Effects of gap size , gap age , and bamboo Fargesia denudata on Abies faxoniana recruitment in South-western China

Aim of study: to study the effects of gap size, gap age and bamboo Fargesia denudata on natural regeneration of Abies faxoniana, both of which are the ubiquitous dominants in our research area. Area of study: subalpine coniferous forests in Wanglang Natural Reserve in Southwestern China. Material and Methods: 10 transect belts were randomly established, and a total of 97 gaps were recorded and used. Main results: (1) the number of bamboos with coverage of <17% significantly increased with increases of gap size and age, but the latter had little influence on the numbers of F. denudata with coverage of >17%. (2) F. denudata strongly inhibited A. faxoniana seedlings and saplings in small, young and old gaps, where the amount of A. faxoniana recruitment was relatively abundant, than in other types of gap. (3) The numbers of A. faxoniana seedlings in A-gaps, significantly decreased with the increases in gap size. However, in gaps where F. denudate was also present, A. faxoniana seedlings and saplings were insensitive to gap size or age. Research highlights: thick F. denudata would not be influenced by gap size or age. Because of the low occurrences of A. faxoniana seedlings and saplings, the negative effect of gap size, gap age and F. denudata on A. faxoniana recruitment was unclear


Introduction
Canopy gaps, known as spatial and temporal environmental heterogeneity in forests (Denslow, 1987;Connell, 1989;Ostertag, 1998), facilitate the establishment and growth of understory vegetation (Ehrenfeld, 1980;Huenneke, 1983) by creating new open spaces and releasing resources (Gray & Spies, 1997;McGuire et al., 2001;De Chantal et al., 2003).Gaps are favorable not only for light-demanding species like bamboos and shrubs (Taylor & Qin, 1988a), but also for shadetolerant coniferous regeneration as well (Gaudio et al., 2008).The crucial effect of gap characteristics on the understory plants has been reported in many studies (e.g.Runkle, 1985;Takahashi, 1997;Narukawa & Yamamoto, 2001).For one thing, gap size is thought to be very important in maintaining the species diversity and tree regeneration success (Brokaw, 1985;Pearson et al., 2003;Li et al., 2005;Albanesi et al., 2008).For example, higher light availability in large gaps primarily facilitates coniferous regeneration's competitors, e.g.shrubs, broad -leaved tree seedlings, and bamboos (Sapkota & Oden, 2009), which would reduce the light received by coniferous regeneration.Moreover, large gaps reduce soil moisture and raise the air temperature, and therefore increase the mortality of young seedlings (Darabant et al., 2007).For another, gap age, which is known as gap phase or temporal heterogeneity of gap disturbance, directly relates to the microsite situations and consequently influences on the establishment and growth of understory vegetation (Barik et al., 1992;Chandrashekara & Ramakrishnan, 1994;Kirchner et al., 2011;Kinny et al., 2012;Zhang et al., 2013).With the gap environment develops, the gradually lower light levels would lead to inter-species 2 (1) Whether the gap size and age would influence the numbers and coverage of F. denudata?(2) What are the effects of gap size and age for A.
faxoniana seedlings and saplings in gaps with bamboos?(3) What are the effects of F. denudata for A. faxoniana seedlings and saplings in different sized or aged gaps?

Layout of the study area
This study was conducted in old-growth Abies faxoniana-Picea purpurea mixed forests in Wanglang Natural Reserve in southwestern China, which covers a total area of 325 km 2 (31 o 43'-33 o 03' N, 103 o 49'-104 o 59' E, altitude 2,300-4,980 m).The mean annual temperature is 11.4-16.8o C, and the mean annual temperature sum (with daily temperature above 0 o C) is about 5,366.6 d.d.The mean annual precipitation is about 1,100 mm, and 58.51% of the annual precipitation takes place from May to September (Zhao et al., 2012).
The study area situated in a typical subalpine coniferous forest in Wanglang Natural Reserve.The terrain is steep and dissected.Limestone is the main parent material, and mountain brown dark coniferous forest soil is the dominant soil type in our study area.The typical vegetation is A. faxoniana-Picea purpurea forest associated with Sabina saltuaria, Betula utilis and F. denudata, etc., and with a ground cover of mosses and grasses (Taylor et al., 2006).The shrub layer mainly consists of Lonicera maackii, Rosa davidii Crep.Var.Davidii, and Sorbus Koehneana Schneid.The herb layer is comprised of Asteropyrum, Kingdonia, etc.The dominant mosses in forests are mostly Pottiaceae, Plagiomnium, Brachytheciaceae, and Hylocomiaceae (Li et al., 2012).Based on our previous studies, the mean stand height and diameter at breast height (D.B.H.) of A. faxoniana trees in our studied areas are 17 m and 20.4 cm, respectively.The stand density is 950 trees N/ha, and the basal area is between 14.8 and 25.7 m 2 /ha.The ages of the A. faxoniana trees ranged from 60 to 230 years.

Field measurements and data analyses
To study the effects of gap size and age on A. faxoniana recruitment and F. denudata, 10 located transect belts were established in Wanglang from May to June in 2013.The transects (20 m × 100 m for each) were established randomly but in regions with the same altitude, slope and aspect.We also recorded those gaps competing for light (Bullock, 2000), and consequently hinder the establishment of seedlings and saplings.To clarify, gap environments determined by various gap characteristics usually contribute to different species coexistence patterns underneath.
The subalpine coniferous forest is of important ecological and economic value in many countries, such as Chile (Veblen et al., 1981;Gonzalez et al., 2002), North America (Runkle, 1981), Japan (Nakashizuka, 1987;Yamamoto, 1993;Narukawa & Yamamoto, 2002), and China (Taylor & Qin, 1988b;Taylor et al., 1996;Liu & Wu, 2002;Dang et al., 2010;Guo et al., 2013).Take the one selected in Southwestern China for example, Abies faxoniana Rehd.et Wils. is the dominant tree species and bamboo Fargesia denudata is an understory dominant.Where this dwarf bamboo is dense, the seedlings and saplings of A. faxoniana are often scarce and concentrated on raised surfaces (Holz & Veblen, 2006;Taylor et al., 2006).Based on these observations, the aim of this work was to study the effects of gap size, gap age and F. denudata on A. faxoniana seedlings and saplings.For this purpose, we randomly established 10 transect belts located in the Wanglang Natural Reserve, and then inventoried the number, coverage of F. denudata, as well as the numbers of A. faxoniana seedlings and saplings in 97 natural formed gaps.We tried to elucidate the following issues: 3 Effects of gap size, age, bamboo on Abies faxoniana recruitment ences.All statistical analyses were performed using R-2.11 software (http://www.R-project.org/).

Effects of gap size and age on the characteristics of F. denudata
Firstly, gap size exerted important effect on F. denudata in Sn-gaps while it had little effect on F. denudata in Ss-gaps and F-gaps.It was clear that the numbers and coverage of F. denudata significantly increased as the Sn-gaps got larger (p<0.05; Figure 1).However, no matter what the gap size was, the numbers and coverage of F. denudata in Ss-gaps and F-gaps were constantly high which distributed near the transect belts (no more than 20 m in distance).In each transect belt, we recorded the sizes and ages of the gaps.The size of each gap was determined by subdividing the irregularly shaped gaps into smaller sections and then measuring each section (de Lima, 2005).Gap age was calculated from the degree of decomposition of the gap makers (Liu & Hytteborn, 1991;Schnitzer & Carson, 2001).
Each gap was divided diagonally into four quadrants (east, north, west, south), then 4 subplots were randomly set in these quadrants and another subplot was selected at the center of the gap.The sizes of sampled plots in small gaps varied from 4 m 2 to 16 m 2 , in medium-size gaps varied between 16 and 49 m 2 , while in large gaps varied between 49 and 100 m 2 (Zhang et al., 2013).Longitude, latitude, altitude, slope and aspect of each gap were recorded at the gap center.The length and diameter of logs, and stumps were also measured.The decomposition grades of logs were recorded according to the eight-grade scale (Liu & Hytteborn, 1991;Kirchner et al., 2011).Meanwhile, in each quadrant plot, the number of A. faxoniana regeneration, and the number and coverage of F. denudata, were recorded.A. faxoniana regeneration was classified into two size classes: seedling and sapling (seedlings: ≥ 10 cm and <50 cm in height; saplings: ≥ 50 cm in height and < 5 cm in DBH) (Zhao et al., 2012).

Data analysis
A total of 103 naturally formed gaps were recorded, with sizes ranging from 20 to 1,860 m 2 and ages from 3 to 60 years.However, since 6 gaps had neither A. faxoniana regeneration nor F. denudata, only 97 gaps were used during the later analyses.The 97 gaps were then classified into three gap size classes [I: < 200 m 2 ('small'), II: 200-1,000 m 2 ('medium-size'), and III: 1,000-2,000 m 2 ('large')], and three gap age classes  (p>0.05).Secondly, the numbers and coverage of F. denudata did not change obviously in young and mediumage Sn-gaps, however, when the gap ages increased from medium to old, the numbers and coverage of F. denudata in Sn-gaps increased significantly (p<0.05;Figure2).
Similarly, the values of the two characteristics in each aged Ss-gap or F-gap were relatively constant (p>0.05).

Effects of gap size and bamboo on the recruitment numbers of A. faxoniana
Firstly, only the numbers of A. faxoniana seedlings in A-gaps decreased sharply with the increases of gap size (p<0.05),while those in Sn-or Ss-gaps, as well as the numbers of A. faxoniana saplings were not influ-enced by gap size (p>0.05; Figure 3).Secondly, the negative effect of F. denudata only occurred in small gaps: the occurrences of A. faxoniana seedlings in small Sn-gaps were significantly lower than in small A-gaps (p<0.05)(small Ss-gaps were not found).In contrast, no matter what the bamboo coverage was, A. faxoniana seedlings in medium-sized or large gaps, and A. faxoniana saplings generally were constantly scarce (Figure 4).As for small gaps, the Mann-Whitney U test was used to compare the numbers of A. faxoniana seedlings between A-gaps and Sn-gaps, while the Independent samples test was used to compare the numbers A. faxoniana saplings between A-gaps and Sn-gaps.As for medium-size gaps, ANOVA was used in order to compare the numbers of A. faxoniana seedlings and saplings in A-gaps, Sn-gaps, and Ss-gaps.As for large gaps, since there was only no A-gap and only one Ss-gap, no statistical test was used in this group.the numbers of A. faxoniana seedlings varied among A-, Sn-, and Ss-gaps (p<0.05).The numbers of A. faxoniana seedlings in young Sn-gaps were statistically lower than in young A-gaps (p<0.05)(the related data in Ss-gaps was insufficient for statistics).Similarly, the numbers of A. faxoniana seedlings in old Sn-and Ss-gaps were significantly lower than in old A-gaps (p<0.05).However, the numbers of A. faxoniana seedlings in medium-age gaps, as well as the numbers of A. faxoniana saplings barely varied (p>0.05; Figure 6).

Discussions
With regard to our first issue of whether the gap size and age would influence the F. denudata underneath, our study confirmed the numbers of F. denudata increased as the size or age of Sn-gaps increased.This finding is similar to the study on F. qinlingensis which also had higher densities in large gaps, and Wang et al. (2006) explained that increased light led to maturation of bamboos.We found that the succession of gaps may also benefit the clonal growth of bamboos.We also found that no matter the gap size or age, the numbers and coverage of F. denudata in Ss-gaps or F-gaps barely varied.It appears that the small amounts of bamboos, as in Sn-gaps, were stimulated by gap size and age.However, as for bamboos with a higher cover-6 With respect to our second issue, our study denied the effects of gap size and age on the numbers of A. faxoniana seedlings and saplings in gaps with bamboos.The numbers of A. faxoniana seedlings in gaps without bamboos (A-gaps) significantly decreased with the increases of gap size.In general, in large gaps the establishment of tree regeneration will be limited by strong light radiation, high temperature and evaporation, and subsequently low soil water availability (Bullock, 2000;Zhang et al., 2013).However, A. faxoniana seedlings in gaps with bamboos (in Sn-and Ss-gaps) were insensitive to increased gap size, which may also result from the scarcity of A. faxoniana seedlings.Taylor et al. (2006) explained similarly that the importance of gap disturbance for Picea seedling establishment was not obvious because of the scarcity of Picea seedlings and saplings in their stands.
With regard to the third issue, our results proved the strong inhibition of F. denudata on A. faxoniana seedlings.In small, young and old gaps, the numbers of A. faxoniana seedlings were relatively abundant but sharply reduced with the increase of F. denudata coverage.The restraint of bamboos on conifer seedlings may results from the rapid growth and large biomass of bamboos (Taylor & Qin, 1989), and has been reported by many studies.For example, in forests in Chile, Japan, and Sichuan Province of China (e.g.Veblen et al., 1981;Runkle, 1981;Takahashi, 1994;Guo et al., 2013), dense bamboos occupy the forest understory or in canopy gaps and greatly impede tree seedlings and saplings.In addition, we found that in gaps with a small number of A. faxoniana seedlings or saplings, the effects bamboo was insignificant.The unclear effects of F. denudata coverage may also result from the low occurrence of A. faxoniana recruitment.After all, A. faxoniana is a fecund but short-lived species, with a high mortality of recruitment (Taylor et al., 2006).

Conclusion
Increased gap size and age would increase the numbers of F. denudata with low coverage.Where A. faxoniana seedlings were relatively abundant, A. faxoniana seedlings were significantly inhibited by increased gap size and F. denudate.In contrast, due to the low occurrences of A. faxoniana recruitment, gap size, gap age and F. denudata exerted little effect on A. faxoniana seedlings and saplings in gaps.
Moreover, whilst our findings are insufficient in sampling plots and investigation on gap size and age are inaccurate, this study could still provide useful information for research on gap characteristics, natural age, as in Ss-gaps and F-gaps, the positive effects of gap size and age got weaker.It implied that the niche of bamboos in larger or older gaps might be constrained by some other environmental variables, for example, topography and disturbance gradients, as well as canopy density and composition gradients which have been shown to influence the distribution and growth of dwarf bamboo, F. nitida, in subalpine forest in southwest China (Wang et al., 2009).denudata cover >=17% Note: Different lower case letters denote significant differences among gaps with the same age but different types (p < 0.05).As for young gaps, since there was only one Ss-gap, the Mann-Whitney U test was used in order to compare the numbers of A. faxoniana seedlings between A-gaps and Sn-gaps, while the Independent samples test was used to compare the numbers of saplings between A-gaps and Sn-gaps.As for medium-age gaps, the Independent samples test was used to compare the numbers of A. faxoniana seedlings and saplings between A-gaps and Sngaps.For old gaps, the Kruskal -Wallis H nonparametric test was used to compare the numbers of A. faxoniana seedlings in A-gaps, Sn-gaps and Ss-gaps, while ANOVA was used to compare the numbers of A. faxoniana saplings in A-gaps, Sn-gaps and Ss-gaps.
regeneration and the relationship between bamboos and tree recruitment in subalpine coniferous forests.In addition, more intensive study should be focused on the distances between conifer recruitment and understory bamboos, since the inhibition effects of bamboos on tree recruitment may directly depend on the distances between the two species.

Figure 1 .
Figure 1.Numbers (mean±S.D., ha -1) and coverage (mean±S.D., %) of F. denudata in three sized gaps.Sn-gap: gaps with F. denudata cover < 17%; Ss-gap: gaps with F. denudata cover >=17%; F-gaps: gaps having F. denudata without A. faxoniana recruitment.Note: Different lower case letters denote significant differences among gaps with the same type but different sizes (p < 0.05).The Kruskal -Wallis H nonparametric test and Games -Howell post hoc test were used in order to compare the numbers and coverage of F. denudata in three sized Sn-gaps.The Independent samples test was also used to compare the numbers and coverages of F. denudata in small and medium -size F-gaps.

Figure 2 .Figure 3 .
Figure 2. Numbers (mean±S.D., ha -1 ) of F. denudata in three aged gaps.Sn-gap: gaps with F. denudata cover < 17%; Ss-gap: gaps with F. denudata cover >=17%; F-gaps: gaps having F. denudata without A. faxoniana recruitment Note: Different lower case letters denote significant differences among gaps with the same type but different ages (p < 0.05).The Kruskal -Wallis H nonparametric test and Games -Howell post hoc test were used in order to compare the numbers and coverage of F. denudata in three sized Sn-gaps and F-gaps.

FirstlyFigura 4 .
Figura 4. Bamboo effects on the numbers (mean±S.D., ha -1 ) of A. faxoniana seedlings and saplings in three sized gaps.A-gaps: gaps having A. faxoniana recruitment without F. denudata; Sngap: gaps with F. denudata cover < 17%; Ss-gap: gaps with F. denudata cover >=17%.Note: Different lower case letters denote significant differences among gaps with the same size but different types (p < 0.05).As for small gaps, the Mann-Whitney U test was used to compare the numbers of A. faxoniana seedlings between A-gaps and Sn-gaps, while the Independent samples test was used to compare the numbers A. faxoniana saplings between A-gaps and Sn-gaps.As for medium-size gaps, ANOVA was used in order to compare the numbers of A. faxoniana seedlings and saplings in A-gaps, Sn-gaps, and Ss-gaps.As for large gaps, since there was only no A-gap and only one Ss-gap, no statistical test was used in this group.

Figure 5 .
Figure 5. Effects of gap age on the numbers (mean±S.D., ha -1 ) of A. faxoniana seedlings and saplings in gaps with different bamboo coverage.A-gaps: gaps having A. faxoniana recruitment without F. denudata; Sn-gap: gaps with F. denudata cover < 17%; Ss-gap: gaps with F. denudata cover >=17%.Note: Different lower case letters denote significant differences among gaps with the same type but different ages (p < 0.05).ANOVA was used to compare the numbers of A. faxoniana seedlings and saplings in three aged A-gaps and Sn-gaps.

Figure 6 .
Figure6.Bamboo effects on the numbers (mean±S.D., ha -1 ) of A. faxoniana seedlings and saplings in three aged gaps.A-gaps: gaps having A. faxoniana recruitment without F. denudata; Sngap: gaps with F. denudata cover < 17%; Ss-gap: gaps with F. denudata cover >=17% Note: Different lower case letters denote significant differences among gaps with the same age but different types (p < 0.05).As for young gaps, since there was only one Ss-gap, the Mann-Whitney U test was used in order to compare the numbers of A. faxoniana seedlings between A-gaps and Sn-gaps, while the Independent samples test was used to compare the numbers of saplings between A-gaps and Sn-gaps.As for medium-age gaps, the Independent samples test was used to compare the numbers of A. faxoniana seedlings and saplings between A-gaps and Sngaps.For old gaps, the Kruskal -Wallis H nonparametric test was used to compare the numbers of A. faxoniana seedlings in A-gaps, Sn-gaps and Ss-gaps, while ANOVA was used to compare the numbers of A. faxoniana saplings in A-gaps, Sn-gaps and Ss-gaps.