Long-term vegetation recovery on reclaimed coal surface mines in the eastern USA


  • Karen D. Holl

    Corresponding author
    1. Department of Environmental Studies, University of California, Santa Cruz, CA 95064, USA
      Karen D. Holl, Department of Environmental Studies, University of California, Santa Cruz, CA 95064, USA (fax 831 459 4015; e-mail kholl@ucsc.edu).
    Search for more papers by this author

Karen D. Holl, Department of Environmental Studies, University of California, Santa Cruz, CA 95064, USA (fax 831 459 4015; e-mail kholl@ucsc.edu).


  • 1While reclamation of degraded landscapes is becoming increasingly common, few studies have investigated the long-term effects of reclamation efforts on plant conservation. The goals of this study were to determine whether vegetation communities on reclaimed mines approximate those of the surrounding forest, and to evaluate how intensive reclamation practices used to address short-term erosion and water quality concerns affect long-term recovery.
  • 2In 1992–93 and 1999 the vegetation on 15 coal surface mines reclaimed between the years 1967–87, and five periodically logged hardwood forest reference sites in south-western Virginia, were surveyed.
  • 3Herbaceous species richness was similar on all sites, whereas woody species richness was higher in reference sites than in reclaimed sites.
  • 4Vegetation community composition on reclaimed sites continued to progress towards reference forest sites between the two sampling periods, but vegetation community composition even on the oldest sites (reclaimed > 35 years prior) still differed substantially from reference sites.
  • 5Herbaceous cover was higher and tree basal area was lower in reclaimed sites compared with reference sites at the first sampling period, but these differences were much less pronounced by the second sampling period.
  • 6The overall composition of the older reclaimed sites was similar to the reference forest sites, suggesting a high level of resilience in the forests studied. But, as with a number of previous studies of long-term recovery on highly disturbed sites, a number of less common forest species still had not colonized reclaimed sites, raising the question of their value for conservation.
  • 7These results concur with a common theme that goals for short-term and long-term recovery of highly disturbed sites may conflict. Planting with aggressive non-native ground cover species to minimize short-term erosion may have slowed long-term recovery on the sites studied. Widespread planting of Pinus strobus has not reduced species richness thus far, but may as the canopy closes. More emphasis should be given to long-term recovery potential in developing mine reclamation plans, and strategies to further this goal should be tested.


Efforts to reclaim degraded landscapes have increased exponentially in the past decade, but few data exist on their long-term effects on plant conservation. The term reclamation is used to describe efforts that aim to improve the quality of the land by restoring some pre-disturbance functions (Bradshaw 1984). The goal of such efforts is not usually to recreate pre-disturbance species composition. Regardless, as legal requirements to reclaim highly disturbed lands are becoming increasingly common, the potential of reclamation efforts to impact conservation of both species and ecosystem services grows.

Most legislative mandates for land reclamation require evaluating the success of such efforts after a relatively short time period, if at all (Holl & Cairns 2002). For example, the success of coal surface mine reclamation efforts in the south-eastern USA is usually evaluated after 5 years (McElfish & Beier 1990). Clearly, this time period is much shorter than that of natural succession in hardwood forest in the eastern USA. Judging success of reclamation efforts after a short period of time encourages land owners to employ strategies that maximize short-term goals, such as providing high ground cover to minimize erosion, rather than restoring a diversity of species.

The effect of initial vegetation composition on succession, particularly in old fields in the eastern USA, has long been discussed (Clements 1916; Egler 1954; Connell & Slatyer 1977; Myster & Pickett 1990) and many studies have highlighted the importance of vegetation composition at the time of abandonment on the successional trajectory (reviewed in Myster 1993). Initial vegetation may facilitate, tolerate or inhibit (sensuConnell & Slatyer 1977) establishment of later-successional vegetation. Callaway & Walker (1997) have argued that there is a tendency towards facilitation in natural recovery of highly degraded systems, such as abandoned mines. Many mine reclamation efforts focus on establishing rapid-growing non-native species that control erosion but may compete with later-successional, native species. A number of authors have suggested that these intensive reclamation efforts may inhibit long-term ecosystem recovery (Chambers, Brown & Williams 1994; Holl & Cairns 1994; Allen, Brown & Allen 2001) but there has been little research to test this hypothesis due to the short period of time since most reclamation projects were initiated.

Areas reclaimed after coal surface mining in the USA provide one of the best opportunities for studying the effect of reclamation on long-term ecosystem recovery. First, in contrast to most reclamation projects, coal surface mines have been reclaimed for more than 35 years, and reclamation has been mandated by US federal law for almost 25 years (Surface Mining Control and Reclamation Act, Public Law 95–87 Federal Register 3 Aug 1977, 445–532), although reclamation practices have changed over time, confounding temporal comparisons. Secondly, surface mining has caused extensive disturbance world-wide. Thirdly, commonly utilized coal surface mine reclamation practices are typical of those of most large-scale land reclamation projects: a few aggressive plant species are used for revegetation in an effort to achieve legal requirements for minimum ground cover and to maintain water quality. The Surface Mining Control & Reclamation Act (SMCRA) of 1977 states that mining operations shall establish ‘a diverse, effective, and permanent vegetative cover of the same seasonal variety and native to the area and capable of self-regeneration and plant succession …’, unless use of non-native species is necessary to achieve the stated post-mining land use. Only recently has it been possible to evaluate the effects of reclamation methods on long-term ecosystem recovery and conservation of native species.

In 1992 and 1993, the vegetation was surveyed on mined sites reclaimed 2–30 years prior and on reference sites in unmined periodically logged forests (Holl & Cairns 1994). This study indicated that the number of plant species was increasing with time since reclamation, and a number of forest species were colonizing reclaimed sites. The composition of the oldest reclaimed sites was approaching that of the adjacent, less-disturbed, hardwood forest, but some forest species were not present on reclaimed sites. The question remained as to whether the full complement of species would colonize over time. Moreover, results of this study and other research (Brenner, Werner & Pike 1984; Burger & Torbert 1990; Torbert & Burger 2000) suggested that some of the aggressive grasses (e.g. Festuca arundinacea) and legumes (e.g. Lespedeza cuneata) commonly seeded for reclamation in the 1980s may have slowed long-term vegetation recovery on these sites. However, it was impossible to answer this question without resurveying sites, as reclamation practices had changed over time in the sites studied (Pickett 1989). These sites were resurveyed in 1999 in order to answer the following questions. (i) Is the vegetation composition of the oldest reclaimed sites (35 year) similar to the periodically logged, surrounding forest? (ii) Is the vegetation composition in the reclaimed sites following a successional trajectory towards the surrounding forest? (iii) Is long-term vegetation recovery impeded by aggressive reclamation efforts to meet short-term legislative requirements to minimize erosion and maintain water quality?

Materials and methods

site description

This research was conducted in Wise County in south-western Virginia, USA. The sites were located on or near the Powell River Project Education Center administered by Virginia Polytechnic Institute and State University. This region receives an average annual precipitation of 1180 mm, distributed relatively evenly throughout the year, and mean temperatures range from a low of −0·1 °C in January to a high of 21·3 °C in July (National Climatic Data Center, Asheville, NC). The pre-mining soils in the area are acidic and infertile, and are derived from sandstones, siltstones and shales (Daniels & Amos 1985). The soils in the unmined hardwood forest are dominantly from the soil order Inceptisols, with well-developed litter layers and A horizon organic matter contents ranging from 2% to 4% (W. L. Daniels, personal communication). After mining, the soil and blasted rock layers are mixed. The resulting ‘spoils’ are usually coarse textured and low in nitrogen (N), phosphorous (P) and organic matter, with high spatial variability in soil chemistry (Daniel & Amos 1985; Li & Daniels 1994).

The pre-mining vegetation was typical of the majority of the central Appalachian coal bearing-region, which is covered by oak Quercus–hickory Carya forests and Appalachian mixed-hardwood forest types, with species composition reflecting local microclimate and soil type. These forests served as a refuge for moist forest species during drier glacial epochs and therefore are home to a large number of floral and faunal species.

The vegetation was surveyed on 15 quarter-hectare reclaimed mine sites, which were grouped into three age classes (n= 5 for each) for most analyses. Site ages on reclaimed sites were determined from mining permit maps and tree cores. Reclamation treatments for the sites surveyed are outlined below and discussed in detail in Holl & Cairns (1994). The sites reclaimed in 1962–67 were hand-seeded at fairly low seeding rates with non-native grasses and legumes, primarily Agrostis alba, Festuca arundinacea, Lespedeza cuneata and Trifolium pratense, as well as the native tree Robinia pseudoacacia; no efforts were made to replace soil or restore the original contour. Sites reclaimed between 1972 and 1977 were hydroseeded in a wood mulch slurry. Plant species included those used previously, as well as other non-native grasses and legumes, Setaria italica, Secale cereale and Melilotus spp. Pinus strobus, a species native to the south-eastern USA, but uncommon in the region, was planted on the top of the slopes. Sites reclaimed in 1980–87 were reclaimed to meet legal requirements, including restoring original contour and providing for c. 1000 trees per hectare and 90% herbaceous cover after 5 years. They were seeded at high rates with the plant species used previously, as well as the non-natives Dactylis glomerata, Phleum pratense, Lotus corniculatus and Trifolium repens. Pinus strobus seedlings were hand-planted throughout the sites due to their rapid growth and commercial value.

Five reference sites in the surrounding, unmined, upland mixed-hardwood forest were also surveyed. The reference forest sites, like the vast majority of this region of the country, had been logged for the past 200 years. Logging records were not available for the specific forest sites surveyed, but most of this area was heavily logged at the turn of the century and large trees have been selectively removed periodically since that time (K. Kyle, Virginia Department of Forestry, personal communication). At the time of the first survey it had been at least 9 years since the forest sites had been subjected to any logging disturbance. Two sites had some large trees removed between the two sampling periods, but vegetation surveys suggested that the overall vegetation composition had changed little. As the reference forest sites contained mixed-aged trees, the age used for analyses was determined by coring a tree in the largest age class that was represented by > 10 individuals in 1992 or 1993. At that time the oldest trees ranged from 25 to 50 years old on the different sites.

Sites were rectangular (62·5 × 40 m), with the longer side perpendicular to the slope. All sites of the same type were separated by a minimum of 0·5 km. Reclaimed sites were located 5–50 m from unmined forest. The sites surveyed ranged in elevation from 700 to 925 m above sea level. All sites were south facing, with slopes ranging from 12·5° to 42·5° and aspects ranging from 140° to 225°.

field methods

The vegetation was sampled on 16 sites during August 1992, and on the four remaining sites during August 1993. Nineteen of these sites were resurveyed in August 1999. One of the 1962–67 reclaimed sites was remined between the two sampling periods, so it was not resurveyed.

Vegetation on the sites was sampled in three vegetational strata: herb (0–0·75 m), shrub (0·75–2 m), and tree (> 2 m). The total percentage cover and percentage cover of individual herbaceous species were recorded in 16 1-m2 quadrats. The cover of individual species was ranked using the Braun–Blanquet cover-abundance scale (Müller-Dombois & Ellenberg 1974). Herb sampling quadrats were systematically distributed along four transects located perpendicular to the slope and separated by 10 m. Herbs were sampled in mid- to late August in each year, which was a time of highest cover for most species. In 1992 and 1993, sites were visited regularly throughout the growing season to collect flowering individuals for identification. Cardamine hirsuta and Galium aparine had wilted by August and therefore were not included in analyses. Species in a few genera, such as Viola spp. and Impatiens spp., could not be reliably identified to species in August and were grouped according to genus. Likewise, Acer rubrum (majority) and Acer saccharum (a few individuals) were analysed together.

Shrubs and trees were sampled in eight quadrats, 4 × 4 and 10 × 10 m, respectively, which were distributed systematically along the same transects as the herb quadrats. Total percentage cover and cover of individual species were estimated in the shrub strata. Because of the difficulty in estimating cover in the canopy, the composition of this layer was quantified by measuring the diameter at breast height (d.b.h) of trees rooted in the quadrat. Shrubs and trees were sampled in June–early July in 1992–93 and in mid-August in 1999.

Nomenclature follows Radford, Ahles & Bell (1968). Gleason & Cronquist (1991) was used as a source for original geographical ranges of species.

numerical analyses

Importance values were calculated for the species in each vegetational stratum (Müller-Dombois & Ellenberg 1974). The importance value was calculated as the sum of the relative cover and relative frequency for herbs and shrubs. For trees, the importance value was calculated as the relative number of individuals and relative basal area. Because a measure of frequency and abundance was included for each layer, the sum of importance values in each layer totalled 200. The species-by-site matrices for each layer were combined for multivariate analyses. For the few species that were recorded in multiple layers, the largest importance value from the three layers was used. Because the shrub layer in the sites studied was characterized by only a few species, these species were combined with the tree strata for calculations of species richness. Vines were included with the strata in which they were most abundantly represented.

Trends in vegetational community composition were explored with detrended correspondence analysis using PC Ord version 4.0 (McCune & Mefford 1999). In correspondence analysis, samples and species are reciprocally averaged to determine the axis that explains the most variation in species distribution; subsequent axes are calculated in the same manner subject to the constraint that they are not correlated to previous axes (Gauch 1982). Detrending was used because of the high turnover in species, which resulted in a high correlation between the second axis and the square of the first axis (Gauch 1982).

Because exact site ages were difficult to determine, ages were ranked and Spearman correlation coefficients, a non-parametric procedure, with DCA axis locations were calculated. Because the interval between sampling periods (6–7 years) was approximately that of the differences in time since reclamation in the different site types, locations along the first DCA axis (strongly correlated with time since reclamation) of older sites at the first sampling period (1992–93) were compared with the next earlier age class in the subsequent sampling period (1999); this provided a rough test of whether succession was occurring at the same rate in different site types. Because samples were repeated on the same sites at two points in time, species richness and total cover measurements in the understorey and overstorey were compared using univariate repeated measures analysis (rmanova, SAS version 8·01; SAS Institute 2000), with sampling period as the repeated effect and site type as the other main effect. When there was a significant interaction between sampling period and site type, changes over time within specific site types were compared using paired t-tests, and means by site type within year were compared using Tukey's studentized range test (α = 0·05). Data were log-transformed when necessary to meet assumptions of normality and homogeneity of variance.

A ranking system was used to quantify less common species that tend to be downweighted in multivariate procedures (Holl & Cairns 1994). Species that were found in one site were given 5 points; those in two sites, 3 points; and those in three sites, 1 point. These scores were summed for each site for each sampling period.


species richness

A total of 159 species was found on reclaimed and reference sites; these included 92 native herbs, 39 native woody species, 22 non-native herbs and six non-native woody species (see Appendix in Supplementary Material). Native, naturally colonizing species comprised the vast majority of both herbaceous and woody species in all sites (Fig. 1). The number of naturally colonizing native herb species did not differ by site type or sampling period (rmanova, P > 0·05 for all effects; Fig. 1a). The number of naturally colonizing non-native and planted (all non-native in the herb layer) herbaceous species in reclaimed sites declined between sampling periods (rmanova, non-native: F1,11 = 8·0, P < 0·05; planted: F1,11 = 5·0, P < 0·05). The number of native tree and shrub species showed a strong site type effect (rmanova, F3,15 = 59·4, P < 0·0001; Fig. 1b), with reference forest sites having the highest number of species and sites reclaimed after 1977 the lowest number. Native tree and shrub species richness did not increase between sampling periods across all sites (rmanova, F1,15 = 0·01, P = 0·91).

Figure 1.

Mean herb (a) and tree and shrub (b) species richness in 1992–93 and 1999. Number of species that are native and naturally colonized (cross-hatch), non-native and naturally colonized (solid) or were planted as part of reclamation efforts (diagonal lines) are indicated by bar shading. Statistical comparisons are discussed in the text.

vegetation composition

Vegetation community composition differed substantially between reclaimed and forest sites. Combining native plant species from both sampling periods, 42 species were found only in reclaimed sites, 27 species were found only in forest sites, and 61 species were found in both reclaimed and forest sites. Of the 103 native plant species found in reclaimed sites, only one (Robinia pseudoacacia) was planted. As expected, the percentage of herbaceous and woody forest species found in reclaimed sites increased from the first to the second sampling periods and was highest in the oldest reclaimed sites (Table 1). The percentage of herbaceous forest species was similar in sites reclaimed from 1980 to 1987 and from 1972 to 1977 in both periods (Table 1).

Table 1.  Percentage of species from forest reference sites that were found in coal surface mined sites reclaimed in 1962–67, 1972–77 and 1980–87 in south-western Virginia. Surveys were conducted in 1992–93 and 1999
Year of reclamationHerbaceous speciesWoody species

Figure 2 shows the locations of the vegetation communities of the sites on the first two DCA axes during each sampling period. The first DCA axis scores (eigenvalue = 0·524) were strongly correlated with site age (Spearman R= 0·86, P < 0·0001). Sites reclaimed in 1962–67 and 1980–87 sites had significantly higher first DCA axis scores in the second sampling period compared with the first sampling period, suggesting that they are following a successional trajectory towards the surrounding forest. This difference was not significant for sites reclaimed in 1972–77 (Table 2). At the second sampling period there was still a substantial separation along the first DCA axis between the oldest reclaimed sites and the reference forest sites (Table 2), indicating that the oldest reclaimed sites still did not host the full complement of forest species. The 1972–77 sites in the second sampling period had lower first axis DCA scores than the 1962–67 sites in the first sampling period, suggesting slower succession in the 1972–77 sites (Table 2). The 1980–87 sites at the second sampling period had higher first axis scores than the 1972–77 sites at the first sampling period (Table 2), suggesting more similarity to the forest sites. The eigenvalue for the second axis (0·198) was much lower than for axis 1 and separated the sites reclaimed between 1980 and 1987 from those reclaimed between 1972 and 1977.

Figure 2.

Vegetation site DCA scores in 1992–93 (open symbols) and 1999 (solid symbols). Site type indicated by symbol shape: reclaimed 1980–87, circles; 1972–77, squares; 1967–72, triangles; forest, diamonds. Eigenvalues: axis 1 = 0·524, axis 2 = 0·198.

Table 2.  Differences in first DCA axis scores (illustrated in Fig. 2) of vegetation communities on reclaimed mines and reference forest sites sampled in 1992–93 and 1999. Values are positive if the 1999 vegetation samples had greater first axis values (i.e. were more similar to the reference forest sites) and negative if the 1992–93 vegetation samples had greater first axis values. Paired t-tests were used to compare the same sites at different sampling periods. Unpaired t-tests were used to compare sites in different reclamation time classes. Values are t-values. *< 0·05, **< 0·01, ***< 0·001
First sampling periodSecond sampling period
1972–772·5* 1·69  
1962–67 −2·72* 3·88* 
Forest  −8·64***−0·95

The sites reclaimed between 1980 and 1987 were dominated in the overstorey by Pinus strobus, with some colonization of native woody species, in particular Acer rubrum. In these sites there was a decrease in most early successional understorey species, both planted (Festuca arundinacea and Lespedeza cuneata) and unplanted (Hierarcium spp., Lysimachia quadrifolia and Solidago rugosum) between 1992 and 1999. The sites reclaimed between 1972 and 1977 also experienced a decrease in the planted species, as well as a number of wind-dispersed herbaceous species, such as Aster pilosus, Erechtites hieracifolia and Gnaphalium obtusifolium. In the overstorey, cover of Robinia pseudoacacia declined between sampling periods and was replaced primarily by Acer spp. (primarily rubrum). The oldest reclaimed sites (1962–67) showed an increase in Acer spp., as well as a number of other naturally colonizing tree and vine species, such as Betula lenta, Carya spp. and Vitis aestivalis between 1992 and 1999. The understorey in these sites was dominated by native species, such as Aster divaricatus, Geum canadense, and Impatiens spp.

A number of forest herbs such as Geranium maculatum, Sanicula canadensis and Galax aphylla were found on the older reclaimed sites but not those reclaimed more recently. Whereas most of the common forest species were present in reclaimed sites, a number of less abundant forest species were not found on reclaimed mines. These were primarily herbs such as Aurora laevigita, Galium circaezans, Trillium grandiflorum and Uvularia pudica, but also included a few woody species such as Amelanchier arborea, Quercus prinus and Vaccinium arboreum. The mean less common species ranking was higher in forest compared with reclaimed sites, suggesting that species were more patchily distributed in forest sites, but this was only marginally statistically significant (rmanova, F3,15 = 2·9, P = 0·07; Fig. 3) due to high variance.

Figure 3.

Mean vegetation less common species ranking by type of site in 1992–93 (solid) and 1999 (cross-hatch). Error bars = 1 SE. Ranking system: species observed in one site = 5 points, two sites = 3 points, three sites = 1 point. There were no significant differences in sites between years using paired t-tests or across sites in the same year using anova.

Most naturally colonizing non-native herb species (e.g. Convolvulus arvensis and Taraxacum officinale) decreased between the first and second sampling period in reclaimed sites. The primary non-native shrub, Rosa multiflora, however, increased in cover in reclaimed sites at the second sampling period. No non-native or planted species were recorded in reference sites.

vegetation cover

There was a strong interaction between the effect of time of reclamation and sampling period on herb cover (rmanova, F3,15 = 18·7, P < 0·0001). At the first sampling period, herb cover was much higher in all reclaimed sites compared with forest sites (one-way anova, F3,16 = 15·0, P < 0·0001; Fig. 4a). By the second sampling period, herbaceous cover in sites reclaimed between 1972 and 1977 and between 1980 and 1987 dropped substantially due to the decline in planted species. In 1999, cover in the oldest reclaimed sites was highest primarily due to cover of native species. Shrub cover was fairly low and variable within sites of the same type (Fig. 4b). At the first sampling period shrub cover was lower in sites reclaimed between 1972 and 1977 and 1962–67 compared with reference forest sites (one-way anova, F3,16 = 4·9, P = 0·0135), but there was no difference by the second sampling period (one-way anova, F3,16 = 0·2, P = 0·92).

Figure 4.

Mean herb cover (a), shrub cover (b) and tree basal area (c) by type of site in 1992–93 (solid) and 1999 (cross-hatch). Error bars = 1 SE. Values were originally compared using repeated-measures anova, but due to significant site type × sampling period interactions values were compared using separate Tukey's LSD (site type) and paired t-tests (sampling period). *Significant differences (< 0·05) in sites between years using paired t-tests. Means with the same letter are not significantly different across site type using Tukey's LSD. Lower case letters 1992–93 comparisons. Upper case letters 1999 comparisons.

In 1992–93 tree basal area was highest in reference forest sites, intermediate in sites reclaimed between 1962 and 1967, and lowest in the more recently reclaimed sites (one-way anovaF3,16 = 5·62, = 0·0087; Fig. 4c). All ages of reclaimed sites had significant increases in basal areas between the two sampling periods (Fig. 4c), with the largest increase in the sites reclaimed in 1980–87. This increase was primarily due to the rapid growth of Pinus strobus, which increased from 1·0 ± 0·7 m2 ha−1 in 1992 to 15·2 ± 2·9 m2 ha−1 in 1999. Basal area in the reference sites did not increase significantly between the two sampling periods, but this was partly due to some selective logging in two of the sites. By 1999, only sites reclaimed in 1972–77 had significantly lower basal area than the reference sites.


One goal of this study was to determine whether the vegetation composition of reclaimed coal surface mines was approaching that of the surrounding forest. Results of this and other studies in the south-eastern USA (Thompson, Vogel & Taylor 1984; Skousen, Johnson & Garbutt 1994; Thompson, Wade & Straw 1996; Rodrigue 2001) suggest that a large number of native species colonize reclaimed coal surface mined sites after 10–15 years. In the current study, the percentage of forest species in reclaimed sites increased in the second sampling period, as expected. Species transitions observed, such as dominance by high light-requiring species (e.g. Liriodendron tulipifera and Prunus serotina) in more recently reclaimed sites and increased establishment of later successional species (e.g. Carya spp. and Quercus rubra) in older reclaimed sites, are consistent with successional patterns after logging (Gilliam, Turrill & Adams 1995). Herbaceous cover and tree basal area were similar in most reclaimed and reference sites by the second sampling period. Given that these are highly disturbed areas with elevated temperature, reduced microtopography and often acidic soil conditions (reviewed in Leopold & Wali 1992), and that few species were planted, these results suggest a high degree of resilience of the south-eastern forest ecosystems studied. Clearly, however, recovery in ecosystems with more extreme temperature variation and periodic rainfall, such as montane and arid systems in the western USA, is likely to be much slower (Chambers, Brown & Williams 1994; Macyk 2000).

While the forest ecosystems studied appear to be relatively resilient, here and in a previous study on Pennsylvania coal surface mines (Schuster & Hutnik 1987), the reclaimed sites tended to be dominated by known habitat generalist species (e.g. Acer rubrum, Clematis virginiana, Polystichum acrostichoides and Potentilla canadensis; Elliott et al. 1999; Beckage et al. 2000). Moreover, a number of the forest species, in particular herbs with ant- or gravity-dispersed seeds, were not found in the reclaimed sites surveyed, consistent with the research of McLachlan & Bazely (2001) who studied recovery on former cottage and road sites in western Ontario. Thompson, Wade & Straw (1996) compiled a comprehensive species list for a coal surface mine in Kentucky and found some rare species. The current study, while not an exhaustive survey, does show that on a per area basis there is lower between-site (beta) diversity on the reclaimed sites studied. This study, as well as a parallel study of lepidopteran communities on reclaimed coal surface mines (Holl 1996), raises the question of the value of such reclamation efforts for conserving rarer species.

While there have been no other replicated studies of both herb and tree community composition on mines in the eastern USA, there has been a great deal of study of eastern forests disturbed by clearcut logging (the vast majority of forest in this region) with comparison to the few unlogged remnants. Most of these studies agree that forest herbaceous species are negatively impacted by clearcut logging (Duffy & Meier 1992; Bratton, Hapeman & Mast 1994; Gilliam, Turrill & Adams 1995; Meier, Bratton & Duffy 1995), although results from Ford et al. (2000) showed less dramatic effects of logging. Meier, Bratton & Duffy (1995) reported that many of the same genera and species were absent from 5-year-old clearcut sites (e.g. Uvularia perfoliata, Disporum lanuginosum and Trillium spp.) as were absent in reclaimed sites in this study, although they were found in periodically logged reference sites. Slow recovery of herbs following clearcut logging and other disturbances has been attributed to extreme microclimate conditions immediately following disturbance, competition with early successional species, and low seed rain and slow growth rates of these species as succession proceeds (Matlack 1994; Meier, Bratton & Duffy 1995; McLachlan & Bazely 2001). Likewise, a number of other authors (Beckage et al. 2000; Ford et al. 2000; Robinson & Handel 2000) note the importance of lack of seed dispersal in limiting plant colonization in eastern hardwood forest. All of these factors, in addition to highly physically and chemically altered soil conditions (Li & Daniels 1994), may explain the absence of a number of forest species on reclaimed sites.

It is important to reiterate that reference sites in this study had been previously logged, although only selectively logged in recent years. Despite this, reference sites still hosted species not found in the oldest reclaimed sites, such as Aurora laevigita, Galium circaezans and Trillium grandiflorum. The oldest reclaimed sites have only had 30–35 years since disturbance and it has been suggested that recovery of eastern hardwood forests from logging (a much less severe disturbance than mining) may take more than a century (Meier, Bratton & Duffy 1995).

A second goal of this study was to assess how increasingly intensive reclamation practices affect long-term recovery. Past research in abandoned eastern USA old fields (Connell & Slatyer 1977; Myster & Pickett 1990) and on mines elsewhere (Bradshaw 1984; Chambers, Brown & Williams 1994) demonstrates that vegetation composition at the time of abandonment strongly influences successional trajectories. It is challenging to interpret the results of the current study within this context, as site age and reclamation practices are necessarily confounded and information on long-term disturbance history is not available; over the past 30 years there have been changes in both the tree and ground cover species planted to reclaim coal surface mines in the eastern USA.

The sites reclaimed between 1972 and 1977 and 1980 and 1987 were planted with aggressive ground covers, such as Festuca arundinacea and Lespedeza cuneata, to maximize erosion control. These species have been shown to inhibit tree growth (Torbert et al. 2000) and it has been suggested that they may inhibit natural seedling establishment (Holl & Cairns 1994; Skousen, Johnson & Garbutt 1994). Indeed, the first DCA axis locations of 1972–77 reclaimed sites in the second sampling period were further from the reference sites compared with the 1962–67 sites at the first sampling period; at these times the sites would have been approximately the same age (c. 25–27 years since reclamation) and the primary reclamation difference was the intensity of ground cover seeding. Other factors, such as differences in soil type or seed rain between sites, which were not studied, however, could explain these differences in vegetation recovery. Controlled experiments are needed to separate the effects of different ground covers on native species establishment. Tree basal area in the 1972–77 sites was lower than any other site type at the second sampling period, probably due to a combination of competition of planted and naturally establishing trees with the aggressive grasses and legumes, and not planting rapid-growing tree species. It will be interesting to note whether recovery occurs more rapidly in the future now that the ground cover is reduced in these sites.

Pinus strobus, which is native to the region although uncommon, has been widely planted in sites reclaimed following SMCRA (1977) in order to meet tree establishment requirements. Clearly this strategy has been successful, given the substantial tree basal area increases in sites reclaimed between 1980 and 1987 compared with sites reclaimed earlier, to the point that these sites had tree basal areas equivalent to sites planted 15 years earlier and not significantly lower than the reference forest sites. The rapid increase in overstorey cover resulted in a concomitant rapid drop in herbaceous cover, which could potentially allow for the establishment of other species, particularly after the trees are logged.

Past research suggests that the dominance of sites by Pinus strobus may affect long-term recovery, as the understorey of P. strobus has been shown not to be amenable to the establishment and growth of other species (Ashby 1964; Schuster & Hutnik 1987; Artigas & Boerner 1989). Results of this study and those of Rodrigue (2001; tree species only), however, show a similar number of native colonizing species in sites planted with hardwoods and pines and a fairly rapid establishment of forest species, suggesting that loss of diversity should not be a concern. It is important to note that in the present study the pine canopy had not yet closed in all sites and most plants in the herb layer were found in gaps between trees (K.D. Holl, personal observation), so species richness may continue to drop as the canopy matures. The separation of the 1972–77 sites from the 1980–87 sites on the second DCA axis largely reflects the difference in planted species, but it is too early to determine to what degree these differences will be maintained over the long-term.

conclusions and management recommendations

Results of this and other studies suggest that coal surface mines in this region can recover a diverse native community fairly quickly, if appropriate site conditions are present. But, it may be much longer than the 35 years of recovery studied before these sites host the entire complement of the local flora. Results of this and other studies suggest a number of management strategies that may facilitate long-term succession on these sites.

First, retention of nearby seed sources is critical to rapid recovery, as few if any native species are planted as part of reclamation efforts in this region. Until the past 5–10 years, most mine sites were relatively small in area, which meant that they were quite near (< 50 m) to a forest that could serve as a seed source. This is probably one reason for the large numbers of plant species recorded on reclaimed mines in this study. Numerous past studies have shown that many eastern USA forest species are dispersal limited (Skousen, Johnson & Garbutt 1994; Ford et al. 2000; McLachlan & Bazely 2001) and that seed rain may drop dramatically only 10–15 m from remnant forest (Matlack 1994). Increasingly, large areas, often entire mountain tops, are being mined; unless remnant patches of forest are retained to serve as seed sources, recovery will certainly be slower than that observed in this study. In other regions, such as the western USA, a higher percentage of native species are used in reclamation efforts (Richards, Chambers & Ross 1998), but still only a small portion of the flora is reintroduced, so recovery is likewise dependent on input of propagules.

Secondly, a common challenge in reclamation efforts is to balance short- and long-term human and ecological needs. For example, research by Torbert & Burger (2000) has demonstrated that less-competitive non-native herbaceous species, such as the annual grasses Setaria italica and Secale cereale, the perennial grasses Agrostis gigantea and Eragrostis curvula, and the legume species Lespedeza striata and Lotus corniculatus, do control erosion effectively, after the first year, but allow for more establishment of native species. As a result, less-aggressive, although non-native, species mixes have increasingly been used for reclamation in recent years (Torbert & Burger 2000).

More research is needed on other native and naturalized ground cover species that might be used to diversify species used in reclamation efforts. Past research suggests that some wildflower species, such as Rudbeckia hirta, Centaurea cyanus and Coreopsis lanceolata, establish when seeded on mine sites (Heckman et al. 1996). Wade (1989) has also found that spreading topsoil from nearby forests on reclaimed mines introduced a large number of native species. Recently increased efforts have been made to move topsoil from areas being mined to areas being reclaimed, increasing the likelihood of re-establishing native species of plants and micro-organisms (C. E. Zipper, personal communication).

Thirdly, most forest reclamation efforts would benefit from planting a wider variety of tree species. At the sites studied, Pinus strobus is valuable for rapid productivity, but planting all sites with P. strobus has resulted in a virtual overstorey monoculture in certain areas. Research has demonstrated that a number of hardwood species are commercially viable for mine reclamation (Brown, Farmer & Splittstoesser 1984; Torbert et al. 1985; Rodrigue 2001) and have wildlife benefits (Leedy 1981; Brenner, Werner & Pike 1984). Although these species may grow more slowly than Pinus strobus in the first few years, they have the potential to provide a greater income over the long-term because of the higher value of their wood. It may also be possible to reforest areas of Pinus strobus with hardwoods after they are logged, or to thin pines and interplant hardwoods. This strategy of using Pinus as a nurse plant for later-successional forest species has been successful in facilitating recovery in highly disturbed tropical moist forest (Ashton et al. 1997).

The short bond release period of current mine legislation serves to encourage use of reclamation strategies proven to address short-term concerns of providing erosion control and minimizing acid mine drainage. But, with increasing recognition of the important conservation value reclaimed mines can have, this is slowly changing. More emphasis should be given to long-term recovery potential in developing mine reclamation plans, and strategies to further this goal should be tested.


This project was supported by funds from the Powell River Project and the University of California, Santa Cruz. I am grateful to Jonathan Beals-Nesmith and Vanessa Mulkey for assistance with field research, and to Jon Rockett and Danny Early for logistical support. Grey Hayes, Julie Lockwood and Carl Zipper provided helpful comments on earlier drafts.

Supplementary material

The following material is available from http://www.blackwell-science.com/products/journal/suppmat/JPE/JPE767/JPE767sm.htm.

Appendix: species list. Nomenclature follows Radford, Ahles & Bell (1968). 1 = average importance value of < 5; 2 = average importance value of 5–25; 3 = average importance value of > 25; E = species not originally native to the eastern USA; P = planted species. For herbs and shrubs, the importance value was calculated as the sum of the relative cover and relative frequency. For trees, the importance value was calculated as the relative number of individuals and relative basal area