Introduction

The riparian area is the boundary between terrestrial and aquatic ecosystems, and riparian vegetation affects many aquatic ecosystem structures and functions (Naiman et al. 2005; Richardson et al. 2005). For instance, riparian vegetation modifies light and temperature regimes, alters water quality, provides food and habitat for aquatic and terrestrial consumers, stabilizes streambanks, and provides coarse wood to streams, which can alter fish habitat, channel morphology, and sediment routing (Kiffney et al. 2003). For many of the above reasons it has become common practice to retain some amount of riparian vegetation cover along streamsides (Richardson et al. 2012).

Partial harvest is one of the forest management tools to provide harvest opportunities and maintain some of the functions of riparian vegetation (Lecerf and Richardson 2010). Partial canopy retention is effective in retaining some biological elements, providing residual structure, refugia, and a local source of propagules for disturbance-sensitive species, and creating barriers to invasion of light-demanding and non-native species (Nelson and Halpern 2005). Because management history and condition of the surrounding managed stand determine the function of vegetation, riparian management practices would be desired to contribute to maintaining ecologically sound wood production. With knowledge of the effects of different harvest disturbances on riparian vegetation, forest managers would be able to choose silvicultural strategies that balance commodity production and conservation objectives.

Seed germination and seedling establishment is one of the most vulnerable transitions in a plant’s life cycle (Gómez-Aparicio et al. 2005), and seedling establishment is affected by many factors, such as climate (Gómez-Aparicio et al. 2005), canopy (Keyes et al. 2001) and groundcover (Sugita and Tani 2001). Environmental factors are highly heterogeneous in space and time, and suitable microsites for seedling growth differ among species. An examination of which microhabitats, under which environmental conditions constitute safe sites for seedling establishment, is essential to consideration of the establishment and trajectories of future forests (Gómez-Aparicio et al. 2005).

The forest floor is composed of various microsites (e.g., bare soil, leaf litter, coarse wood, mound, and pit). Coarse wood, such as fallen logs, large branches and stumps, is an important establishment site for tree regeneration, serving as a “nurse log” (Harmon and Franklin 1989; Kennedy and Quinn 2001). Because fallen logs are elevated above the ground surface, they have some advantages, i.e., seedlings growing on the logs can avoid competition with other plants, being buried in litter accumulation, or being consumed (Callaway 1995; Swanson et al. 2011).

The partial harvest of the overstory trees would bring some changes, and it could have significant effects on the understorey. First, some fallen logs and stumps resulting from the harvest would create favorable sites for tree regeneration. Second, resource availability and physical disturbance would increase with formation of canopy gaps. The increases in resource availability change interspecific competition, and promote germination from the seed bank and regeneration of suppressed trees (Narukawa and Yamamoto 2001), and this change promotes invasion of ruderal species, shade-intolerant species, and introduced species (Nelson and Halpern 2005; González-Alday et al. 2009).

Rubus spectabilis Pursh (Rosaceae) and Gaultheria shallon Pursh (Ericaceae) are common shrubs in the understory in coastal forests of the Pacific Northwest of North America (Naiman et al. 2000), and their dense undergrowth affects forest vegetation. Rubus spectabilis, a deciduous shrub, is common in moist sites, Alnus rubra stands and riparian stands (Pojar and MacKinnon 2004). Gaultheria shallon, an evergreen shrub, occurs on somewhat drier, more interior sites than R. spectabilis. A greater amount of light reaches the forest floor following disturbance to the overstory. G. shallon and R. spectabilis rapidly respond to such disturbances and can produce pure, dense (>20,000 stems ha−1) cover (Tappeiner et al. 1991; Mallik 1995; Tappeiner et al. 2001). These species also expand their cover from the seed bank. Tappeiner and Zasada (1993) documented that the emergence of G. shallon and R. spectabilis was greater on mineral soil than on undisturbed soil because of their small seed with relatively low stored energy.

Despite several studies of light intensity related changes to forest floor vegetation have been conducted (Mallik 1995; Tappeiner et al. 2001), influences of alteration of light availability caused by a partial harvest in riparian vegetation on the dynamics of shrubs that form dense cover is not studied. Also, there is insufficient knowledge of the competitive relationship between shrub cover and tree seedlings, and of the difference in the responses among species of tree seedlings there. The present study addresses this lack of knowledge by discussing the following questions: What are changes to forest floor vegetation, particularly shrub species, after 50 % partial harvest? What is the difference in distribution of the species along a stream to upland gradient? Finally, what is the potential difference in tree seedling establishment and growth with gap formation? We hypothesised that dense cover of G. shallon and R. spectabilis could be found in partial harvest sites and these shrub species could constrain the distribution of conifer seedlings. This may consequently affect the structure and species composition of forest stands (Fraser et al. 1995; Mallik 1995; Kennedy and Quinn 2001; Villarin et al. 2009). Based on these underlying assumptions, we examined how partial harvest affected G. shallon and R. spectabilis densities, the relationship between the differences of G. shallon and R. spectabilis densities, and the distribution of conifer seedlings.

Materials and methods

Study site

We conducted this study in the Malcolm Knapp Research Forest of the University of British Columbia (122°34′W, 49°16′N), Maple Ridge, BC, Canada (Fig. 1). The forest is in the foothills of the Coast Mountains, approximately 60 km east of Vancouver. The forest lies in the coastal western hemlock biogeoclimatic zone. The dominant forest species are Tsuga heterophylla (Raf.) Sarg. (Pinaceae), Thuja plicata Donn ex D. Don (Cupressaceae), and Pseudotsuga menziesii (Mirb.) Franco (Pinaceae) (Kiffney et al. 2003). The forest was logged before 1931 and the current stand resulted from a stand-replacing fire in 1931 (The University of British Columbia 2010), which is now dense second-growth forest (550–650 trees ha−1, average diameter at breast height 40 cm; average stand height 45 m; Kiffney et al. 2003). Soils are shallow and dominantly podzols formed in ablation until colluviums were overlying relatively impermeable basal till or granitic bedrock (Gomi et al. 2006). The forest receives much precipitation from ca. 2,200 mm per year at the southern end of the forest to ca. 3,000 mm per year at the north end (The University of British Columbia 2010). More than 70 % of the total annual precipitation falls between October and March, mostly in the form of rain (Kiffney et al. 2003).

Fig. 1
figure 1

Location of the Griffith Creek study area in MKRF, southwest British Columbia. Three bold bars along the creek show each study site, the Harvested, Border and Control, from the bottom

Full size image

The study site at Griffith Creek is part of a riparian management experiment at the Malcolm Knapp Research Forest (Richardson et al. 2010). Griffith Creek is a typical headwater stream. As a part of the riparian project to test the effectiveness of riparian management treatments, one treatment was a partial harvest (removal of 50 % of the stems based on basal area in the riparian and adjacent areas of the experimental streams), which was applied along Griffith Creek in 2004. The harvested trees were skidded from the site to load on trucks at the road, but some coarse woody debris (branches or small logs) remained. We chose this area as the study site to examine the effects of a partial harvest on regeneration of conifers. Multiplots should be required for replication, nonetheless, the present study was based on a single sample at a single location, because there was no site with similar conditions around the study site.

We established three sites; harvested, border, and control sites (unreplicated), along the creek (Fig. 1). The intersite distance was approximately 100 m at ca. 400 m asl. The harvested site is the partial-harvested area in 2004. We chose an unharvested area as the control site that was adjacent to and the same age and same vegetation (Richardson et al. 2010) as the partial-harvested area because pre-harvest data were unavailable. The border site is set along the edge of the harvested site (upper end of the cutblock), and we established the border site where harvest effects might be indirect, or some changes such as edge effects would affect seedlings. These three sites were set on the left bank of the creek and their mean slope angles were 8.4 % (the harvested site), 13.9 % (the border site), and 11.8 % (the control site), respectively.

Sampling design

We established a 1 m × 45 m transect along the environmental gradient from the creek to the upper slope on each site. Small plots (1 m × 1 m) were established along each transect for counting seedlings and other major understorey plants. The first small plot was established immediately adjacent to the creek and successive plots were established at 4-m intervals, for a total of ten small plots in each transect. All seedlings encountered on each small plot were counted, and species, height, and their substrate types as a microsite for their growth were recorded. Seedlings were defined roughly as individuals <25 cm in height (ground to apical tip; Hughes et al. 2009). The substrate type consisted of leaf litter, fallen log, or stump. Major understorey plant covers were estimated visually as the percentage of each small plot’s ground area covered by the plants. Percentage of canopy openness was quantified at four points (10, 20, 30, and 40 m from the creek) on each transect based on a spherical densiometer (Model-A, Forest Densiometers, USA). Then we calculated percentage of mean canopy openness of each transect.

There were many patches of G. shallon and R. spectabilis in the harvest site. In order to evaluate the potential influence of G. shallon and R. spectabilis on seedlings, we recognized patches in which either one shrub species or the other was observed. We also classified patches of the shrub species based on percent cover. The classification into five cover classes was as follows: patches with <10 % total cover of both species (open); patches with 10–80 % cover of G. shallon or R. spectabilis (SG, sparse G. shallon; SR, sparse R. spectabilis); and patches with ≥80 % cover of G. shallon or R. spectabilis (DG, dense G. shallon; DR, dense R. spectabilis). Ten additional small plots (1 m × 1 m) were placed randomly on the patches with each cover class (open, SG, SR, DG, DR) within the harvested site (50 small plots in all, not on the transects). We recorded species and height of seedlings established in each small plot. We measured soil water content with an auto-analyzer (Hydrosense CD620 + CS620, Campbell Scientific, Australia) in the 50 small plots. Samples of the soil were collected from each small plot for analysis of soil chemistry. The samples were dried and sieved with a 2 mm mesh sieve in the laboratory. The samples of 20 g (dry mass) were analyzed for ammonia–N, nitrate–N, and phosphate–P in the Applied Environmental Research Laboratories at Vancouver Island University, in Nanaimo, BC.

Statistical analysis

We tested the difference in the height of each coniferous seedling established on the harvested, border, and control sites, and also the difference in the total number of seedlings and the number of each coniferous seedling established in three different cover classes of G. shallon (open, SG, DG) and R. spectabilis (open, SR, DR), respectively, using the Steel-Dwass test, which is a nonparametric multiple comparison test. We also tested the differences in the soil chemistry between three different cover classes of G. shallon and R. spectabilis, using the Tukey-Kramer test, which is a parametric multiple comparison test. All statistical analysis was conducted with KyPlot (version 5.0, Kyenslab Inc., Tokyo, Japan).

Results

The highest mean canopy openness was found in the harvested site (37.05 ± SD 5.80 %), and it was lower in the border (25.09 ± SD 7.58 %) and control (22.83 ± SD 3.98 %) sites.

Gaultheria shallon and R. spectabilis were the most common species, appearing in all sites. G. shallon had the highest cover in the border and control sites, but R. spectabilis showed the highest cover in the harvested site (Table 1). Vaccinium spp. were also observed in all sites, but the cover was not high. We found the mean community height of R. spectabilis (74.59 ± SD 47.32 cm) with upright stems was much taller than G. shallon (28.35 ± SD 12.90 cm) with sprawling stems in our study site.

Table 1 Percent mean cover (SD) of understory species in the study sites
Full size table

Cover of G. shallon varied markedly between small plots on all sites regardless of the distance from the creek (Fig. 2). R. spectabilis was found within 20 m from the creek (small plot # 4), and the highest cover (90 %) was found at 15 m point from the creek in the harvested site (Fig. 2). In addition, the cover decreased with proximity to the upper slope from the creek in all sites.

Fig. 2
figure 2

Difference in cover (%) of major two shrub species in each small plot established according to the distance from the creek in each study site

Full size image

The greatest number of seedlings could be observed in the harvested site, 81 individuals·10 m−2 of T. heterophylla, 146 individuals·10 m−2 of T. plicata, and 45 individuals·10 m−2 of P. menziesii were found (Fig. 3). There also were many T. heterophylla seedlings (83 individuals·10 m−2) in the border site. Seedlings of T. plicata and P. menziesii were rarely found in the border site, and no seedling in the control site. There were almost all T. plicata and P. menziesii seedlings on leaf litter, and only one individual of both species was found on fallen logs. T. heterophylla seedlings were also found on leaf litter commonly, and 15 % of the seedlings were also found on woody substrates in the border site.

Fig. 3
figure 3

Number of individuals of three coniferous species established on microsites. H, B and C show harvested, border, and control sites, respectively. No coniferous seedlings were found in the control site

Full size image

The mean height of T. heterophylla in the harvested site was significantly higher than in the border site (4.8 ± SD 3.0 cm, 2.4 ± SD 2.5 cm, P < 0.001, Steel–Dwass test), and the mean height of T. plicata in the harvested site was also significantly higher than in the border site (2.9 ± SD 2.5 cm, 1.3 ± SD 1.0 cm, P < 0.05, Steel–Dwass test), but the mean height of P. menziesii showed no significant difference between sites (5.4 ± SD 3.8 cm, 3.4 ± SD 1.9 cm for the harvested and border sites, respectively, Steel–Dwass test). The distribution of all species in the harvested site showed an almost inverse-J shape (Fig. 4). In the border site, a weak inverse-J shaped distribution of sizes can be found for T. heterophylla but not for the other two species.

Fig. 4
figure 4

Height class (cm) distribution of the three coniferous species in the Harvested and Border sites. As no coniferous seedlings were found in the Control site (Fig. 2), bars were not depicted in the figure

Full size image

Tsuga heterophylla and T. plicata seedlings showed an almost inverse-J shaped distribution in the open, SG, and SR cover class, but the number of P. menziesii seedlings in <5 cm height class was fewer than the other two species (Fig. 5). While a number of seedlings were observed in the DG cover class, almost no seedlings could be found in the DR cover class. Significant differences were found in T. heterophylla and T. plicata with <5 cm height class between the SR (10.4 ± SD 12.6 and 14.9 ± SD 14.5) and DR (0.2 ± SD 0.6 and 0.4 ± SD 1.3) cover classes, and open (14.3 ± SD 25.6 and 9.3 ± SD 13.6) and DR (0.2 ± SD 0.6 and 0.4 ± SD 1.3) classes (P < 0.01 in T. heterophylla, P < 0.05 in T. plicata, Steel–Dwass test). In the case of all seedlings of three coniferous species together, significant differences were also found between SR (32.7 ± SD 25.1) and DR (2.3 ± SD 3.3) cover classes, and open (36.5 ± SD 43.6) and DR (2.3 ± SD 3.3) classes (P < 0.05, Steel–Dwass test).

Fig. 5
figure 5

Frequency distribution of height (cm) of seedlings associated with shrub cover, open, SG, DG, SR, DR sites (each with n = 10). White, dark gray and black bars show T. heterophylla, T. plicata, and P. menziesii, respectively. Abbreviations for each cover class are SG sparse G. shallon, DG dense G. shallon, SR sparse R. spectabilis and DR dense R. spectabilis. Sparse and dense were defined as 10–80 % cover and ≥80 % cover, respectively, and open was defined as <10 % cover of both species. Same letters on bars indicate significant differences (a and c, P < 0.01; b and d, P < 0.05, Tukey–Kramer test)

Full size image

The soil properties did not significantly differ among the five sites, except for Nitrate–N among SR (4.25 ± SD 7.51 mg kg−1) and DR (29.48 ± SD 32.68 mg kg−1) cover classes, and open (4.42 ± SD 9.14 mg kg−1) and DR (29.48 ± SD 32.68 mg kg−1) cover classes (P < 0.05, Tukey–Kramer test) (Table 2). Water content, Ammonia–N and Nitrate–N, were relatively high in the DG and DR cover classes compared with the open SG and SR cover classes but not with significant differences (P > 0.05).

Table 2 Mean (SD) of some chemical characteristics of the soils associated with shrub cover; SG, SR, DG, DR, and open sites (each with n = 10)
Full size table

Discussion

The comparison of total number of seedlings and their height observed in the three sites showed that the light availability might promote seedling establishment and growth. Canopy openness is a good surrogate of light availability (Rich et al. 1993; Easter and Spies 1994; Machado and Reich 1999); therefore, the highest mean canopy openness in the harvested sites and lower in the other sites show that light availability to the understory layer increased through the manipulation of the harvest. The overstory vegetation inhibited regeneration of seedlings (Keyes et al. 2001), and high-light conditions created by gap formation or harvest promoted seedling establishment and growth (Narukawa and Yamamoto 2001; Boudreau and Lawes 2005). We found that only 15 % of variation in canopy openness affected distribution of conifer seedlings (Fig. 3).

The different distribution pattern among the tree species to the partial harvest could be due to different tolerance to water and light. Previous studies (Minore and Weatherly 1994; Villarin et al. 2009) conducted in riparian areas in the Pacific Northwest documented that the absence of P. menziesii seedlings might be attributed to intolerance of flooding and shallow water tables and preference to dry sites. In addition, P. menziesii has the lowest shade-tolerance of the three species (Harrington 2006). Higher canopy openness brought by the 50 % partial harvest in our study may cause increased light availability (Machado and Reich 1999). It was relative enough to enhance of P. menziesii seedlings, but the small size seedlings (<5 cm) were not found to be as many as others. Thus, the partial harvest promoted an increase of especially T. heterophylla and T. plicata seedlings. It is known that light requirements for survival and growth is in order from P. menziesii to T. plicata to T. heterophylla (Harrington 2006).

Leaf litter was favorable for growth of T. heterophylla, T. plicata, and P. menziesii. Many studies have documented the importance of rotten wood for seedling establishment (Christy and Mack 1984; Harmon and Franklin 1989; Minore and Weatherly 1994; Hibbs and Giordano 1996; Beach and Halpern 2001). The forest floor condition would be one of the important factors that determine suitable substrates for seedling growth. Indeed, fallen logs were not suitable because many logs felled by the harvest were still fresh, and they were not valuable as the nurse logs in our study sites. Leaf litter (Hovstad and Ohlson 2009) and moss (Harmon and Franklin 1989; Nakamura 1992; Katsumata et al. 2008) have both facilitative and inhibitory effects on seedling establishment. The effect of leaf litter on seedling establishment depends on edaphic conditions, climate, and type and quantity of litter (Hovstad and Ohlson 2009). In our study site, litter accumulation was formed by small needles of T. heterophylla, and the litter layer was not thick. Moss cover was small and not thick in our study site. Therefore, the litter layer would not cause inhibitory effects such as physical obstruction and dark condition for germination.

When light levels increase, not only will tree seedlings grow well, but other plant species will also accrue. After retention harvest and gap formation in the Pacific Northwest, observations generally describe a decrease in forest herb and shrub richness, and an increase in early seral species and introduced species, which would be competitors with tree seedlings (Nelson and Halpern 2005; Fahey and Puettmann 2007; Ares et al. 2009). We recognized great cover of R. spectabilis in the harvested site. As Minore and Weatherly (1994) and Villarin et al. (2009) showed, R. spectabilis develops dense cover with increased proximity to the creek. In addition, G. shallon often forms dense cover with a wide distribution in our study site. Establishment of the seedlings appeared to be restricted by the competition with these shrubs.

The shrub species, particularly R. spectabilis cover decreased along the creek to upland, and seedlings could rarely be found in the higher cover patches of the shrubs. This indicated that the dense shrub cover may interfere with conifer regeneration near the creek. The possible causes of the poor establishment of coniferous seedlings under dense shrub would be their competition for resources, such as light conditions. Dosono and Nyland (2006) described that Rubus spp. have interference effects on tree regeneration because they shade the ground, intercept and transpire water, and reduce the rate of litter decomposition and nutrient cycling. Other studies have indicated that the inhibitory effect was primarily due to competition for nutrients (Fraser et al. 1995; Mallik and Prescott 2001; Bennett et al. 2003), but soil chemical properties did not differ significantly between three sites, i.e., neither sparse nor dense cover of G. shallon and R. spectabilis have a potential to affect soil nutrient availability in the present study. This result suggested that a competitive influence of these shrubs on coniferous seedlings was to shade ground (Dosono and Nyland 2006). We have gained new insights that increasing light availability has not always facilitated growth of coniferous seedlings, such as P. menziesii as confirmed in this study, if anything, it may accelerate expansion of shrub cover rather than the regeneration of tree seedlings so that P. menziesii seedlings, especially the shorter individuals, have not been observed in the dense shrub cover caused by their competition demanding light resources.

As observed previously by Minore and Weatherly (1994), competition from dense G. shallon cover did not seem to be more severe than dense R. spectabilis cover. R. spectabilis could strongly affect tree seedlings establishment due to two factors. First, differences in seed dormancy would affect their growth after disturbance and inhibitory effects. G. shallon seeds have little or no dormancy, but R. spectabilis seeds have deep dormancy and remain in a buried seed bank for many years (Tappeiner and Zasada 1993). When light levels are increased by a disturbance, R. spectabilis would regenerate more quickly, and the effect can result in its persistence as the dominant understory species for long periods of time. Second, the ability of R. spectabilis to shade seedlings would be stronger than G. shallon. Thick cover of R. spectabilis might strongly reduce the access of seedlings to light as compared with G. shallon. Both factors may be responsible for the inhibitory effect.

Dosono and Nyland (2006) reported that interference effects of Rubus spp. decreased at 5–7 years, so seedlings could grow through the Rubus spp. However, they also pointed out that very poorly drained sites would likely prove difficult for regenerating tree seedlings. In our study site, we found that R. spectabilis developed densely along the creek side. Thus, shading by shrub species is considered a possible reason for limiting the regeneration of conifer species; however, further experiments are needed to test this.

We recognized that dense cover of G. shallon and R. spectabilis was found in the harvested site, and also there were lesser numbers of conifer seedlings in the dense shrub cover. These results supported out hypothesis that dense cover of shrubs could be found in partial harvest sites, and they could constrain the distribution of conifer seedlings. In the dense R. spectabilis cover observed near creek side, we have considered that distribution of conifer seedlings was restricted by competition with especially exuberant foliage of R. spectabilis for light conditions. This suggests that harvesting near creek side should be constrained, at least 30 m from a stream, in riparian management.