Natural disturbance impacts on ecosystem services and biodiversity in temperate and boreal forests
Thom, Seidl · 2015
공개 학술지 Biological Reviews of the Cambridge Philosophical Society 에 실린 총설 논문입니다(2015년). CC BY 4.0 으로 공개된 글이고, 그림 · 표 · 참고 문헌 표시를 뺀 본문을 실었습니다.
출처 Thom & Seidl (2015), Biological Reviews of the Cambridge Philosophical Society · 라이선스 CC BY 4.0 · 원문
이 장은 원문만 있습니다. 문장별 해석은 앞 장부터 차례로 붙입니다.
2장 · Natural disturbance impacts on ecosystem services and biodiversity in temperate and boreal forests — MATERIALS AND METHODS
색을 입힌 곳이 TOEFL® 읽기 문제에 실린 대목입니다. 그 앞뒤의 글을 이어 읽어 보세요. 대목으로 가기
Literature review
We searched the literature for studies on disturbance by fire, wind and bark beetles, and their impacts on ecosystem services as defined by the Millennium Ecosystem Assessment (MEA, 2005), as well as their effects on biodiversity, focusing on species richness and habitat quality as well as on indices of diversity (e.g. Shannon-Index, Simpson-Index, etc.). We restricted our literature review to boreal and temperate forest ecosystems as subtropical and tropical forests differ considerably in ecological processes and anthropogenic impacts. In particular, extratropical forests are generally less diverse than tropical forests, and share a common set of genera as well as drivers of forest dynamics (e.g. temperature) (Thomas & MacLellan, 2002). Furthermore, land-use history and recent management differ strongly between tropical and extratropical regions, with a long history of intensive human use and several decades of sustainable management in the temperate and boreal zone (Siry, Cubbage & Ahmed, 2005; Canadell & Raupach, 2008). Focusing solely on the boreal and temperate subset of the literature controlled for these broad differences in our analysis, and thus increased the inferential potential with regard to disturbance effects. The literature search was performed using the Scopus database (SciVerse Scopus, 2013), and the cutoff date for the inclusion of publications was June 6th, 2013. The search terms and synonyms used are listed as supporting online information in Appendix S1. In total, 1958 papers were identified for screening. From this overall body of literature, reviews and syntheses were excluded in order to avoid double counting and the potential transfer of artifacts or errors from one review to the next (Whittaker, 2010). Furthermore, we excluded articles which did not compare disturbed forests with long-lasting undisturbed ‘control’ sites. Depending on the study scale and context, either the state before a disturbance, an undisturbed reference, or an assumption about an equilibrium condition was assumed as a reference to determine the disturbance effect. From the 1958 papers screened initially 478 were selected for further analysis. For each of these studies we collected information on geographical location, spatial and temporal scales, assessment methodologies and management treatments (Tables 1 and 2, see online Appendix S2). We furthermore recorded whether the reported disturbance effect is related to single or multiple disturbance events (i.e. disturbance regime). If studies included expert opinions on certain disturbance effects they were initially included in our database, but were subsequently omitted from quantitative analyses. We allowed multiple entries per study, for instance if a study examined more than one disturbance agent, ecosystem service or biodiversity indicator. Furthermore, considering that ecological effects can change over time, we also recorded the temporal time frame for every study. In order to alleviate potential autocorrelation issues, effects were grouped into four different time horizons (i.e. short term: 1–5 years, mid term: 6–25 years, long term: 26–100 years, very long term: >100 years). The final database for analysis contained 887 entries of disturbance effects on ecosystem services and biodiversity.
Analysis
We analysed our literature-derived database of disturbance effects in two steps. First, we assessed the disturbance effect on indicators of ecosystem services and biodiversity. To that end, a descriptive classification of the disturbance impact was made based on the findings reported in the literature (i.e. negative, neutral, mixed, or positive impact of disturbance on the respective indicator). This classification allowed us to synthesize results consistently from different methodological approaches. It furthermore enabled a comparison of disturbance impacts between ecosystem services measured on different scales (e.g. recreational value versus carbon storage in a forest landscape), as well as between the impacts on ecosystem services and biodiversity. Initially, we tested whether the observed distribution of studies over response categories differed significantly from a random distribution, i.e. we assessed whether a significant disturbance effect can be established from the literature. Subsequently, we tested for differences in disturbance impact among agents, biomes, and study approaches, evaluating the variation of disturbance impacts with these categories. In an attempt to confirm or reject the hypothesized diverging impacts of disturbance on criteria of relevance for ecosystem management (disturbance paradox hypothesis) we also tested whether disturbance impacts differ between indicators of ecosystem services and biodiversity. Another controversial issue in the context of disturbance management is the effect of salvage harvesting after disturbance, i.e. partial or complete removal of disturbance-killed trees from a site (Donato et al., 2006; Lindenmayer, Burton & Franklin, 2008; Thorn et al., 2014). We thus also tested the hypothesis that disturbance effects after salvage differ significantly from unsalvaged conditions. Finally, we also compared impacts of prescribed burning to those of wildfires, in order to test for differences in disturbance impacts from intended and unintended fires. All these tests were conducted using independence tests, a powerful, permutation-based approach to test the null hypothesis that two variables (measured on arbitrary scales) are independent of each other (Hothorn et al., 2008), using the package coin (Hothorn et al., 2013) within the R language and environment for statistical computing (R Development Core Team, 2014).
In a second step, in order to determine effect size, we conducted a meta-analysis based on quantitative information on disturbance impact for two particularly well-researched criteria: biodiversity and carbon storage. For biodiversity, we analysed disturbance-induced changes in species richness (S′, N = 57) and species entropy (H′, N = 28), the latter represented by the Shannon-Index of diversity. Due to the limited sample size further subdivision into the effects of disturbance on specific taxonomic groups was not possible. With regard to carbon storage, we distinguished between disturbance effects on total ecosystem carbon (TEC, N = 27), aboveground live carbon (ALC, N = 38), dead aboveground carbon (DAC, N = 25), and soil organic carbon (SOC, N = 39) in our meta-analysis. For all variables the effect size was calculated as the per cent change induced by disturbance relative to the reference condition (control). Only entries from single disturbance events without subsequent salvage logging were considered in this second analysis step. We used multiple linear regression analysis to examine the size and statistical significance of disturbance effects on indicators of carbon storage and biodiversity. To generalize the disturbance regime and allow a comparison across studies we used time since disturbance (in years) and disturbance severity (i.e. proportion of timber volume, basal area, or forest area affected by disturbance, using a scale of 0–1) as covariates in the analysis. These parameters were recently used by Miller, Roxburgh & Shea (2011) in an attempt to generalize disturbance effects on diversity. We analysed the residuals of our regression models for trends as well as for temporal autocorrelation (using a Durbin–Watson test), and found support for the assumptions of homoscedasticity and independence. From these regression models we analysed both the intercepts (i.e. the standardized effect at fixed severity and time since disturbance) and slopes (i.e. how the disturbance effect changes with time and severity). To aid the interpretation of the former we transformed severity to 1–severity in our analysis, making the intercept a standardized effect of 100% severity. Additionally, we fitted multiple linear regression models with disturbance agents and biomes as covariates in order to test for the generality of our findings across agents and geographical locations.
RESULTS
Disturbance effects on ecosystem services and biodiversity
Overall, 478 studies from the boreal (34.9%), cool (47.1%) and warm temperate (18.0%) biomes addressing effects of disturbances on forest ecosystems were reviewed. The overwhelming majority of articles originated from North America (63.8%), followed by Europe (21.3%) and Australasia (8.8%) (Fig. 1, Table 1). With regard to disturbance agents the effects of forest fires were addressed most frequently (78.0%), while only 15.4% of studies investigated impacts of wind and 6.6% of bark beetles. 60.9% of the research results compiled in our database were empirical, while 19.3% were based on expert opinion, 16.0% derived from simulation studies, and the remaining 3.8% either investigations based on remote sensing, public questionnaires or a combination of different approaches (Table 2). Studies from recent years were overrepresented in our database, with publications on disturbance impact increasing at a rate of approximately 3.1 papers per year between 1996 and 2012 (before 1996 the number of studies was sparse and irregular). This rate of increase of +11.9% year⁻¹ is considerably higher than that of the general literature on, e.g. ecosystem management, which was +7.0% over the same period (Seidl, 2014).
Overall, there is strong evidence for a distinct impact of disturbances on criteria relevant to ecosystem management, with only 19.3% of entries in our database showing no or mixed effects of disturbance. The fact that in our sample of the literature negative impacts (45.1%) and positive effects (35.6%) were nearly equally distributed confirms the hypothesized disturbance paradox in ecosystem management. These divergent impacts are primarily driven by the disparity of disturbance effects on biodiversity and ecosystem services (Fig. 2). We found that all ecosystem service categories [i.e. supporting, provisioning, regulating and cultural services (see online Fig. S1)] were affected predominately negatively by disturbance (P < 0.001). At the level of individual ecosystem service indicators, the only investigated aspect that was positively influenced was albedo (Fig. 3), as related to the climate change mitigation function of forest ecosystems (Jin et al., 2012). Timber and primary production, fresh-water provisioning as well as protection against gravitational natural hazards were found to be predominately negatively affected by disturbances. Moreover, the large majority of studies reported a negative disturbance impact on carbon storage, mainly due to a reduction of live biomass in the ecosystem. However, there were also some examples of a positive disturbance effect on carbon storage: in a boreal forest ecosystem in Ontario, ALC peaked 92 years after disturbance then declined to a significantly lower level during the following decades, stabilizing 140 years after disturbance (Seedre & Chen, 2010). For the same forest, SOC peaked between 29 and 140 years after disturbance, before decreasing by approximately one-third over the next 63 years (Chen & Shrestha, 2012). This suggests that not only direct disturbance-related C losses in ALC but also the enhanced growth of a regenerating forest as well as the rate of decomposition of dead organic matter need to be considered for a comprehensive assessment of disturbance effects on forest C budgets. Overall, however, 96.3% of 27 observations on C cycle impacts indicated a negative effect of disturbances on TEC.
By contrast, we found an overall positive effect of disturbances on biodiversity (P < 0.001). Species richness, habitat quality, and diversity indices were equally positively affected by disturbances. However, the disturbance effect is less consistent for biodiversity than for many ecosystem service indicators, and a number of studies also report negative impacts of disturbances on the indicators of biodiversity investigated here. Hingston & Grove (2010), for example, reported reduced bird species richness in Tasmanian lowland wet eucalypt forests during the first 50 years after wildfire. By contrast, Klaus et al. (2010) found a positive effect of fire on the number of bird species in southern Appalachian upland forests. This illustrates that some species groups might react differently to disturbances depending on the context and specific ecosystem investigated. Also belowground diversity is affected by disturbances, yet dedicated studies are still rare to date. Negative impacts on earthworm biomass and diversity at sites with uprooted trees were reported from areas as different as Belgium and northern Iran (Nachtergale et al., 2002; Kooch & Hosseini, 2010). Another belowground species group that was reported to be negatively affected by windthrow (salvaged) and fire disturbance was Oribatida in the Slovakian High Tatra Mountains (Lóšková et al., 2013). However, a positive impact of fire was reported on soil collembolan diversity in a northern hardwood forest (Huebner, Lindo & Lechowicz, 2012) as well as on soil microbial communities in Spain (Fontúrbel et al., 2012), indicating that disturbances can have both positive and negative impacts on soil diversity. Overall, however, 73.1, 69.8 and 65.3% of studies reported either a positive or neutral response of diversity, species richness and habitat quality, respectively, to disturbance.
At the level of different disturbance agents we found no support for significant differences between the effects of fire, wind, and bark beetles on indicators of biodiversity. With regard to ecosystem services, however, the impacts of fire differed significantly from those of wind and bark beetles (P < 0.001 and P = 0.006, respectively), with the latter agents being more frequently reported to have no influence on ecosystem services. This indicates that bottom-up disturbances such as fire (i.e. susceptibility decreasing with tree size and/or age) might have different impacts than top-down disturbances such as wind and bark beetles (where susceptibility increases with tree size and/or age). Differences in disturbance impacts between biomes were evident in our data: the effect of disturbances on ecosystem services differed among the boreal and temperate biomes (P < 0.001 and P = 0.005 for cool- and warm-temperate biomes, respectively), while boreal and cool-temperate biomes differed with regard to disturbance impacts on biodiversity (P = 0.022). Generally, disturbance effects were least distinctive in the boreal biome, with negative disturbance impacts on ecosystem services more pronounced in the temperate biomes compared to boreal ecosystems. However, disturbances also had a stronger positive effect on biodiversity in the cool-temperate biome than in the boreal biome.
By comparing results across different types of methodologies, e.g. simulation studies versus empirical approaches, we found some noteworthy deviations from the null hypothesis of consistent disturbance impacts across study methods. Concerning the impacts of disturbances on ecosystem services we found a significant difference between empirical studies and simulation studies (P = 0.030) as well as an indication for differences between empirical studies and expert opinions (P = 0.057), with simulation studies and experts reporting a stronger negative effect than empirical analyses. With regard to the effects of biodiversity, we found that both simulation studies (P = 0.007) and expert opinions (P < 0.001) differed significantly from empirical studies. Here, our data indicate that simulation studies underestimate the positive effects of disturbance on biodiversity compared to empirical analyses, while experts overestimate this positive effect. It is also interesting to note that neutral effects (i.e. no disturbance impact on biodiversity) were more commonly reported in empirical studies than in any other methodological approach.
The effect of salvage logging and prescribed burning
We tested whether the reported disturbance impacts of prescribed burning differed relative to those of wildfires, hypothesizing that controlled burns will have fewer negative effects on ecosystem service provisioning. We found no support for this hypothesis: prescribed burns were more frequently reported to have a negative impact on ecosystem services than wildfires (P < 0.001). Yet, this result must be interpreted with caution as it is based only on a small sample of studies for the effect of prescribed burning (N = 13). With regard to the predominately positive effects of fire on indicators of biodiversity, prescribed burns did not differ significantly from wildfires (P = 0.413).
Another frequently discussed management intervention in the context of disturbance management is salvage logging. Based on previous findings, we hypothesized a negative impact of salvage logging on biodiversity (Lindenmayer et al., 2008). Although a slight trend was evident in our data (i.e. the positive disturbance effect on biodiversity indicators was more pronounced for non-salvaged forests), it was not significant in our comparison of 38 observations on salvage logging with 145 observations of unsalvaged disturbance effects (P = 0.205). Moreover, with regard to the impact on ecosystem services no significant differences between salvaged and unsalvaged studies were found (P = 0.168), however the data reveal a negative trend for salvaged forests.
The size of disturbance effects on biodiversity and forest carbon storage
Disturbance effects on forest ecosystems differ greatly with disturbance severity and time since disturbance, which is why we studied effect sizes using these two variables as covariates. Time since disturbance significantly explained disturbance effects for all investigated carbon compartments (Table 3). Effects on ALC and DAC were particularly strongly related to this variable, and differences to undisturbed conditions (−91.3 and +155.5% in the first year after disturbance for ALC and DAC, respectively) decreased by +0.6% (ALC) and −1.4% (DAC) on average with every passing year following disturbance. Disturbance severity was not significant in any model, but was retained in the analysis due to its ecological relevance (see also Miller et al., 2011). While the analysis of disturbance impacts on indicators of C storage yielded acceptable coefficients of determination (R ² from 0.736 to 0.124), the explanatory value of disturbance regime covariates was poor with regard to species richness and entropy. Neither species richness nor entropy was found to differ significantly with time since disturbance and disturbance severity. Tests for differences between agents and biomes overall supported a common global meta-analysis under consideration of disturbance regime covariates for both response variables (data not shown).
The analysis of the standardized disturbance effect (i.e. the calculated impact for a year of an event with 100% severity) showed that indicators of biodiversity as well as deadwood C stocks increased with disturbance, while aboveground and soil carbon stocks decreased (Fig. 4). The mean ± 95% C.I. standardized effect of disturbance on total ecosystem carbon was −38.5 ± 8.3% (P < 0.001), while species richness was significantly increased by +35.6 ± 32.3% (P = 0.035).
TOEFL® 읽기 문제 5문항. 이 장의 한 대목을 그대로 옮기고 문항을 붙였습니다. 대목을 읽고 문제를 풉니다.
이 글은 위 출처의 원문을 그 라이선스(CC BY 4.0)에 따라 옮겼습니다. 참고문헌 목록과 인용 번호, 그림 · 표 · 각주 · 수식 · 부록은 빼고 절별로 나누어 실었으며, 그 밖의 문장은 원문 그대로입니다. 원 저작자와 발행 기관은 이 화면을 만들거나 검토하지 않았고, 이곳의 해석과 낱말 풀이는 공식 번역이 아닙니다.
