Edward A. Johnson

University of Calgary

Kiyoko Miyanishi

University of Guelph

INTRODUCTION

Natural or anthropogenic disturbance was traditionally viewed as an event that initiated primary or secondary succession, and succession explained the development of vegetation in the absence of disturbance. Thus, the concepts of disturbance and succession are inextricably linked in plant ecology.

Succession has been used in so many different ways and situations that it is almost useless as a precise idea. However, no matter whether succession has been considered a population (Peet and Christensen, 1980), community (Cooper, 1923a; 1923b; Clements, 1916), or ecosystem (Odum, 1969) phenomenon or process, it has contained certain common ideas. Succession is an orderly unidirectional process of community change in which communities replace each other sequentially until a stable (self-reproducing) community is reached (see definitions in Abercrombie et al. 1973; Small and Witherick, 1986; Allaby, 1994). The explanation of why and how succession is directed has changed over its more than hundred-year history, but most arguments share the notion that species are adapted to different stages in successions and in some way make the environment unsuited for themselves and more suited for the species in the next stage. This group selection argument was first instilled into succession in the Lamarckian ideas of Warming, Cowles, and Clements.

Succession arose at the end of the 1800s and early 1900s out of a naturalist observation tradition when quantitative methods were almost nonexistent, Aristotelean essentialism (Hull, 1965a, 1965b; Nordenskiöld, 1928) still had a firm grip on how nature should be understood, and meteorology, soil science, biology, and geology were very poorly developed. Further, and equally important, spatial and temporal scales of observation were limited to the scale of a naturalist’s sight.

DISTURBANCE AS THE NEMESIS OF SUCCESSION

By the beginning of 2000, most of the original classical examples of succession (e.g., Cowles, 1899; Shelford, 1911; Cooper, 1923b) given in textbooks had been restudied and found not to support the original arguments.

The first example in North America of primary succession was that on sand dunes (Cowles, 1899). The spatial sequence of plant communities as one moves away from the lake was interpreted by Cowles (1899; 1901) and Clements (1916) as representing a temporal succession of communities from dune grasses to cottonwoods, then pines and oaks to the climax beech-sugar maple forest (see Fig. 1 in Chapter 8). Olson’s (1958) study of the same dunes using techniques that allowed actual dating of the dunes produced a much more complex picture of community changes than the previously proposed simple sequence from grasses to mesophytic forest. Olson found that dunes of similar age supported a wide range of plant communities, depending on the location as well as the disturbance history of the site.

A second classic example of primary succession was that on glacial till left by the retreating glacier at Glacier Bay, Alaska (Cooper, 1923a; 1923b; 1926; 1931; 1939). Again, the spatial pattern of vegetation on areas deglaciated at varying times was interpreted as representing the temporal stages of communities through which each site would pass from herbaceous Dryas and Epilobium to shrubby willow and alder thickets, then Sitka spruce forest, and finally the spruce-hemlock climax forest. Subsequently, Crocker and Major’s (1955) study of soil properties at the different aged sites concluded that occupation of each site by the shrubs, particularly the nitrogen-fixing alders, allowed subsequent establishment of the later successional tree species through soil alteration (changes in pH and addition of carbon and nitrogen). However, Cooper’s original study sites were reexamined by Fastie (1995), who found that the tree ring record from spruces in the oldest three sites did not indicate early suppression of growth with subsequent release once the spruces had exceeded the height of the alder or willow canopy. In other words, these oldest sites apparently had not experienced a succession from a community dominated by alders and willows to one dominated by spruce. Furthermore, the oldest sites showed a much more rapid colonization by a dense stand of trees soon after the sites were deglaciated compared to the younger sites. The differences between the different aged sites in their vegetation history (i.e., the order and rate of species establishment) as shown by Fastie’s reconstructions were explained primarily by the availability of propagules (distance to seed source) at the time the retreating ice exposed the bare substrate. Interestingly, Cooper (1923b) had also noted that “establishment of the climax does not depend upon previous dominance of alder, for in the areas of pure willow thicket the spruces were found to be invading with equal vigor,” “almost any plant of the region may be found among the vanguard,” and “even the climax trees make their first appearance with the pioneers.” Despite such observations, the lasting legacy of the early studies of primary succession at Glacier Bay has been the classic successional idea of sequential invasion and replacement of dominants driven by facilitation. As Colinvaux stated in his 1993 textbook Ecology 2: “The record from Glacier Bay shows that a spruce-hemlock forest cannot grow on the raw habitat left by the glacier, but that spruce trees and hemlock can claim habitats that have first been lived on by pioneer plants and alder bushes…. [I]t is undeniable that primary succession on glacial till at Glacier Bay is driven by habitat modification.”

A third example of primary succession was the hydrarch succession of bogs and dune ponds. As with both of the previous examples, the spatial pattern of vegetation outward from the edge of bogs was interpreted as representing successional stages, leading to the conclusion by Clements (1916) that the open water would eventually become converted to a mesophytic forested site. However, the paleoecological reconstruction by Heinselman (1963) of the Myrtle Lake bog in Minnesota indicated that, despite deposition of organic matter and mineral sediments into the bog since deglaciation, the open water has persisted and has not been filled in and invaded by the surrounding forest because of the rising water table with the accumulation of peat.

Shelford (1911; 1913) used the spatial sequence of ponds in the Indiana Dunes of Lake Michigan to develop a model of temporal change in vegetation resulting from hydrarch succession. Jackson et al. (1988) tested this classic hydrosere model by using paleoecological data spanning 3000 years and found no evidence of significant change in vegetation until the early 1800s, when rapid change occurred following European settlement. They concluded that the spatial differences in vegetation along the chronosequence reflected differential effects of disturbance rather than any temporal successional pattern.

A fourth example, this time of secondary succession in forests, by Stephens (1955) and Oliver and Stephens (1977) concerned whether forest canopy composition resulted from the continuous recruitment of new stems of more shade-tolerant species. What they found in the old, mixed-species, northern hardwoods Harvard Forest in Massachusetts was the overriding influence of small- and large-scale disturbances, both natural and anthropogenic. While small disturbances allowed release of suppressed understory trees that might otherwise never make it to the canopy, large disturbances resulted in seedling establishment of new trees. Thus, the canopy composition was determined by disturbance processes. Foster (1988) came to the same conclusions about the old-growth Pisgah Forest in New Hampshire.

Poulson and Platt’s (1996) long-term study of Warren Woods, the classic example of a climax beech-maple forest (Cain, 1935), led them to conclude that natural disturbances were chronic, occurring dependably on an ecological time scale and producing continual changes in light regimes. Because tree species respond differentially to the changing light conditions, different species are favored under different light regimes. Thus, the relative abundance of tree species in the understory at any given time cannot be used to predict the composition of the canopy at some later time. Poulson and Platt (1996) presented data to show that the relative abundance of beech and maple (as well as other species) in the canopy fluctuates in response to spatial and temporal fluctuations in frequency and sizes of treefall gaps. This is despite Cain’s (1935) tentative conclusion, based on the abundant maple reproduction he observed, that “maple seems destined to increase in importance.” In other words, the system is neither in, nor tending toward, an equilibrium climax community dominated by the most shade-tolerant species growing from...