Canada: Ontario: Trial Restoration of Cliff-Edge Forest in Bruce Peninsula National Park


Ecological restoration in parks and conservation areas poses a unique challenge, as the need to maintain public access often precludes gradual, undisturbed vegetative succession and in many cases results in continuing processes of degradation. In this study conducted in Bruce Peninsula National Park, researchers tested whether the natural canopy species, Thuja occidentalis, can be restored to degraded coastal cliff edges without reducing the number of visitors to the park. A combination of seed and nursery-reared seedlings 4 years of age and 10 years of age was used to evaluate eight different treatments designed to test the effects of planting site (distance from cliff edge and pathways) and supportive measures (soil, water, protective cages, or signs) on survival, growth, and damage. After four growing seasons, trees planted away from the cliff edge and away from pathways had the greatest establishment success, 4-year survival, and general health. Spot counts of visitors to the park showed that relative visitor density had a greater effect on restoration success than any other factor. In fact, the results suggest a threshold of visitor density above which restoration may be impossible. Consequently, it is concluded that restoration of cliff-edge forest habitats is possible without total visitor exclusion but that some restrictions on visitor numbers or activities are necessary.

Quick Facts

Project Location:
Bruce Peninsula National Park, Cyprus Lake Road, Tobermory, ON, Canada, 45.1890624, -81.52977069999997

Geographic Region:
North America

Country or Territory:

Temperate Forest

Temperate Forest - Coniferous

Area being restored:

Organization Type:


Project Stage:

Start Date:

End Date:

Primary Causes of Degradation

Urbanization, Transportation & Industry, Other

Degradation Description

In Bruce Peninsula National Park, Ontario, hiking trails along the cliff edges of the Niagara Escarpment offer a spectacular view of the waters of Georgian Bay and may be used by more than 20,000 visitors in a single season (Kettle 1998). Heavily visited areas along the cliffs show visible degradation of the cliff-edge forest, and the most disturbed sections of cliff edge are virtually denuded of vegetation (Larson 1990; Parikesit et al. 1995).

Reference Ecosystem Description

In undisturbed parts of the park, the cliff edges support a characteristic stunted old-growth forest community, which is dominated by the woody species Thuja occidentalis (eastern white cedar) (Larson et al. 1989). Fierce winds, ice, rock fall and searing sun exposure torment the trees and likely cause their dwarfed and twisted shapes. The cedars are part of a much more complex ecosystem than was previously imagined. The barren-looking cliff face is actually covered with mosses and lichen, while countless caves and crevices provide homes for ravens, turkey vultures, swallows and bats. Other commonly seen wildlife on the Bruce Peninsula includes: chipmunk, squirrel, raccoon, porcupine, snowshoe hare, skunk, white-tailed deer, snakes and frogs. Black bear, fox, fisher, martin and the Eastern Massasauga rattlesnake are less commonly seen.

The Massasauga rattlesnake, an endangered species, is now reduced to a few scattered populations. The snake was once found throughout southern Ontario, and Bruce Peninsula National Park, the largest remaining extension of natural habitat in the region, plays an important role in preserving suitable habitat for this species.

Project Goals

To evaluate the possibility of restoring the canopy species T. occidentalis to degraded cliff edges in Bruce Peninsula National Park without reducing visitor numbers.


The project does not have a monitoring plan.


This project was conducted with the support of Parks Canada as a small-scale trial intended to help guide larger-scale restoration efforts envisaged for the near future.

Description of Project Activities:
The study was conducted along a 360 m long and 10 m wide section of Niagara Escarpment cliff edge that included two of the major points of interest within Bruce Peninsula National Park: the Grotto and Indian Head Cove, as well as adjacent less frequently visited parts of the shoreline. To stratify for differences in current disturbance levels, the sampling area was divided into 10 blocks of 36 m each, and the relative disturbance within each was quantified by conducting visitor counts. Over a 2-hr period on three separate dates (21 July, 5 August, and 1 September 1999), simultaneous counts were taken every 5 minutes of the number of people present within each block. All observation periods were then summed to provide a relative measure of visitation rate for each block. Each block was divided into 24 sections of 1.5 m length, and these were randomly assigned to three tree age classes and eight treatments. A planting site, consisting of a soil-filled depression or crevice in the rock, was selected within each section. Such planting sites reflect the natural habitats of the trees in this ancient forest (Matthes-Sears & Larson 1995). Sites were located either within a meter of the cliff edge or 10 m away from the edge, depending on the treatment. At the 10-m distance, where visitor traffic has created a dense network of small footpaths, planting sites selected were either disturbed (on pathways) or undisturbed. No such distinction was possible for cliff edge sites where disturbance was uniformly heavy. Each planting site was further categorized by canopy cover as either shady, open, or intermediate. Three age classes of T. occidentalis were used for planting: 10-year-old trees, 4-year-old trees, and seeds. All seed had been collected in the autumn of 1996 from cliffs in southern Ontario and was stored at 4º C. Some of the seed was pre-germinated in a growth chamber at 30º C/8-hr light and 20º C/16-hr dark on moist filter paper and was used when the radicle was starting to protrude. The trees had been grown from seed and kept in pots outdoors at the University of Guelph. The eight treatments were designed to test the effects of ameliorating growth conditions by adding soil, water, or soil and water; protecting trees with brightly colored signs or cages; and varying the planting location with respect to disturbance and distance from the cliff edge. Soil addition took place at planting and consisted of 500 mL of commercial potting mix (Pro-Mix; Premier Horticulture, RiviÁ¨re-du-Loup, Québec) for 4-year-old trees and seeds and twice that amount for 10-year-old trees. Watering was performed during the first growing season only and consisted of a once weekly soaking of the soil surrounding the tree or seeds. Cages were open at the top and were constructed of poultry netting 80 cm high for 10-year-old trees, 60 cm high for 4-year-old trees, and 40 cm high for seeds. Signs were of yellow plastic, 15 cm in diameter, and inscribed "Replanted tree"”please do not disturb." They were placed immediately adjacent to the planting site. Each planting location received one tree or several seeds. The trees were planted during the last week of May 1997. Five pre-germinated seeds per site were planted during the first week of June 1997. When few seedlings emerged, another five germinated seeds were added to the same sites in half of the blocks 2 weeks later. A third planting of 10 ungerminated seeds per site was performed in April 1998. All trees and seeds were watered once immediately after planting. Each planting site was visited for data collection at the following intervals: weekly, biweekly, and monthly in the first, second and third growing seasons, respectively, and at the beginning (May) and end (September) of the fourth growing season. Damage and shoot growth were assessed on five (10-year-old trees) or three (4-year-old trees) randomly selected branches plus the main leader of each tree. New branches were selected and tagged at the beginning of each growing season. At each visit, a score between 0 and 5 was assigned to each branch in three categories: breakage, representing damage caused by humans (most often by trampling); herbivore damage; and browning (also including yellowing or wilting) as a sign of general poor health of the tree. To measure shoot growth, a small mark of nontoxic paint was placed on the leader of the branch and the distance from the mark to the growing tip was measured (Matthes-Sears et al. 1995). Also at each visit, the angle of the main axis with respect to the horizontal plane and the presence of exposed roots were scored on a scale of 1 - 3. At the end of the study, each tree was assigned to one or more fate categories that took into account not only whether the tree survived but also its health, exposure to disturbances, and likely reason for mortality. These fate categories were as follows: Fate 1: healthy throughout Fate 2: variable health Fate 3: survived partial uprooting Fate 4: survived trampling Fate 5: survived herbivory Fate 6: survived cage damage Fate 7: failed to become established Fate 8: died without external factors Fate 9: uprooted Fate 10: trampled Fate 11: herbivory Fate 12: cage damage

Ecological Outcomes Achieved

Eliminate existing threats to the ecosystem:
The first planting of seeds in 1997 resulted in the emergence of five seedlings from the 400 seeds planted, an emergence rate of 1.25%. None of the seedlings survived past the second week. Emergence rate was 17.5% for the second planting in 1997, with the seedlings evenly distributed over all blocks and treatments. All of these seedlings died within a few weeks except for one that survived until midsummer 1998. No seedlings emerged from the ungerminated seed planted in April 1998, although several natural seedlings were observed in the vicinity of the plots. Mortality of planted trees in both age classes was greatest in the year of planting and decreased over time. Overall, 64.4%, 44.4%, 39.4%, and 38.8% of trees survived the first, second, third, and fourth years, respectively. Life table analysis revealed no significant difference in survivorship curves between 4-year-old and 10-year-old trees. There were, however, significant differences among treatment groups. Survival was highest for trees planted 10 m away from the cliff edge in undisturbed locations (treatment 8); 95% of these trees survived the first year, and 85% were alive at the end of the study. Lowest survival (35% after year 1, 20% at the end of the study) was seen in trees planted at the cliff edge without supplements (treatment 1). All other treatments were intermediate, and only the difference between the two extremes was significant. Significant differences in survivorship curves were also found among blocks. Survival was highest in block 1, with 94% of trees surviving the first year and 87.5% the fourth year. Block 1 was significantly different from blocks 9 and 10, where 31% of trees survived year 1 and only 6.25% (corresponding to a single tree out of the 16 planted in this block) survived to the end of the study. Twenty-one percent of all trees became well established and were healthy throughout the study period (fate category 1). Another 21% alternated between health and decline but were alive at the end of the study (fate 2). Fifteen percent of trees failed to become established (fate 7), and another 9% died without obvious external causes after alternating periods of health and decline (fate 8). Shoot growth in surviving trees showed the typical seasonal course with fastest rates early in the year and a slowing down by midsummer. Total seasonal growth showed no significant differences among treatments, blocks, and tree ages for each of the 3 years that measurements were taken. Neither block nor treatment affected seasonal growth patterns of surviving trees, but small and large trees differed in their growth curve in the year of planting. Contrasts for the individual time intervals showed that the age x time interaction was only significant for the very first time interval in 1997, the month immediately after planting. Growth rate during this period was significantly faster in 4-year-old trees than in 10-year-old trees; thereafter, growth rates were statistically the same for both age groups. For the following 2 years, there were no significant differences in growth rates among age groups, but 10-year-old trees tended to grow slightly faster.

Factors limiting recovery of the ecosystem:
The results of the study show that a low survival rate of T. occidentalis is correlated with high visitor numbers in the more popular sections of cliff edge in Bruce Peninsula National Park. The visitor counts showed a more than 150-fold range in visitor numbers as a result of the visitor flow patterns within the study area. The corresponding range in restoration success was approximately 15-fold, from a 4-year survival rate of 87.5% in block 1 to 6.25% in blocks 9 and 10. The large effect of disturbance on restoration potential is underlined by the result that for almost all measures of restoration success, differences among blocks within the study area exceeded all differences among treatments. Indeed, the patterns of visitor density explain much of the variation in tree survival, fate, and damage seen within the study area. Block 1, with the lowest visitor density, had the highest 4-year survival rate, the fewest trees that failed to become established and the lowest scores for browning, breakage and root exposure. Block 10 had the lowest survival rate and the highest damage scores. However, although visitor impact was clearly the major factor causing mortality of planted trees, it was not the only factor, and not all results are explained by visitor numbers. For example, survival rates in block 9 were as low as in block 10, even though block 9 had much lower visitor numbers. Conversely, block 6 had a low frequency of trees that failed to become established and a high frequency of trees that were healthy throughout, even though it received the second highest number of visitors. Visitor residence time may play a role in these discrepancies, as visitors lingering in an area are likely to have a greater impact than those passing through quickly, as most do in block 6. In all, approximately 95% of trees died as a result of external causes such as uprooting, trampling or herbivory, while damage from dislodged or trampled cages (fate 12) probably contributed to the death of the other 5%. Thirty-seven percent of all trees were partially or completely uprooted at some point during the study period, and this was the most frequent external cause of tree death (fate 9). Sixteen percent were completely uprooted and either died immediately or disappeared (fate 9a and b), and in another 8% partial uprooting led to a slow decline and eventual death (fate 9c). Twelve percent of trees survived a partial uprooting (fate 3). Twenty-five percent of all trees were severely or repeatedly trampled at some point during the study period. For 7%, this was almost certainly the reason for their subsequent death (fate 10a); 7% survived it (fate 4), and for 11% it was a contributing factor but not necessarily the only reason for their death (fate 10b). Seven percent of trees experienced serious herbivory at some point during the study period; herbivory was the most likely reason for the death of 3% of trees (fate 11). Four percent experienced similar levels of herbivory but survived (fate 5). The different fates were not homogeneously distributed over the eight treatments, and tests of independence confirmed that some of these differences were significant. Comparisons of the number of trees that survived and were healthy throughout (fate 1) versus the number of trees that died without external causes (fates 7 and 8) for different treatments show how the establishment and survival of trees are affected by the treatments in the absence of external reasons for mortality. Trees were more likely to be successful in the presence of added water, added soil, or both, whereas failure was more likely in the absence of these factors. Establishment success was significantly lower at the cliff edge than 10 m away from the edge; away from the edge, it did not differ between pathways and undisturbed locations. Herbivore damage, whether fatal or not, was rare at the cliff edge but much more common 10 m away from the cliff edge: 22.5% of all trees in treatments 7 and 8 experienced herbivory, as opposed to only 2.5% of all trees in the other six treatments. The frequencies of the different fates also differed across the study area. Blocks 1 and 2 as well as 5 and 6 had a low proportion of trees that failed to become established or died without external causes and a high proportion of trees that were healthy throughout, whereas the opposite was true for blocks 3 and 4 as well as 7 - 10. Blocks 8 - 10 had a high incidence of trees trampled or uprooted but a low incidence of herbivory. There was also a significant relationship between establishment success (as measured by the relative frequencies of fate 1 vs. fates 7 and 8) and the characteristics of the planting site. In fact, the planting site was the second most important factor behind visitor numbers affecting the survival, fate and damage scores of planted trees. Trees planted in the presence of shade from nearby canopy trees were much more likely to be healthy throughout and much less likely to die quickly without external causes than trees planted in sites classified as open. Another important aspect of the planting site--this one controlled by the treatment--was the distance from pathways and the distance from the cliff edge. The effect of trampling was only studied away from the cliff edge because no undisturbed sites were available at the cliff edge. Planting sites on pathways were less favorable for the successful establishment of T. occidentalis than were untrampled planting sites. Besides ongoing disturbance, this may be due to decreased soil depth and litter cover and increased soil bulk density (Parikesit et al. 1995). Soils compacted by trampling are generally considered unfavorable for seedling establishment and early growth (Tuttle et al. 1988). Trees planted at the cliff edge were more likely to die soon after planting or suffer root exposure and poor health and had lower 4-year survival rates than trees planted in equally disturbed locations 10 m away from the cliff edge. A number of factors may be responsible for this pattern, which is also seen in natural seedling establishment at cliff edges (Bartlett et al. 1991). Shallow soils near the cliff edge may exacerbate root exposure due to trampling, which in turn makes trees more susceptible to drought and wind throw (James et al. 1979). Microclimatic conditions become more severe at the cliff edge, in particular water availability, temperature fluctuations, and wind speed (Bartlett et al. 1990). In Bruce Peninsula National Park, this natural gradient may have been further increased by the absence of canopy-forming trees from the most disturbed cliff edges. The analysis showed that trees planted in shady conditions, regardless of treatment or block, were more likely to remain healthy throughout the study period than trees planted in open areas. In blocks 5 and 6, where survival rates were better than expected from the high visitor density, more than half the trees had been planted in the shade; conversely, all trees in block 9, which had lower than expected survival rates, had been planted in open locations. All of the trees in the fate category "failed to become established" had been planted in open conditions and none in the shade. Although a tree canopy is generally known to create conditions favorable for tree establishment (McChesney et al. 1995), such a large effect was not expected for T. occidentalis, which is said to regenerate best on exposed mineral soils with abundant light (Johnston 1990). The only advantage of the cliff edge as a planting site was the absence of herbivory by rabbits or rodents; however, herbivory was not a major cause of mortality in this study. Getting visitors to refrain from damage-causing activities is a common challenge on public land with high visitor traffic (Mortensen 1989), and efforts to protect plantings from human damage were of only limited success in this experiment. Regardless of treatment group, many planted trees were actively and in some cases violently dislodged or uprooted by visitors. Visitors appeared unresponsive to the signs conveying appeals not to disturb the trees: Signs had no effect at all on tree survival, fate, damage, or growth. The high frequency with which the signs were trampled or dislodged suggests they may need to be larger to be noticed. Futhermore, the physical protection of trees by caging had mixed success, as cages were frequently found at the bottom of the cliff or suspended in the canopy of neighboring trees after having apparently been removed by visitors. For the smaller trees, cages significantly decreased the amount of breakage, most likely because visitors were encouraged to walk around rather than step over the trees. However, the survival rate of these trees was unaffected by caging, as were fate, browning, and growth rate. For large trees the enclosures were detrimental, increasing the amount of breakage compared with uncaged trees. Most likely, this was because these trees were too tall to be stepped over even when uncaged, and the taller cages were more likely than shorter ones to be removed by visitors or blown over by the wind, causing breakage to the trees within them.

Socio-Economic & Community Outcomes Achieved

Key Lessons Learned

In conclusion, the results of this study suggest that the planting of smaller (e.g., 4-year-old) container-grown trees without supportive measures of any kind is the best approach for cliff-edge restoration in Bruce Peninsula National Park. Very few T. occidentalis emerged from seed (regardless of treatment, disturbance, and whether or not seeds were pre-germinated), and none of them survived for more than a year. Although water and soil addition made a small improvement in survival, this improvement was disproportionately small considering the effort involved in carrying water or soil to the planting sites. The same was true for protective cages, whose small improvement to survival was far outweighed by the effort required for setup and maintenance, in addition to the aesthetic drawback. The size and age of container-grown trees are unimportant, but smaller trees are recommended because they are easier to grow, transport, and plant. The broader implication of these results is that, in other settings, the age-related patterns of mortality should be carefully examined in the context of the site conditions before the restoration materials and protocols are selected.

Long-Term Management

The large differences in tree survival between trampled and untrampled planting sites suggest that, in conjunction with reforestation, the current network of small pathways should be blocked off using boulders or brush and replaced by a single clearly marked main trail or boardwalk. The effectiveness of such visitor channeling is apparent from the high planting success in the heavily visited blocks 5 and 6, where a dense forest in combination with topography has so far largely prevented the networking of pathways so evident in other parts of the study area. Since shade from existing canopy trees was shown to favor restoration efforts, one management priority should be to prevent the further degradation of cliff edges to the point where few or no canopy trees remain. Where such trees are absent, as they are currently in the most disturbed parts of the study area, restoration may be difficult under current circumstances and a temporary reduction in visitor admissions may need to be considered. Such a reduction would be most beneficial during the summer holiday season and into the autumn, when our data show visitor impacts to be greatest.

A drastic reduction of visitor admissions to Bruce Peninsula National Park is only one possible way to improve the success of a future cliff-edge restoration project. The fate data indicate that a large proportion of tree mortality results from careless (trampling of trees) or even intentionally destructive (uprooting of trees) visitor behavior. If park managers find a way to specifically target these behaviors, for example, through increased efforts at visitor education in combination with increased surveillance at peak times, it may be possible to have successful site restoration coexist with large visitor numbers enjoying the park in appropriate ways.

Sources and Amounts of Funding

This research was supported by Parks Canada, the Natural Sciences and Engineering Research Council (NSERC) of Canada, and Human Resources Development Canada (Job Creation Partnerships Program).

Other Resources

Matthes, U., J.A. Gerrath, and D.W. Larson. 2003. Experimental restoration of disturbed cliff-edge forests in Bruce Peninsula National Park, Ontario, Canada. Restoration Ecology 11(2):174-184.

S.L. Ross Environmental Research Ltd., Mosquin Bio-Information Ltd., and Horler Information Inc. 1990. Bruce Peninsula National Park Biophysical Survey. Canadian Parks Service. 225 pp.+ appendices.

Parks Canada profile

Primary Contact

Organizational Contact