FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2890623015> ?p ?o ?g. }

• W2890623015 endingPage "594" @default.
• W2890623015 startingPage "582" @default.
• W2890623015 abstract "生境片段化研究大多聚焦于老龄林破碎化之后产生的效应，但对次生林演替过程受片段化影响的研究还很少。本研究以千岛湖片段化生境中次生马尾松林群落为研究对象，分别于2009–2010年和2014–2015年调查了29个岛屿上的木本植物物种多样性、功能性状和群落结构等，探讨了生境片段化对次生林演替过程中的群落构建过程的影响。结果显示，小岛上的森林群落（相对大岛而言）仍处于演替较早期阶段：物种少且以演替早期物种为主，而动物传播和耐阴的物种较少；小岛具有和大岛上相似的单位面积地上生物量，主要归因于快速生长的优势种马尾松的贡献；岛屿面积是次生演替的关键因素，而隔离度的作用相对较小；边缘效应影响物种和功能组成。总之，人工湖泊（水库）建设造成的生境片段化对植物群落的构建过程具有重要影响，使得小岛上具有相对较少的演替后期物种。因此，维持较大面积的森林斑块是快速演替的关键，在生态恢复实践中需要考虑景观格局，以加快次生林的演替速率。本研究还表明，基于成熟林得出的生境片段化效应范式并不一定适用于次生林。 The overwhelming majority of forest fragmentation studies focus on the effects of carving up old-growth forests into small and isolated patches. These studies reveal that habitat fragmentation leads to elevated mortality of large, late-successional tree species at forest edges (Laurance, Delamônica, Laurance, Vasconcelos, & Lovejoy, 2000), non-random loss of specialist seed-dispersing animals (Cramer, Mesquita, & Williamson, 2007) and the proliferation of pioneer species with low wood density and short-lifespan (Tabarelli, Aguiar, Girão, Peres, & Lopes, 2010). Hence, fragmentation of old-growth forests leads to a rapid shift in species composition towards early-successional (e.g. shade-intolerant) species (Laurance et al., 2006; Tabarelli, Lopes, & Peres, 2008), a change in plant functional traits with less animal-dispersed, hard-wooded, large-seeded species (Benchimol & Peres, 2015; Magnago et al., 2014), and a decline in forest carbon stocks (Laurance et al., 1997, 2002 ) (Table 1). Much less is known about fragmentation effect on regenerating forests, which represents an important knowledge gap given that so much of the world's forest is regenerating on abandoned lands or recovering from logging (Arroyo-Rodríguez et al., 2017; Chazdon et al., 2016; Poorter et al., 2016). For example, 1.17 million km2 of forest in China that was clear-cut in the 1960s is currently regenerating (FAO, 2015). These regenerating forests are of great importance to biodiversity conservation and carbon sequestration (Barlow et al., 2007; Chazdon et al., 2016), but the recovery of original vegetation composition by natural successional processes can take over a century (Liebsch, Marques, & Goldenberg, 2008). How to accelerate successional rates and understanding the predictability of successional pathways are key issues in forest restoration, but few studies have linked these issues to landscape variables (Arroyo-Rodríguez et al., 2017). Investigating this link with landscape ecology is important because secondary forests are often highly fragmented in terms of patch size, the absence of interior forests and their degrees of isolation (Haddad et al., 2015; Taubert et al., 2018). In contrast to the pattern observed in old-growth forest fragments where habitat fragmentation can drive forest communities towards early successional stages (Tabarelli et al., 2008), these regenerating forests are “assembling systems” in which secondary succession and fragmentation effects are occurring simultaneously (Collins et al., 2017). As summarised in Table 1, studies of secondary succession show that species diversity, functional diversity and structural properties change significantly as succession proceeds (e.g. Craven, Hall, Berlyn, Ashton, & Breugel, 2018). For example, structural complexity and species richness increase rapidly while community mean specific leaf area (SLA) decreases during succession (Chazdon, 2008; Muscarella et al., 2017), although it is also recognised that “priority effects” resulting from differences in the timing of species arrival among sites can also have major influences on community assembly (Fukami, 2015). However, few studies have investigated forest secondary succession in fragmented landscapes (Arroyo-Rodríguez et al., 2017; Crouzeilles et al., 2016), and paradigms drawn from old-growth forest studies are may not hold for regenerating forests. The species richness of regenerating forests is determined by the interplay between colonisation and extinction rate (Chazdon, 2008, 2014 ), secondary succession in fragmented landscapes has many parallels with the Theory of Island Biogeography (MacArthur & Wilson, 1967). Colonisation and extinction are, in turn, related to the processes of dispersal, competition and other interactions, and ecological drift. Numerous studies have shown how patch-configuration influences these processes in successional systems (Sams et al., 2017). For example, fragment size can influence the local species pool and the diversity of niches available to colonising species, and affects plant functional group assembly in terms of species and trait composition (Cook, Yao, Foster, Holt, & Patrick, 2005; Moreno-Mateos, Power, Comín, & Yockteng, 2012). Spatial isolation can decrease the arrival rate of new species, as undisturbed habitats act as seed sources for adjacent fragments (Goosem et al., 2016), and is known to be a key driver of community assembly on ocean islands (Negoita et al., 2016). Meanwhile, large forest fragments can provide more habitat diversity for species and have lower rates of local extinction (Gibson et al., 2013). Several studies based on small shrubs and herbs indicate that landscape contexts affect community assembly during secondary succession (Helsen, Hermy, & Honnay, 2013; Turley, Orrock, Ledvina, & Brudvig, 2017). For example, small forest fragments often support communities comprised of early-successional species with more acquisitive traits (e.g. high specific leaf area) compared to large forest patches (Magnago et al., 2014; Muscarella et al., 2017), and harbour low species richness (Chazdon, 2014). In contrast, some other studies have found little evidence for edge effects in regenerating forests (Slik et al., 2011). Consequently, the effects of habitat fragmentation processes on regenerating forests should be different from those reported on old-growth forests, which are mostly related to mortality of late-successional species near edges. However, there is no systematic investigations into the effects of habitat fragmentation on the secondary succession in regenerating forests. Here we quantify the secondary succession stages of 29 islands located in the Thousand Island Lake in China created by a hydroelectric dam, to evaluate the influences of fragmentation on succession. This system has several advantages: (a) all forest fragments are of the same age, having been clear-felled in the late 1950s, and then left to regenerate naturally; (b) environmental conditions are relatively similar across the islands; (c) islands can be regarded as independent samples, because they have clear boundaries and a uniform open-water matrix; (d) the abundance of islands provides statistical power and a range of fragment sizes and spatial configurations. Therefore, this site represents an excellent opportunity to investigate the influences of habitat fragmentation on secondary succession, by excluding the effect of confounding environmental factors, time since disturbance and matrix type. We assess the impacts of habitat fragmentation on various aspects of secondary forest succession including species diversity, functional traits and structural properties. We hypothesise that community assembly during secondary succession on small and isolated islands proceeds differently than on large and connected islands, resulting in lower species richness, more pioneers, the preponderance of species with acquisitive functional traits, and less structural complexity than larger, better-connected, islands. Our analyses allow us to evaluate which landscape variable matters most for secondary succession in regenerating forest fragments. This study was conducted in the Thousand Islands Lake in eastern China (Latitude: 29°22′–29°50′N, Longitude: 118°34′–119°15′E) (Figure 1). This region has a subtropical climate with an annual temperature of 17.0°C (ranging from −7.6°C in winter to 41.8°C in summer) and rainfall of 1,430 mm (Wang, Bao, Yu, Xu, & Ding, 2010). The zonal vegetation is subtropical broad-leaved forest (Wu, 1980). In the late 1950s, trees of this mountainous area were completely or near-completely clear-cut prior to the construction of a dam and subsequently inundated, leading to the formation of 1,078 islands (each >0.25 ha) when the water reached its highest level at 108 m above sea level (Hu, Feeley, Wu, Xu, & Yu, 2011). Community assembly on these islands might not be strongly affected by historical contingency effects (Fukami, 2015) because they share common history, and have developed via secondary succession over 50 years from almost bare land, as evidenced by tree ring studies of Pinus massoniana trees (J. Liu, D. Wu, & M. Yu, unpublished data). However, we cannot rule out that seed banks and perennating tissues persisted in the soils, result in differences in the order and timing of arrivals. These islands are typically <2 ha in size (the largest being 1,158 ha), mainly covered by an orthic acrisol (Ao18-3bc according to the FAO soil classification), and have narrow elevational ranges (ranged from 109 to 299 m, M ± SD is 134 ± 31 m; according to Hu et al., 2011). This site has been strictly protected since the 1980s, and has rarely been disturbed hereafter. Currently, island vegetation is dominated by Pinus massoniana along with a mixture of broad-leaved tree species. To ensure that sampling plots were representative of the range of forest fragment attributes, we chose 29 islands that have different landscape attributes including fragment size (ranged from 0.1 to 1,158 ha), and degree of isolation, defined as the shortest edge-to-edge distance from each island to the nearest island (ranged from 11 to 136 m) and the shortest distance to the mainland (ranged from 694 to 3,725 m). More complex landscape-based indices of isolation were explored, based on calculating non-focal fragment area within circles centred on a focal fragment (Diver, 2008), but simple distance-based metric were found to correlate with these and proved to have greater explanatory power, because of the influence of the one very large island included in our study. We measured these landscape attributes by manually delineating islands from SPOT-6 imagery (using ArcGIS 10.4) and then applying FRAGSTATS 4.2 (McGarigal, Cushman, Neel, & Ene, 2015). A total of 71 vegetation plots were established on the islands. Plots were classed as “edge” if they were within 40 m of an island's edge, or otherwise as “interior” (see Laurance et al., 2002). All plots established on 20 small islands (<1.4 ha) were “edge” plots because all were close to the lake edge; two replicates plots were established on 15 of these small islands, but just one on the five smallest islands. On nine larger islands (>1.4 ha), we established two replicated plots to sample both interior and edge forests (i.e. four plots per island). At each sampling location, we conducted vegetation surveys within a 20 × 20 m plot according to the field protocol of the Centre for Tropical Forest Science (Condit, 1995). All woody plants with a diameter at breast height (DBH) ≥1 cm were identified and mapped in 2009–2010 and 2014–2015 separately. We therefore have plot data for regenerating forests of ~50 and ~55 years of age. For each individual, DBH was measured and a visual estimation of height was made. For further details of study sites, see Hu et al. (2011) and Yu, Hu, Feeley, Wu, and Ding (2012). We recorded 68 woody plant species on the 71 plots. For each species, we measured several frequently used functional traits, which indicate position in secondary succession (Table 1) using the standardised measurement protocols (Cornelissen et al., 2003). Specific leaf area (SLA; cm2/g) was measured on 10 fully expanded current-year leaves under the sun-exposed conditions from five individuals per species distributed across our study system. Wood density (WD; g/cm3) was measured after oven drying to constant mass at 80°C and divided by fresh volume determined by water displacement. We used wood from branches as many shrub species are too small to get wood cores (Osuri, Kumar, & Sankaran, 2014). We defined whether a species was a tree or a shrub species, its shade tolerance ability (“shade-tolerant” vs. “shade-intolerant”) according to Hu et al. (2011), and extracted its maximum plant height value from the Flora of China and local floras (Chen & Gilbert, 2006). Seed dry mass (SM; g) was measured for 63 species. The remaining five species occurred at such low abundances that we could not collect seeds, so we used the mean genus values from the Kew Seed Information database (https://data.kew.org/sid/). Fruit type (flesh and non-flesh fruits) and seed dispersal method (animal, wind, gravity and ballistic) were extracted from published references and field observations. To separate the main dispersers of the seeds, we identified species as being mainly bird- or mammal- dispersed (Table S1). To quantify the effect of habitat fragmentation on successional stage, we calculated the average value of the successional indicators listed above for all plots on each island, and used these averages as response variables. Three indices of fragmentation were used as predictor variables in regression models: island size (log transformed) and isolation measured in two ways, as distance to the mainland and distance to the nearest island. Environmental data, including aspect and elevation were included in initial analyses, but had no discernible effect, so results are not included in this paper. Multivariate least-squares regression models were used to assess the relationships between successional indicators and explanatory variables, which were scaled to have a mean of zero and standard deviation of one using the scale function in R (R Core Team, 2015), in order to compare the relative effect of the island size and isolation. The variance inflation factors of the predictor variables (i.e. the vif function in the car package (Fox & Weisberg, 2011)) were less than 1.5, indicating collinearity was not a major issue. None of the pairwise correlations among explanatory variables were significant (Pearson's test: p > 0.05). Nine large islands had both edge and interior plots, and for these we checked whether successional indicators differed between edge and interior plots, using t test. All statistical analyses were performed in r version 3.4.1. Because we make multiple comparisons, only results with p < 0.01 are described in the results. The magnitude of effects given in the results is based on the effect size multiplied by four (i.e. two standard deviations either side of mean). Our longitudinal analysis during the 5 year period (2010–2015) did not provide enough persuasive evidence to evaluate effects of habitat fragmentation on rates of secondary succession, and thus was not included in the analyses. Forest dynamics were further explored by plotting changes in stem density and mean basal area over 5 years on log-log axes (Coomes & Allen, 2007). These plots are associated with self-thinning theory, originally developed to describe the dynamics of even-aged stands. This classic, and controversial, theory argued that regenerating trees increase in size without loss of stems until canopy closure is complete, and thereafter the basal-area growth of some individuals is matched by a loss of similar basal area through mortality (i.e. competitive exclusion; see Coomes & Allen, 2007 and references therein). For even-aged forests, this self-thinning process leads to data falling along a line with −1 slope on these log-log graphs, once canopy closure is completed. Quantile regression (with tau = 0.95) was used to fit an upper boundary to the dataset to explore whether natural succession followed this classic self-thinning trajectory. Separate thinning plots were created for shade-tolerant and light-demanding species, to test whether shade-tolerant species were increasing in size and abundance over time, and whether light-demanding species were dying off, as expected in the later stages of succession. This approach has been applied to diverse natural forests (e.g. Wright et al., 2012) but never to secondary forest succession. Higher proportions of shade-tolerant and flesh-fruited plants were found on larger islands (Table 2). These plants, especially those dispersed by birds, were often shorter-statured and small-seeded (Supporting Information Figure S1). Therefore, larger islands had lower community-weighted seed mass and maximum height. Wood density and SLA did not vary with island size (Table 2). There was no evidence that isolation (distance to the nearest island and the mainland) affected plant functional trait composition (at p > 0.01 for all tests; Table 2). Comparing edge and interior plots on nine islands, we found a greater proportion of fleshy-fruited and shade-tolerant woody plants in interior plots (Supporting Information Table S2, Cohen's d = 0.899 and 0.96, respectively), perhaps suggesting that dispersers preferred the interior of islands. Species richness of woody plants increased strongly with island size, rising from 15 to 22 across the gradient at the plot scale. The number of shade-tolerant species was greater on large islands (range 8.1–13.7), but the number of shade-intolerant species did not differ (range 6.4–8.3). Again, there was no evidence that isolation affected species richness of woody plants. Comparing edge and interior plots on nine islands, species composition varied with location (paired t test of NMDS axis 1 scores, p = 0.01, Cohen's d = 0.94), with a greater abundance of shade-tolerant species in the interior. The above-ground biomass of forests was not affected by island size, despite the compositional differences explained above (Table 2). Most forest biomass was held in shade-intolerant species (83.9% on average), and the biomass of these species were not affected by island size. A much smaller amount was held in shade-tolerant species; although these species were more common on large islands and consequently had greater biomass on large islands (Table 2). Biomass change was mostly attributed to tree growth, and was not affected by island size (Figure 2). However, a lower biomass gain of shade-intolerant species was found on larger islands. Plants on small islands have relatively high average DBH (Table S2). Isolation did not significantly affect structural properties (all p > 0.01, Table 2). Above-ground biomass, mean DBH and mean height of woody plants all increased significantly between two surveys, as expected of forests undergoing succession (see intercept values in Table 2). Figure 3a is a classic “self-thinning” plot, where trends of mean basal area and total stem density are shown on a bi-plot. Most stands show the characteristics of competitive thinning during succession (Figure 3a) with a clear upper boundary shown by the quantile regression line, indicating strong competitive interactions are at play (slope = −1.52). The shade-intolerant species are increasing in their average basal area but losing stems (Figure 3b), while shade-tolerant species are gaining basal area with little mortality except in two plots (Figure 3c). Thus shade-intolerant species are increasingly dominating the biomass of the forests, despite the development of a dense layer of shade-tolerant subordinates beneath them. Islands experienced a shift in species composition towards late-successional species (Table 2 and Figure 3c). We found significant increases in community weighted value of wood density and proportion of shade-tolerant plants, but decreases in SLA and species richness between two investigations (Table 2). There were no significant differences in maximum height and animal-dispersed species (t test, p > 0.05). Forests on islands in the Thousand Island Lake are undergoing secondary succession following clear-cutting in the late 1950s, indicated by increasing biomass, decreasing stem density over the 5 years of measurement (Figure 3). Even small and isolated forest fragments are undergoing rapid succession, unlike the retrogressive succession frequently reported in the old-growth forests upon which most researches on fragmentation has focussed (Ewers et al., 2017). Plant trait analyses indicate that habitat fragmentation skews secondary succession. Large islands house a much higher proportion of animal-dispersed, fleshy-fruited and shade-tolerant species that are characteristic of old-growth forests (Feeley, Davies, Perez, Hubbell, & Foster, 2011; Reid, Holl, & Zahawi, 2015). These traits are linked because fleshy-fruited species are mainly dispersed by animals (Garcia, Zamora, & Amico, 2010). Having a greater abundance of bird and mammal species on large islands is consistent with island biogeography theory (MacArthur & Wilson, 1967). Large islands have more frugivorous birds that disperse the fleshy-fruited seeds, because these islands are able to support large populations of birds that have low extinction risks (Si, Baselga, Leprieur, Song, & Ding, 2016; Yu et al., 2012). These birds, in turn, promote the accumulation of bird-dispersed, fleshy-fruited plant species on large islands. Plot-scale species richness of woody plants and island area were positively related, as was total vascular plant species richness on the same islands (Hu et al., 2011; Yu et al., 2012). Larger islands harbour greater habitat heterogeneity, and have a greater probability of intercepting new species during the colonisation stage (Moreno-Mateos et al., 2012). As such, species accumulation rates are often slower in small forest fragments relative to large ones. Specifically, large islands contain more interior habitats that are suitable for the survival of shade-tolerant species (Table 2). We found significant edge effects on species composition indicated by ordination analysis. This is consistent with a study conducted in reservoir systems of central Brazilian Amazonia, where edge effects increased the probability of compositional shifts towards the early-successional assemblages (Benchimol & Peres, 2015). To sum up, regenerating forests on large islands have accumulated more species with late-successional traits compared with smaller islands. In terms of structural properties, we found small islands have similar AGB to larger islands despite the marked differences in species composition. The shade-intolerant, pioneer species that still dominate the forests after 50 years of succession (e.g. Pinus massoniana) are stress-tolerant species that even benefited from fragmentation with increased growth rates (McDonald & Urban, 2004; Reinmann & Hutyra, 2017). Recruitment and mortality contribute little to the biomass growth on the islands (Figure 2a), consistent with other studies showing that biomass dynamics are driven by tree growth in early secondary succession (Muscarella et al., 2017; Rozendaal & Chazdon, 2015). Hence, the productivity of the shade-intolerant overstorey dominates the accumulation of biomass, and it is relatively unaffected by fragmentation. Note that shade-tolerant plants have a greater biomass on larger islands, as expected from their greater contribution to species diversity on these islands (Figure 2), but this biomass remains small, and indeed is increasing at a slower rate than the overstorey layer. Differences in biomass among islands may well arise in another 50–100 years, when the pioneer community dies allowing differences in the shade-tolerant communities to become apparent (Lin et al., 2012; Figure S2). Our result is consistent with reports from forests in Madagascar that biomass remains constant across an edge-to-interior gradient (Razafindratsima et al., 2017), perhaps because biomass can recover rapidly during the early stages of succession (Martin, Bullock, & Newton, 2013). One Japanese study found that biomass growth was greatest in small mature forest fragments, with lower biomass loss due to mortality and higher biomass gain from tree growth (Tomimatsu, Yamagishi, Suzuki, Sato, & Konno, 2015). Therefore, small forest fragments have similar biomass, but different composition, to larger fragments. It is well established that fragmentation of old-growth forests causes a shift towards early-successional vegetation stages along the newly created edges (Laurance et al., 2006; Tabarelli et al., 2008), but little attention has been paid to the effect of fragmentation on regeneration following land abandonment (Arroyo-Rodríguez et al., 2017; Chazdon et al., 2016). Contrary to previous studies of old-growth forests (Laurance et al., 1997), we show that small fragments gain biomass at a similar rate than large fragments (Figure 2b). In old-growth forest fragments, edge effects cause high mortality of large trees which hold the vast majority of biomass (Laurance et al., 2000; Slik et al., 2013), leading to significant loss of biomass and rapid changes in ecosystem functioning (Magrach, Laurance, Larrinaga, & Santamaria, 2014). These studies of old-growth forests have focussed on the large tree (with DBH >10 cm), but 95% of the stems in our study system had DBH <10 cm, and mortality processes make little contribution to biomass dynamics (Figure 2a; see also Tomimatsu et al., 2015). In contrast, long-lived pioneers (mostly Pinus massoniana, which can persist for over 150 years (Lin et al., 2012)) have a dominating influence on biomass dynamics, and habitat fragmentation promotes their growth due to the increased light availability (McDonald & Urban, 2004). In this way, biomass gain from the rapid growth of these large pioneer trees is negatively correlated with island size (Figure 2c). Similarly, rapid growth of large pioneer trees on small islands have contributed to the higher average diameters (Table 2), contrast with the findings in the old growth forest fragments (Osuri et al., 2014). We also found a different pattern of functional composition in regenerating compared to old-growth forest fragments. Plots on large islands have a greater tendency to hold more small-statured and small-seeded species (Figure 4). This contrasts with findings from old-growth fragmentation studies, which report that large-seeded and big tree species go extinct in small fragments, from which dispersers are lost (Costa, Melo, Santos, & Tabarelli, 2012; Cramer et al., 2007). In our system, birds are main dispersers for animal-dispersed plants (80.3% of the individuals), the stem density of animal-dispersed plants is significantly influenced by island size (Table 2), but the stem density of species with unassisted-dispersal was unaffected by island size (p > 0.05). Seeds of bird-dispersed species are significantly smaller than species of other dispersal types in this region (Table S1). Considering that more birds are found on large islands (Yu et al., 2012) and bird-dispersed species have no problem in seed dispersal, large islands may accumulate small-seeded plants. In addition, larger seeds have stronger ability to establish and compete as seedlings on small islands (Coomes & Grubb, 2003). Together, this helps explain why larger islands have more bird-dispersed species and hence smaller seed sizes (Figure 4). Our study shows that fragmentation of regenerating forests has different effects on some structural properties and functional composition than reported in old-growth forests, potentially because different processes are involved: old-growth forests are closer to dynamic equilibrium and fragmentation moves them away from that state, while habitat fragmentation in regenerating forests alters the community assembly process of secondary succession. Patch size had a much greater influence than isolation effects on successional processes in the Thousand Islands Lake study (Table 2), because of its influences on species arrival, colonisation, establishment and extinction processes (Cook et al., 2005; Moreno-Mateos et al., 2012). The dominant effect of patch size on diversity patterns has also been reported for birds and snakes (Si, Pimm, Russell, & Ding, 2014; Wang et al., 2015; Yu et al., 2012). Large fragments have greater habitat diversity (Cook et al., 2005), support larger populations with lower extinction risk (MacArthur & Wilson, 1967), provide more habitats for potential seed dispersers and promote seed dispersal efficiency and recruitment success (Gibson et al., 2013; Turley et al., 2017). The long-term effects of habitat fragmentation on colonisation processes (in this case related to patch size) can lead to different successional pathways of forest communities (Arroyo-Rodríguez et al., 2017), with fewer late-successional species on smaller islands. Lack of strong isolation effects on species composition is against of our predictions that spatial isolation can decrease the arrival of new species. There are at least two explanations for why isolation had little discernible effect on most successional indicators. Firstly, birds are the main seed dispersers among islands, but diversity patterns of birds are not influenced by isolation as a result of their high mobility (Si et al., 2014). Secondly, the islands would have harboured seeds and perennating stems/roots after clear-cutting about 60 years ago, and these species may have helped species persist. These two reasons may explain why no dispersal limitation is apparent in our study system. Edge effects are critically important when old-growth forests are fragmented (Ewers et al., 2017; Magnago et al., 2014). Pioneer species with acquisitive traits, e.g. small seeds, low wood density and high SLA occupy edge niches (Slik et al., 2008; Tabarelli et al., 2010). However, edge effects influenced species composition, biomass, and the proportion of flesh-fruited, shade-tolerant plants in our study, but not other successional indictors. Similarly, some studies have shown that edge effects are limited in regenerating forest fragments (Lawes, Lamb, & Boudreau, 2005; Slik et al., 2011). This is probably due to the rapid formation and closure of forest canopy (mainly pine trees), which result in similar microclimates in the edge and interior habitats at the current stage of succession. Effects of habitat fragmentation in regenerating forest are of increasing interest (Arroyo-Rodríguez et al., 2017; Chazdon et al., 2016; Poorter et al., 2016). The Thousand Island Lake provides a “natural experiment” for studying the effects of habitat fragmentation on forest secondary succession. Our study shows that habitat fragmentation slows down the accumulation of species with late-successional traits in regenerating forests. In particular, paradigms on habitat fragmentation effects drawn from old-growth forest studies are unlikely to hold in regenerating forest fragments. The practical value of our study lies in its recognition that large forest patches will recover most rapidly from disturbance by humans, and small forest fragments may have different successional pathways from those of large fragments. To accelerate succession of regenerating forests, landscape patterns should be considered when designing restoration activities. This project was supported by the National Natural Science Foundation of China (31361123001, 31500382, and 31570524), US National Science Foundation (DEB-1342754, DEB-1342757), and 521 Talent Project of Zhejiang Sci-Tech University. We particularly thank Jinfeng Yuan, Yuexia Wang, Ge Nan, Lei Zhong and local farmers for their assistance with field work. We are grateful to staff at the David Attenborough Building for facilitating Jiajia Liu's visit at the University of Cambridge. We thank handling editor Peter Bellingham, Madelon Lohbeck and an anonymous reviewer for helpful comments on previous drafts of this manuscript. We also thank Ed Tanner, Tommaso Jucker, Laura Bentley, Jianguo Wu, Maxwell Wilson, Lin Jiang and Ping Ding for their suggestions. We have no conflict of interest to declare. J.J.L., G.H. and M.J.Y. led conceptualisation of the project; J.J.L., G.H., J.L.L., J.J.Y., Y.Q.L. and M.J.Y. collected the data; J.J.L. and D.A.C. analysed the data; J.J.L., D.A.C. and M.J.Y. led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication. Species data used in these analyses are provided in the Supplementary Materials. Data available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.7727rc0 (Liu et al., 2018). Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article." @default.
• W2890623015 created "2018-09-27" @default.
• W2890623015 creator A5032169401 @default.
• W2890623015 creator A5038252804 @default.
• W2890623015 creator A5053535927 @default.
• W2890623015 creator A5061137912 @default.
• W2890623015 creator A5081805002 @default.
• W2890623015 creator A5085399047 @default.
• W2890623015 creator A5089619012 @default.
• W2890623015 date "2018-10-15" @default.
• W2890623015 modified "2023-10-18" @default.
• W2890623015 title "Larger fragments have more late‐successional species of woody plants than smaller fragments after 50 years of secondary succession" @default.
• W2890623015 cites W1506250860 @default.
• W2890623015 cites W1811969986 @default.
• W2890623015 cites W1817539162 @default.
• W2890623015 cites W1818856478 @default.
• W2890623015 cites W1821968718 @default.
• W2890623015 cites W1893383824 @default.
• W2890623015 cites W1958909456 @default.
• W2890623015 cites W1965335994 @default.
• W2890623015 cites W1974273209 @default.
• W2890623015 cites W1977944376 @default.
• W2890623015 cites W1979165317 @default.
• W2890623015 cites W1984000480 @default.
• W2890623015 cites W1990553630 @default.
• W2890623015 cites W1994643738 @default.
• W2890623015 cites W1997761453 @default.
• W2890623015 cites W2002390308 @default.
• W2890623015 cites W2007401784 @default.
• W2890623015 cites W2017431346 @default.
• W2890623015 cites W2021050140 @default.
• W2890623015 cites W2026336219 @default.
• W2890623015 cites W2031090217 @default.
• W2890623015 cites W2037356735 @default.
• W2890623015 cites W2041396888 @default.
• W2890623015 cites W2047672923 @default.
• W2890623015 cites W2048895396 @default.
• W2890623015 cites W2056203170 @default.
• W2890623015 cites W2060321526 @default.
• W2890623015 cites W2073014258 @default.
• W2890623015 cites W2092431848 @default.
• W2890623015 cites W2095019927 @default.
• W2890623015 cites W2100345851 @default.
• W2890623015 cites W2105507965 @default.
• W2890623015 cites W2123865160 @default.
• W2890623015 cites W2124124339 @default.
• W2890623015 cites W2127597920 @default.
• W2890623015 cites W2133171915 @default.
• W2890623015 cites W2136615258 @default.
• W2890623015 cites W2136875645 @default.
• W2890623015 cites W2138310003 @default.
• W2890623015 cites W2142128121 @default.
• W2890623015 cites W2144042033 @default.
• W2890623015 cites W2153248731 @default.
• W2890623015 cites W2155292717 @default.
• W2890623015 cites W2162570317 @default.
• W2890623015 cites W2163966812 @default.
• W2890623015 cites W2166409662 @default.
• W2890623015 cites W2175186564 @default.
• W2890623015 cites W2186041856 @default.
• W2890623015 cites W2189206582 @default.
• W2890623015 cites W2254290345 @default.
• W2890623015 cites W2261784637 @default.
• W2890623015 cites W2276465998 @default.
• W2890623015 cites W2384950079 @default.
• W2890623015 cites W2395618046 @default.
• W2890623015 cites W2398582793 @default.
• W2890623015 cites W2525158211 @default.
• W2890623015 cites W2548525853 @default.
• W2890623015 cites W2567021640 @default.
• W2890623015 cites W2607306204 @default.
• W2890623015 cites W2750113781 @default.
• W2890623015 cites W2754299151 @default.
• W2890623015 cites W2769097220 @default.
• W2890623015 cites W2787970086 @default.
• W2890623015 cites W2791571805 @default.
• W2890623015 cites W4206171885 @default.
• W2890623015 doi "https://doi.org/10.1111/1365-2745.13071" @default.
• W2890623015 hasPublicationYear "2018" @default.
• W2890623015 type Work @default.
• W2890623015 sameAs 2890623015 @default.
• W2890623015 citedByCount "39" @default.
• W2890623015 countsByYear W28906230152019 @default.
• W2890623015 countsByYear W28906230152020 @default.
• W2890623015 countsByYear W28906230152021 @default.
• W2890623015 countsByYear W28906230152022 @default.
• W2890623015 countsByYear W28906230152023 @default.
• W2890623015 crossrefType "journal-article" @default.
• W2890623015 hasAuthorship W2890623015A5032169401 @default.
• W2890623015 hasAuthorship W2890623015A5038252804 @default.
• W2890623015 hasAuthorship W2890623015A5053535927 @default.
• W2890623015 hasAuthorship W2890623015A5061137912 @default.
• W2890623015 hasAuthorship W2890623015A5081805002 @default.
• W2890623015 hasAuthorship W2890623015A5085399047 @default.
• W2890623015 hasAuthorship W2890623015A5089619012 @default.
• W2890623015 hasBestOaLocation W28906230151 @default.
• W2890623015 hasConcept C128758860 @default.
• W2890623015 hasConcept C150117547 @default.

• next