Species Structure, Diversity, and Tree Regeneration on Stumps in Second-Growth Temperate Rainforests of British Columbia, Canada

Background: The signicance of stumps and other coarse woody debris (CWD) in maintaining biodiversity has been widely recognized. However, there is a paucity of research on the role of stumps in tree regeneration. We studied vascular plant structure, diversity, and tree regeneration on stumps in temperate rainforests across three sites in British Columbia, Canada (Malcom Knapp Research Forest, MKRF; Pacic Spirit Regional Park, PSRP; and Stanley Park, SP). Results: 1) There were 19 vascular plant species found on stumps, including eight tree species, 2) Overall seedling abundance was higher on stumps than the nearby ground, 3) The number of established plants showed a positive linear (MKRF, SP) or Gaussian (PSRP) relationship with stump basal diameter, 4) Vegetation abundance varied with site (MKRF > SP > PSRP), 5) The overall species established on stumps were positively associated in MKRF (Variance ratio = 2.74) and SP (Variance ratio = 1.37), but negatively associated in PSRP (Variance ratio = 0.57), and 6) Tree species appeared to compete with each other on stumps and were likely to co-occur with understorey species. Conclusion: Our results highlight community and species associations on stumps. We found that stump diameter is a major factor affecting tree regeneration in these second-growth temperate rainforests. To aid future research on stump-vegetation relationships, we synthesize our results in a schematic of vascular plant biodiversity and tree regeneration on stumps. Our work and this schematic can be used to stimulate ideas for new hypothesis generation and for studies relevant for conservation, management and basic science research. cottonwood (Populus trichocarpa), wild cherry (Prunus avium), red alder (Alnus rubra), Pacic yew (Taxus brevifolia), Cascara (Rhamnus purshiana), and Pacic dogwood (Cornus nuttallii). Thousands of trees were uprooted and blown over in several extreme weather events, especially wind storms in 1934–1960, and 2006 (Kheraj 2007). Our study plots in SP were located near Beaver Lake.


Background
Rainforests are important for carbon and nitrogen cycling (Hamaoui et al. 2016; Mackey et al. 2017), timber production, biodiversity, ecosystem services (Catterrall et al. 2005;Nahuelhual et al. 2007), and even beauty, mystery and spirituality (DellaSala 2011). While the majority of rainforests occur within the tropics, temperate rainforests are roughly 0.2% of the earth's land area, with a signi cant portion occurring in British Columbia, Canada (Farr 2003; Mackey et al. 2017). British Columbia's temperate rainforests are composed mainly of coniferous trees, with Douglas-r (Pseudotsuga menziesii), western redcedar (Thuja plicata), and western hemlock (Tsuga heterophylla) as the dominant species (Jenkins 2004). The number of species able to grow within these coastal forests is limited by dense overstorey canopy (Kimmins 2004), low temperature and limited winter daylight. These conditions also cause slow vegetation succession processes (Super et al. 2013; Defrenne et al. 2016).
The important role of coarse woody debris (CWD) in maintaining biodiversity has been widely recognized (Hörnberg et al. 1995(Hörnberg et al. , 1997, and has been the subject of intensive research (Chmura et  created by disturbance events such as wind, lightning, harvesting or re (Oliver and Larson 1996). They are the largest coarse woody debris (CWD) component in second-growth forests (Hörnberg et al. 1995;Nordén et al. 2004;Rackham 2008;Wirth et al. 2009), and play an important role in nutrient cycling (Kubin 1977;Finér et al. 2003). Generally, nutrient availability increases on stumps after moss mortality (During 1979) and is further enhanced by fungal activity (Johansson et al., 2002). However, stump decomposition can be slow, e.g., one study indicated that a net release of nitrogen from pine stumps may take as long as 40 years (Palviainen et al. 2010). Tree seedlings tend to establish in higher concentrations on CWD than the nearby ground (Kumar et al. 2018). While stumps have limited surface area, they are a particularly important component of CWD for tree and understory species establishment (Hörnberg et al. 1995(Hörnberg et al. , 1997. Despite this, management considerations for tree stumps have often focused their provisions of nests for animals or habitat for bryophytes and fungi (Lindelöw et al. 1993;Hörnberg et al. 1997; Waldien et al. 2000;Prescott 2002; Konuk et al. 2007;Laitila et al. 2015Laitila et al. ) 1979. Surprisingly, we nd very few or no studies aimed at estimating vascular plant species associations during tree regeneration and species establishment on stumps, despite the abundance of stumps.
Tree regeneration on stumps can be classi ed into two types: sprouting (regeneration re-sprouting, or new growth originating from the stump) and new establishment seedlings (not from sprouting, different origin, e.g., regeneration from seed rain of a nearby tree) (Oliver and Larson 1996). Tree seedlings established on stumps can be affected by stump structures (e.g., basal diameter and height), which may, in turn, relate to nitrogen and water content availability (Vonhof and Barclay 1996), and in uence the survival and tness of vascular plants species. Only limited research related to tree regeneration on stumps has been reported. A study on Norway spruce (Picea abies) regeneration in Italy found stump diameter to be the most important factor maximizing the regeneration potential of tree species (Motta et al. 2006). Similarly, such a trend was found with re-sprouting of stumps (sprouting as mentioned above) discovered in India (Khan and Tripathi 1986). However, these studies exclusively dealt with the understorey species or tree species on stumps, and thus present an incomplete assessment (e.g., species diversity) as both understorey plants and tree species should be concurrently considered to represent the entire community.
To date, the effect of intraspeci c tree species competition and the competition with other vascular plants on stumps remain unclear. Furthermore, spatial relationships of vegetation growth on stumps seem unexplored. Therefore, the objectives of our study were to: 1) quantify temperate rainforest vascular plant diversity on stumps; 2) assess vegetation associations and competition related to tree regeneration on stumps; and 3) synthesize ideas into a schematic illustrating the conceptual process of tree regeneration on stumps useful for future research on stump-vegetation relationships.

Study Sites
Three study sites were selected in southwestern British Columbia, Canada. These were Malcolm Knapp Research Forest, Paci c Spirit Regional Park, and Stanley Park (Fig. 1). Subplots were situated at these main study sites.
Malcolm Knapp Research Forest (MKRF), which is owned and operated by The University of British Columbia (UBC), is 5,157 hectares and located in Maple Ridge, British Columbia (Farahbakhchian 2017). MKRF has coastal forest stands characteristic of the Paci c Northwest with naturally regenerated and plantation forests, including variable retention, as well as other forms of management for research and education purposes. Portions of MKRF experienced logging between 1920 and 1931. The second-growth areas surveyed in this study were approximately 80-year-old forest stands that were naturally regenerated (personal communication Ionut Aron, MKRF, 2019) and composed mostly of western hemlock and western redcedar, with a smaller amount of Douglas r. The site history is evident, including massive cedar stumps.
Paci c Spirit Regional Park (PSRP) is an urban greenbelt valued for recreation, education, and biodiversity. The park is located in UBC Endowment Lands, Point Grey to the west of Vancouver, British Columbia, and is bordered by the UBC campus (Artibise and Meligrana 2005;Super et al. 2013). PSRP has a maritime climate with warm, dry summers and mild, wet winters (Meidinger and Pojar 1991). PSRP is within the Coastal Western Hemlock zone of the provincial Biogeoclimatic Ecosystem Classi cation (Krajina and Brooke 1965), and has typical temperate rainforest tree species such as western hemlock and western redcedar (Goward 1994). Vegetation throughout PSRP has been impacted by past anthropogenic disturbances, and has secondary growth regeneration in many places (Super et al. 2013). Our study plots were located in an area that was clear-cut and burned in the year 1910.
Stanley Park (SP) is approximately 400 hectares and located in the Coastal Western Hemlock zone in the city-centre of Vancouver, one of the largest citycentre parks in North America (McDonald 1984). It is regarded as an "invaluable commercial and advertising asset" for nature, recreation, education and history for local people and many tourists (Kheraj 2007) and has impacted people-wildlife relationships in Canada (Kheraj 2012). The most abundant species include western hemlock, western redcedar, Douglas-r, bigleaf maple (Acer macrophyllum), black cottonwood (Populus trichocarpa), wild cherry (Prunus avium), red alder (Alnus rubra), Paci c yew (Taxus brevifolia), Cascara (Rhamnus purshiana), and Paci c dogwood (Cornus nuttallii). Thousands of trees were uprooted and blown over in several extreme weather events, especially wind storms in 1934-1960, and 2006(Kheraj 2007). Our study plots in SP were located near Beaver Lake.

Data collection
Permits were secured from UBC to survey MKRF and PSRP as well as from SP management. For each site, 10-11, 30 × 30 m plots were delineated. To minimize disturbance effects from trail-use, all plots were established at least 5 m into the forest from the trail edge. For each plot, all the species on the ground and overall canopy coverage were recorded. For each stump in each plot, its maximum height, basal diameter, decay stage, and the height of trees growing on it (height: 5 to 200 cm) were measured. In addition, species names and the number of individuals of all plants (tree and understory species) growing on stumps were recorded.

Data analysis
To identify factors affecting vegetation growth on stumps, we used linear regression to analyze the association between the number of individuals on stumps and other possible variables, such as stump basal area, stump height, and canopy coverage. We did not nd a clear pattern for stump height, and canopy coverage contributed to species abundance on stumps. A regression analysis was also used to test Individual Species Area Relationship (ISAR) respect to relationship between the number of species on stumps and the stump basal area. All analyses were conducted in R 3.3.1 (R Development Core Team 2014). To further evaluate the species relationship on stumps, we used the variance ratio (VR) (Vroh et al. 2016), which indicates that the total species relationship and total species association will be positively and negatively associated when VR > 1 and VR < 1, respectively. We measured the strength of the linear association between species pairs growing on stumps by using the Pearson product-moment correlation coe cient (Sedgwick 2012). Species relationships on the same stump species were further analyzed using the R "spaa" package (Zhang and Zhang 2013; Gri th et al. 2016).

Vascular plant distribution, abundance, and community composition
A total of 23 vascular species (9 tree and 14 understorey species), belonging to 15 plant families, were found in the three study sites (Table 1), with 19 species (8 tree and 11 understorey species) found on both stumps and the ground. Species abundance was highest in Malcolm Knapp Research Forest (MKRF), followed by Stanley Park (SP) and then Paci c Spirit Regional Park (PSRP). In the MKRF site, the major stump species were western redcedar and western hemlock. A total of 2,381 individuals belonging to 17 vascular plant species (of 13 families) grew on 93 (76.23%) out of the 122 sampled stumps; on stumps, the most abundant tree species was western hemlock, while the most abundant understorey species was red huckleberry (Vaccinium parvifolium) ( Table 2, Fig. 2). MKRF and SP sites had similar species. In the PSRP site, the major stump species were western redcedar, western hemlock, and Douglas-r. A total of 155 individuals belonging to nine vascular plant species (of seven families) grew on 38 (35.19%) out of the 108 sampled stumps. The most abundant tree species was western hemlock, and the most abundant understorey species was salal (Gaultheria shallon) growing on stumps ( Table 2, Fig. 2). In the SP site, the major stump species were western redcedar and western hemlock. A total of 842 individuals belonging to 14 vascular plant species (of 13 families) grew on 92 (81.41%) out of the 113 sampled stumps; on stumps, the most abundant tree species was western hemlock, while the most abundant understorey species was red huckleberry (Table 2, Fig. 2).   (Table 2). Regression analysis showed that basal diameter was signi cantly correlated with the species abundance on stumps (MKRF: P < 0.05; PSRP: P < 0.001; and SP: P < 0.001) (Fig. 2). In MKRF and SP the number of individuals increased with increasing basal diameter of stumps. Whereas in PSRP, there was a Gaussian distribution of plant abundance with the basal diameter of stumps; stumps with medium basal diameter had more individuals (Fig. 2).
Vegetation association on stumps Vascular plant species had signi cant co-occurrence patterns on stumps. The overall species established on stumps were positively associated in MKRF (variance ratio = 2.74) and SP (variance ratio = 1.37), but negatively associated in PSRP (variance ratio = 0.57). In MKRF, there were two highly signi cant (P < 0.01) and eight signi cant (P < 0.05) positive associations, and four highly signi cant (P < 0.01) and 50 signi cant (P < 0.05) negative associations between species pairs were observed (Table 3a). In PSRP, no positive association between species pairs but ve negative associations pairs (P < 0.05) were observed (Table 3b). In the SP site, four highly signi cant (P < 0.01) and two signi cant (P < 0.05) positive associations, and one highly signi cant (P < 0.01) and 49 signi cant (P < 0.05) negative associations between species pairs were observed (Table 3c). Table 3: Vascular plant species associations on stumps based on Pearson product-moment correlation coe cients. a) MKRF, b) PSRP and c) SP. Negative and po associations are assumed to suggest species competition during seed germination and before seedling establishment respectively (see Table 1

Vascular plants kingdom on stumps
Previous studies related to tree regeneration on stumps in temperate forests have been restricted to only a few tree and stump species such as Picea abies (Hörnberg et al. 1997), Abies alba, and Fagus sylvatica (Szewczyk and Szwagrzyk 1996). Thus, we have expanded the current level of knowledge through reporting for the rst time on the growth of different herbaceous, tree and shrub species on conifer stumps. We have examined the composition of tree species on stumps in diverse understorey plant communities. Of the total 23 vascular plant species that we identi ed in the three sites, 19 were found growing on stumps. It is likely that these stumps created a suitable habitat for the establishment of these plant species by providing additional moisture through the water-holding mosses that regularly colonize coarse woody debris (During 1979). As stumps are elevated from the ground, they provide vascular plants with growing conditions that receive more sunlight and have higher temperatures than the forest oor, which may help to promote additional plant growth (Nelson 1951). In addition to this, animals have been shown to select the best seeds of vascular plants to store in stumps (Breen-Needham 1994). These factors may help tree and understorey regeneration on stumps within second-growth temperate rainforests in spite of the thick overstorey canopy and cool temperatures Vegetation-stump basal area relationship The relationship between vegetation-stump basal area has not been fully studied and it was among the goals of the present study. Re-sprouting on stumps research has shown that the age class of a stump is one of the major factors affecting tree regeneration. A study on the relationship between vegetationstump basal area conducted in India has indicated that the median basal diameter may have maximized the regeneration of tree species (Khan and Tripathi 1986). This pattern was detected in one out of the three study sites, PSRP, suggesting that stumps of median basal area had maximized established vascular plant quantities. In other words, for PSRP we did not nd an explicit "individual species-area relationship", i.e., the number of individuals increase with increasing habitat area as reported by "Individual Species Area Relationship (ISAR)" by Tsai et al. (2015). The Gaussian distribution of vascular plant species with basal stump area may be due to the understorey species mortality caused by competition with the canopy of tree species on larger stumps, as the mean basal diameter of stumps in PSRP (42.77 cm) is much smaller than that in MKRF (89.78 cm) and SP (74.90 cm) ( Table 2). In contrast, for MKRF and SP, there was a clear ISAR, with increasing numbers of individuals as basal area increases. To our knowledge, this is the rst time that a study has examined the ISAR concept in a small area (74.90-89.78 cm in diameter). Stumps are ubiquitous in many forests, but still under researched, especially with respect to vascular plants and tree regeneration; thus, we suggest additional research efforts be dedicated to epixylic communities on stumps. Interestingly, our data did not show a correlation between vegetation diversity and stump height. However, this may be due to our exclusion of stumps greater than 200 cm, which we described as snags. Further research is also needed to illustrate the role of high stumps on the various steps and processes in vascular plant establishment.

Species competition patterns on stumps
Species interactions on stumps can be in uenced by factors related to natural enemy, density-dependence, inter and intra-speci c competition, and species coexistence (Chesson 2013). Competition is one of the most fundamental interactions of ecological organization (Solé et al. 1992;Chesson 2013). Species competition patterns on stumps may help unveil tree regeneration processes on stumps. In the present study we focused on interspeci c competition and species coexistence. As previously mentioned, the overall species associations on stumps were positive in MKRF and SP but negative in PSRP which could be caused by factors related to the smaller average basal diameter in PSRP. Within all the studied sites, tree species seemed to signi cantly compete with each other, such as ACCI vs. ILAQ (species abbreviations in Table 1), ACCI vs. TSHE, ILAQ vs. TSHE, ILAQ vs. PSME, ACCI vs. PSME, and THPL vs. ACCI (Table 3).
Tree species also compete with other highly occurring species such as GASH and VAPA. However, in SP, two positive pairs of tree species are ACMA and TSHE, THPL and TSHE; this may due to the large occurrence of TSHE. It should to be noted that ILAQ, an invasive species, successfully colonized stumps; mitigating invasive species is under management consideration in Vancouver (Mosquin 1997).

Tree regeneration process
Tree regeneration and vascular plant biodiversity on stumps have intricate processes and patterns. Our literature searching uncovered that little is known about stump-vegetation relationships other than non-vascular plants, and linking processes to patterns is relatively rare. Our research suggests that stumps can play an important role in tree regeneration in Paci c Northwest temperate rainforests, and a conceptual model for future stump-vegetation research would be useful. To ll this gap, we have included a schematic depicting the regeneration process (Fig. 3), which can be described as: 1. Disturbance: Stumps are initially created by abiotic disturbances (e.g., wind, lightning, harvesting, or re) synergistically with biotic disturbance (e.g., a weakened tree by disease could be more likely to be broken by wind). Canopy gaps resulted from the disturbances start allowing for shifts in succession on stumps.
2. Stump decay: Stumps begin the decay process, which is enhanced by the activity of agents such as insects, fungi, bacteria, etc. (Palviainen et al. 2010). At this stage, the chemistry of the stump and nutrient cycling and accumulation matter especially. After consistent rainfall, bryophytes and lichens appear on stumps, and then, the habitat on the stump becomes suitable for the establishment of tree species and vascular plant biodiversity.
3. Seed or propagule dispersal: Tree seeds are dispersed on to stumps passively (e.g., falling from nearby trees, wind), or actively by animals (e.g., squirrels, birds, including hiding by animals). Seed survival can be affected by multiple processes, e.g., pathogens, seed predation, facilitation, mutualism, etc. The seeds that move into germination and subsequently other stages will be impacted by stochastic and deterministic processes. manipulative experiment. We focused mainly on plant species patterns relevant to stages ve and six; whereby we found ISAR with respect to individual plants and stump basal area for two study sites (MKRF and SP) and a Gaussian distribution instead ISAR for PSRP. To our knowledge, our study is the rst to evaluate the ISAR with stumps with to plants and any organism on stumps. Another pattern across these stages was species associations on stumps, which suggested species interactions such as competition.
Stump history from stage one to stage three may play a crucial role in setting up the habitat conditions needed to create the association patterns that we found in this study. Further experiments could do manipulative transplant and growth experiments to study this further and look at other patterns such as plant tness (or survival ratio) and species cover (how much of stump surface area covered by vegetation). These appear to not yet be measured by any researchers to our knowledge. Furthermore, studies could compare how different stump species at different sizes impact regeneration processes. The patterns and processes mentioned with our schematic are not simple and could potentially have downstream effects, with not all stages being mutually exclusive and may happen together. For example, some seeds may land when other seedlings are already established. Very early processes such as insects could affect the entire trajectory; for example, mountain pine beetle attack could modify the nutrition of tree which affect it as a stump relative to non-beetle kill stumps, which potentially affect all the other processes. The nutrition processes of stumps could be further studied with isotope methods that tease apart the different substrate (e.g., redcedar stump vs. other species, etc.) and chemistries. The habitat matrix around stumps potentially has a signi cant impact on seedling regeneration and vascular plant biodiversity on stumps. Many factors could impact the start of a stump's trajectory for tree seedling regeneration and vascular plant biodiversity.

Conclusion
This study elucidated vascular plant biodiversity on stumps in Paci c Northwest temperate rainforest sites, including the individual species-area relationship (ISAR), and inferred species interactions (e.g., competition) on stumps. Furthermore, we generated a schematic that integrates our ndings; this synthesis can help with further research related to understanding the role of tree regeneration on stumps beyond the present study. Vascular plant diversity and tree regeneration arise due to complex processes. To test and explore the implications of our schematic, one could explore tree regeneration and vegetation succession on stumps in second-growth forests across various climatic zones in British Columbia, or elsewhere. Such research could lead to novel discoveries relevant to management, conservation, and to an overall better understanding of tree regeneration processes. Availability of data and materials The datasets used during the current study are available from the corresponding author on reasonable request.
Ethics approval and consent to participate Not applicable

Consent for publication
Not applicable

Competing interests
The authors declare that they have no competing interests.

Figure 2
Frequency distribution of vascular plant species occurrence on stumps: a) MKRF, b) PSRP, and c) SP (see Table 1