利用遙測技術探討台灣福山地區森林之動態

全文

(1)國立臺灣師範大學理學院生命科學系中央研究院生物多樣性國際研究生博士學位學程博士論文 Department of Life Science, College of Science Taiwan International Graduate Program on Biodiversity, Academia Sinica. National Taiwan Normal University Doctoral Dissertation. 利用遙測技術探討台灣福山地區森林之動態 Assessment of forest dynamics through remote sensing in Fushan, Taiwan 白喬楠 PEEREMAN, Jonathan Manuel Eric 指導教授：林登秋博士 Advisor: LIN, Teng-Chiu, Ph.D. 中華民國 109 年 6 月 June 2020.

(2) Acknowledgments This work is the product of three years of study in the laboratory of forest ecology of the NTNU, financed by the TIGP program supported by the Academia Sinica and the NTNU. This experience has been very enriching personally and scientifically, and I am thankful to everyone who have made this possible. I would like to express my gratitude to Prof. Teng-Chiu Lin, my supervisor and head of the NTNU laboratory of forest ecology, for his support, and valuable teachings regarding forest environments and the typhoons in Taiwan. Prof. Lin provided great guidance regarding analyses, and helped improving my writing skills. His optimism has been of great help during the complicated steps of this work. I am thankful to my proposal committee for their helpful comments and advices regarding my research objectives and data analysis. I appreciate J. Aaron Hogan for his contribution to our research, and for being our co-author for the two publications. To Dr. Chung-Te Chang for providing advices and data for this work. Many thanks to my lab colleagues for their kindness, cheerfulness, and assistance. I am grateful to students and friends of the TIGP program, friends from Taiwan and Belgium, especially Shez and Lorraine, for these good times that helped me remain sane and appreciate the Taiwanese culture. My family’s love and support has helped me going through these three years. Many thanks to my sister, Lisa, and my parents who are always up for interesting discussions, and who helped me vent out during stressful times. I also appreciate their comments regarding data presentation. Last, I am grateful to my fiancé Yu-chan for being herself and supporting me during these three years despite the distance.. i.

(3) 摘要永久樣區與遙測為森林生態學研究提供新的可能性，使我們對森林的動態有更多的了解。頻繁而大規模的擾動透過對森林的結構與過程影響而成為森林動態變化的重要趨動力。然而現有森林樣區對其所在的地景系統的代表性、不同熱帶氣旋對森林影響是否可相類比、以及由遙測影像發展出的上些常用的植生指數的可對比性仍有很大不確定性。本研究以遙測技術利用常用的植生指數探討五個颱風對台灣北部亞熱帶雨林--福山試驗林的影響。研究首先檢測 25 公頃的福山森林動態樣區對所在的地景系統的代表性，並檢驗涵蓋較廣海拔梯度以及不同形狀的替代樣區設計的代表性。其次本研究比較五個颱風對福山試驗林的影響及其與颱風風速風以及地形特徵(如坡度坡向以及海拔高度)的關係。研究亦分析颱風造成的植生損失的頻率和損失程度之間的關係，最後研究檢測五個颱風對四個植生指數造成的變動的一致性。研究結果顥示，現有的永久樣區僅涵蓋現有植生覆蓋情形的一小部分。雖然颱風造成的植生指數變動在永久樣區和地景系統間不一致，但動態樣區並無高估或低估颱風對地景系統影響的趨勢。以最小歐幾里得距離(Euclidean distance)為代表性的指標的分析結果顯示，現有的樣區比其他替代的樣區設置更能代表颱風造成的影響，且分析亦發現不同颱風造成的植生損失的空間分布差異很大(即相關性低)。在福山試驗林地形和颱風造成的植生損害之間的關係相當複雜，即使路徑與強度相似的颱風所造成的損害的空間樣貌亦不相同。然而颱風的影響仍有一些相似性，如均增加植生覆蓋的異質性，颱風前的植生覆蓋度皆和颱風影響的程度有正相關，以及植生損失的頻率和損失的程度亦多有正相關。又研究結果顯示不同植生指數對颱風擾動偵測的敏感度在不同颱風間並不一致，故在探討颱風對植生造成的影響時，宜同時使用多個植生指標。森林動態必然會受到擾動特性(如頻率、強度與發生時間)改變的影響，總結而言本研究發現相較於替代的樣區設計方式，現有的福山森林動態樣區對整個福山森林生態系有不錯的代表性，但要將樣區所得結果外推到整個地景系統仍應小心為之。由於熱帶氣旋對植生的影響在不同事件間並不一致，未來對擾動與森林動態的研究宜包括多個熱帶氣旋事件，才能對複雜的熱帶氣旋和地景系統的交互作用有較完整的了解。此外欲探討擾動對植生影響，最好要有定期的地面調查，以驗證利用遙測技術發展出來的植生指數所測得的結果。. 關鍵詞: 森林擾動;颱風;熱帶氣旋;森林永久樣區;代表性;地景生態;林冠覆蓋;植生指數;森林動態. ii.

(4) Abstract Permanent plots and remote sensing have permitted studies of many aspects of forests ecology and led to numerous advancements in our understanding of forest dynamics. Frequent large scale disturbances are important drivers of forest dynamics through their effects on forest structure, and processes. However, the landscaperepresentation of the current forest plots, and the comparability of tropical cyclone effects, and the comparability of commonly used vegetation indices based on remote sensing in assessing disturbance effects remain unclear. This study used remote sensing techniques and common vegetation indices to assess the effects of five typhoons on the Fushan Experimental Forest (FEF), a subtropical rainforest in northern Taiwan. First, it examined the landscape-scale representativeness of the 25-ha Fushan Forest Dynamics Plot (FFDP) and alternative plot designs based on wider altitudinal gradient and different shape. Then, the effects of the five typhoons on the FEF were compared in relation with their characteristics (e.g., wind speed and direction) and topographical variables (e.g., slope, elevation). The relationship between typhoon damages frequency – as defined by two detection thresholds – and intensity was also assessed. Finally, the consistency of four vegetation indices responses was examined across the five typhoons. The current permanent plot represented a small part of the vegetation cover diversity across the FEF. Although vegetation indices did not change similarly in the FEF and the FFDP after typhoon passages, there was no trend for under- or overexposure of the forest plot to typhoon damages across the five events. The use of minimal Euclidean distances as a representativeness metric demonstrated that the current plot was a better approach to landscape disturbances representation than alternative sampling designs. Furthermore, the comparison of the five typhoons indicated that spatial distributions of damages within the FEF varied greatly across events, as most correlations were low among events. Topography-damage relationships appear to be complex within the FEF as typhoons with similar tracks and iii.

(5) strengths did not have comparable damage distributions. Nevertheless, typhoons also had consistent effects such as increasing vegetation heterogeneity, and positive relationships between pre-typhoon vegetation cover and disturbance severity, and between typhoon frequency and disturbance severity. Finally, this study showed that the use of multiple vegetation indices is necessary when assessing typhoons disturbances as sensitivity of individual indices to vegetation change did not remain consistent among events. Forest dynamics are bound to be affected by changes in disturbance regime (e.g., frequency, intensity and timing). This study concludes that the current plot in northern Taiwan is a good representation of typhoon disturbances in its immediate landscape but that findings from the plot should be upscaled with caution. However, further studies focused on multiple cyclones are necessary to understand the complexity of cyclone-landscape interactions that vary among events, even when they have comparable tracks. In addition, routine surveys for studies that aim to assess disturbance effects on vegetation whenever possible are strongly recommended for the validation of assessments made with vegetation indices via remote sensing techniques.. Keywords: forest disturbance; typhoon; tropical cyclone; forest permanent plots; representativeness; landscape ecology; canopy cover; vegetation index; forest dynamics. iv.

(6) Table of Contents Acknowledgments................................................................................................................................. i 摘要 ......................................................................................................................................................... ii Abstract ................................................................................................................................................. iii List of Tables ......................................................................................................................................viii List of Figures........................................................................................................................................ x Chapter 1 – Introduction ..................................................................................................................... 1 1.1. Forest Environments and Disturbances ................................................................................. 1 1.1.1. Forests Dynamics................................................................................................................ 1 1.1.2. Cyclones Disturbances ....................................................................................................... 1 1.2. Large Scale Studies of Forest Dynamics ................................................................................. 3 1.2.1. Forest Permanent Plots ...................................................................................................... 3 1.2.2. Remote Detection of Vegetation and of its Properties .................................................. 6 1.2.3. Studies on Forest Landscapes Disturbances through Remote Sensing ...................... 8 1.3. Objectives.................................................................................................................................... 8 1.3.1. Representativeness of a Permanent Plot ......................................................................... 8 1.3.2. Effects of Multiple Cyclones ............................................................................................. 9 Chapter 2 – Landscape Representation by a Permanent Forest Plot and Alternative Plot Designs in a Typhoon Hotspot, Fushan, Taiwan ........................................................................... 11 Abstract ............................................................................................................................................ 11 2.1. Introduction.............................................................................................................................. 12 2.2. Materials and methods ........................................................................................................... 14 2.2.1. Site, Plot, and Disturbances............................................................................................. 14 2.2.2. Data Sources ...................................................................................................................... 16 2.2.3. Pre-Processing ................................................................................................................... 17 2.2.4. Processing .......................................................................................................................... 17 2.2.5. Analysis of FFDP-FEF Representation .......................................................................... 19 2.2.6. Testing Alternative Forest Plot Designs ........................................................................ 22 2.3. Results ....................................................................................................................................... 23 2.3.1. Overall Plot Differences between FEF and FFDP ........................................................ 23 2.3.2. Disturbances Effect and Representativeness ................................................................ 25 2.3.3. Alternative Plot Designs .................................................................................................. 28 2.4. Discussion ................................................................................................................................. 31 2.4.1. Overall Representativeness ............................................................................................. 31 2.4.2. Typhoons Damages Intensity in the Plot and the Reserve ......................................... 33 2.4.3. Vegetation Cover and Topographical Representativeness with Alternative Strategies ...................................................................................................................................... 34 2.4.4. Disturbances Representativeness with Alternative Strategies ................................... 34 2.4.5. Comparison of Strategies ................................................................................................ 35 2.5. Conclusions .............................................................................................................................. 36 Funding ............................................................................................................................................ 37 v.

(7) Acknowledgments.......................................................................................................................... 37 Conflict of Interest .......................................................................................................................... 37 Permission ....................................................................................................................................... 37 Chapter 3 – Assessing Typhoon-Induced Canopy Damage Using Vegetation Indices in the Fushan Experimental Forest, Taiwan .............................................................................................. 38 Abstract ............................................................................................................................................ 38 3.1. Introduction.............................................................................................................................. 39 3.2. Materials and Methods ........................................................................................................... 42 3.2.1. The Fushan Experimental Forest and Typhoons ......................................................... 42 3.2.2. Satellite Images ................................................................................................................. 45 3.2.3. Pre-Processing ................................................................................................................... 46 3.2.4. Processing .......................................................................................................................... 47 3.2.5. Analysis of Disturbances Among Vegetation Indices ................................................. 48 3.2.6. Typhoon Damages and Topography ............................................................................. 49 3.2.7. Disturbance Frequencies and Intensity ......................................................................... 49 3.2.8. Vegetation Recovery ........................................................................................................ 50 3.3. Results ....................................................................................................................................... 50 3.3.1. Vegetation Indices, Typhoons, and the Effect of Prior Vegetation Cover ................ 50 3.3.2. Variation of ΔVIs Among Typhoons ............................................................................. 52 3.3.3. Effects on Vegetation Heterogeneity ............................................................................. 52 3.3.4. Topography and Disturbance Severity ......................................................................... 53 3.3.5. Disturbance Frequency and Severity ............................................................................. 54 3.3.6. Recovery ............................................................................................................................ 56 3.4. Discussion ................................................................................................................................. 57 3.4.1. Consistency in the Damage Effects Among Typhoons ............................................... 57 3.4.2. ΔVIs in Relation to Topography ..................................................................................... 60 3.4.3. Typhoon Disturbance Frequency ................................................................................... 61 3.4.4. Consistency Among Vegetation Indices ....................................................................... 62 3.5. Conclusion ................................................................................................................................ 64 Funding ............................................................................................................................................ 65 Acknowledgments.......................................................................................................................... 65 Conflict of Interest .......................................................................................................................... 65 Permission ....................................................................................................................................... 65 Chapter 4 – Conclusions and Future Works ................................................................................... 66 References ............................................................................................................................................ 70 Appendices .......................................................................................................................................... 87 Appendix A – List of Abbreviations ............................................................................................ 87 Appendix B – Supplementary Data for Chapter 2: Comparison of the FFDP and FEF ....... 89 Appendix C – Supplementary Data for Chapter 2: Comparison of the Alternative Plot Designs ............................................................................................................................................. 92. vi.

(8) Appendix D – Supplementary Data for Chapter 3: Color composites of the FEF before and after disturbances ........................................................................................................................... 99 Appendix E – Supplementary Data for Chapter 3: Correlations ........................................... 100 Appendix F – Supplementary Data for Chapter 3: Relationships between topographical variables and vegetation change ................................................................................................ 101. vii.

(9) List of Tables Table 2.1. Basic information of Landsat images of the Fushan Experimental Forest and the Fushan Forest Dynamic Plot. ................................................................................. 17 Table 2.2. Topographical variables and vegetation indices used in this study along with their meanings and calculations based on Landsat bands and digital elevation model. .............................................................................................................................. 18 Table 2.3. Ranks of mean minimal Euclidean distances of four alternative plot design strategies based on vegetation indices for representativeness of difference effects and of the overall vegetation cover for the Fushan Experimental Forest. ............. 31 Table 3.1. Total rainfall (mm) and maximum wind speed (m s-1) measured at the Fushan Experimental Forest during the periods associated with the five typhoons passages. .......................................................................................................................... 44 Table 3.2. Basic information on Landsat images used in this study. ............................ 46 Table 3.3. Conversion thresholds of topographic position index to slope positions and based on standard deviation and slope steepness. ................................................... 47 Table 3.4. Average change (standard deviation) in vegetation indices over five typhoon intervals for the Fushan Experimental Forest. ........................................... 51 Table 3.5. Spearman’s correlations among the four ΔVIs measured over five typhoons for the Fushan Experimental Forest. ........................................................................... 51 Table 3.6. β Coefficients of ordinary least squares linear models between topographical variables and changes in vegetation index values (ΔVIs) associated with Typhoons Nari and Herb. .................................................................................... 54 Table B1. Overall analysis Spearman’s ρ (and p value) for correlation between vegetation indices and topographical variables for the FFDP and the FEF. ......... 90 Table B2. Mean (SD) ΔVIs for the five studied typhoons in the FFDP and FEF. ........ 90 Table B3. Coefficient of variation of four variables before and after disturbance. ..... 90 Table B4. Percentage of cells included in each damage sum class (0 to 5 disturbances) based on two thresholds and the two vegetation indices, NDVI and NDII for the Fushan Forest Dynamics Plot and the Fushan Experimental Forest. ..................... 91 Table C1. The 95% confidence intervals (CIs) from bootstrapped comparisons on means between alternative plots and the FEF in the overall analysis for vegetation indices, topographical variables. ................................................................................. 95 Table C2. 95% confidence intervals from bootstrapped comparisons on mean ΔVIs between the alternative plot designs and the FEF for Typhoon Nari. ................... 96 Table C3. 95% confidence intervals of bootstrapped comparison on means of FFDP and alternative plots for minimal Euclidean distances based on ΔVI for the four typhoons and three alternative plot strategies. ......................................................... 97 viii.

(10) Table C4. Wilcoxon tests p values (with Bonferroni adjustment) for the comparison of mean minimal Euclidean distances between each plot designs for all typhoons (except Typhoon Aere), and the overall vegetation analysis. ................................. 98 Table E1. Spearman's ρ for the correlation between pre-disturbance VIs and ΔVI associated with the typhoons. .................................................................................... 100 Table E2. Spearman’s ρ (p-value) between corresponding ΔVIs of the five typhoons (p adjustment for multiple comparisons using the Bonferroni method). ................ 100 Table F1. β Coefficients of ordinary least squares linear models between topographical variables and changes in vegetation index values caused by Typhoons Aere, Herb, Nari, and Soudelor in relation to topographical variables. ................................... 101 Table F2. Multiple comparisons on mean disturbances for different aspect and topographic position index derived categories. ...................................................... 106. ix.

(11) List of Figures Figure 1.1. Paths of the cyclones that passed at less than 100 km from land between 1980 and 2020. ................................................................................................................... 2 Figure 1.2. Locations of Forest Dynamics Plots from the Smithsonian ForestGEO network in three regions with frequent cyclones. ....................................................... 5 Figure 1.3. Spectrum of leaf reflectance over the light spectrum from blue to medium infrareds. ............................................................................................................................ 7 Figure 1.4. Main questions addressed in this study. .......................................................... 9 Figure 2.1. Trajectories of five typhoons analyzed in this study. .................................. 15 Figure 2.2. Overview of the steps for minimal Euclidean distances calculation based on vegetation indices and topographical variables. ................................................. 21 Figure 2.3. Boxplots of vegetation indices and topographical variables in the Fushan Forest Dynamics Plot and the Fushan Experimental Forest based on Landsat-8 and digital elevation model. ................................................................................................. 24 Figure 2.4. Minimal Euclidean distance between the Fushan Forest Dynamics Plot and the Fushan Experimental Forest calculated from VIs and topographical variables; and the disturbance effect associated with Typhoon Nari (ΔVI-based). ............... 25 Figure 2.5. ΔNDII and ΔNDVI for Typhoon Nari in the Fushan Forest Dynamics Plot and the Fushan Experimental Forest. .......................................................................... 26 Figure 2.6. Spatial representations of disturbance occurrences for NDVI and NDII or 5 typhoons based on ΔVI < 0 and ΔVI < mean – 0.5 × SD thresholds. .................... 28 Figure 2.7. Minimal Euclidean distances measured between the Fushan Experimental Forest and the Fushan Forest Dynamics Plot for the four-rectangular plot design strategy. ........................................................................................................................... 29 Figure 2.8. Minimal Euclidean distances for changes of vegetation indices associated with Typhoons Herb and Nari between the Fushan Forest Dynamics Plot and the four-rectangular plot design strategy within the Fushan Experimental Forest. ... 30 Figure 3.1. Tracks and intensities (on the Saffir-Simpson scale) of the five studied typhoons affecting the Fushan Experimental Forest. ............................................... 43 Figure 3.2. Mean hourly wind direction and speed, and direction of daily strongest wind during five typhoons that affected the Fushan Experimental Forest. .......... 45 Figure 3.3. Changes in coefficient of variation of four vegetation indices following the five typhoons................................................................................................................... 52 Figure 3.4. ΔNDII and ΔEVI for five typhoons and across the six damages frequency classes. .............................................................................................................................. 55 Figure 3.5. Percent of pixels of the eight aspects in different typhoon damage frequency classes (0 to 5) based on NDII and EVI. ...................................................................... 56 x.

(12) Figure 3.6. One-year recovery of the Fushan Experimental Forest as shown by the vegetation indices measured before typhoon passages and a year later at the same season. .............................................................................................................................. 57 Figure B1. Cloud coverage in satellite imagery in the Fushan Experimental Forest and Fushan Forest Dynamics Plot for 5 typhoon disturbances. ..................................... 89 Figure B2. Percentage of slope aspects in the FEF and the FFDP. ................................. 89 Figure C1. Location of forest dynamics plots created following four alternative strategies for the overall and disturbance analysis. .................................................. 92 Figure C2. Minimal Euclidean distances measured between the FEF and the FFDP for the four alternative plot designs. ................................................................................. 93 Figure C3. minED for changes of vegetation indices associated with four typhoons between the FFDP and the three alternative plot design strategies within the FEF. ........................................................................................................................................... 94 Figure D1. True colors composites of the forest cover and clouds of the Fushan Experimental Forest before and after each of the five studied typhoons. ............. 99. xi.

(13) Chapter 1 – Introduction 1.1. Forest Environments and Disturbances 1.1.1. Forests Dynamics Forests cover 43% of the global land surfaces (Roy et al., 2001) and represent half of the terrestrial net primary production (Saugier et al., 2001). Among forest biomes, the tropical and temperate biomes are the most productive with respectively 2500 and 1550 g m-2 year-1 of net primary production (Saugier et al., 2001). Forests are key in the global carbon cycle as they represent around 92% of terrestrial carbon stocks, with 52% in wet tropical forests which cover only 13% of the emerged surfaces (Roy et al., 2001). In addition, forests are critical for water cycle, and the conservation of fresh water resources (Ellison et al., 2017). Forest environments contain most of the terrestrial diversity, with many of them being in the world biodiversity hotspots. Although being increasingly protected (Morales-Hidalgo et al., 2015), high extinction rate in forests diversity is expected (Pimm et al., 2014). Benefits provided by forests are numerous and are not limited to carbon stocks and accumulation (Foley et al., 2005; Bonan, 2008; Watson et al., 2018), and forest conservation is necessary in order to maintain these ecosystems services (Chazdon, 2008). 1.1.2. Cyclones Disturbances Forest ecosystems are subject to several types of disturbances such as fires, ice storms, pathogens, or wind blows. These disturbances can create landscape mosaics of patches in different stages of stand development, with complex dynamics and processes (Turner, 2010). Tropical cyclones are large scale disturbances occurring at varying frequencies depending on the cyclone basins (Lin et al., 2020), which also coincide with areas of high tree density (Figure 1.1). The designation of tropical cyclones varies among basins, they are called “typhoon” in the western Pacific, and “hurricane” in the North Atlantic and eastern Pacific. Within forest landscapes, cyclone damages distribution is a complex result of topography (Metcalfe et al., 2008; 1.

(14) Turton, 2008; Tanner et al., 2014; Hu & Smith, 2018), cyclone properties (i.e., path, speed, rainfall, Hu & Smith, 2018), species sensitivity (Webb et al., 2014), and other factors such as soil types (e.g., nutrient richness, Herbert et al., 1999) and predisturbance vegetation state (Elmqvist et al., 1994; Harrington et al., 1997; Herbert et al., 1999; Burslem et al., 2000; Gannon & Martin, 2014; Tanner et al., 2014; Webb et al., 2014).. Figure 1.1. Paths of the cyclones that passed at less than 100 km from land between 1980 and 2020, and tree density (per pixel) as calculated by Crowther et al. (2015). Cyclone strength is on the Saffir-Simpson scale, data from the NOAA IBTrACS archive (Knapp et al., 2010; Knapp et al., 2018).. Defoliation is the most common damage caused by intense winds and rainfall to vegetation, followed by branch snap, uprooting, and bole breaking (Lugo, 2008). The cyclone litterfall can represent more than 50% of a non-cyclone year litterfall (Ostertag et al., 2003; Imbert & Portecop, 2008) and an important part of the inter-annual variation (Lin et al., 2017). The cyclone-induced litter represents nutrient influxes because it often contains higher concentrations of nutrients than the regular litter (Lodge et al., 1991; Herbert et al., 1999; Xu et al., 2004) although it varies among forests (Lin et al., 2003a). As a result, local dynamics in the carbon cycle are affected by the creation of gaps and the release of large amount of matter on the forest floor ranging. 2.

(15) from fine litterfall, coarse woody debris, to trees. Forest recovery begins fast after cyclone passage, with variation between tree species (Walker et al., 1992), and it is manifested through increased growth rates (Bellingham et al., 1995). Nevertheless, cyclones cause long-lasting changes in forest structures and diversity by damaging taller and less flexible trees, and inducing growth of new individuals in gaps which increase local diversity (Bellingham et al., 1995; Tanner & Bellingham, 2006; Heartsill Scalley et al., 2010). Consequently, cyclones have been described as selection pressure that could change forest resistance, resilience, and dynamics (Bellingham et al., 1995; Lin et al., 2011). Cyclone intensity and frequency are expected to change in the near future, with many models predicting more frequent intense events (Emanuel, 2013; Knutson et al., 2015; Sobel et al., 2016; Sugi et al., 2017). In addition, their paths have already shifted poleward (Kossin et al., 2014; Altman et al., 2018). In turn, forest processes in these regions are likely to be affected by changes in the disturbance regime (Johnstone et al., 2016), thus potentially affecting the carbon cycle. It appears from numerous field and remote sensing studies that cyclone damages are not distributed evenly across the landscape (Bellingham, 1991; Inagaki et al., 2008; Zhang et al., 2013; Gannon & Martin, 2014; Negrón-Juárez et al., 2014; Tanner et al., 2014), and that their effects in relation to topographical variables are not constant across studies (see Bellingham, 1991; Ostertag et al., 2005). Although studies have compared the effects of multiple cyclones on a same forest (≈ 1 km, Ostertag et al., 2005; Webb et al., 2011; McLaren et al., 2019; Uriarte et al., 2019), fewer studies have compared the effects of different tropical cyclones across the same landscape (≈ 10 km, Boose et al., 2004; de Beurs et al., 2019). 1.2. Large Scale Studies of Forest Dynamics 1.2.1. Forest Permanent Plots Forest dynamics have been extensively studied through ground surveys of permanent plots (Forest Dynamics Plot, FDP) that constitute several research networks of different focuses and geographical ranges, such as RAINFOR in tropical-equatorial 3.

(16) South America and the Smithsonian’s ForestGEO across different forest biomes (Malhi et al., 2002; Anderson-Teixeira et al., 2015). Data acquired from the FDPs have permitted numerous advancements in forest ecology either locally, or at the regional and world scales by using observations from several plots (Lewis et al., 2004; John et al., 2007; Laurance et al., 2009; Cleveland et al., 2011; Wright et al., 2011; Chisholm et al., 2014; Yu et al., 2019). Multiple surveys allow researchers to draw temporal trends. For instance, results from decades of plot censuses demonstrated the changes in tropical forest dynamics (see review of Lewis et al., 2009a; Phillips & Lewis, 2014). The FDPs are ideal cases to observe the effects of unpredictable events, and large plots are practical to study infrequent disturbances, such as forest fires (Lutz et al., 2018) or snow (Song et al., 2017), that could be missed by small plots. Several plots of the ForestGEO network are located in regions affected by cyclones (Figure 1.2), and as such, they permitted to identify effects of cyclone disturbances on forest characteristics such as tree mortality and forest resistance (Hogan et al., 2018; Johnson et al., 2018). To upscale results derived from the FDPs, the representativeness of the FDPs need to be validated. Many scientists expressed concerns regarding location of FDPs within their landscapes (e.g., Körner, 2009; Lewis et al., 2009a; Marvin et al., 2014), as biased location may invalidate generalization of their results at broader scales as discussed by Saatchi et al. (2015) regarding biomass (see also Chave et al., 2004), and by Clark (2002) about carbon sinks. Indeed, plot size, shape, and location has often been carefully chosen based on different criteria such as soil homogeneity (Ashton, 1995; Phillips et al., 2003; Magnusson et al., 2005), forest age (‘majestic forest’), or accessibility (Phillips et al., 2003; Ostertag et al., 2005), often in attempt to minimize possible biases. Among networks, ForestGEO and RAINFOR plots have followed guidelines to improve their representation of surrounding forests, which include plot size to reach a good representation of tree demographics.. 4.

(17) Figure 1.2. Locations of Forest Dynamics Plots from the Smithsonian ForestGEO network (Anderson-Teixeira et al., 2015) in three regions with frequent cyclones. Tree density from Crowther et al. (2015). Cyclone strength is on the Saffir-Simpson scale, data from the NOAA IBTrACS archive (Knapp et al., 2010; Knapp et al., 2018).. Different methods have been used to assess and improve plot representativeness. First, suspect plots can be removed from the analysis (Lewis et al., 2004; Phillips et al., 2004; Lewis et al., 2009b). Second, orientation with topographical gradient allows to increase representation of the environment variability (Tuomisto et al., 2003), which may be necessary when species diversity change with the terrain (e.g., Valencia et al., 2004). Third, under-sampled sites can be remotely sensed, as areas with different reflectance likely have different vegetation types (e.g., species, Rocchini et al., 2015). Langford et al. (2016) used several images from different phenological stages and Euclidean distances based clustering to assess the representativeness of their plots in a tundra landscape, a method which was first developed to study regional networks representativeness (Hargrove et al., 2003; Hoffman et al., 2013). Di Vittorio et al. (2014) used remote sensing to show that existing FDPs in the Amazon forest did not represent 5.

(18) tree mortality, as mortality may be spatially clustered (e.g., blowdown). Particularly, large disturbance events were not well represented by plots in the Amazon (Chambers et al., 2013), and the representation of disturbance events in this region has been thoroughly discussed (see Fisher et al., 2008; Lloyd et al., 2009; Muller-Landau et al., 2014) as failing to represent forest mortality can lead to over- or under-estimation of carbon stock. 1.2.2. Remote Detection of Vegetation and of its Properties Common remote sensing techniques used to study vegetation are either based on passive optical sensing (light reflectance, spectral) of the vegetation surface, or active such as the LiDAR that can go through the forest cover, and these techniques have been extensively used to improve the understanding of forest structure (e.g., Asner et al., 2008; Vitousek et al., 2009). Besides its use in local studies, optical remote sensing also permits detection of vegetation cover at the global scale (Crowther et al., 2015), and its application to follow vegetation change through time (Hansen et al., 2013; Song et al., 2018) has provided insights on world forests dynamics (Baccini et al., 2017). Remote sensing of plant properties lies in their varying reflectance in different parts of the light spectrum (Figure 1.3). Because of chlorophyll and leaf structure, healthy mature leaves have a high absorbance of blue and red wavelengths, whereas their reflectance is higher in the green wavelengths and dramatically increases past the red wavelengths, in the red-edge and the near infrareds (NIR, see Figure 1.3). Besides photosynthetic pigments, leaves moisture can also be monitored as water has a great absorbance in the short-wave infrareds (SWIR).. 6.

(19) Figure 1.3. Spectrum of leaf reflectance over the light spectrum from blue to medium infrareds. Reflectance decreases correspondingly to the presence of chlorophyll a and b in the blue and red wavelengths, water absorbs light at different infrared wavelengths. Corresponding bands from sensors of Landsat 5, 7, and 8 are indicated. Figure adapted from Pu (2017).. The change in reflectance over different wavelengths have been used to detect vegetation and as proxies for leaf and vegetation canopy structural and physiological properties using vegetation indices (VIs) that involve multiple wavelengths and coefficients, such as ratio of red-NIR (chlorophyll) and NIR-SWIR (water). Sensitivity to vegetation characteristics vary greatly among VIs (Huete et al., 1997a, review by Ollinger, 2011), and some have been largely used, such as the Enhanced Vegetation Index (EVI), to study large areas. For example, Huete et al. (2006) used EVI to confirm plots observations in the Amazon, and demonstrated that the amount of sunlight was more determinant for rainforest dynamics than rainfall during the dry season. Eddy flux measurements further confirmed these observations and suggested that VIs could be proxies for the estimation of gross primary productivity. In the Amazon and the Andes, Asner et al. (2015) demonstrated that leaf chemical traits could be monitored 7.

(20) using remote sensing (LiDAR and spectroscopy), so as to avoid the lack of landscape representativeness of plots in that region (see also Marvin et al., 2014). 1.2.3. Studies on Forest Landscapes Disturbances through Remote Sensing Remote sensing techniques are appropriate to monitor forest disturbances (reviewed by Frolking et al., 2009), and among them, vegetation indices, used to compare pre- and post-disturbances images, have been the basis of numerous studies. VIs have been used to follow several disturbance types such as herbivory (Townsend et al., 2012), deforestation (Schultz et al., 2016), and fire (Kane et al., 2014). Forest change caused by cyclones has also been the focus of several studies using either VIs or other derives of aerial images, such as the identification of non-photosynthetic vegetation (NPV) or the disturbance index based on tasselled cap transformation (Ayala-Silva & Twumasi, 2004; Wang et al., 2010; Rossi et al., 2013; Negrón-Juárez et al., 2014; Hu & Smith, 2018; de Beurs et al., 2019; Hall et al., 2020 among others). In combination with ground surveys, studies have permitted to identify VIs that are sensible to vegetation change such as the EVI (Rossi et al., 2013), and NDII (Normalized Difference Infrared Index, Wang et al., 2010). However, analyses of forests dynamics have also been done using aerial images alone, such as the assessment of the effects of two major hurricanes in Puerto Rico showing the effect of distance to cyclone eye and the surface type on hurricane effects (Hu & Smith, 2018). In Taiwan, Normalized Difference Vegetation Index (NDVI) has been used to assess typhoon damages in Lienhuachih and Fushan forests, and their relationships with topography (Kang et al., 2005; Lee et al., 2008). 1.3. Objectives 1.3.1. Representativeness of a Permanent Plot Disturbance representativeness of current FDPs is not well understood for forests with short disturbance return interval (< 10 y) whereas their data have been used to draw conclusions for these regions (e.g., Johnson et al., 2018). Better understanding of their representation of disturbance effects is necessary because the effects of 8.

(21) disturbance on vegetation structure and composition have been shown to vary with topography (Chang et al., 2012; Lin et al., 2012; Shen et al., 2013; Di Vittorio et al., 2014). Poor disturbance representativeness would imply that forest dynamics over the broader area is not well represented. Hence, Chapter 2 focuses on the disturbance representativeness assessment of a 25-ha FDP in northern Taiwan (the Fushan Forest Dynamics Plot, FFDP) through remote sensing (Figure 1.4). Moreover, although plot designs strategies have been largely discussed (e.g., Alder & Synnott, 1992; Clark & Clark, 2000; Reese et al., 2005; Lutz, 2015), further research is needed to understand how the various designs perform in capturing landscape pattern of disturbance effects. Thus, Chapter 2 compares the representativeness of diverse alternative plotting designs with the FDP for their representativeness of the landscape vegetation cover and its variation following typhoons (Figure 1.4).. Figure 1.4. Main questions addressed in this study.. 1.3.2. Effects of Multiple Cyclones Although several studies have been focused on the effects of cyclones on forests, few compared multiple events (e.g., Burslem et al., 2000; Boose et al., 2004). Hence, the second part of this study, Chapter 3, assesses how spatial distribution of cyclone damages in a forest landscape varies among five typhoon events in relation with terrain, as well as with wind speed and direction, and rainfall, in order to determine 9.

(22) the consistency of disturbance effects among typhoons (Figure 1.4). Moreover, Ostertag et al. (2005) have observed that disturbed trees in Puerto Rico were also likely to be redisturbed by the next hurricane, while, in the same island, Hall et al. (2020) showed that changes in soil condition caused by the first hurricane of a hurricane season could explain the variation in damages caused by the second one. Hence, Chapter 3 also examines whether frequently disturbed sites also show greater typhoon effects passage. Finally, the consistency of vegetation indices to monitor cycloneinduced damage is not well understood. In fact, the consistency of a VI response to disturbances is necessary for cross-studies comparisons and their integration to broadscale models. Zhang et al. (2016) and Wang et al. (2010) have already compared common VIs, and identified the NDII as the best index to follow vegetation changes following disturbance events in flat landscapes, however, there has been no comparison of commonly used VIs for multiple disturbances in subtropical forests of more complex topography. Here, Chapter 3 compares four common VIs across a same landscape and five typhoons to identify consistencies across events, and whether there is a single VI that is the most sensitive to typhoon disturbance across the five events.. 10.

(23) Chapter 2 – Landscape Representation by a Permanent Forest Plot and Alternative Plot Designs in a Typhoon Hotspot, Fushan, Taiwan† Abstract Permanent forest dynamics plots have provided valuable insights into many aspects of forest ecology. The evaluation of their representativeness within the landscape is necessary to understanding the limitations of findings from permanent plots at larger spatial scales. Studies on the representativeness of forest plots with respect to landscape heterogeneity and disturbance effect have already been carried out, but knowledge of how multiple disturbances affect plot representativeness is lacking—particularly in sites where several disturbances can occur between forest plot censuses. This study explores the effects of five typhoon disturbances on the Fushan Forest Dynamics Plot (FFDP) and its surrounding landscape, the Fushan Experimental Forest (FEF), in Taiwan where typhoons occur annually. The representativeness of the FFDP for the FEF was studied using four topographical variables derived from a digital elevation model and two vegetation indices (VIs), Normalized Difference Vegetation Index (NDVI) and Normalized Difference Infrared Index (NDII), calculated from Landsat-5 TM, Landsat-7 ETM+, and Landsat-8 OLI data. Representativeness of four alternative plot designs were tested by dividing the FFDP into subplots over wider elevational ranges. Results showed that the FFDP neither represents landscape elevational range (<10%) nor vegetation cover (<7% of the interquartile range, IQR). Although disturbance effects (i.e., ΔVIs) were also different between the FFDP and the FEF, comparisons showed no under- or over-exposure to typhoon damage frequency or intensity within the FFDP. In addition, the ΔVIs were of the same magnitudes in the plots and the reserve, and the plot covered 30% to 75.9% of IQRs of the reserve ΔVIs. Unexpectedly, the alternative plot designs did not lead to increased representation of Content of this chapter published in Remote Sensing 2020, 12, 660; doi:10.3390/rs12040660; co-authored with James Aaron Hogan, and Teng-Chiu Lin. Permission regarding the use from MDPI as appeared on the MDPI website is attached at the end of this chapter.. †. 11.

(24) damage for 3 out of the 4 tested typhoons and they did not suggest higher representativeness of rectangular vs. square plots. Based on the comparison of mean Euclidean distances, two rectangular plots had smaller distances than four square or four rectangular plots of the same area. Therefore, this study suggests that the current FFDP provides a better representation of its landscape disturbances than alternatives, which contained wider topographical variation and would be more difficult to conduct ground surveys. However, upscaling needs to be done with caution as, in the case of the FEF, plot representativeness varied among typhoons. Keywords: forest disturbance; typhoon; forest permanent plots; representativeness; landscape ecology 2.1. Introduction Forest studies commonly use permanent forest dynamics plots, which may serve as reference sites for the larger studied environment. Their uses range from a single vegetation census to multiple censuses spanning decades, as is the case in long-term forest research such as those of the Smithsonian’s ForestGEO network (Condit, 1995; Anderson-Teixeira et al., 2015). In long-term research, permanent plots (PP) can be large continuous blocks (e.g., ForestGEO plots) or consist of several smaller plots dispersed over the landscape (e.g., RAINFOR plots, Malhi et al., 2002). The PPs permit important discoveries in plant demography, community dynamics, to ecosystem change ecology (Lewis et al., 2004; John et al., 2007; Laurance et al., 2009; Cleveland et al., 2011; Wright et al., 2011; Chisholm et al., 2014; Yu et al., 2019, among numerous others), made possible by standardized regular censuses that require a tremendous amount of time and effort. As a result, PP design is the product of trade-offs between the ecological questions being asked, and practical constraints of regularly censusing them, such as accessibility and plot size. However, researchers must consider the limitations inherent to the PP design if PP-based findings are to be scaled up to the landscape and regional scales, uses that may be beyond initial research goals.. 12.

(25) Studying the representativeness of PPs is necessary to verify if observations from them can represent their environments at the landscape scale, and also to know which parts of the landscape remain under-represented (Neyland et al., 2000; RodríguezGonzález et al., 2017). For example, a study comparing plots and landscape through LiDAR imagery and image spectrometry in the Amazon found biases for forest biomass and canopy height in both lowland and montane forests (Marvin et al., 2014). A similar study in Hawaii revealed that PPs underestimated the variation in forest height and had a slight bias for taller forests (Vitousek et al., 2009). PPs provide a unique approach to study disturbance ecology, however, disturbances often occur in clusters across the landscape (Fisher et al., 2008, discussed by Lloyd et al., 2009; MullerLandau et al., 2014). Thus, plots may miss clustered disturbance events, potentially over-estimating biomass and carbon stocks in forests (Di Vittorio et al., 2014) or lie in the center of a disturbance cluster and overestimate disturbance effects at the landscape level. It is obvious that heterogeneities within the landscape should be considered in studies that aim to address questions relevant to the ecosystems in which the PPs are situated. The effect of topography is evident not only in the spatial variation of plant species distributions and vegetation structure but can also be seen from the spatial pattern in disturbance effect (e.g., fire, storms, and cyclones). For tropical cyclone disturbance, although the distance to the cyclone’s eye plays a key role in the severity of forest damage across the landscape, topographic exposure is a principle factor explaining spatial patterns of cyclone-induced tree damage (Hopkins & Graham, 1987; Bellingham, 1991; Walker et al., 1992; Metcalfe et al., 2008; Turton, 2008). For example, cyclone-induced tree damage often varies with elevation, resulting in change of biomass and tree height along the altitudinal gradient (McEwan et al., 2011; Chi et al., 2015), depending on cyclone strength (see Inagaki et al., 2008). In addition, wind exposure, which is heavily influenced by topographic position, also explains some of the spatial heterogeneity in cyclone tree damage (Boose et al., 1994; Yap et al., 2016) and vegetation structure (Noguchi, 1992) across the landscape. In general, valley and ridge 13.

(26) vegetation are more damaged than slope vegetation (Ostertag et al., 2005; Tanner et al., 2014), but ridgetops can be less susceptible than slopes to landslides caused by heavy rainfall (McEwan et al., 2011). Therefore, the interactions among all the topographical features lead to complex spatial patterns of damage across the landscape, which in turn affect forest resistance to cyclone disturbance (Boose et al., 1994; Webb et al., 2014). The representativeness of PPs for the larger landscape likely varies among PPs, as they can vary greatly in physical environment (e.g., topography, soils), community composition, and agents of disturbance (Anderson-Teixeira et al., 2015). Thus, more PP-specific studies are needed for evaluating the representativeness of each PP to its greater landscape, before generalizations across PPs can be made. The Fushan Forest Dynamics Plot (FFDP) is a 25-ha PP in Taiwan, and studies from the FFPD have provided important insights into forest dynamics in relation to topography, extreme climate events, and cyclone disturbance (McEwan et al., 2011; Hogan et al., 2018; Johnson et al., 2018). However, the representativeness of FFDP for the larger landscape has not been carefully examined. In this study, we examine how topography relates to the magnitude of typhoon-induced changes in vegetation cover between the FFDP and the broader landscape (i.e., the Fushan nature reserve). We first assess the representativeness of the FFDP for larger landscape in terms of the topography and vegetation and evaluate how the FFDP represents variation in landscape-scale. typhoon. disturbance. effects.. Then,. we. explore. if. plot. representativeness varies by plot design (i.e., shape) and placement within the reserve. 2.2. Materials and methods 2.2.1. Site, Plot, and Disturbances The 1097 ha Fushan Experimental Forest reserve (FEF) is located in northeastern Taiwan (Figure 2.1), with elevation ranging from 400 to 1400 m above sea level (asl). It has an annual mean temperature of 18.2°C, mean annual precipitation of 4270 mm, and mean relative humidity of 95% (Lin et al., 2018). The forest is described as an oldgrowth sub-montane evergreen broadleaf forest (Su et al., 2007; Lin et al., 2018). The 14.

(27) FEF is subject to winter monsoons, and is frequently hit by typhoons between June and October, with an average of 0.74 typhoon per year from 1951 to 2005 (Lin et al., 2011). In 2004, the FFDP was established following the CTFS ForestGEO standardized protocol for forest dynamic plots (Su et al., 2007). It is located in the western part of the FEF and ranges between 600 and 750 m asl, with approximately 84% of its area lying between 650 and 750 m asl (Su et al., 2007).. Figure 2.1. Trajectories of five typhoons analyzed in this study. Dotted lines delineate a 100 km range from the path of the typhoon eye. The location of the Fushan Experimental Forest (FEF) is labelled in the left panel and enlarged in the right panel, in which the Fushan Forest Dynamics Plot (FFDP) is also labelled. Typhoon tracks were acquired from the NOAA IBTrACS archive (Knapp et al., 2010; Knapp et al., 2018).. 15.

(28) Five typhoons were selected for this study based on the availability of high-quality Landsat images with low amounts of cloud cover. Landsat images with little cloud cover are rare because it usually rains more than 200 days per year at the FEF (Lin et al., 2011). Thus, images with <50% cloud cover and within five weeks before and after typhoon impact were selected when available in order to minimize phenological changes and canopy regrowth that can take place within weeks (Walker et al., 1992; Hu & Smith, 2018). The five typhoons were all of category 2 or 3 on the Saffir-Simpson scale (Simpson & Riehl, 1981) during landfall or at their nearest point to the FEF if they did not make landfall (i.e., Typhoon Aere). Moreover, for each of the five typhoons, the maximal distance between the reserve and the typhoon eye was always less than 100 km when it was nearest to the FEF so that the FEF was mostly within the radius of maximum wind of the typhoon (Figure 2.1). 2.2.2. Data Sources Landsats 5, 7, and 8 data were acquired from the USGS’s Earthexplorer website (USGS, 2019) as level-2 data surface reflectance, which were atmospherically and terrain corrected. The USGS also provides a cloud mask along with level-2 data that relies on the CFmask detection algorithm (Foga et al., 2017). The spectral features of the three sensors have been shown to be comparable, such that their data can be used in continuity to monitor forests (She et al., 2015; Vogelmann et al., 2016). A 30 m resolution DEM (digital elevation model) was acquired from the JAXA’s website (JAXA, 2019) and 30 m resolution global forest cover raster from the Global Forest Change dataset version 1.5 (GFC, Hansen et al., 2013) was used to remove non-forested surfaces. Table 2.1 summarizes basic information of the images used in the study.. 16.

(29) Table 2.1. Basic information of Landsat images of the Fushan Experimental Forest (FEF) and the Fushan Forest Dynamic Plot (FFDP). All images are of 30 m spatial resolution. Null cells were cloud covered or non-forested surfaces. Note that because analysis-related typhoon disturbance is based on pre- and post-disturbance images within five weeks of each typhoon except for Dujuan and Soudelor, substantial cloud cover on some parts of the FEF is inevitable (see Figure B1, in Appendix B). Files. Acquisition Dates. Sensors. overall Fushan Typhoon Herb Typhoon Nari Typhoon Aere Typhoon Soudelor Typhoon Dujuan. 2018-03-13 1996-07-06 & 08-23 2001-09-14 & 10-08 2004-07-12 & 09-30 2015-06-09 & 08-12 2015-08-12 & 12-02. OLI TM5 TM5 ETM+ TM5 OLI OLI. Null cells FFDP (%) FEF (%) 0 2.5 0 19.4 0 17.5 0 30.8 1.4 48.8 0 48.1. 2.2.3. Pre-Processing Landsat data was pre-processed following Young et al. (2017). All reflectance rasters were topographically corrected with the topcor function of ‘RStoolbox’ package (Leutner et al., 2019) in R (version 3.6) by applying a C correction on each band with JAXA’s DEM. All disturbance images were partially cloudy (Table 2.1, see Figure B1) and clouded surfaces were removed from both pre- and post-disturbance images, and from the DEM in subsequent analysis. Visual inspection of true color composites showed that the USGS’s mask removed all clouds and shadows except for Typhoon Nari, for which clouded areas were removed manually. Across all images, cloud coverage was skewed toward higher elevations (mean elevation 861 m for clouded sections against 783 m for the entire reserve). Non-forested surfaces were excluded from the analysis by masking areas with forest cover below 75% in the GFC dataset, as has been done in studies of other tropical moist forests (Achard et al., 2014; Vieilledent et al., 2018). 2.2.4. Processing Two vegetation indices (VIs), NDVI (Normalized Difference vegetation index, Rouse et al., 1974) and NDII (Normalized Difference Infrared Index, Hardisky et al., 1983) were used for the overall and disturbance analysis using QGIS 3.4.2. NDVI was. 17.

(30) chosen because it is the most widely used VI for various studies around the globe (Kerr & Ostrovsky, 2003; Pettorelli et al., 2005; Huete, 2012) and has been shown to have a close relationship with leaf area index (LAI, Carlson & Ripley, 1997), and therefore forest productivity (Clark et al., 2008). The NDII was chosen because SWIR-NIR-based indices such as NDII can track defoliation better than does NDVI (Townsend et al., 2012). The topographic position index (TPI), slope steepness and aspect were calculated with the terrain function from R’s ‘raster’ package (Hijmans, 2019). Table 2.2 summarizes the meaning and calculation of topographical and vegetation indices. Changes of VIs following each disturbance event, ΔVI, were calculated for each VI and disturbance events as: ΔVI =. −. (2.1). Numerical values of slope aspect (0 to 360) were converted into eight cardinal orientations (i.e., N, NE, E, ES, S, SW, W, and NW). Table 2.2. Topographical variables and vegetation indices (VIs) used in this study along with their meanings and calculations based on Landsat bands and digital elevation model. R: red band; NIR: near infrared; and SWIR1: short-wave infrared. Index, variable Topographical Elevation. Meaning. Calculation Method. Altitude above sea level. -. Slope aspect. Surface orientation (e.g., facing North). Orientation of each cell. Slope steepness. Angle to the horizontal (°). Comparison of cell value with surrounding cells. TPI. Topographic Position Index. Landforms, positive values associated with ridges, negative values with valleys1. Comparison of a cell elevation with surrounding cells. Spectral. NDVI2. Normalized Difference Vegetation Index. It quantifies vegetation by measuring the difference between near-infrared (which vegetation strongly reflects) and red light (which vegetation absorbs), with high values indicating dense vegetation cover.. − +. Normalized Difference Infrared Index. It is sensible − 1 to vegetation water content and vegetation changes NDII + 1 associated with cyclone disturbance4 or defoliation5. 1 Weiss (2001); 2 Rouse et al. (1974); 3 Hardisky et al. (1983); 4 Wang et al. (2010); 5 Townsend et al. (2012). 3. 18.

(31) 2.2.5. Analysis of FFDP-FEF Representation The analyses of the general FFDP-FEF representation (one Landsat image) and the representation for each typhoon disturbance (two images per event, before and after) were carried out similarly. First, mean values for VIs, ΔVIs (for typhoon disturbances only), and the three topographical variables of FEF and FFDP were statistically compared using the bootstrapped difference (via 5000 iterations, n = 289) as: ! "# $%. &. −. ! "# $%. '. .. (2.2). There was no significant difference between the FEF and the FFDP if the 95% confidence interval of their mean difference (CI) includes 0. Slope aspects of FEF and FFDP were compared with a χ2 test. Spearman’s ρ was used to examine correlations between VIs, and between VIs and topography of the FFDP and FEF. For each bootstrapped comparison of means, we calculated the coefficient of variation (CV). The post- and pre-disturbance CVs were compared between the FEF and FFDP as: ( where ). *. ) ). *. ). &. −. (. +. is pre-typhoon CV and ). +. ) ). *. ). (2.3). '. +. is post-typhoon CV. A bootstrapped. comparison of means was used again to assess if typhoon-induced changes of heterogeneity of vegetation indices were different between the FFDP and the FEF. Two thresholds were used to evaluate if a section of the FEF and FFDP was disturbed by each of the five typhoons: ΔVI < 0 ΔVI <. ΔVI. &. (2.4). − 0.5 × 2. &. (2.5). Although a cell with ΔVI < 0 could be considered as a disturbed cell, a lower threshold (i.e., more negative ΔVI) helps minimize mis-identification (e.g., due to differences in image quality). Thus, the zone model threshold, Equation 2.5, was also used to define disturbed cells (Jin et al., 2013). Spatial variation in typhoon frequency, defined as the proportion of pixels experiencing different frequencies of typhoon disturbance (0 to 5), in the FEF and FFDP were compared with a χ2 test. 19.

(32) The Euclidean distance measure (ED) has been used as a metric to estimate representativeness of eddy flux towers (Hargrove et al., 2003; He et al., 2015) and sampling networks (Hoffman et al., 2013) at the continental scale, and observation plots at the landscape scale (Langford et al., 2016). We used ED to pinpoint well or underrepresented areas of the FEF by the FFDP. The VIs (NDVI, NDII) and topographical variables (elevation, slope, TPI) were normalized (i.e., scaled to 0–1)—hence leading to a distance metric akin to the Mahalanobis distance—and then used to compute multidimensional distances between all FEF cells and FFDP cells. Several EDs were calculated for each reserve cell because there was more than one cell within the FFDP. Thus, the minimal values among these distances—the minimal ED (minED)—was kept for each of the reserve cells on the basis that one cell of the FFDP is more representative of a given FEF cell than the others (Figure 2.2). Two types of minED maps were produced: VI-based ED (a combination of NDVI and NDII) and topography-based ED (elevation, slope steepness, TPI). Correlations between topography-based and VIbased minEDs were conducted using Spearman’s ρ.. 20.

(33) Figure 2.2. Overview of the steps for minimal Euclidean distances (minED) calculation based on vegetation indices (VIs) and topographical variables. This is a modification of regular Euclidean distance (ED) computation in which there is more than one reference cell (the plot contains x cells), where only the minimal value is kept. Steps leading to minED maps are described here: After grouping several data layers (step 1), Euclidean distances were computed between each plot cell and each reserve cell based on variable values standardized on a 0–1 scale (step 2). Then, step 3 involved keeping only the minimal Euclidean distance value computed for each reserve cell (among the n values, where n = the number of cells within the plot), on the basis that there was one plot cell that offered the best representation of the considered reserve cell. Finally, step 4 led to the construction of georeferenced rasters with minED values computed either with vegetation indices or topographical variables. MinEDs show how well the reserve is represented based on the inserted variables. 21.

(34) 2.2.6. Testing Alternative Forest Plot Designs Because vegetation cover varies with elevation and the FFDP spans a limited subset of the FEF elevational range, four alternative plot designs were tested to evaluate if increasing plot elevational range and splitting the large plot to several smaller plots, without changing the total plot area, would improve the plot’s landscape representativeness. The four plot designs we tested are: one large-square plot (one 500 m × 500 m, ≥30% of FEF elevational range), four square plots (four 250 m × 250 m, ≥50% of FEF elevational range), two rectangular plots (two 250 m × 500 m transects, ≥50% of FEF elevational range), and four rectangular plots (four 125 m × 500 m, ≥50% of FEF elevational range). The first strategy is like the FFDP, except that it covers a broader elevational range of the FEF, whereas the other three strategies use subplots that are dispersed across the landscape and cover even greater elevational ranges of the FEF. Ten plot replicates per plot design were created by randomly generating plot locations and orientations within the FEF. Plot locations were selected based on the pre-defined elevation range threshold and by limiting the number of cells with no data to <10%. Because pre-typhoon and post-typhoon images for each typhoon event were less than 30 days apart from the typhoon event (except for Dujuan), it was difficult to get disturbance-related images that lack cloud cover as it rains more than 220 days annually at the FEF (Chang et al., 2017). After removing clouds, maintaining an elevation range >50% was not achievable. Thus, in the analysis of disturbance representativeness, the elevation range threshold was set to ≥30% of the FEF range; however, the analysis for Typhoon Aere and for the one-square strategy were excluded because clouds limited the image area available for the analyses. Ten to twelve replicates were used for each strategy. The same spatial areas were used to study the four typhoons even though they had varying cloud cover. This permitted us to compare typhoons effects in the exact same locations, and thus to maintain integrity in comparing alternative plot designs by keeping the other variables (such as slope steepness or other unmeasured factors) fixed across different typhoons. Some of the. 22.

(35) replicates of the same strategy overlap partially in areas, however subplots of each replicate do not overlap (Figure C1). The analysis of the landscape level representativeness of the four strategies was done similar to the FFDP-FEF comparisons. Elevation was excluded from the minED calculation because this variable was used to create alternative plots. Finally, VI-based minED values for each strategy were compared through a pairwise Wilcoxon test, with Bonferroni adjustment, for each typhoon and the overall FEF. They were then ranked on a 1 to 4 scale for the overall analysis, or 1 to 3 for the disturbance analysis based on the mean minimal Euclidean distance calculated across all replicates, where the strategy with the lowest mean gets first rank. 2.3. Results 2.3.1. Overall Plot Differences between FEF and FFDP The difference between the FEF and FFDP was significant for both VIs as well as slope and elevation (Figure 2.3). Vegetation cover was greater in the FFDP, with both mean NDII and NDVI being significantly higher for pixels with the FFDP boundary (NDII = 0.25 ± 0.01, NDVI = 0.80 ± 0.009) than the greater FEF (NDII = 0.21 ± 0.03 and NDVI = 0.78 ± 0.033). Mean elevation and slope steepness were significantly higher in the reserve, whereas there was no difference for TPI between the FEF and FFDP. The proportion of the cells with different slope aspects were also significantly different between the FFDP and the FEF (χ2 = 41.86, df = 7, p < 0.001, Figure B2), with a greater proportion of NW-N slope aspects and a lower proportion of SW-S aspect in the FFDP than in the FEF. The range of values represented by the FFDP in relation to the FEF was greater than 66% for the TPI, 58% for slope steepness, but only 10% for elevation, 9% for NDVI, and 23% for NDII. The interquartile range (IQR i.e., Q1–Q3) of the FFDP represented only 6.3% of the NDVI, 0% of the NDII, and 9.5% of the elevation, but 33.7% (slope steepness) and 81.1% (TPI) of the FEF’s respective IQRs.. 23.

(36) Figure 2.3. Boxplots of vegetation indices and topographical variables in the Fushan Forest Dynamics Plot (FFDP) and the Fushan Experimental Forest (FEF) based on Landsat-8 and JAXA digital elevation model (30 m spatial resolution). Significant differences between FFDP and FEF are shown with an asterisk (*), based on bootstrapped comparison on means.. In both the FFDP and FEF, the NDII and NDVI were moderately correlated (ρ > 0.6, p < 0.001, Table B1). However, the topographical variables were only weakly correlated (ρ < 0.2), except for TPI and elevation in the FFDP (ρ = 0.5, p < 0.001). There were no significant correlations between VIs and topographical variables except for elevation, which was negatively correlated with NDVI (ρFEF = −0.56, ρFFDP = −0.34, p < 0.001) and NDII (ρFEF = −0.32, ρFFDP = −0.21, p < 0.001) in both the FFDP and FEF. The minimal Euclidean distance varied across the reserve (Figure 2.4). There was a significant, but very weak relationship between VIs-based minED and topographical variable-based minED (ρ = 0.05, p < 0.001).. 24.

(37) Figure 2.4. Minimal Euclidean distance between the Fushan Forest Dynamics Plot (FFDP) and the Fushan Experimental Forest (FEF) calculated from (a) VIs (left) and topographical variables (right); (b) and the disturbance effect associated with Typhoon Nari (ΔVI-based). White surface within the FEF was either obscured by cloud cover or is non-forested area. The FFDP boundary is shown via the red square.. 2.3.2. Disturbances Effect and Representativeness The mean ΔVIs were mostly negative for the five disturbances, indicating that all typhoons led to vegetation loss (Table B2). For Soudelor, Herb, and Nari, there was less vegetation loss in the FFDP than in the reserve (Figure 2.5, Table B2). There was, however, no significant difference for ΔNDVI with Dujuan and ΔNDII with Nari between the FFDP and the FEF, but there was a greater vegetation loss within the FFPD than the FEF for Typhoon Aere (positive 95% CI, Table B2). Among the five typhoon disturbances, ΔNDVI of the FFDP encompassed 12.5% (Soudelor) to 47.7% (Aere) of the ΔNDVI range of the FEF, and ΔNDII of the FFDP covers 17.6% (Soudelor) to 35.4% (Dujuan) of the ΔNDII range of the FEF. However, looking at IQRs, the FFDP contained between 31.4% (Soudelor) and 75.9% (Dujuan) of the reserve’s IQR for ΔNDII, and between 34.2% (Aere) and 71.0% (Dujuan) for ΔNDVI.. 25.

(38) Figure 2.5. ΔNDII and ΔNDVI (calculated as the difference between post- and predisturbance values) for Typhoon Nari in the Fushan Forest Dynamics Plot (FFDP) and the Fushan Experimental Forest (FEF). Negative values indicate loss of vegetation. Significant differences between FFDP and FEF based on bootstrap comparison on means are shown with an asterisk (*).. ΔNDVI and ΔNDII were moderately correlated (0.4 < ρ < 0.7) for Typhoons Herb, Nari, and Soudelor for the FEF, were strongly correlated (0.7 < ρ < 0.9) for Typhoon Soudelor within the FFDP and for Typhoon Dujuan for the FEF and were very-strongly correlated (ρ > 0.9) for Typhoon Dujuan for both the FFDP and the FEF. However, for Typhoon Aere, the correlation between the two VIs was weak (ρ = 0.05 and ρ = 0.28 in the FFDP and the FEF, respectively). Topographical variables were weakly, if ever, correlated with ΔVIs with absolute Spearman’s ρ always below 0.2 (e.g., for elevation). NDVI-based CV ratios were different in FFDP and FEF except for Typhoons Herb and Nari (Table B3). For both the FFDP and FEF, the ratio was below 1 (i.e., greater post- than before-typhoon CVs), indicating that typhoons caused an increase in the variation of vegetation cover. In addition, the CV ratios of the FEF were significantly closer to 1 than those of the FFDP, signifying that there was a greater difference between pre- and post-disturbance vegetation states within the FFDP than for the 26.

(39) greater FEF (Table B3). For NDII-based CVs, significant differences between CV ratios in the FEF and FFDP were seen for Typhoons Nari, Dujuan, and Soudelor. The magnitude of change in NDII-based CVs relative to zero (i.e., no change) varied in both FFDP and FEF, with either increasing, decreasing, or equal values depending on the typhoon considered (Table B3). The χ2 tests indicated different patterns in damage frequencies between FFDP and FEF based on ΔVI < 0 threshold (χ2 = 22.48, df = 3, p < 0.001 for NDVI; χ2 = 16.26, df = 4, p = 0.003 for NDII). For NDVI, the FFDP had more cells within the intermediate damage frequency classes (2, 3) than the FEF, whereas the proportion of higher frequency (4, 5) and no damage (0) were lower in the FFDP than in the FEF (Figure 2.6). The NDII also showed damage of higher intermediate frequency (3) within the FFDP than within the FEF (Table B4). Using the ΔVI < mean − 0.5 × SD threshold, the differences between the FFDP and the FEF remained (χ2 = 49.00, df = 5, p < 0.001 for NDVI; χ2 = 33.88, df = 5, p < 0.001 for NDII), even though less disturbance frequency was detected, and there was an increased proportion of undisturbed areas (class 0), especially in the FFDP.. 27.