Restoring conifers by natural regeneration on slopes exposed during highway reconstruction, Glacier National Park, Montana, USA RAYMOND C. SHEARER1* AND JENNIFER M. ASEBROOK2 1 Forestry Sciences Laboratory, US Department of Agriculture, Forest Service, Rocky Mountain Research Station, Missoula, MT 59807-8089, USA 2 Glacier National Park, US Department of Interior, National Park Service, West Glacier, MT 59936, USA * Corresponding author. E-mail: rshearer@fs.fed.us Summary In 1991, about 70 years after construction, the 16-km Lake McDonald section of the Going-to-theSun Road in Glacier National Park was rebuilt. The road, which is at the base of steep north-westfacing slopes, follows the 975 m contour above Lake McDonald within forests originating after wildfires. Tree composition is mostly western red cedar (Thuja plicata Don ex D. Don) and western hemlock (Tsuga heterophylla (Raf.) Sarg.) with a minor component of six other conifer species. Trees were expected to disperse enough seeds to naturally regenerate road cuts on four land-types with developed soils, but not on the fifth land-type with rock and scree. From 1987 to 1995, red cedar and hemlock dispersed 4466 potentially viable (filled) seeds m–2 on cut slopes while other species added 74 filled seeds m–2. From 1991 to 1995, after reconstruction, an average of 3294 filled red cedar seeds fell m–2, in addition to 511 western hemlock m–2 and 61 m–2 for six other species. On cut slopes from 1992 to 1998, of land-types with developed soils, seeds germinated after snowmelt, usually from early May until the surface dried. Almost no seeds germinated and no seedlings survived on the rock/scree land-type. Due to the harsh environment on cut slopes, most seedlings died. In spring 1999, red cedar and hemlock seedlings were small and inconspicuous and required an average of 312 and 173 filled seeds for each surviving seedling. Introduction Successful restoration of disturbed forestlands is a worldwide concern. Afforestation of abandoned agricultural lands (Lockhart et al., 2003; © Institute of Chartered Foresters, 2003 Jõgiste et al., 2003) are examples of large-scale restoration, while this article on conifer natural regeneration on cut slopes above a mountain road exemplifies small-scale restoration of degraded sites. Because of its broad scope, the study of Forestry, Vol. 76, No. 2, 2003 200 F O R E S T RY natural processes within the field of restoration ecology is an important and complex topic. The recovery and maintenance of processes is the key to ecosystem resilience and repair (Bradshaw, 1997; Whisenant, 1999). In 1982, the United States Congress passed the National Surface Transportation Assistance Act to upgrade and rehabilitate deteriorating road systems in national parks within the US Department of Interior (Shearer et al., 1996). Reconstruction of the spectacular 84 km Goingto-the-Sun Road in north-western Montana’s Glacier National Park was a high priority. Scheduled first was the 16-km Lake McDonald section (west central) built in the 1920s and the only road within the park kept open all year. A continuous forest, principally western red cedar (Thuja plicata Don ex D. Don) and western hemlock (Tsuga heterophylla (Raf.) Sarg.), grows on both sides of the road. Park planners developed a strategy to revegetate treated areas, including road cuts, using recommendations made by an interagency core team of managers and scientists from the Park and the US Department of Agriculture, Forest Service. Photographic and written records document natural recovery of vegetation, including conifers, on cut slopes after initial construction 70 years earlier. However, specific seedfall, seed germination and seedling development were not recorded. Except on rocky sites, conifers were anticipated to naturally regenerate road cuts exposed during reconstruction because of the abundant seed source, moderate climate and favourable soil conditions. Both western red cedar and western hemlock produce frequent heavy cone crops in the Upper Columbia River Basin (Gashwiler, 1969; Minore, 1990; Packee, 1990; Feller and Klinka, 1998). Associated conifer species also periodically produce substantial cone crops (Haig et al., 1941; Boe, 1954; Schmidt et al., 1976; Graham, 1990). Natural regeneration of mixed conifers in cedar–hemlock forests is usually prompt and abundant (Ferguson, 1994), especially when the litter/duff layer is reduced to expose mineral soil (Haig et al., 1941). For example, following clearcut and burn treatments on a western red cedar forest cover type in western Montana every established seedling required 13 filled seeds (Shearer, 1976). But, restoring vegetation on fresh road cuts is much less certain than on surfaces exposed by fire or following forest management treatments. Objectives of this opportunistic case study were to document conifer seed-fall and subsequent regeneration on road cuts by estimating (1) the number of conifer seeds dispersed from 1987 to 1995 in forest above and below cut slopes; (2) the number of dispersed conifer seeds that germinated on cut slopes from 1992 to 1998; and (3) the number of conifer seedlings that survived on cut slopes in 1999. Materials and methods Study area The Lake McDonald section of the Going-to-theSun Road (48° 35 N lat., 113° 55 W. long.) follows the 975-m contour above the east shore of the lake and is at the toe of slopes that average about 35°. Forests surrounding the road were initiated by three fire events (Barrett, 1988): 84 per cent after 1735, 10 per cent after 1813 (centre) and 6 per cent after 1929 (southern). The forest is classified as western red cedar–western hemlock (Eyre, 1980), both at the eastern extreme of their natural ranges (Habeck, 1968). Basal area is composed of 85 per cent western red cedar and western hemlock; 10 per cent Rocky Mountain Douglas-fir (Pseudotsuga menziesii var. glauca (Beissn.) Franco) and western larch (Larix occidentalis Nutt.); and 5 per cent western white pine (Pinus monticola Dougl. Ex D. Don), spruce (probably hybrids of Engelmann (Picea engelmannii Parry ex Engelm.) and white (Picea glauca (Moench) Voss), lodge-pole pine (Pinus contorta var. latifolia Engelm.) and sub-alpine fir (Abies lasiocarpa (Hook.) Nutt.) (Shearer and Potter, 1994). Habitat types as classified by Pfister et al. (1977) are Thuja plicata/Clintonia uniflora, Tsuga heterophylla/Clintonia uniflora and Abies lasiocarpa/Clintonia uniflora. Lake McDonald creates an environment favourable for plant growth and soil development (Habeck, 1968). Mean annual precipitation is ~800 mm, maximum temperatures average –3°C in January and 26°C in July, and the frost-free period is about 104 days (Finklin, 1986). Soils were formed from ice and water deposition associated with glaciers (Martinson and Basko, R E S T O R I N G C O N I F E R S B Y N AT U R A L R E G E N E R AT I O N 201 Table 1: Land-type number and description, land-type characteristics and percentage of road length within each land-type, Lake McDonald section, Going-to-the-Sun Road, Glacier National Park, USA Land-type number and description Characteristics % 1. 2. 3. 4. 5. Moist soil Moist, clayey soil Droughty surface soil; cool, good air drainage Dry soil; cool, poor air drainage Little soil 35 10 7 41 7 Silty and fine sandy glacial till Clayey glacial till Sub-irrigated alluvial fan Glacial drift and outwash Rock and scree 1983). Glacial tills originated from Precambrian argillite, shale and limestone. Volcanic ash deposits, up to 25 cm deep, improved soil development, fertility and moisture-holding characteristics, which promoted vegetation growth in these red cedar–hemlock forests. Martinson and Basco (1983) identified several ‘land-types’ (soils integrated with vegetation, drainage and climate) along this section of road. These (Table 1) were condensed into five general land-types (Shearer and Potter, 1994). Methods Ten sample points with two on each of the five land-types were randomly selected along the Lake McDonald section of the Going-to-the-Sun Road (Shearer and Potter, 1994). Figure 1 shows the approximate location of seed traps and regeneration plots in relation to the road. Conifer seed-fall was estimated using 0.4-m2 rectangular seed traps, four placed on the forest floor at each of the 10 sample points (Shearer and Potter, 1994). Two seed traps spaced ~15 m apart were positioned ~8 m from the edge of the road prism, within the forest above the road cut, and two below the road fill (Figure 1). Depending on availability of personnel and climatic factors, contents of seed traps were emptied from three to nine times per year during the snow-free period from 1987 to 1995, placed into labelled paper sacks and stored at room temperature. Later, seeds were separated by species and X-rayed. From X-rays, seeds were classified as filled (potentially viable with a developed embryo), empty, or damaged. Unless specified, this paper refers only to filled seeds. In 1991, near each of the 10 groups of seed traps, four 0.25-m2 plots were randomly estab- lished approximately at road level on fresh cuts above the ditch (Figure 2). Germination of conifer seeds and seedling mortality were counted once a month, May to September, 1992–1998. Plastic toothpicks were placed at the right of new seedlings for identification, a different colour for each measurement. When seedlings died these markers were removed. At the first measurement in the spring, the multicolour toothpicks marking surviving seedlings from the previous year were replaced with a single colour to identify the year of germination. A final seedling count was made in May 1999. Analysis Seed dispersal data for each conifer species were not normal and were highly variable between years. Each species was analysed using the following: (1) the number of seeds from paired traps averaged for a plot mean; (2) plot means ranked within each year to address large differences between years; (3) yearly ranks for each plot averaged across years to produce an aggregate rank; and (4) possible differences in seed-fall among land-type/road position analysed using KruskalWallis test statistics (Daniel, 1990). Because there were only two plots for each land-type/road position combination, Monte Carlo methods were used to estimate significance levels for Kruskal-Wallis test statistics (Mehta and Patel, 1996). Large numerical differences in seed germination and seedling survival on cuts defied meaningful statistical analysis because of low sample sizes. Hence, data were only summarized. Ratios of number of filled seeds per germinant and per surviving seedling for 1999 data were calculated for each land-type. 202 F O R E S T RY Figure 1. Schematic of Lake McDonald section, Going-to-the-Sun Road, Glacier National Park, Montana, USA shows location of seed traps and conifer natural regeneration plots. Mean road dimensions: pavement width, 6.7 m; shoulders, 0.7 (0.3–0.9) m; front slope to ditch, 1.2 m; ditch, 0.13 m; back-slope measures 1.5–1.8 m at a ratio of 3 : 1 (2 : 1 maximum). Results Seed dispersal About 70 per cent of the western red cedar seeds and 90 per cent of all other species dispersed in September and October before snow began to accumulate. From 1987 to 1995, an average of 6989 total seeds m–2 of all species fell along the road, and 4519 seeds m–2 (65 per cent) of these were filled (Table 2), ranging from 2342 seeds m–2 in 1995 to 0.3 seeds m–2 in 1989. Composition of filled seeds during the study was: red cedar, 74.4 per cent; hemlock, 23.9 per cent; Douglas-fir, 0.7 per cent; larch, 0.5 per cent; spruce, 0.2 per cent; and combined lodge-pole pine, western white pine and sub-alpine fir, 0.2 per cent (Table 2). These percentages were similar to the composition of the basal area: red cedar, 63 (294 m2 ha–1); hemlock, 22 (103 m2 ha–1); Douglas-fir, 6 (28 m2 ha–1); larch, 4 (21 m2 ha–1); and the other four conifers, 5 (23 m2 ha–1). Both basal area and percentage of seeds were greatest below the road for red cedar and spruce, and greatest above the road for hemlock, Douglas-fir and western larch. Western red cedar and western hemlock produced over 98 per cent of all conifer seeds during the 9 years of this study (Table 2). From 1987 to 1995 red cedar dispersed 97 per cent of its seeds in 1993 and 1995, while hemlock distributed 99 per cent in 1987, 1992, 1993 and 1995. Seed-fall from other conifers occurred mostly in 1993 and 1995. Few seeds matured in 1989 or 1994. From 1987 to 1995, significant variation in seed-fall occurred by land-type and road position combinations (Table 3) for western red cedar R E S T O R I N G C O N I F E R S B Y N AT U R A L R E G E N E R AT I O N 203 Figure 2. Section of the Going-to-the-Sun Road showing typical shoulders, ditch and cut slope where four 0.25 m2 natural regeneration plots were established to count conifer seed germination and seedling survival, 7 May 1997, Glacier National Park, Montana, USA. Seed traps were placed in the forest above and below the road. Table 2: Mean number (standard deviation) of filled seeds m–2 dispersed by species and year of cone production in forests paralleling Lake McDonald section, Going-to-the-Sun Road, Glacier National Park, USA Year WRC WH 1987 0.6 (1.5) 565.3 (751.9) 1988 0.1 (0.4) 0.6 (1.6) 1989 0.0 0.0 1990 69.6 (108.1) 4.4 (7.2) 1991 2.5 (5.7) 4.0 (13.2) 1992 32.1 (68.0) 196.4 (285.2) 1993 1107.6 (1337.7) 141.5 (204.4) 1994 5.1 (8.4) 0.7 (1.2) 1995 2146.4 (1740.6) 168.7 (191.6) Total 3364.1 1081.6 DF WL 0.9 (1.6) 2.4 (4.8) 0.1 (0.5) 0.8 (2.0) 0.0 0.0 0.9 (2.5) 1.9 (4.0) 7.1 (8.4) 3.4 (7.4) 0.0 0.3 (0.1) 3.8 (6.6) 11.2 (18.0) 0.0 0.1 (0.4) 18.0 (21.3) 4.1 (8.6) 30.8 24.2 SPR LPP WWP 1.8 (2.5) 0.2 (0.7) 0.0 2.8 (11.8) 0.2 (0.7) 0.0 0.9 (2.6) 0.1 (0.4) 1.5 (2.8) 7.5 0.1 (0.5) 0.4 (1.7) 0.3 (2.0) 0.0 3.1 (5.1) 0.3 (0.8) 0.7 (1.9) 0.2 (0.9) 1.4 (5.7) 6.5 0.0 0.1 (0.5) 0.0 0.1 (0.4) 1.7 (5.0) 0.1 (0.4) 0.4 (1.1) 0.0 1.6 (3.1) 4.0 SAF Total 0.1 (0.4) 571.2 0.0 2.3 0.0 0.3 0.0 79.7 0.0 22.0 0.0 229.2 0.1 (0.5) 1266.2 0.0 6.2 0.4 2342.1 0.6 4484.6 Species abbreviation: WRC, Western red cedar; WH, western hemlock; DF, Douglas-fir; WL, western larch; SPR, spruce; LPP, lodge-pole pine; WWP, western white pine; SAF, sub-alpine fir. seeds (P < 0.01) and for western hemlock seeds (P = 0.02) but not for other conifers (P > 0.15). Red cedar and hemlock composed nearly 99 per cent of filled seeds that fell on land-types 1–4 and 70 per cent of seeds that fell on land-type 5. Red cedar seed-fall ranged from a high of 5836 m–2 on land-type 4 below the road to a low of 44 m–2 on land-type 5 above the road. During the same period hemlock seeds ranged from an average high of over 2700 m–2 on land-types 2 and 4 above the road to an average low of 2 filled seeds m–2 on land-type 5 below the road. 204 F O R E S T RY Table 3: Mean sum (standard deviation) of filled conifer seeds m–2 by species that fell from 1987 to 1995 by land-type and location of seed traps in relation to Lake McDonald section, Going-to-the-Sun Road, Glacier National Park, USA Land-type Location 1. Silty, fine sandy till Above Below Above Below Above Below Above Below Above Below 2. Clayey till 3. Sub-irrigated alluvium 4. Drift and outwash 5. Rock and scree WRC WH DF 3722 (980) 5746 (1298) 3178 (877) 5700 (1180) 4570 (1137) 1802 (386) 2871 (703) 5836 (1292) 44 (8) 97 (18) 1710 (332) 947 (104) 2933 (495) 306 (49) 1004 (152) 432 (61) 2783 (547) 444 (88) 4 (1) 2 (0.4) 14 (2) 28 (6) 88 (20) 41 (8) 33 (7) 45 (7) 14 (3) 27 (7) 10 (1) 7 (2) After road reconstruction in 1991, conifers growing adjacent to the road dispersed abundant seeds in 1992, 1993 and 1995 (Table 2). During the 5 years after treatment, red cedar accounted for 85 per cent of dispersed filled seeds (3294 m–2), hemlock 13 per cent (511 m–2), Douglas-fir 0.7 per cent (29 m–2), larch 0.5 per cent (19 m–2) and the four other species 0.7 per cent (13 m–2). Substantially more filled seeds (4766 m–2) fell on land-types 1–4 than on land-type 5 (87 m–2). There was no statistical difference by land-type for other conifers. Seed germination Germination on road cuts began after snowmelt, usually the second week of May. In 1997, germination began later in May because of more snow. Western red cedar seeds germinated more rapidly than western hemlock seeds. An average of 62, 30 and 5 per cent of red cedar seeds germinated in May, June and July while 23, 52 and 22 per cent of the hemlock seeds germinated in the same period. Too few seeds of other species fell on these sites to make similar comparisons. In spring 1992, few red cedar seeds and no hemlock seeds germinated following the poor cone crop in 1991 (Table 4). Germination increased for both species in 1993 after greater seed-fall in 1992. In 1994 and 1996, 8 per cent of filled red cedar seeds and 10 per cent of filled hemlock seeds germinated following abundant cone production the previous years. For each seedling that germinated in the spring WL 56 (7) 8 (2) 4 (1) 3 (0.4) 48 (9) 6 (1) 61 (9) 46 (5) 7 (2) 3 (1) from 1992 to 1996 on land-types with developed soils, an average of eight filled western red cedar seeds (range 4–48) and 13 filled western hemlock seeds (range 8–21) fell from 1991 to 1995 cones (Table 4). Fewer seeds fell on land-type 5, and only one hemlock and one Douglas-fir seedling were counted there from 1992 to 1996. The percentage of subplots with germinants on landtypes 1–4 increased each year of the 5-year period. For example, red cedar germinated on 50 per cent of the plots in 1993, 88 per cent in 1994, and 100 per cent in 1998; hemlock germinated on 69 per cent of the plots in 1993 and 1994 and 88 per cent in 1996. However, on replication 1, land-type 3, red cedar seeds germinated there only in 1996 and hemlock seeds germinated there in 1993, 1996 and 1998. Seedling survival By 1999, vegetation covered cut slopes treated during reconstruction on land-types 1–4, but not on rocky land-type 5. Surviving conifers were inconspicuous among the post-treatment vegetation. All red cedar and hemlock appeared small and frail. However, the few spruce and Douglasfir seedlings present were more robust on these adverse sites. Conifer seedling survival on road cuts in May 1999 varied greatly. Survival ranged from a high of 130 seedlings m–2 (51 per cent of all living seedlings including three-quarters of all red cedar and over one-third of all hemlock and spruce) on land-type 1 (replication 1) to none on land-types R E S T O R I N G C O N I F E R S B Y N AT U R A L R E G E N E R AT I O N 205 Table 4: On land-types 1–4, the mean number of filled western red cedar (WRC) and western hemlock (WH) seeds m–2 that fell on cut slopes from 1991 to 1995, the number of those seeds that germinated m–2 from 1992 to 1996, and their seeds : germination ratio; also seedling survival m–2 in May 1999 and their seeds : seedling survival ratio, Lake McDonald section, Going-to-the-Sun Road, Glacier National Park, USA Year of dispersal/ germination Seeds dispersed m–2 Seeds germinated m–2 Ratio 1999 seedling survival m–2 Ratio WRC 1991–1992 1992–1993 1993–1994 1994–1995 1995–1996 Total 3.1 40.1 1384.4 6.3 2680.8 4114.7 0.6 5.9 173.9 0.0 354.0 535.3 1:5 1:7 1:8 – 1:8 1:8 0.0 0.1 0.2 – 12.9 13.2 – 1 : 401 1 : 6922 – 1 : 208 1 : 312 WH 1991–1992 1992–1993 1993–1994 1994–1995 1995–1996 Total 4.9 245.5 176.9 0.9 210.9 639.1 0.0 11.9 10.9 0.1 27.6 50.5 – 1 : 21 1 : 16 1:9 1:8 1 : 13 – 0.5 0.0 0.0 3.2 3.7 – 1 : 491 – – 1 : 66 1 : 173 Species 3 (replication 1) and 5 (both replications). Survival was 15 per cent on land-type 2, 9 per cent on land-type 3 and 16 per cent on land-type 4. Each red cedar germinant required an average of eight filled seeds in 1994 and 1996, and each hemlock germinant needed 16 seeds in 1994 and eight in 1996 (Table 4). By May 1999, the number of red cedar seeds per living seedling inflated to 6922, and 208 for seeds dispersed in 1993 and 1995, while each surviving hemlock seedling required 491 and 66 for seeds dispersed in 1992 and 1995. Red cedar and hemlock made up 95 per cent of surviving conifer seedlings in 1999, and spruce, Douglas-fir, and larch composed the remainder. From 1991 to 1995 on landtypes 1–4, seeds : seedling ratios were 312 : 1 for red cedar and 173 : 1 for hemlock. Seed and seedling mortality After snowmelt on land-types 1–4, a slow-moving film of water, resulting from subsurface water disrupted by the road cuts, flowed freely down many exposed cut slopes. This run-off persisted from a few days to several weeks. Initially, many seeds washed from the cuts into the ditch. As vegetation re-established and organic matter accumulated, most seeds were held in place. While soils were saturated, many seedlings died from causes associated with excess water and possibly disease. Between monthly examinations, some seedlings disappeared while others succumbed to disease or animal feeding. When soils dried, usually during July and August, some seedlings perished from lack of moisture. But the greatest decrease in seedlings occurred from October to May when 53 per cent of the red cedar and 40 per cent of the hemlock seedlings died. Some frost heaved before winter snow covered the plots, others were wrenched from the soil or broken by snow creep, and many died from causes associated with excess water in the spring. About 25 per cent of western red cedar and western hemlock seedlings died the year they germinated. Discussion During this 9-year study western red cedar dispersed seed from two outstanding, one moderate, five poor and one failure cone crops, while western hemlock had one outstanding, three good, three poor and two failure cone crops. In 1989, nearly all conifer seed and pollen cones were killed by low temperature and wind chill associated with an extreme Arctic front that swept throughout the Northern Rockies on 1 February (Shearer and Potter, 1994; Shearer and Kempf, 1999). In 1993 and 1995 an average of 1627 red cedar seeds m–2 fell from stands 206 F O R E S T RY composed of 367 m2 ha–1 basal area of red cedar. In 1987 an average of 565 hemlock seeds m–2 were dispersed from the same stands containing 129 m–2 basal area of hemlock. At two sites near Vancouver, British Columbia, Feller and Klinka (1998) report high seed-fall for combined viable red cedar and hemlock seeds m–2 dispersed in 1990 and again in 1994 was about 1000 for one stand with basal area of 173 m2 ha–1 and 200 for a second stand with basal area of 82 m2 ha–1. Ratios of red cedar and hemlock seed-fall to their basal area between the Montana and British Columbia sites are similar for years of good cone crops. After road reconstruction, western red cedar and western hemlock dispersed over 3700 filled seeds m–2 from 1991 to 1995. Most of these seeds fell before mid-November when residual snow usually began to accumulate (Shearer and Potter, 1994). A few seeds continued to disperse through the winter and the next growing season. Seed germination on plots may have been delayed slightly because snow cleared from the road by park crews added to the accumulation on lower cut slopes and required more time to melt (Figure 2). Road cuts are more hostile environments for germination of and survival of conifer seeds than seedbeds on nearby natural surfaces because seeds and seedlings are often subjected to more extreme physical and biological conditions that limit survival. Although many viable seeds dispersed from heavy cone crops in 1993 and 1995, few germinated and only occasional seedlings survived through 1999. Dense mats of mosses that developed on cut slopes seemed to prevent continuing germination of conifer seeds within these clumps. Until cuts re-established vegetation, many seeds were carried off the plots by water that flowed gently across the surface for up to several weeks after snowmelt. High seedling losses from October to May were attributed to frost heaving before snow cover and to snow creep that pulled young seedlings from the ground. During the five years of measurements, eight and 12 filled seeds were required for each germinating red cedar and hemlock. However, by May 1999, that ratio expanded to over 300 red cedar and 160 hemlock seeds for each surviving seedling. Characteristics of moisture and vegetation observed on cut slopes closely fit the land-type definitions shown in Table 1. Land-type 1 maintained wet or moist cut slopes much of the summer and mosses became a major component of the vegetation. Sixty per cent of the surviving conifer seedlings in 1999 grew on land-type 1. Surfaces of land-types 2 and 4 also remained moist much of the summer and sustained a high component of mosses among the vegetation, but each had ~15 per cent of the surviving conifer seedlings in 1999. Land-type 3 was much drier with sparse vegetation and 9 per cent of the surviving conifer seedlings in 1999. Little vegetation of any kind and no conifer seedlings were observed on land-type 5. Once-a-month site visits from May to September were not frequent enough to identify specific causes of most seedling mortality. However, we know a majority of them died over winter and excess water was associated with death of many seedlings in May and June. Other seedlings failed to survive drying soils from late July to early September most years. Conclusions Cut slopes exposed in 1991 during reconstruction of the Lake McDonald section, Going-to-the-Sun Road, created uncertain environments for restoring conifers on five land-types. In the first 5 years following treatment, plenty of potentially viable seeds fell on the exposed backslopes. Only a small percentage of these seeds germinated, early seedling survival was low, and initial seedling development of surviving conifer seed was slow. Eight years after road reconstruction, conifer seedlings were not conspicuous among dominant shrubs and forbs developing on the cut slopes, and it is likely that it will be many years before they become evident and add greater variety to the vegetation along this scenic road. The hypothesis that seed-fall from bordering forest trees is sufficient to regenerate conifers that rapidly add texture to other vegetation is rejected if the object is to quickly attain this goal. Tree planting with nursery stock that is compatible with these harsh sites will be necessary. However, if speed is not an objective and the manager is satisfied to know natural conifer regeneration is present and increasing slowly below overtopping vegetation, then this hypothesis is accepted. R E S T O R I N G C O N I F E R S B Y N AT U R A L R E G E N E R AT I O N Acknowledgements We are indebted to Rachel Potter and Laurie Kurth for supervising data collection; Leon Theroux and Jack Schmidt for seed identification, X-ray and analysis, and data handling; Rudy King for statistical advice and methods for data analysis; Kristen Schaffer and Carol Wolfe for helping with manuscript preparation; and James Cramer for figures. References Barrett, S.W. 1988 Fire history of Glacier National Park, McDonald Creek Basin. Unpublished report. US Department of the Interior, National Park Service, Glacier National Park, West Glacier, MT, 32 pp. Boe, K.N. 1954 Periodicity of cone crops for five Montana conifers. Montana Acad. Sci. Proc. 14, 5–9. Bradshaw, A.D. 1997 What do we mean by restoration? In Restoration Ecology and Sustainable Development. K.M. Urbanska, N.R. Webb and P.J. Edwards (eds). Cambridge University Press, Cambridge, pp. 8–14. Daniel, W.W. 1990 Applied Nonparametric Statistics, 2nd edn. PWS-Kent Publishing Co., Boston, MA. Eyre, F.H. (ed.) 1980 Forest Cover Types of the United States and Canada. Society of American Foresters, Bethesda, MD. Feller, M.C. and Klinka, K. 1998 Seedfall, seed germination, and initial survival and growth of seedlings of Thuja plicata in southwestern British Columbia. Northw. Sci. 72, 157–169. Ferguson, D.E. 1994 Natural regeneration following timber harvest in interior cedar-hemlock-white pine forests. In Proceedings Interior Cedar-HemlockWhite Pine Forests: Ecology and Management Symposium. D.M. Baumgartner, J.E. Lotan and J.R. Tonn (compilers and eds). 2–4 March 1993, Spokane, WA, pp. 249–256. Finklin, A.I. 1986 A climatic handbook for Glacier National Park—with data for Waterton Lakes National Park. USDA Forest Service, Intermountain Research Station, General Technical Report INT-204. Gashwiler, J.S. 1969 Seed fall of three conifers in westcentral Oregon. For. Sci. 15, 290–295. Graham, R.T. 1990 Pinus monticola Dougl. Ex D. Don, western white pine. In Silvics of North America. I. Conifers. R.M. Burns and B.H. Honkala (tech. coordinators). Agric. Handbook 654, USDA Forest Service, Washington, DC, pp. 385–394. Habeck, J.R. 1968. Forest succession in the Glacier Park cedar-hemlock forests. Ecology 49, 872–880. Haig, I.T., Davis, K.P. and Weidman, R.H. 1941 Natural regeneration in the western white pine type. Tech. Bull. 767. US Department of Agriculture, Washington, DC, pp. 99. Jõgiste, K., Vares, A. and Sendrós, M. 2003 Restoration 207 of former agricultural fields in Estonia: comparative growth of planted and naturally regenerated birch. Forestry 76, 209–219. Lockhart, B.R., Keeland, B., McCoy, J. and Dean, T.J. 2003 Comparing regeneration techniques for afforesting previously farmed bottomland hardwood sites in the Mississippi Alluvial Plain, USA. Forestry 76, 169–180. Martinson, A.H. and Basko, W.J. 1983 Flathead County, Land System Inventory: a soil resource inventory and analysis for land-use planning and resource allocation. Unpublished report. USDA Forest Service, Flathead National Forest. Kalispell, MT. Mehta, C.R. and Patel, N.R. 1996 SPSS Exact Tests 7.0 for Windows. SPSS, Inc. Chicago, IL. Minore, D. 1990 Thuja plicata Dougl. Ex D. Don, western redcedar. In Silvics of North America. I. Conifers. R.M. Burns and B.H. Honkala (tech. coordinators). Agric. Handbook 654, USDA Forest Service, Washington, DC, pp. 590–600. Packee, E.C. 1990 Tsuga heterophylla (Raf.) Sarg., western hemlock. In Silvics of North America. I. Conifers. R.M. Burns and B.H. Honkala (tech. coordinators). Agric. Handbook 654, USDA Forest Service, Washington, DC, pp. 613–622. Pfister, R.D., Kovalchik, B.L., Arno, S.F. and Presby, R.C. 1977 Forest habitat types of Montana. USDA Forest Service, Intermountain Forest and Range Experiment Station Gen. Tech. Rep. INT-34, 174 pp. Schmidt, W.C., Shearer, R.C. and Roe, A.L. 1976. Ecology and silviculture of western larch forests. USDA Forest Service Tech. Bull. 1520, 96 pp. Shearer R.C. 1976 Early establishment of conifers following prescribed broadcast burning in western larch/Douglas-fir forests. In Proceedings Tall Timbers Fire Ecology Conference No. 14, Missoula, Montana. Tall Timbers Research Station, Tallahassee, FL, pp, 481–500. Shearer R.C. and Kempf, M.M. 1999 Coram Experimental Forest: 50 years of research in a western larch forest. USDA Forest Service, Rocky Mountain Research Station Gen. Tech. Rep. RMRS-GTR-37, 66 pp. Shearer, R.C. and Potter, R.W. 1994 Conifer seedfall in the cedar-hemlock forest near Lake McDonald, Glacier National Park, Montana. In Proceedings Interior Cedar-Hemlock-White Pine Forests: Ecology and Management Symposium. D.M. Baumgartner, J.E. Lotan and J.R. Tonn (compilers and eds). 2–4 March 1993, Spokane. WA. Pullman, Washington: Washington State University, pp. 249–256. Shearer, R.C., Potter, R.W., Kurth, L.L. and Asebrook, J.M. 1996 Cooperation enhances revegetation efforts in Glacier National Park. Park Sci. 16, 20–21. Whisenant, S.G. 1999 Repairing Damaged Wildlands: a Process-Oriented Landscape-Scale Approach. Cambridge University Press, Cambridge, 312 pp.