little influence on vegetation in pre-Columbian times (Vale 2002). In extensive areas, including the mixed conifer–hardwood forests of the Great Lakes–St. Lawrence corridor and the conifer forests and woodlands of the Rocky Mountains, pre-Columbian human societies were hunter-forager-gatherer, supplemented in some cases with cultivation on relatively restricted spatial and temporal scales. Thus, paleoecological studies of pre-Columbian vegetation in these regions can provide insights into the natural response of vegetation to natural forcings, particularly climate variation and change. Does the natural vegetation observed in paleoecological records represent PNV?
Vegetation history in North America
Post-glacial paleoecological records from throughout North America show substantial, repeated changes in vegetation composition and structure (Webb 1988; Betancourt et al. 1990; Williams et al. 2004). Few locales have not experienced substantial vegetational turnover in the past 10,000 years, and all have undergone transformation since the last glacial maximum. Many modern plant communities are hundreds to a few thousand years old, with no antecedents anywhere on the continent (Jackson 2006; 2012). Many communities of the past have no modern counterpart (Williams et al. 2001). For example, late-glacial vegetation (ca. 17,000–12,000 years ago) of the mid-continent was dominated by Picea, Fraxinus, Ostrya–Carpinus and Quercus, unlike any vegetation of the North American Holocene (Williams et al. 2001; Gill et al. 2009), at a time when climate also had no modern counterpart. Clearly, natural vegetation evolves in response to change, in the long-term.
Viewed at a subcontinental scale and millennial perspective, this vegetational flux represents the tracking of climatically determined targets that change through time (Webb 1986). Under this view, vegetation at any given time is determined by the prevailing climate (contingent on local site conditions), and when the climate shifts (e.g. increasing or decreasing temperature or precipitation, or shifting seasonality), vegetation composition adjusts accordingly. Transient disequilibrium may arise during or just following periods of climate change, but given sufficient time, vegetation should settle into an equilibrium state with prevailing climate, based on individualistic responses of plant species to the environment and the competitive matrix in which they are embedded. Support for this perspective comes from the spatial and temporal coherence of Holocene pollen assemblages: pollen assemblages persist at individual sites for extended time periods, and are spatially autocorrelated (Webb 1988; Williams et al. 2004). A series of 1000-year time-slice maps of pollen-assemblage composition and vegetation type (Williams et al. 2004) can be viewed as successive, specific realizations of natural vegetation, and presumably of PNV, throughout the Holocene. As climate changed, so did PNV, and vegetation shifted accordingly to match the PNV determined by the climate. Based on the argument presented in Webb (1986), PNV could be perceived as the equilibrium vegetation under a particular climate regime given a particular regional flora (and absence of extraneous disturbance by humans or other factors).
Although PNV may correspond to actual vegetation when spatial and temporal scales are broad and smoothed (e.g. subcontinental and millennial), disturbance-dependent alternative vegetation realizations often persist at these scales. For example, Pinus contorta forests have persisted for ten millennia over extensive parts of the Rocky Mountains (Whitlock 1993), although from Küchler's argument, PNV of these regions would have been Picea–Abies forest throughout this period. The P. contorta forests have been maintained by a natural, high-frequency, stand-replacing fire regime under the climate of this period (Millspaugh et al. 2000), leading to a persistent, multi-millennial divergence between actual natural vegetation and PNV across a vast area.
Further challenges to PNV arise at temporal and spatial scales closer to what is experienced by humans and human societies–i.e. at local to regional spatial scales and annual to multi-centennial temporal scales. Accumulating paleoclimate evidence indicates that ecologically significant climate variability occurs at timescales ranging from years to millennia, and that abrupt climate changes can occur (Jackson et al. 2009; Shuman et al. 2009a). Climate variability at relatively high frequencies (annual to multi-decadal) imposes legacy effects and contingencies on ecological systems: the state of an ecosystem at a particular time is a function not only of its prevailing environment, but of its history (and the history of the environment) (Jackson et al. 2009). Actual vegetation is often a contingent outcome of ecological processes (mortality, disturbance, recruitment, competition, herbivory) superimposed on a highly variable environment.
A consequence of this historical imprint is that multiple stable equilibrium states–even “climax” states–may exist on identical sites under a particular “average” climate. These multiple states may persist for decades to hundreds of years, or more. Recent paleoecological studies of species migrations illustrate this phenomenon, revealing that migrations are paced by climate variability, and that a species range at any particular time may be a manifestation of its particular migration history as well as the prevailing environment (Jackson et al. 2009). For example, Juniperus osteosperma has been expanding its range across the Bighorn Basin in the central Rocky Mountains for the past 5000 years (Lyford et al. 2003). In this process, scattered J. scopulorum individuals are replaced by high-density J. osteosperma woodlands, leading to changes in ecosystem properties. The migration consisted of rapid colonization and expansion phases, alternating with extended quiescent periods, during which established populations persisted, but new sites were not colonized (Lyford et al. 2003). The Bighorn Basin includes extensive suitable habitats that remain uncolonized by J. osteosperma; these are currently occupied by grassland and steppe with low-density J. scopulorum, but could easily be replaced by J. osteosperma woodlands given sufficient time and climate variability to allow colonization. Gray et al. (2006) discuss a similar case for establishment of Pinus edulis woodlands.
Transitions from one vegetation state to another can be triggered by a number of factors, including immigration or extirpation of dominant species, transient disturbances (or cessation of disturbances) and changes in herbivory. Replacement of Quercus savanna by mesic hardwood forest in Minnesota a few centuries ago coincided with a drought interval, apparently leading to reduction in fire disturbance that allowed establishment of tree seedlings (Shuman et al. 2009b). Forests might have been established centuries before had not the prevailing disturbance regime prevented tree establishment. Gill et al. (2009, 2012) recently showed that a late-glacial transition from Picea-dominated forest to woodlands co-dominated by Picea, Quercus, Fraxinus and Ostrya–Carpinus (part of the no-analogue vegetation discussed above) did not coincide with a climate transition, but did follow immediately after a decline in megaherbivore populations. The climate at the time was capable of supporting either vegetation type, and the reduction of herbivory led to a rapid transition in vegetation composition and structure.
The longevity of many woody plants and non-woody perennials, and the time (and sometimes disturbance) required for establishment of new individuals and species, can impart vegetational inertia. Many vegetational realizations are anachronisms, representing past disturbances and establishments that occurred under a different climate to that found today. In the past 1000 years, much of North America has experience multi-decadal droughts during the Medieval Climate Anomaly, followed by the Little Ice Age, followed by 19th and 20th century warming. These and other recent events have left durable imprints on vegetation composition and structure (Webb 1981; Swetnam & Betancourt 1998; Hotchkiss et al. 2007; Shuman et al. 2009b; Booth et al. 2012). As Cowles (1901) noted, we are often confronted with “a variable pursuing a variable,” and more than one legitimately natural state sustainable under a given environment can be identified.