臺北市都市公園中地上與地下部交互作用與土壤生態功能運作之關係

國立臺灣大學理學院地理環境資源學研究所 碩士論文

Department of Geography College of Science

National Taiwan University Master Thesis

臺北市都市公園中地上與地下部交互作用 與土壤生態功能運作之關係

Above-belowground interactions on

soil ecosystem functioning in Taipei urban parks

謝嘉 Chia Hsieh

指導教授﹕李美慧 博士 Advisor: Mei-Hui Li, Ph.D.

中華民國 106 年 7 月 July 2017

Acknowledgement

This thesis summarizes my four-year research life in Department of Geography, National Taiwan University and takes a giant stride forward to my research goal in urban ecology. Throughout this journey, I would like to acknowledge the people who give me pertinent advice, unselfish assistance, and kindly supports. Without their help, I could not accomplish what I achieved.

First of all, I would like to give my deepest appreciation to my advisor, Prof. Mei- Hui Li (李美慧) in Department of Geography, National Taiwan University. During my days in the Environmental Ecology Lab, Prof. Li not only guided me to find my research goal but also taught me a rigorous attitude to do independent research. Especially, I would like to appreciate Prof. Li for having discussions with me. Whenever I ran into obstacles, she always helped me out with hers sincere words and thoughtful advice. Therefore, I could make progress and accomplish what I achieved with the five-year training in the Environmental Ecology Lab. It is a great privilege to meet Prof. Li, like a friend and mentor, at the start of my academic journey.

Secondly, I would like to thank Friends of Daan Forest Park Foundation (大安森林 公園之友基金會) for the research funding, and Prof. Yang-Hsin Shih (施養信) for the technical assistance to analyze TOC. I would also like to give my profound gratitude to Prof. Ping-Shih Yang (楊平世) and Prof. Shih for being my master thesis oral defense committee and giving the appreciate advice to complete this thesis. Moreover, I would like to give my acknowledgment to Parks and Street Lights Office of Taipei City Government, Chiang Kai-shek Memorial Hall, and National Theater & Concert Hall for approving the soil and arthropod sampling in Daan Forest Park, 228 Peace Memorial Park, and Chiang Kai-shek Memorial Hall.

Thirdly, I would like to give my great thanks to my college friends, K.T. (愷庭), X.C.

(顯程), Tzu-Hsin (慈忻), Hsin (歆穎), and the Environmental Ecology Lab members, Paul Wu (瑞槟學長), Ball Ball (球球學姊), and Wen-Chieh (雯潔) for the spiritual supports and fieldwork assistance in my master research. Especially, it is grateful to have my boyfriend, Ying-Jen (穎任), in my life. His continuous accompany, encouragement, and inspiring academic discussion gives me bravery and confidence to face any difficulty.

I would like to thank the fieldwork helpers, Yu-Sheng (宇昇), Hsiang-Kai (祥楷), Leo (逸峰), Peter (博群), Tzu-Yu (姿又), 信之, and 嘉鴻 for helping measure and record the environmental factors during field sampling work. I would also like to thank Kae-An (凱安) for helping modify the drEEM toolbox used in this study.

Finally, I must give my profound gratitude to my family. Their constant warm words and support in my life give me the strength to be myself and go on my academic journey.

I would like to give my heartfelt appreciation to those who have helped me. Without your help, I could not have done this research. Words are not enough to express my gratitude.

Hope everything is going well with all of you.

Chia Hsieh (謝嘉) Environmental Ecology Lab, Department of Geography, National Taiwan University, July 2017

摘要

都市綠色基礎建設所具之土壤生態系統功能，可協助減緩人為活動對於都市生態 系統中土壤生物多樣性與生地化循環所造成的影響。然而目前對於綠色基礎建設 中，尤其是都市公園，植被與土壤交互作用關係與其所具之生態系統作用過程尚未

有完整地了解。因此，本研究於2015 年 12 月至 2017 年 1 月間，調查了副熱帶地

區台北市三個都市公園中，樹與草以及草坪等兩種棲地型態之地上─地下部特性與 溶解性有機物質特性。本研究使用紫外光─可見光光譜學與三維螢光光譜學，以了 解土壤溶解性有機物質的組成特性，探討其與地上─地下部特性在兩種棲地型態間 的差異，及其與土壤微生物胞外酵素活性間的關係性。在相同的枯落物移除頻率、

割草強度以及相近之遊憩活動強度下，植被組成、表層土壤 (0-10 cm) 物化特性與 土壤生物組成於三個公園間不具顯著差異。兩種棲地型態中，地表節肢動物組成也 不具有顯著差異。另一方面，草坪相對於樹與草之棲地型態，則具有較高的土壤水 分含量與土壤微生物胞外水解酵素活性，以及較高比例之降解過的溶解性類腐植 酸物質、氮營養物質濃度與土壤有機質含量。此外，前四者間更具有顯著正相關性。

本研究結果顯示，在實際具有經營管理與民眾遊憩使用等人為活動下，透過規劃種 植相似之植被型態，可使不同土地利用歷史的公園綠地具有相似的土壤食物網組 成與生態系統作用速率。由於木本植物枯落物的移除，地表草本植物成為影響地表 棲地結構組成、表土層之有機物質資源輸入，以及土壤含水量等重要的因子。兩個 棲地型態間，相似的地表植被組成使土壤食物網組成趨於相近，而土壤中有機物質 分解速率則主要受到土壤水分含量所影響。基於本研究的結果，建議未來如欲藉由 規劃與經營管理公園綠地中之植被，以提升土壤生物多樣性或土壤有機物質含量 時，應進一步了解植被組成多樣性與植被管理強度對於土壤生物組成的影響與其 系統功能運作之情況。

關鍵字：人為活動、綠色基礎建設、地表節肢動物、微生物胞外酵素活性、溶解性 有機物質、三維螢光激發─發散矩陣、平行因子分析方法

Abstract

Soil ecosystem functions in urban green infrastructures (GIs) could help mitigate the impact of human activities on soil biodiversity and biochemical cycle in urban ecosystems.

However, plant-soil interactions on soil ecosystem processes in GIs, especially urban parks, are unclear. In this study, the above-belowground properties and soil dissolved organic matter (SDOM) properties of two habitat types, namely tree with grass (TG) and grass (G), from three subtropical urban parks in Taipei City were investigated from December 2015 to January 2017. Ultraviolet-visible spectroscopy and three-dimensional excitation-emission matrix fluorescence spectroscopy were used to characterize SDOM.

The differences between the above-belowground properties in TG and those in G were investigated. The relationships between SDOM properties and soil microbial extracellular enzyme activities were also analyzed. With identical litter removal, grass clipping intensities, and similar recreational activities intensities, there were no significant differences in plant compositions, the physicochemical properties of surface soils (0-10 cm), and soil food web compositions. Between two habitat types, ground arthropod compositions were not significantly different. In addition, G had higher soil water content, soil microbial extracellular hydrolases activities, decomposed humic acid-like substances in SDOM, nutrients levels, and SOM content than TG did. Furthermore, the former four parameters had strong positive correlations with each other. The results of this study suggest the formation of soil food web and the rate soil ecosystem processes could correspond to vegetation planting under different land use histories in the field. With the woody litter removal, ground grass cover becomes an important factor in the composition of ground habitat structure, soil organic resources input, and the soil water retention in two habitat types. Between two habitat types, the comparable composition of grass plants

could form similar soil food web compositions in two habitat types, and the rates of decomposition and mineralization could be majorly induced by soil water content. From this study, if we want to promote the soil biodiversity or the soil organic matter content by plant planning and management in GIs, the effect of plant diversity and management intensity on soil ecosystem functioning should be further investigated.

Keywords: human activity, green infrastructure, ground arthropod, microbial extracellular enzyme activity, dissolved organic matter, three-dimensional excitation- emission matrix, parallel factor analysis

Contents

Acknowlegement………...………i

摘要……….…...……...…...iii

Abstract………..………..………...…iv

List of Figures………….……….………....ix

List of Tables………..…………...xiii

1. Introduction………..1

2. Literature review………..3

2.1 Relationships between plant, soil, and soil food web ... 3

2.2 Anthropogenic effects on above-belowground interactions in GIs ... 7

2.2.1 Land use/cover change and history ... 7

2.2.2 Plant managements ... 9

2.2.3 Recreational activity ... 10

2.2.4 Summary ... 10

2.3 Soil dissolved organic matter (SDOM) and above-belowground interactions 12 2.3.1 SDOM, and its sources and properties ... 12

2.3.2 Properties of SDOM and microbial extracellular enzyme activities (EEAs) ……….14

2.3.3 Spectroscopic methods for SDOM properties ... 17

2.3.4 Summary ... 21

3. Materials and methods………...22

3.1 Study sites ... 23

3.2 Soil sampling and storage ... 26

3.3 Ground arthropod sampling ... 27

3.4 Plant survey ... 27

3.5 Recreational activity survey ... 28

3.6 Soil physical properties analyses ... 28

3.7 Soil chemical properties analyses ... 29

3.7.1 pH ... 29

3.7.2 Nitrate and ammonium extraction ... 29

3.7.3 Extractable nitrate concentration ... 30

3.7.4 Extractable ammonium concentration ... 30

3.8 SDOM analyses ... 31

3.8.1 SDOM extraction and storage ... 31

3.8.2 Dissolved organic carbon ... 31

3.8.3 Colorimetric properties of SDOM... 31

3.8.4 Fluorescent properties of SDOM ... 32

3.9 Soil microbial extracellular enzyme assays ... 34

3.10 Statistics ... 38

4. Results and discussions………..40

4.1 Park effect on the above-belowground properties ... 40

4.1.1 Aboveground habitat characteristics ... 40

4.1.2 Ground arthropod composition ... 42

4.1.3 Belowground soil physiochemical properties and microbial activities ... 47

4.2 Effects of habitat structures on above-belowground interactions ... 51

4.2.1 Aboveground properties of plant, recreational activity and ground arthropod ……….51

4.2.2 Soil physiochemical properties ... 57

4.2.3 Grass height, soil water content and soil nutrient levels ... 59

4.2.4 The quantity and quality of SDOM ... 63

4.2.5 Soil microbial EEAs and ecosystem processes ... 69

5. Conclusion………...77

Reference………79

Appendix………....94

Appendix A Land covers and GPS locations of sampling sites in three parks..….94

Appendix B Photos of sampling sites in three parks………...98

Appendix C Photos of impounded surface waters in Daan forest park... 101

Appendix D Soil textures of two habitat types in three parks….………...101

Appendix E Weather and climate conditions in 2016 from the nearest Taipei weather station……….……….………..102

Appendix F The bimonthly data of SDOM properties………..103

List of Figures

Figure 2.1. The pathways and interactions between organic matter resources and soil food web with ecosystem processes during organic matter decomposing and nutrient cycling………..……...…..…………3 Figure 2.2. The composition, ecological role and interactions in currently known soil food web.………...….…...……5 Figure 2.3. The conceptual model of soil dissolved organic matter to link the above- and belowground properties………...…………...………...…..11 Figure 2.4. The pathways between labile/recalcitrant soil dissolved organic matters, the catalysis of microbial extracellular oxidases and hydrolases, passive soil organic content, and nutrient concentrations.……….….………….……….….16 Figure 2.5. The identified fluorophore peak positions from natural water in a three- dimensional excitation-emission matrix (peak A and C: humic acid-like ； peak M;

precursor of peak C; peak B: tyrosine-like; peak C: tryptophan-like; FI: fluorescence index; BIX: biological index; HIX: humification index)……….…….…19 Figure 3.1. The flow diagram of research process in this study………...……..22 Figure 3.2. (a) The location of Taipei City in Taiwan, (b) the position of study sites in Taipei City and their surrounding land use types; DFP: Daan forest park, CKS: Chiang Kai-Shek memorial hall, PMP: 228 peace memorial park (the land use map is from National Land Surveying and Mapping Center, 2nd land use investigation in 2006)……23 Figure 3.3. The mean monthly temperature and monthly total precipitation of nearest Taipei weather station from 2006 to 2015 (data from Taiwan Central Weather Bureau:

http://www.cwb.gov.tw/V7/climate/watch/watch.htm)....………...…………25

Figure 3.4. The flow diagram for the calibration of excitation-emission matrices……...33 Figure 4.1. Non-metric multidimensional scaling (NMDS) ordination plot of Bray-Curtis dissimilarity on the abundance data of ground arthropod communities in the three urban parks (n=5, stress value: 0.11); the ellipse shows 95% confidence intervals around the centroid of the ground arthropod community positions for each park site; the red hollow point and the red font depict the position of ground arthropod taxon in the respective ordination plot………...………....………...…44 Figure 4.2. The relationship between grass coverage and (a) tree DBH in tree with grass (n=12), or (b) exercise activities intensities in two habitat types (n=24). r represents the Pearson’s correlation coefficient and ρ represents the Spearman’s correlation coefficient………...……….52 Figure 4.3. Seasonal ground arthropod community properties (mean ± SD) for (a) individual abundance, and (b) taxa richness in 5 traps per 100 m² in two habitat types in 2016 (n=12; February: Winter, April: Spring, August: Summer, November: Fall, December: Winter)………...………....…53 Figure 4.4. The relationship between grass coverage and (a) individual abundance, or (b) taxa richness in 5 traps per 100 m² in two habitat types (n=24). r represents the Pearson’s correlation coefficient.………..………...……54 Figure 4.5. Non-metric multidimensional scaling (NMDS) ordination plots of Bray- Curtis dissimilarity on taxa abundance in the two habitat structures (n=12) for five seasons, (a) Winter, (b) Spring, (c) Summer, (d) Fall, and (e) Winter in 2016; the ellipse shows 95% confidence intervals around the centroid of the ground arthropod community positions for each park site; the red hollow point and the red font depict the position of ground arthropod taxon in the respective ordination plot.……….……..56

Figure 4.6. Bimonthly soil physiochemical properties (mean ± SD) for (a) soil bulk density, (b) soil organic matter content, and (c) soil pH in two habitat types in 2016(n=12)………...………...58

Figure 4.7. The relationship between grass coverage rate and (a) soil bulk density, or (b) soil organic matter content in two habitat types (n=24). r represent the Pearson’s correlation coefficient.………..……59 Figure 4.8. Bimonthly ground plant properties (mean ± SD) for (a) grass height, surface soil properties (mean ± SD) for (b) soil water content, and for nutrient concentrations of (c) nitrate and (d) ammonium in two habitat types in 2016 (n=12)………60 Figure 4.9. The relationship between soil water content and (a) grass coverage, or (b) soil bulk density in both habitat structures (n=24). r represent the Pearson’s correlation coefficient………...………. 62 Figure 4.10. The excitation-emission matrix (above) and excitation-emission spectral loading (below) of three components of SDOM from park soil by PARAFAC method;

Comp.1: aromatic tryptophan-like component, Comp.2: humic acid-like component, and Comp.3: tryptophan-like component………..………64 Figure 4.11. Bimonthly surface soil dissolved organic matter properties (mean ± SD) for (a) quantity, and relative abundances of three groups of organic components, (b) aromatic tryptophan-like component, (c) humic acid-like component, and (d) tryptophan-like component in two habitat types in 2016 (n=12)………..……….65 Figure 4.12. Bimonthly values (mean ± SD) of (a) molecular weight index S275-295, and (b) aromaticity index SUVA254 for soil dissolved organic matter in two habitat types in 2016 (n=12)………...……….……..67

Figure 4.13. Bimonthly values (mean ± SD) of (a) fluorescence index, (b) biological index, and (c) humification index for soil dissolved organic matter in two habitat types in 2016 (n=12)………...………...68 Figure 4.14. Bimonthly soil microbial extracellular oxidase activities (mean ± SD) for (a) PO: phenol oxidase, and (b) PER: peroxidase in two habitat types in 2016 (n=12)..70 Figure 4.15. Bimonthly soil extracellular microbial hydrolase activities (mean ± SD) for (a) BG: β-glucosidase, (b) NAG: β-N-acetylglucosaminidase and (c) LAP: leucine aminopeptidase, and (d) the value of ln(BG)/ln(NAG+LAP) in two habitat types in 2016 (n=12)………...72 Figure 4.16. The relationship between soil water content and (a) hydrolase activities, or (b) oxidase activities; the relationship between dissolved organic carbon concentration and (c) hydrolase activities, or (d) oxidase activities in both habitat structures in June, August, October and December 2016 (n=24 x 4); ρ represents the Spearman’s correlation coefficient………....73 Figure 4.17. The relationship between the ratio of humic acid-like substances to tryptophan-like substances (a) hydrolase activities, or (b) oxidase activities; the relationship between nutrient concentrations and (c) hydrolase activities, or (d) oxidase activities in both habitat types in June, August, October and December 2016 (n=24 x 4);

r represents the Pearson’s correlation coefficient, and ρ represents the Spearman’s correlation coefficient………...………...75

List of Tables

Table 2.1. Processes of catalysis from five soil microbial extracellular enzymes...…...15 Table 2.2. The formulas and the interpretation of fluorescent indices for soil dissolved organic matter properties…………...………...……...18 Table 2.3. The identified fluorophore peak positions in three-dimensional excitation- emission matrix of dissolved organic matter.……….……...…..20 Table 3.1. The characteristics of three study sites.……….……….24 Table 3.2. Instrument settings for excitation-emission matrix and fluorescent

indices………...33 Table 3.3. The substrates and incubation times in the soil microbial extracellular

enzyme activities assays.………...………..37 Table 4.1. The aboveground habitat characteristics among three parks.………41 Table 4.2. The abundance and taxa of total captured ground arthropods in three

parks………..………...………43 Table 4.3. The characteristics of ground arthropod communities (mean ± SD) in three parks in 2016…...…..………..……44 Table 4.4. The soil physicochemical properties (mean ± SD) in three urban parks…....47 Table 4.5. The soil microbial extracellular enzyme activities (mean ± SD) in three parks………..……….……….…………....…… 49 Table 4.6. The aboveground characteristics (mean ± SD) in two habitat types………. 51 Table 4.7. Summary of two-way repeated measures ANOVA for ground arthropod communities between two habitat types and seasons………...…………...……53 Table 4.8. Summary of two-way repeated measures ANOVA for soil physicochemical properties between two habitat types and months...………57

Table 4.9. Summary of two-way repeated measures ANOVA for the properties of grass height, soil water content, and soil nutrient levels between two habitat types and

months………...………..………...………… 59 Table 4.10. Summary of two-way repeated measures ANOVA for the characteristics of soil dissolved organic matter between two habitat types and months….………65 Table 4.11. Summary of two-way repeated measures ANOVA for two spectral indices between two habitat types and months………....66 Table 4.12. Summary of two-way repeated measures ANOVA for the values of three fluorescent indices between two habitat types and months………...…...68 Table 4.13. Summary of two-way repeated measures ANOVA for soil microbial

extracellular oxidase activities between two habitat types and months. …………...….70 Table 4.14. Summary of two-way repeated measures ANOVA for soil microbial

extracellular hydrolase activities between habitat types and months…………...…...…71

1.

Introduction

The impact of human activities on biogeochemical cycle is growing with urbanization increasing (Seto et al., 2012). The soil in green infrastructures (GI), especially urban park, can help mitigate the impact of urbanization on biogeochemical cycle as the basic component of ecosystem services and the largest carbon stock for urban planning (Chiesura, 2004; Pataki et al., 2011; Scharlemann et al., 2014).

The major carbon stocks in urban parks include biomass of different plant covers (Enloe et al., 2015; Strohbach and Haase, 2012) and soil organic matter (SOM) contents under different types of land cover (Bae and Ryu, 2015; Edmondson et al., 2014a;

Livesley et al., 2015). Soil carbon stock is influenced by plant-soil interactions on soil food web composition and soil ecosystem processes (Wardle et al., 2004). Recalcitrant litter from slow-growing plants (e.g., woody plants) with fungal food web could have slow ecosystem processes of decomposition and nutrient mineralization having high carbon sequestration as SOM (De Deyn et al., 2008; Vauramo and Setälä, 2010). However, previous studies in GIs showed that fast-growing clipped grass cover could have higher SOM content than urban tree cover (Huyler et al., 2014; Pouyat et al., 2006) or than native plant cover (Pouyat et al., 2008). These contrast results might be caused by the influences of organic matter input from litter removal and clipping, but they did not further explain the relationships between organic matter inputs and ecosystem processes. These indicate that the comprehensive understanding of above- and belowground interactions on soil ecosystem functioning in urban parks is still unclear.

Compared to plant litters, soil dissolved organic matter (SDOM) from plant and animal is the most mobile and active fraction of SOM (Marschner and Kalbitz, 2003;

Osler and Sommerkorn, 2007). The decomposition and polymerization of SDOM by microbial extracellular enzymes regulate the decomposition and nutrient mineralization

in soil ecosystem processes (Kalbitz et al., 2003; Marschner and Kalbitz, 2003;

Sinsabaugh et al., 2008). Thus, SDOM is an indicator reflecting the effect of plant types and soil managements on soil organic resources input and ecosystem processes (Aitkenhead-Peterson et al., 2009; Chantigny, 2003; Kalbitz et al., 2007). Therefore, SDOM could help understand the relationships between above-belowground properties, and the relationships between organic matter inputs and ecosystem processes in urban parks.

This study aimed to examine the role of SDOM properties in the interactions between above- and belowground properties, and the relationships between organic resources and ecosystem processes under slow-growing and fast-growing plant covers in three subtropical urban parks. The objectives of this study are as follow:

(1) the differences in above-belowground properties and SDOM properties among three urban parks in Taipei City

(2) the differences in above-belowground properties and SDOM properties between two common habitat types with slow-growing and fast-growing plants in urban parks, namely tree with grass (TG) and grass (G)

(3) the relationships between SDOM properties and soil ecosystem processes.

2.

Literature review

2.1 Relationships between plant, soil, and soil food web

An above-belowground system comprises producer subsystem and decomposer subsystem linked by plants. In a producer subsystem, plants produce litters from shoots and fine roots, and root exudates in the soil. Plant litters and root exudates are decomposed into particular organic matters or dissolved organic matters (DOMs) by soil fauna in decomposer subsystem. Then, a portion of DOM would be oxidized and/or hydrolyzed into inorganic forms under microbial extracellular enzymatic catalysis. These processes include the mineralization and nitrification for N-nutrients, and organisms’ respiration for energy (Figure 2.1). The remain organic matter from litter, microbially processed (e.g., decomposed or polymerized) organic matter and dead microbial biomass, would be stabilized in soil as passive SOM, or leach to other sites as SDOM (Cotrufo et al., 2015;

Osler and Sommerkorn, 2007; Sollins et al., 1996).

Figure 2.1. The pathways and interactions between organic matter resources and soil food web with ecosystem processes during organic matter decomposing and nutrient cycling Source: adapted from Osler and Sommerkorn, 2007

The flux of SOM is regulated by the properties of organic resources and the activities of soil food web. Formation and activities of bacterial or fungal food web under different types of organic resources are based on the difference in the carbon and nitrogen use efficiency between bacteria and fungi (Six et al., 2006), and the bacterial or fungal feeding groups in soil (Lavelle et al., 2006) (Figure 2.2). Element use efficiency is the ratio of element invested in producing microbial biomass over total element uptakes such as producing microbial extracellular enzymes, available resources uptake and increased microbial biomass (Sinsabaugh et al., 2013). In general, the carbon use efficiency of fungi (C: N ratio 10 in average and 8 in grass leaf) is higher than that of bacteria (C: N ratio 4 in average and 6 in grass leaf) (De Deyn et al., 2008; Mouginot et al., 2014).

Moreover, microbial extracellular enzyme activities (EEAs) for resources acquisition in soil is influenced by the composition of soil microbial community and the top-down effects from grazers (e.g., nematodes, springtails, and oribatid mites) and predators (e.g., spiders, carabid beetles, and rove beetles) in ground arthropod community.

In addition, detritivores (e.g., isopods and dung beetles) and ecosystem engineers (e.g., termites and ants) in ground arthropod community can accelerate ecosystem processes by converting litters into available resources (e.g., feces), or bioturbation in soils (Lavelle et al., 2006; Osler and Sommerkorn, 2007).

Figure 2.2. The composition, ecological role and interactions in currently known soil food web

Source: adapted from Wardle, 2002

Theoretically, aboveground litters produced by fast-growing plants (i.e., herbaceous plants), which allocate organic resources mainly in growth and produce labile litters, could form bacterial food web with fast decomposition rate and low carbon sequestration in fertile nature ecosystem. Labile litter means the high content of nitrogen and low content of structural carbohydrate in litters. Recalcitrant litter from slow-growing plants (i.e., woody plants) could form fungal food web with slow decomposition rate and high carbon sequestration by litter layer. Then the litter layer provides more organic resources and habitat for ground arthropod than herbaceous plants (De Deyn et al., 2008; Wardle et al., 2004).

Compared to aboveground litter, the properties of belowground root litters and exudates could differ from leaf litters. Craine et al. (2005) and Tjoelker et al. (2005) provide evidence that the nitrogen concertation in root litter is lower than that in leaf litter from herbaceous plants. In an in situ degradation experiment, Mambelli et al. (2011) found that the microbially altered fine root litter had a higher contribution to the passive

fraction of SOM than leaf litter did. These indicate that the root litters could be more recalcitrant organic resources for a microbial community than leaf litters, and could have a potential contribution to the passive fraction of SOM. On the other hand, the effects of root exudates containing labile resources on stimulating soil ecosystem processes are still controversial (Kuzyakov, 2010).

In brief, belowground soil physicochemical properties and rates of ecosystem processes could be predicted by aboveground plant cover types in natural ecosystems.

Soil under slow-growing tree cover may have higher SOM content, and lower nutrient content than that under fast-growing grass cover. The theoretical relationships between above- and belowground properties under these two types of plant covers are based on natural soils which have been through long-term soil evolution. However, the plant-soil interactions in urban ecosystems are usually altered during the construction of infrastructures and plant management, and these interactions remain to be investigated.

2.2 Anthropogenic effects on above-belowground interactions in GIs 2.2.1 Land use/cover change and history

Under urban planning, land use changes alter the structures of plants and soils in urban ecosystems, and further change the soil food web and ecosystem processes.

Topsoils in urban ecosystems usually consist of artifacts (e.g., bricks or trash) from former land uses, or fill soils from other sites with young ages during the constructions of infrastructures in urban areas. These characteristics of urban soils generally contribute to alkaline soil pH and high soil bulk density (Lehmann and Stahr, 2007). The destruction of soil structure during infrastructure construction, or the derived high soil pH and bulk density after infrastructure construction would cause impacts on soil biota. In fact, this provides an opportunity to understand how belowground soil food web response to disturbances, and how we could manage the plant-soil interactions to restore destructed soil (Heneghan et al., 2008; Pavao-Zuckerman, 2008) or promote ecosystem services (Gómez-Baggethun and Barton, 2013; Kremen, 2005) in urban ecosystems.

After infrastructure constructions, the belowground soil food web composition and ecosystem processes could be affected by different land use types. Pouyat et al. (2002) observed that the surface soil organic carbon (SOC) contents (at 15 cm depth) under the forest land use were lower than the ones under low-density residential land use in Baltimore city. However, in a more recent study about the SOM contents of different land cover types in Baltimore city, there were no significant differences between clipped grass (59 ± 3.7 g kg^-1 SOM in parks or golf courses, and 50 ± 1.7 g kg^-1 SOM in residential areas), unmanaged forest (51 ± 4.6 g kg^-1 SOM), and high impenetrable surface areas (~54 g kg^-1 SOM) (Pouyat et al., 2007). These indicate that land covers could be the dominant driver on ground arthropods and ecosystem processes in GIs.

Compositions of decomposer subsystem and rates of ecosystem processes could be driven by the vegetation planting in GIs. Through an experimental study, Vauramo and Setälä (2010) indicated that the surface soil food web and ecosystem processes in two different land use history sites were driven to bacterial- and fungal-energy channel corresponding to the planting of grass and tree (both in 1 m² experimental sites) within three years. In addition, this phenomenon was observed in 41 urban parks with different ages under cold climate (Hui et al., 2017; Setälä et al., 2016). The higher contents of surface SOM under tree covers than in grass covers were also observed in other urban ecosystems (Bae and Ryu, 2015; Edmondson et al., 2014a). Furthermore, plant type effects on soil organic content and nutrient content could be enhanced by park ages (Pouyat et al., 2008; Setälä et al., 2016).

Local and landscape factors would influence the composition and structure of ground arthropod community. Local habitat features (within 100 m x 100 m), such as litter layer, plant configuration and impervious surface, dominantly explain the abundance and richness of ground arthropods in different land use types (McIntyre et al., 2001; Philpott et al., 2014; Riedel et al., 2009; Sattler et al., 2010). In addition, park age could have positive effects on the diversity of ground arthropods by the increase in plant composition diversity, or longer time for species dispersal (McIntyre, 2000; Sattler et al., 2010). Landscape factor of the proportion of urban land use cover in surrounding landscape context could hinder the dispersal of less mobile taxa such as ground-dwelling predator with estimated movement range for 500 m to 1 km (Egerer et al., 2017).

2.2.2 Plant managements

Types and frequencies of plant managements in GIs could alter the above- belowground interactions by influencing the ground arthropod compositions and organic resource properties. The similar or higher SOM contents in clipped grass covers than in tree covers have been observed in temperate urban areas (Edmondson et al., 2014b), and in subtropical urban areas (Pouyat et al., 2008, 2007; Weissert et al., 2016). These studies proposed that plant managements might contribute to the discrepancy of SOM contents under these two types of plant covers such as mulch clipping or litter removal, fertilization, and irrigation.

Clipping could affect soil food web composition and ecosystem processes by the changes in organic resources allocation in plants. Bardgett et al. (1998) suggested two broad pathways of foliar herbivory on forming bacterial food web and fast decomposition rate. First, the increase in root exudates and root carbon allocation for short-term and long-term. Second, the increased nitrogen content in root through the investment of secondary metabolites in leaves. These two pathways would positively affect sizes and activities of soil microbial communities. Compared to the effects of foliar herbivory in pasture grass covers, Lilly et al. (2015) observed that the root biomass was slightly increased with increased clipping intensities and irrigation in turfgrass. Byrne et al. (2008) also observed the clipped grass cover had higher soil microbial mass and respiration in the soil than unclipped grass cover did, as well as the nitrogen concentrations (Byrne and Bruns, 2004). Yao et al. (2009) found the positive effects of clipping addition and N- fertilization on soil microbial extracellular hydrolase activities under the clipped grass cover, but no effects on soil microbial extracellular oxidase activities in the surface soils under managed golf course.

Furthermore, grass clipping and pesticide addition could change the soil food web composition and ecosystem processes by influencing the abundance and structure of ground arthropod community. Increased clipping intensity could cause a decrease in the abundance of large size detritivores (Byrne and Bruns, 2004), or low mobility ground- dwelling predators which are at a higher trophic level in the soil food web (Birkhofer et al., 2017; Sattler et al., 2010). On the other hand, pesticide addition could have a short- term negative impact on the abundance and activity of predators (e.g., spiders and rove beetles) approximately three weeks (Larson et al., 2012), which may contribute to a temporally changes in the ecosystem processes.

2.2.3 Recreational activity

The recreational activities in urban parks could change the soil properties and organic resources by stepping and N-richer organic resources input. With increasing stepping in urban parks, soil bulk density would increase and cause a decrease in grass coverage and soil water content (Sarah and Zhevelev, 2007; Sarah et al., 2015). The decrease in grass coverage and soil water content would have a negative impact on ground arthropods and soil microbial activity. Moreover, N-richer organic resources such as food residues from eating meals and feeding wildlife, or dungs from pets and mammal wildlife may provide additional organic resources to dung detritivores (e.g., dung beetles) and microbial community (Carpaneto et al., 2005).

2.2.4 Summary

Previous observations from experiments and field surveys found the theoretical above-belowground interactions could happen in urban ecosystems even the soils have different land use histories. However, plant managements and recreational activities in

the real world may alter the theoretical above-belowground interactions on organic resources input and soil food web composition. The relationships between organic resources input and soil ecosystem processes under plant managements and recreational activities are rarely investigated. As the most mobile and active fraction of SOM, SDOM could reflect the effect of plant types and human activities (planning, management, and recreational activity) on organic resources input and ecosystem processes, and help us understand the above-belowground interactions on soil ecosystem functioning in urban parks (Figure 2.3).

Figure 2.3. The conceptual model of soil dissolved organic matter to link the above- and belowground properties

Source: adapted from Byrne, 2007

2.3 Soil dissolved organic matter (SDOM) and above-belowground interactions 2.3.1 SDOM, and its sources and properties

Dissolved organic matter (DOM) is defined as the soluble fraction of organic matters passing through a 0.45 μm filter practically (Zsolnay, 2003). Based on the carbon is the most abundant element of organic matter, DOM is usually quantified as dissolved organic carbon (DOC) (Bolan et al., 2011). DOM is recognized as the most mobile and active fraction of organic matter (Kalbitz et al., 2000). DOM in soils comes from photosynthates in shoot system (e.g., leaf litter, throughfall, and stemflow) and in root system (e.g., root exudates and decaying fine root), as well as microbially processed organic matters (e.g., metabolic by-products during decomposition and leachates of microbially processed SOM). Indeed, SDOM would majorly accumulate in topsoil, and a small portion of SDOM would store in subsoil or leach to an aquifer (Bolan et al., 2011).

The major source of SDOM in clipped grass covers is plant residues, and in forest tree covers is litter and litterfall (Ghani et al., 2007; Laik et al., 2009). Chantigny (2003) suggested that SDOM concentrations in different land covers could decrease in order:

coniferous forest, deciduous forest, and managed herbaceous plant. This decrease represents the contribution of aboveground litters to SDOM concertation. However, belowground root organic resources also play an important role to affect the distribution and composition of SDOM. Uselman et al. (2007) suggested that the SDOM in the rhizosphere, associated with a high turnover rate of fine root and microbial biomass, could contribute to deep soil horizon than leaf litter. Moreover, the distribution and composition of SDOM are susceptible to human managements such as biological wastes input and cultivation in agricultural lands (Bolan et al., 2011; Provin et al., 2008).

Based on chemical compositions, SDOM can be grouped into labile organic matters and recalcitrant organic matters. Marschner and Kalbitz (2003) suggested that the labile or recalcitrant organic matters could be classified by the aromaticity, molecular weight, and polarity of DOM. The labile DOM includes simple carbohydrate compounds (i.e., glucose and fructose) and low molecular weight organic matter (i.e., organic acids, amino acids, and protein) for which the hydrophilic neutral fraction. In contrast, the recalcitrant DOM consists of aromatic, high molecular weight and hydrophobic components such as polysaccharides (i.e., cellulose and hemicellulose), other plant compounds (e.g., phenolic acids and carboxylic acids), and microbially processed products.

Properties of terrestrial DOM are mostly investigated in temperate forest ecosystems and agricultural ecosystems. In forest ecosystems, DOMs from tree leaf litters usually consist of low-molecular-weight hydrophilic neutral and weak hydrophobic (i.e., phenolic) components; leachates of SOM are usually composed of recalcitrant compound such as organo-metal complexes (Kiikkilä et al., 2013; Uselman et al., 2012; Yano et al., 2005).

DOMs from fine roots with more phenolic components are less labile than the ones from leaf litter (Uselman et al., 2007; Yano et al., 2005). In contrast, DOMs from agricultural soils could contain a higher proportion of labile DOM than the leachates from forest floors (Delprat et al., 1997). However, the proportion of recalcitrant components, such as fulvic acids in SDOM, could increase with intensive human management (Kalbitz, 2001;

Leinweber et al., 2001). For example, the higher aromatic components in the SOM or the SDOM under clipped grass covers than the ones under tree covers in urban ecosystems were observed by Beyer et al. (1995) and Cioce and Aitkenhead-Peterson (2015).

2.3.2 Properties of SDOM and microbial extracellular enzyme activities (EEAs) Organic matters in terrestrial ecosystems are primarily decomposed and mineralized by microbial communities to provide energy and nutrients for metabolism and growth.

Environmental conditions and organic matter resources influence the microbial community composition and resources requirement (e.g., carbon and nitrogen). To meet the community-level resources demand, a microbial community would produce and excrete enzymes to the environment for degrading complex organic compounds into assimilable subunits such as DOM and ammonium. These microbial EEAs are free in soil water and associated with organic matters (i.e., a specific substrate, SOMs, and condensed tannins), or clay minerals (Burns et al., 2013). Moreover, types and activities of these soil microbial extracellular enzymes can be an indicator to investigate the microbial community composition and their response to organic resources (Caldwell, 2005; Skujiņš and Burns, 1976).

The flux and ultimate fate of SDOM are influenced by the relationships between plant organic matter input and soil microbial EEAs. To carbon resources, polysaccharides in SDOM such as cellulose and chitin would be hydrolyzed under the catalysis of β- glucosidase (BG) and β-N-acetylglucosaminidase (NAG); recalcitrant lignin and phenolics would be oxidized under the catalysis of phenol oxidase (PO) and peroxidase (PER) (Table 2.1). To nitrogen resources, peptides would be hydrolyzed under the catalysis of leucine aminopeptidase (LAP) (Table 2.1). Sinsabaugh et al. (2008) investigated the relationships between two groups of soil microbial EEAs (i.e., hydrolases and oxidase) and SOM contents at the global scale. This study suggested that the activities of hydrolases (i.e., BG, NAG, and LAP) were positively correlated with the SOM contents. The activities of oxidases (i.e., PO and PER) had no significant relationships with SOM contents but were strongly related to soil pH. Furthermore, the ratio of

ln(BG)/ln(NAG+LAP) can be considered as an indicator for microbial community acquisition activity for C: N resources preference.

Table 2.1. Processes of catalysis from five soil microbial extracellular enzymes

Enzyme EC number Process of catalysis Element

Peroxidase (PER)¹ EC 1.11.1

Oxidization of phenolic compound by using H2O2 as the terminal electron

acceptor (a group of EEAs in practical assays)

C

Phenol oxidase (PO)¹ EC 1.10.3

Oxidization of phenolic compound by using O2 as the terminal electron acceptor (a group of EEAs in practical assays)

C

β-glucosidase (BG)² EC 3.2.1.21 Hydrolysis of short-chain

cellulose bonds into glucose C β-N-

acetylglucosaminidase (NAG)³

EC 3.2.1.30

Hydrolysis of chitobiose or chitooligosaccharides with the production of N- acetylglucosamine

C, N

Leucine aminopeptidase

(LAP)⁴ EC 3.4.11.1 Hydrolysis of polypeptide chains into amino acids N Source: ¹Sinsabaugh, 2010; ²Eivazi and Tabatabai, 1990; ³Parham and Deng, 2000;

4Saiya-Cork et al., 2002

The quantity of SDOM can induce soil microbial activities by increasing the quantity of enzyme substrate (Jones et al., 2008). On the other hand, the quality of DOM shows to have different relationships between hydrolase and oxidase activities. Relationships between microbial activities and properties of DOM usually are investigated in manipulated experimental studies. Addition of N-resources is suggested to induce the hydrolase activities but inhibit the oxidase activities, which might cause the increased

proportion of recalcitrant components in SDOM in natural forest, grassland and golf course (Ajwa et al., 1999; Cleveland et al., 2013; Huang et al., 2016; Yao et al., 2009).

However, Jiang et al. (2013) found that the phenol oxidase activity was induced and hydrolases activities were inhibited by N-fertilization addition, which caused the release of recalcitrant DOM. Tian et al. (2010) observed that the peroxidase activity was negatively related to the proportion of labile DOM, but positively related to the N- mineralization under different land covers in an agricultural land.

In brief, previous studies indicate that there could be dual pathways in the relationships between microbial EEAs and properties of SDOM, especially the formation of recalcitrant components. Based on these studies, the conceptual model of pathways between properties of SDOM and microbial extracellular enzymes in this study is shown in Figure 2.4.

Figure 2.4. The pathways between labile/recalcitrant soil dissolved organic matters, the catalysis of microbial extracellular oxidases and hydrolases, passive soil organic content, and nutrient concentrations

2.3.3 Spectroscopic methods for SDOM properties

Ultraviolet-visible spectroscopy and fluorescence spectroscopy have been applied to study the properties of DOMs. The absorption by the functional groups of DOM at different wavelengths can be used as an indicator for relative aromaticity and molecular weight. Weishaar et al. (2003) suggested an indicator to estimate the aromaticity of DOM by the absorbance at 254 nm normalized to the DOC concentration (specific ultraviolet absorbance, SUVA254) of the DOM, which is based on the principle of aromatic structure would absorb light at short wavelengths. Therefore, a larger value of SUVA254 reflects a larger level of aromaticity of DOM. To the molecular weight of DOM, Helms et al. (2008) proposed a good proxy by the spectral slope of log-transformed absorption within 275- 295 nm (S275-295). A higher value of S275-295 represents a smaller molecular weight of DOM.

Potential sources and ages of SDOM can be characterized with derived fluorescent indices (Table 2.2). Recalcitrant compounds in the DOM from plant and SOM usually contain a higher level of aromatic matters (e.g., lignin, phenolic, and humic substance) than the ones in the microbially derived DOM. Aromatic compounds would have higher long wavelength emitting fluorescence than simple structural substances does. With this characteristic, potential sources of SDOM from soil or biological activities can be characterized by fluorescence index (FI) and biological index (BIX). FI is calculated by the ratio of short (450 nm) to long wavelength (500 nm) emitting fluorescence and BIX is calculated by the ratio of 380 nm to 430 nm emitting fluorescence. Consequently, the value of FI or BIX of SDOM from passive SOM would be lower than the one from biological activities.

Humification index (HIX) is determined from the ratio of integrated emitting fluorescence in the long wavelength region (435-480 nm) to the short wavelength region (300-345 nm). This index is based on the theoretical procedure of humification in which

the molecules decrease in H/C ratio and result in a shift of emission to longer wavelengths.

Therefore, a value of HIX generally increases with the degree of decomposition. These fluorescent indices have been used as representatives of SDOM properties from different land covers (Fellman et al., 2008; Kalbitz, 2001) and aquatic ecosystems (Birdwell and Engel, 2010).

Table 2.2. The formulas and the interpretation of fluorescent indices for soil dissolved organic matter properties

Index λex λem Formula Potential source and age interpretation from index value

FI¹ 370 nm 450 nm, 500 nm

𝐼_{𝜆𝑒𝑚450} 𝐼_{λ𝑒𝑚500}

Fluorescence index

≤ 1.4 allochthonous from terrestrial source

≤ 1.9 autochthonous from microbial source

BIX² 310 nm 380 nm, 430 nm

𝐼_λem380 𝐼_λem430

Biological index

0.8-1.0 fresh and autochthonous from microbial source

≤ 0.6 low autochthonous from terrestrial source

HIX³ 254 nm 300-345 nm, 435-480 nm

∫_λ^λem480𝐼_λem

em435

∫_λ^λem345𝐼_λem

em300

Humification index

< 5 autochthonous derived from plants or animals

10-30 allochthonous from soil Source: ¹McKnight et al., 2001; ²Huguet et al., 2009; ³Zsolnay et al., 1999

On the other hand, three-dimensional excitation-emission fluorescence spectroscopy is commonly applied to characterize the components in DOM. Three-dimensional fluorescence excitation-emission matrix (EEM) is a result of merging a series of emission scans from excitation over a range of wavelengths. EEM contains information of

abundance, numbers, and types of fluorophores (Fig 2.4) (Birdwell and Engel, 2010;

Chen et al., 2003; Coble, 1996). By peak identification, DOM can be identified as recalcitrant components (e.g., humic-like and fulvic-like compounds) and labile components (e.g., tryptophan-like and tyrosine-like substances) (Table 2.3). However, with visually peak identifying, much information in overlapped regions between identified peaks in EEM would be ignored.

Figure 2.5. The identified fluorophore peak positions from natural water in a three- dimensional excitation-emission matrix (peak A and C: humic acid-like ； peak M;

precursor of peak C; peak B: tyrosine-like; peak C: tryptophan-like; FI: fluorescence index; BIX: biological index; HIX: humification index)

Source: Birdwell and Engel, 2010

Table 2.3. The identified fluorophore peak positions in three-dimensional excitation- emission matrix of dissolved organic matter

Peak λex (max) λem (max) Properties

A 260 nm 380-460 nm Humic-like

C 350 nm 420-480 nm Humic-like

M 312 nm 380-420 nm Precursor of peak C

B 275 nm 310 nm Tyrosine-like

T 275 nm 340 nm Tryptophan-like

Source: Birdwell and Engel, 2010; Coble, 1996

Recently, three-dimensional excitation-emission matrices can be decomposed by multi-way parallel factor analysis (PARAFAC analysis). The advantage of PARAFAC analysis is that each component of PARAFAC model has specific ranges of excitation wavelength, and still follows the Beer’s law. The concept of PARAFAC analysis for excitation-emission matrices is that the data signal can be decomposed into a set of trilinear terms and a residual array (Murphy et al., 2013; Stedmon and Bro, 2008) as shown in the following equation (1):

𝑋_𝑖𝑗𝑘 = ∑^𝐹_𝑓=1𝑎_𝑖𝑓𝑏_𝑗𝑓𝑐_𝑘𝑓+ 𝑒_𝑖𝑗𝑘 (1) where i = 1, …, I; j = 1, …, J; k = 1, …, K; f = 1, …,F; i, j and k represent the i^th sample, j^th emission and k^th excitation wavelength; f corresponds to the modeled component that has fluorescence intensity scores (a-values) of samples, emission wavelength loading (b- value), and excitation wavelength loading (c-value); eijk corresponds to a residual array and Xijk corresponds to modeled sample.

Three-dimensional fluorescence spectroscopy with PARAFAC analysis has been applied to trace changes in compositions of DOM during decomposition (Hunt and Ohno, 2007), and detect effects of land covers and managements on changes in compositions of

DOM (Fellman et al., 2008; Hernandez-Soriano et al., 2015). Therefore, this technique with PARAFAC analysis could a powerful tool to characterize components in SDOMs and trace their sources for investigating their role in the above-belowground interactions in urban parks.

2.3.4 Summary

The SDOM under tree covers may have a higher proportion of recalcitrant components than that under managed grass covers. However, the relationships between SDOM properties and microbial EEAs in GIs without manipulations or experiments are rarely investigated. Spectroscopic methods with PARAFAC analysis for SDOM properties could be a useful tool to characterize labile and recalcitrant components in SDOM and trace their sources. This could not only help reveal the role of labile and recalcitrant components of SDOM in above-belowground interactions but also in the relationships between organic resources properties and ecosystem processes under the effects of plant types, management, and recreational activities in urban parks.

3.

Materials and methods

The above- and belowground properties in subtropical urban parks were investigated by field samplings and surveys, as well as laboratory analyses from December 2015 to January 2017. The study designs and procedures in this study are summarized in Figure 3.1.

Figure 3.1. The flow diagram of research process in this study

3.1 Study sites

The urban parks in subtropical Taipei City (25° 02' 52" N, 121° 31' 55" E;

approximately 2.7 million people) (Figure 3.2) were selected by the following criteria: (i) the park area is larger than 5 hectares to provide enough sampling sites covered with tree or grass; (ii) the distance from the park border to mountain forests or river sides is larger than 1 km to minimize the soil biota from different sources (Egerer et al., 2017; Small et al., 2006); (iii) the depth of park soil is deeper than 10 cm. Three Taipei urban parks that met the criteria were chosen as studied sites, namely Daan forest park (DFP), Chiang Kai- Shek memorial hall (CKS), and 228 peace memorial park (PMP).

Figure 3.2. (a) The location of Taipei City in Taiwan, (b) the position of study sites in Taipei City and their surrounding land use types; DFP: Daan forest park, CKS: Chiang Kai-Shek memorial hall, PMP: 228 peace memorial park (the land use map is from National Land Surveying and Mapping Center, 2^nd land use investigation in 2006)

Three chosen parks have different land use histories (Table 3.1). Instead of in situ pedogenesis, the topped park-soils were from the filling soils or garden soils during park construction and plant planting. Although with different park histories and sources of soils,

they all have similar plant covers (i.e., evergreen tree and grass) and management types in park green spaces.

Table 3.1. The characteristics of three study sites Daan Forest Park

(DFP)^1,2

Chiang Kai-Shek Memorial Hall (CKS)³

228 Peace Memorial Park (PMP)^1,2 Background

Area (m²) 259,293 251,500 76,180

Purpose urban forest, disaster prevention

memorial, artistic and cultural activities

memorial, disaster prevention

Former land use residential land residential land

(for troop use) park land Year of

establishment

start building in 1992; open in 1994

start building in 1976; open in

1980 open in 1889

Green cover accessibility in park

Concrete curbing No Yes; part of area

with hedge Yes; part of area with hedge Management*

Irrigation (tree) during summer or drought

Fertilizer addition^$ No^# No No

Pesticide addition^$ No No No

Litter/clipping

removal Yes Yes Yes

Grass mowing every 20 days in summer or 30 days in winter

* The information on urban green space management was obtained from interviewing the officer of Parks and Street Lights Office, Taipei City Government in March 2016. ^$The addition of fertilizer and pesticide were not carried out in the sampling sites but might happen in other areas. ^# The trees in the north area of Daan forest park were fertilized during the soil improvement project during October 19 to December 7, 2015.

Source: ¹Parks and Street Lights Office, Taipei City Government, 2016a; ²Parks and Street Lights Office, Taipei City Government, 2016b; ³Chiang Kai-Shek Memorial Hall, 2017.

As for the climatic conditions for these parks, the mean annual temperature recorded by nearest Taipei weather station (121°30' 24〞E, 25°02' 23〞N) from 2006 to 2015 ranges from 22.7 to 23.8℃. The monthly temperature ranges from 16.4℃ (January) to 30.1℃

(July) (Figure 3.3). The mean total annual precipitation ranges from 1669.2 to 3015.9 mm, which is mainly affected by frontal rain (in winter and spring) and typhoon events (in summer and fall).

Figure 3.3. The mean monthly temperature and monthly total precipitation of nearest Taipei weather station from 2006 to 2015 (data from Taiwan Central Weather Bureau:

http://www.cwb.gov.tw/V7/climate/watch/watch.htm)

The sampling sites for two habitat types, tree with grass (tree height larger than 4 m and ground plant) and grass (only ground plant), were four 100 m² (10 m x 10 m or 5 m x 20 m) rectangular plots per habitat type (Byrne et al., 2008) in each park. The positions of sampling sites were determined by plant cover map (Appendix A) with the following criteria: (i) plant cover area is larger than 100 m²;(ii) bare area is smaller than 1 m² (Philpott et al., 2014); (iii) soil depth is larger than 10 cm.

To examine effects of the habitat heterogeneity at different spatial scales on ground arthropods (Philpott et al., 2014), the area ratios of land uses and land covers of these parks were estimated. The surrounding land use types within 1 km buffer zones of these parks were classified into six types (impervious surface, agriculture, greenspace¹, grass, forest, other) by using the land use investigation in 2006 (acquired from National Land Surveying and Mapping Center in Taiwan). The land cover types of parks were also digitized to four types (water body, impervious surface² , grass, and tree) with digital colored orthophoto map in July 2016 (acquired from Aerial Survey Office in Taiwan).

Then the area ratios of the land use types in 1 km buffer zone, and land cover types of each park were calculated as the representatives of habitat heterogeneity by Esri ArcGIS 10.0.

3.2 Soil sampling and storage

The surface soils were sampled to a depth of 10 cm bimonthly from February to December 2016 by using a 3.8 cm diameter stainless steel corer. Four cores were sampled by selecting four quadrats from 1 m² grid system with tagging and selecting random numbers³. One core sample was used to analyze soil water content, bulk density, and organic matter content. The other three core samples were mixed as one homogenous composite sample and sieved to 2 mm after overnight air-drying for the analyses of SDOM properties, chemical properties, and soil EEAs. All samples were stored at 4℃

before further analyses. To record the potential disturbance of clipping events, the grass heights at each sampling site were measured in each field sampling.

1 Greenspaces in this study include urban parks, street verges, and other open spaces.

2 The tree well which is smaller than 1 m², or the concrete raised planter in the parks were classified as impervious surface.

3 The tagging and selection of random numbers were performed by choosing numbers from a vector (1 to

3.3 Ground arthropod sampling

In February, April, August, November and December 2016, the ground arthropods were collected by pitfall traps left open for 72 h (McIntyre, 2000). The sampling times reflect their corresponding seasons, i.e., winter, spring, summer, fall, and winter. In each rectangular sampling site, five pitfall traps with 20 mL 75% ethanol in 50 mL plastic centrifuge tubes (3 cm diameter) were placed at the center and on the diagonal with a 3- m (or 4-m in 5 m x 20 m sampling site) distance from the center (modified from (Larson et al., 2012). After left open 72 h, five pitfall traps in each sampling sites were combined into one sample and stored in 75% Ethanol for taxonomic classification. The ground arthropod samples were then identified to taxonomic groups (orders and families) or feeding groups (Mueller et al., 2016; Scheu, 2002; Vauramo and Setälä, 2010). For the comparison of ground arthropod compositions between parks, ground arthropods at each sampling sites were then combined as one park sample.

3.4 Plant survey

The plants in our sampling sites were investigated in December 2015, July 2016, and January 2017. Each survey investigated the tree and the ground plants. For the trees in the sampling sites, tree species, tree number, diameter at breast height (DBH; at 1.3 m above grounds), and tree height were recorded. For the ground plants in all sampling sites, coverage and taxon number were observed by quadrat sampling with a 50 cm x 50 cm quadrat (dividing into 25 sections) and photography. In each sampling site, five 1 m x 1 m grids were randomly chosen for taking a photo of ground plants with the square sampler place at the center of each plot.

3.5 Recreational activity survey

The recreational activities surveys were conducted to estimate stepping effect of active agents (people or pets) by point observation in March, July and November 2016.

One week-day and one weekend day were included in each survey. The number and behaviors of active agents inside a sampling plot were recorded in 10 minutes during morning (08:00-10:00), afternoon (13:00-15:00) and evening (17:00-19:00) in each day modified from Sarah et al. (2015). The numbers of active agents during three time periods were averaged for daily estimation. Then the daily estimations for a weekday and for a weekend day were averaged with the weight of occurring frequencies (five vs. two), and calculated into the unit in an hour as an index of averaged stepping intensity.

3.6 Soil physical properties analyses

The analyses of soil physical properties of soil water content, bulk density, and SOM content were determined with one unprocessed core sample. All physical analyses were completed within two weeks after sampling. Soil water content (gravimetric method) and bulk density (core method; Blake, 1986) were measured by the soil dry mass (oven drying for 72 h at 105℃) and volumes of core and other materials (diameter > 2 mm):

soil water content (%) =^{𝑤w−𝑤d}

𝑤d × 100%

bulk density (g cm⁻³) = ^{𝑤d−𝑤o}

(𝑣1−𝑣2)

where ww (g) is the wet sample mass, wd (g) is the dry sample mass, wo (g) is the other materials mass, v1 (cm³) is the volume of soil core (1.9²𝜋 × 10 cm ≅ 113.41 cm³), and v2 (cm³) is the volume of other materials.

Followed by the bulk density analysis, soil organic content was evaluated by igniting 5 g sieved and dried soil at 360℃ for 2 h (loss-on-ignition method; Salehi et al., 2011) in

the furnace (48000 Furnace, Thermolyne):

SOMLOI (%) = ^{𝑤ds−𝑤i}

𝑤_ds × 100%

where wds (g) is the sieved and oven-dried soil mass, and wi (g) is the soil mass after ignition.

3.7 Soil chemical properties analyses

The measurements of pH and concentration of extractable nitrate and ammonium were performed in duplicates with the sieved composite sample within two weeks after sampling.

3.7.1 pH

Soil pH was measured with 10 g air-dried soil in 10 mL Milli-Q water suspension (1:1 soil: water ratio) by pH meter (CyberScan pH 510, EUTECH instrument).

3.7.2 Nitrate and ammonium extraction

Extractable nitrate (NO3-) and ammonium (NH4+) were extracted by 0.8 g air-dried soil in 8 mL 1 M KCl (1:10 soil: 1 M KCl ratio; Keeney and Nelson, 1982) with shaking at room temperature for 30 minutes. Then the soil suspensions were centrifuged at 900 x g for 30 minutes ( Universal 32 Benchtop non-refrigerated centrifuge, Hettich). The clean supernatants were collected for the colorimetric analyses of nitrate and ammonium.