<%BANNER%>

Effect of Disturbance on Ecosystem Processes

Permanent Link: http://ufdc.ufl.edu/UFE0042450/00001

Material Information

Title: Effect of Disturbance on Ecosystem Processes
Physical Description: 1 online resource (182 p.)
Language: english
Creator: Lee, Seungjun
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: disturbance, ecosystem, er, gpp, idh, maturity, microcosm, model, processes, productivity, pulsing, resilience, resistance, respiration, selforganization, simulation
Environmental Engineering Sciences -- Dissertations, Academic -- UF
Genre: Environmental Engineering Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: One of the ecosystem management goals is to understand how disturbance regimes influence an ecosystem's traits and integrate the regimes into management criteria. This study contributes to the understanding of ecosystem energetics under disturbance by investigating the relationships between disturbance and ecosystem processes in the context of the current ecological paradigms: intermediate disturbance hypothesis (IDH), ecosystem maturity, and pulsing. The responses of gross primary productivity (GPP) and ecosystem respiration rate (ER) to water motion disturbance regimes were tested using open top aquatic microcosms, and the mechanisms of the resultant patterns from the microcosms were suggested by computer simulation models. Each microcosm developed a unique system and maintained a balance between GPP and ER through the self-organizing processes. Each microcosm showed a distinct relationship between disturbance regime and GPP. The mechanisms for the distinct relationships were explained by changes of efficiencies in energy flow pathways of a system under disturbances and by a disturbance threshold above which the efficiencies are permanently altered. The tests of the disturbance effects on GPP and ER under different maturities of microcosms supported the current theory on the ecosystem's development arguing that an ecosystem reinforces internal structures and thus becomes more resistant to disturbances over time. GPP and ER oscillated over time in the microcosms, and the wavelength and amplitude of the GPP and ER pulsing patterns were amplified by disturbance in microcosms' early stages of development, compared with those of an undisturbed system. As originally intended, the combined study of microcosms and simulation models provided new hypotheses on disturbance and ecosystem processes, which need to be tested further using microcosms, simulation models, or in the field.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Seungjun Lee.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Brown, Mark T.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042450:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042450/00001

Material Information

Title: Effect of Disturbance on Ecosystem Processes
Physical Description: 1 online resource (182 p.)
Language: english
Creator: Lee, Seungjun
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: disturbance, ecosystem, er, gpp, idh, maturity, microcosm, model, processes, productivity, pulsing, resilience, resistance, respiration, selforganization, simulation
Environmental Engineering Sciences -- Dissertations, Academic -- UF
Genre: Environmental Engineering Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: One of the ecosystem management goals is to understand how disturbance regimes influence an ecosystem's traits and integrate the regimes into management criteria. This study contributes to the understanding of ecosystem energetics under disturbance by investigating the relationships between disturbance and ecosystem processes in the context of the current ecological paradigms: intermediate disturbance hypothesis (IDH), ecosystem maturity, and pulsing. The responses of gross primary productivity (GPP) and ecosystem respiration rate (ER) to water motion disturbance regimes were tested using open top aquatic microcosms, and the mechanisms of the resultant patterns from the microcosms were suggested by computer simulation models. Each microcosm developed a unique system and maintained a balance between GPP and ER through the self-organizing processes. Each microcosm showed a distinct relationship between disturbance regime and GPP. The mechanisms for the distinct relationships were explained by changes of efficiencies in energy flow pathways of a system under disturbances and by a disturbance threshold above which the efficiencies are permanently altered. The tests of the disturbance effects on GPP and ER under different maturities of microcosms supported the current theory on the ecosystem's development arguing that an ecosystem reinforces internal structures and thus becomes more resistant to disturbances over time. GPP and ER oscillated over time in the microcosms, and the wavelength and amplitude of the GPP and ER pulsing patterns were amplified by disturbance in microcosms' early stages of development, compared with those of an undisturbed system. As originally intended, the combined study of microcosms and simulation models provided new hypotheses on disturbance and ecosystem processes, which need to be tested further using microcosms, simulation models, or in the field.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Seungjun Lee.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Brown, Mark T.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042450:00001


This item has the following downloads:


Full Text

PAGE 1

1 EFFECT OF DISTURBANCE ON ECOSYSTEM PROCESSES By SEUNGJUN LEE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORID A IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Seungjun Lee

PAGE 3

3 To my family

PAGE 4

4 ACKNOWLEDGMENTS I owe my deepest gratitude to Dr. Mark Brow n for his guidance as an advisor during my PhD program. His great passi on and unique philosophy on advisement have driven me to freely explore the intellectual world, develop my own research project, and mature as an independent scholar over the last four and a half years. The completion of this dissertation would not have been possi ble without his intellectual and financial support. I was fortunate to have as one of my committee members Dr. Clay Montague, who was always delighted to provide me with invaluable ideas and suggestions on microcosm experiments. His extensive k nowledge and experience on microcosms were crucial to the development and completion of my research projec t. Dr. Matt Cohen enriched my research project with original questions and suggestions, and encouraged me to acquire diverse theoretical perspectives. Dr. Paul Zwick gui ded me in the right direction until the completion of the pr ogram with fundamental questions and comments regarding my theories and thoughts. My system-oriented thin king has been fostered by teaching and advisement from Drs. Brow n, Montague, Cohen, an d Zwick, who have been the best systems ecologists I have ever met. I have been honored to work with all of my previous and current colleagues of the Systems Ecology group at the University of Florida. Divers e thoughts and discussions they shared with me in the systems seminars, research meetings, and individual talks have been cornerstones for this dissertation. In particular, David Pfahler helped me by reviewing and commenting on the final revision of this dissertation. Most importantly, I would like to thank my family. I thank my parents for their patience and support with uncondi tional love. I would like to address my special thanks to Hyewon for her love, ideas, and emotiona l support as a wife, colleague, and friend

PAGE 5

5 throughout my doctoral study. She shared t he best and worst moments of the journey of the last several years with me, and her never-failing support and encouragement have been essential to the completion of this dissertation. The birth and growth of my son, Shimok, during the last year and a half have given me great joy and inspiration for the self-organization of man and nature. The construction of the microcosm system s was supported in part by Glick and Wilkes scholarships awarded through the department of Envir onmental Engineering Sciences at the University of Florida. I thank Sejin Youn, who is working with Dr. JeanClaude Bonzongo, for gladly providing me wit h purified water for the maintenance of microcosms. Charles H. Meyer, a tutor at the University of Flor ida Reading and Writing Center, helped me with proofreading during the revision of this dissertation.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDG MENTS .................................................................................................. 4 LIST OF TABLES............................................................................................................ 9 LIST OF FIGURES ........................................................................................................ 10 LIST OF ABBR EVIATIONS ........................................................................................... 13 ABSTRACT................................................................................................................... 14 CHA PTER 1 INTRODUCTION.................................................................................................... 16 Statement of the Pr oblem ....................................................................................... 16 Research Questions ............................................................................................... 17 Studies on Ecological Distu rbance ......................................................................... 18 Progress in the Theory of Ecological Disturbance ............................................ 18 Disturbance and Ecosyst em Stru ctures ........................................................... 22 Disturbance and Ecosyst em Proc esses ........................................................... 26 Water Motion Disturb ance ................................................................................ 28 Ecosystem Theories and Distu rbance .................................................................... 29 Self-Organization and Hypotheses on Ecosystem-Level Strategies ................. 29 Ecosystem Traits during Succession ................................................................ 31 Pulsing Pa radigm ............................................................................................. 34 Microcosm Studies ................................................................................................. 35 Microcosm Overview ........................................................................................ 35 Ecosystem-Level Traits and Measurements ..................................................... 38 A pH Method for Continuous Monitori ng of Ecosystem-Level Metabolism ....... 39 Simulation Models of Distu rbance .......................................................................... 40 Research Plan ........................................................................................................ 43 2 METHODS.............................................................................................................. 51 Microcosm Ex periments .......................................................................................... 52 System Design ................................................................................................. 52 Sampli ng.......................................................................................................... 53 Maintenance and Measurem ents ..................................................................... 53 Disturbance Regimes ....................................................................................... 55 Data Processi ng and Analysis ................................................................................ 57 Data Processing: From Raw Data to Gross Primary Pr oductivity (GPP) and Ecosystem Respirat ion Rate (ER) ................................................................. 57 Ranges of pH and Alkalin ity ............................................................................. 60

PAGE 7

7 Analysis of Disturbance Int ensity and Frequen cy Effects ................................. 61 Analysis of GPP and ER Pulsi ng Pattern ......................................................... 62 Analysis of Resistance and Resi lience ............................................................. 63 Analysis of Uncertainty in the Resultant Patterns ............................................. 64 Simulation Models .................................................................................................. 65 Model of a Microcosm ...................................................................................... 66 Simulation of Disturbance ................................................................................. 66 Validation of Models ......................................................................................... 68 3 RESULT S............................................................................................................... 78 Structures and Proces ses of Mi crocosms ............................................................... 78 Microcosm Features ......................................................................................... 78 Ranges of pH and Alkalin ity ............................................................................. 79 GPP and ER of Microc osms............................................................................. 80 Effect of Distur bance Int ensity ................................................................................ 81 GPP and Disturbanc e Intens ity ........................................................................ 81 Pulsing Patterns of GPP and ER under Disturbance ........................................ 84 Effect of Distu rbance Fr equency ............................................................................. 85 Effect of Disturbance under Diffe rent Ecosystem Ma turities................................... 87 Relationship between MGPP and Disturbanc e Regime...................................... 88 Resistance and Resili ence of G PP and ER ...................................................... 90 GPP and ER Pulsing ........................................................................................ 90 Uncertainty of Re sultant Pa tterns ........................................................................... 91 Uncertainty from [GCO2] Correction................................................................. 91 Uncertainty from pH M easuremen t Error.......................................................... 93 Simulation Models .................................................................................................. 95 Model Equations and Calibr ations .................................................................... 95 Simulation Results ............................................................................................ 95 Validation of the Mod els................................................................................... 98 4 DISCUSSION ....................................................................................................... 128 Conclusi ons .......................................................................................................... 128 Discuss ion ............................................................................................................ 129 Credibility of MGPP-Disturbance Relationships in the Microcosms.................. 129 Self-Organization ............................................................................................ 134 Effect of Distur bance Int ensity ........................................................................ 135 Effect of Distu rbance Fr equency .................................................................... 138 Disturbance and Ecos ystem Ma turity............................................................. 139 Disturbance and Pulsing................................................................................. 140 Possible Mechanisms of Disturbance Effects from Simulation Models.......... 142 Trade-Offs in the St udy .................................................................................. 145 Future Wor k.......................................................................................................... 147 APPENDIX

PAGE 8

8 A GPP, MGPP, AND GPP/ER OF MICROCOSMS.................................................... 149 B R CODES FOR SIMU LATION MODELS .............................................................. 165 LIST OF RE FERENCES ............................................................................................. 170 BIOGRAPHICAL SKETCH .......................................................................................... 182

PAGE 9

9 LIST OF TABLES Table page 2-1 Experimental duration plans and test types of microcosms. ............................... 69 2-2 Sampling site informa tion. .................................................................................. 69 2-3 Chemical and biological proper ties of the sa mple la kes. .................................... 70 2-4 Disturbance regimes appl ied to micr ocosms. ..................................................... 70 3-1 Ranges of pH, alkalinity, and GPP in the 16 microcosms. ................................ 103 3-2 Pearsons correlation coefficient (r) bet ween GPP and ER in the four sections of each microcosm during the distur bance and post-disturbance periods. ....... 103 3-3 Average GPP/ER ratio in the four se ctions of each microcosm during the disturbance and post-di sturbance per iods. ....................................................... 104 3-4 MGPP-intensity relationship of the 11 micr ocosms in the intensity tests............. 104 3-5 Ratio of post-disturbance to disturbance in wavelength ( ) and peak amplitude (PA) of the pulsing GPP (GPPP) and pulsing ER (ERP) of microcos ms...................................................................................................... 105 3-6 Rank of MGPP among the four sections in each microcosm under disturbance frequency regimes............................................................................................ 105 3-7 Resistance (RS) and resilience (RL) of GPP and ER in the three microcosm sets S1, S2 and S6. ......................................................................................... 106 3-8 MGPP differences in the MGPP-disturbance relationships and potential [TCO2] changes by 0.01 pH error in the 16 mi crocosms.............................................. 107 3-9 Equations and calibrations of the basic steady-state model of a microcosm. ... 108 B-1 R code for a basic steady-sta te model of a microcosm .................................... 165 B-2 R code for disturbanc e intensit y tests ............................................................... 166 B-3 R code for distur bance frequency tests ............................................................ 168

PAGE 10

10 LIST OF FIGURES Figure page 1-1 Diversity-disturbance relationship in the intermediate dis turbance hypothesis... 47 1-2 Conceptual diagram of the pH-alka lin ity method in a freshwater aquatic microc osm.......................................................................................................... 48 1-3 Model of destr uctive pulses. ............................................................................... 49 1-4 Schematic plan of disturbance regimes. ............................................................. 50 2-1 Microcosm tank design and pl an for each micr ocosm p air................................. 71 2-2 Theoretical relationship between pH and [TCO2] (25C, zero salinity)............... 72 2-3 [PCO2] diel pattern and calcul ation of G PP and ER........................................... 72 2-4 Representation of MGPP-disturbance relationship from GPP time series............ 73 2-5 Procedure to transform GPP to GPPP................................................................ 74 2-6 Wavelength ( ) and peak amplitu de (PA) of GPPP............................................. 75 2-7 Energy systems diagram and equations of the basic microcosm model. ........... 76 2-8 Change of fd over the experimental peri ods according to power (p)................... 77 3-1 Ranges of pH-alkalinity pairs of the eight mi crocosm se ts............................... 109 3-2 Time series of pH in t he four sec tions of S8-A.................................................. 110 3-3 MGPP-intensity relationships of the 11 microcosms for the intensity tests......... 111 3-4 Top view of S6 microcosms on Day 45. ........................................................... 112 3-5 MGPP-frequency relationships of the five microcosms for the frequency tests... 113 3-6 Sensitivity of MGPP-intensity relationships of the 15-day post-disturbance period under the different k (cm/h) va lues in the 11 microcosms for the intensity tests.................................................................................................... 114 3-7 Sensitivity of MGPP-frequency relationships of the 15-day post-disturbance period under the different k (cm/h) val ues in the five microcosms for the frequency te sts................................................................................................. 115 3-8 Time series of [PCO2] and GPP in Sec 1 of S6-B under the different k (cm/h) values ............................................................................................................... 116

PAGE 11

11 3-9 Time series of [PCO2] and GPP in Sec 1 of S3-A under the different k (cm/h) values ............................................................................................................... 117 3-10 [PCO2] patterns during Day 10 in Sec 1 of S3-A under the different k (cm/h) va lues.................................................................................................... 118 3-11 Change of [TCO2] per 0.01 pH error in the different pH and [CA] levels........... 119 3-12 Flows and storages with numbers in the basic steady-state model of a microc osm. ....................................................................................................... 120 3-13 Simulation results of the variables from the basic stead y-state model of a microc osm. ....................................................................................................... 121 3-14 Patterns of GPP time series from t he simulation of the disturbance intensity and frequency models under the differ ent combinations of pfd signs between k2-k3 and k4-k5.................................................................................... 122 3-15 Effect of fd-5 values on GPP and MGPP patterns from the simulation of the disturbance int ensity m odel.............................................................................. 123 3-16 Effect of water column nutrient percentage on GPP and MGPP patterns from the simulation of the dist urbance intens ity model............................................. 124 3-17 Possible MGPP-intensity relationships from the simulation models for the intensity tests.................................................................................................... 125 3-18 Possible MGPP-frequency relationships from the simulation models for the frequency te sts................................................................................................. 127 A-1 Time series of S1-A. ......................................................................................... 149 A-2 Time series of S1-B. ......................................................................................... 150 A-3 Time series of S2-A. ......................................................................................... 151 A-4 Time series of S2-B. ......................................................................................... 152 A-5 Time series of S3-A. ......................................................................................... 153 A-6 Time series of S3-B. ......................................................................................... 154 A-7 Time series of S4-A. ......................................................................................... 155 A-8 Time series of S4-B. ......................................................................................... 156 A-9 Time series of S5-A. ......................................................................................... 157 A-10 Time series of S5-B. ......................................................................................... 158

PAGE 12

12 A-11 Time series of S6-A. ......................................................................................... 159 A-12 Time series of S6-B. ......................................................................................... 160 A-13 Time series of S7-A. ......................................................................................... 161 A-14 Time series of S7-B. ......................................................................................... 162 A-15 Time series of S8-A. ......................................................................................... 163 A-16 Time series of S8-B. ......................................................................................... 164

PAGE 13

13 LIST OF ABBREVIATIONS AMI Average Mutual Information BACI Before After Control Impact CA Carbonate Alkalinity ER Ecosystem Respiration GPP Gross Primary Productivity IDH Intermediate Disturbance Hypothesis MPP Maximum Power Principle PA Peak Amplitude PAR Photosynthetically Active Radiation TA Total Alkalinity TST Total System Throughput

PAGE 14

14 Abstract of Dissertation Pr esented to the Graduate School of the University of Florida in Partial Fulf illment of the Requirements for t he Degree of Doctor of Philosophy EFFECT OF DISTURBANCE ON ECOSYSTEM PROCESSES By Seungjun Lee December 2010 Chair: Mark T. Brown Major: Environmental Engineering Sciences One of the ecosystem management goals is to understand how disturbance regimes influence an ecosystems traits and integrate the regimes into management criteria. This study contributes to the understanding of ecosystem energetics under disturbance by investigating the relati onships between disturbance and ecosystem processes in the context of the current ec ological paradigms: intermediate disturbance hypothesis (IDH), ecosystem maturity, and puls ing. The responses of gross primary productivity (GPP) and ecosystem respiration rate (ER) to water motion disturbance regimes were tested using open-top aquatic microcosms, and the mechanisms of the resultant patterns from the microcosms were suggested by computer simulation models. Each microcosm developed a unique system and maintained a balance between GPP and ER through the self-organizing processe s. Each microcosm showed a distinct relationship between disturbance regime and G PP. The mechanisms for the distinct relationships were explained by changes of e fficiencies in energy flow pathways of a system under disturbances and by a disturbanc e threshold above which the efficiencies are permanently altered. The tests of t he disturbance effects on GPP and ER under different maturities of microcosms support ed the current theory on the ecosystems

PAGE 15

15 development arguing that an ecosystem reinforces internal structures and thus becomes more resistant to disturbances over time. GPP and ER oscillated over time in the microcosms, and the wave length and amplitude of the G PP and ER pulsing patterns were amplified by disturbance in microcosms early stages of development, compared with those of an undisturbed system. As orig inally intended, the combined study of microcosms and simulation models pr ovided new hypotheses on disturbance and ecosystem processes, which need to be tested further us ing microcosms, simulation models, or in the field.

PAGE 16

16 1 CHAPTER 1 INTRODUCTION Statement of the Problem Natural disturbance has been recognize d as an important component of ecosystems for maintaining ecological inte grity in rece nt decades (Grumbine, 1994; Ward, 1998). In accordance with this rec ognition, ecologists have published a number of studies on how natural disturbance influence s ecosystem structures, such as species diversity or patch dynamics (see review s and analyses by Mackey and Currie, 2001; Pickett and White, 1985; Sousa, 1984). Consi dering that ecosystem processes, such as primary production, respir ation, and decomposition, ultimately build ecosystem structures, the fundamental relationship between natural disturbance and ecosystem processes has not been discussed or studied much in spite of its potent ial significance. Although recent studies (Cardinale et al ., 2005; Haddad et al., 2008; Kondoh, 2001; Worm et al., 2002) demonstrated the comple x relationship among species diversity, productivity, and disturbance, they regar ded disturbance and productivity as two interacting independent variables affecting s pecies diversity. Because energy flows related to productivity (or another process) in an ecosystem are likely to be influenced by disturbance regimes, such as how often di sturbances occur or how strong they are, certain patterns may be inherent in the relationship between disturbance regimes and the impacted ecosystem processes. For this reason, it is important to investigate how ecosystem processes respond to disturbance regimes in light of current ecological paradigms potentially related to disturbance. In addition, as Shea et al. (2004) pointed out, it is also important to identify the mechanistic basis of any resultant patterns of disturbance effects.

PAGE 17

17 Research Questions This study investigated how ecosystem proce sses respond to disturbance regimes to answer the following f our research questions: What are the relationships between di sturbance intensities and ecosystem processes? Mackey and Currie (2001) re cently analyzed the published literature on the relationship between disturbance regimes and ecosystem structures, arguing that the relationships may be various in contrast to the conjecture of the intermediate disturbance hypothesis (IDH) earlier proposed by Grime (1973) and Connell (1978). The IDH is still contro versial among contempor ary ecologists. Meanwhile, it has rarely been discussed w hether such a hypothesis as the IDH is applicable to the relationship between di sturbance intensities and ecosystem processes. Fundamental relationships between disturbanc e intensities and ecosystem processes would allow bette r understanding of how a disturbance influences energy flows in an ecosystem. What are the relationships between disturbance frequenc ies and ecosystem processes? It is questionable whether certain disturbance frequencies increase energy flows in an ecosystem. Also if disturbance frequencies influence ecosystem processes, the effects of disturbances on energy flows of an ecosystem may differ depending on whether the disturbances are continuous or discrete. How do ecosystem processes respond to disturbance regimes under different ecosystem maturities? Margalef ( 1963) proposed general trends of ecosystems according to their successional stages (ma turities), where a mature ecosystem tends to be strongly influenced by inter nal successional trajectories, whereas a less mature counterpart tends to be more affected by external forces such as disturbances or environmental fluctuati ons. Distinct responses of ecosystem processes to disturbance events under di fferent maturities would allow better understanding of how an ecosystem self-o rganizes and reinforces its ability to resist and recover from external disturbances. How does disturbance influence ecosystems internal pulsing regimes? Pulsing has recently been recognized as being prev alent in ecosystems (Odum et al., 1995). If internal processes of an ecosyst em fluctuate over time, a disturbance may influence the oscillating patterns. To answer the four questions regar ding disturbance effects on ecosystem processes, freshwater aquatic microcosm experiments were conducted under controlled water motion disturbance regimes in the labor atory. Unlike in a seawater environment,

PAGE 18

18 where turbulence is prevalent, freshwater organisms in the littoral zone of Florida lakes from which the microcosm water and sedim ent samples were transplanted were expected to be well adapted to the still-water condition, so the water motion generated in the experiments was regarded as disturbance. As variables of ecosystem processes, gross primary productivity (GPP) and ecosyste m respiration rate (ER) were estimated by monitoring water column pH and alkalini ty changes of the microcosms over time. Four different regimes of water motion dist urbance were applied to the four replicated sections in a microcosm tank to test how GPP and ER of microcosms respond to the different disturbance regimes through the pr ocess of self-organization. A modeling study followed the microcosm experiments to suggest possible mechanisms of resultant patterns from the microcosms. A steady-s tate model of an aquatic microcosm was established, and parameters were modifi ed to include disturbance factors based on hypotheses regarding disturbance effects. A final synthesis of microcosms and simulation models suggested the patte rns and their possible mechanisms. Studies on Ecological Disturbance Progress in the Theory of Ecological Disturbance Disturbanc e is a major source of ecosystem dynamics and heterogeneity in temporal and spatial scales (Sousa, 1984). As the concept of ecosystem has shifted from a static to a dynamic pattern-and-pr ocess view of nature (Grumbine, 1994), disturbance has been regarded as an integral part of ecosystems (Ward, 1998). As Laska (2001) noted, however, researchers have studied ecological disturbance with alternative definitions and concepts, which ultimately impeded the generation of a unified theory on disturbance.

PAGE 19

19 The alternative definitions and concepts of disturbance were partially attributed to the scale of a system in question. Whereas studies of the effects of a disturbance on individual species (e.g., Larsson et al., 1986) focused on the physiological change of the species under the disturbance, studies on la ndscape-scale impact of large disturbance (e.g., Turner et al., 1998) explained the ro le of disturbance as an opening of a patch and the colonization of new species. Levin and Paine (1974) emphasized the effects of disturbance on spatial patch dynamics, wher eas Sousa (1984) later defined disturbance from the perspective of indi vidual organisms. From the per spective of scale, Pickett et al. (1989) suggested a concrete description of ecological disturbance based on the concept of hierarchical structure of ecosyst ems and of the minimal structure by which a system of interest persists. They proposed a new definition of disturbance as follows: Disturbance is a change in the minimal struct ure caused by a factor external to the level of interest. This definition emphasizes the importance of scale of observation in disturbance studies. For instance, a dist urbance at an individual level may not be regarded as such at an ecosystem level. As Sparks et al. (1990) pointed out, whether a certain force is a disturbance to a system can be determined by identifying normal condition in the system and by monitoring a threshold above which it is regarded as a disturbance in the long term. The type of an ecosystem studied or the subject of a study also influenced the alternative concepts of disturbance. For example, Sousa (1979) explained the mechanisms of species diversity in an interti dal boulder field in southern California with the opening of habitat by intermediate frequency of wave disturbances, whereas Boerner (1982) addressed the im pacts of fire in temperate ecosystems with regard to

PAGE 20

20 nutrient cycling. Recent studies (Bond-Lamberty et al., 2007; Randerson et al., 2006) investigated the effects of fire from the perspective of carbon cycle and subsequent global climate change. The def inition of disturbance proposed by Pickett et al. (1989), thus, suggests that no unifying theory of di sturbance exists and that the theory of disturbance needs to be addressed in each scale of interest and in each subject of study. Although many studies investigated the effects of one disturbance type on ecosystem traits (e.g., floods by Bornette and Amoros, 1996; temperature by Gates et al., 1992; wind by Nowacki and Kramer, 1998; waves by Sousa, 1980; hurricanes by Vandermeer et al., 2000), ecosystems are gener ally under the influence of compounded disturbances of different types superimposed or in a rapid sequence, and the mixture of disturbances may have multiplicative effe cts on the ecosystems structures and processes (Paine et al., 1998). As an exampl e of the interactions between different disturbance types and their emerging effects on ecosystems, Platt et al. (2002) studied the effects of hurricane-fire in teractions on the mortality of trees. They investigated how different fire types influenced the potential effe cts of hurricanes on the mortality of slash pine trees. In the study, the mo rtality of the slash pine tree s was higher during and after the hurricane when the ecosystem was pre-disturbed by dry-season anthropogenic fires than when the system was unburned or pre-disturbed by wet-season natural fires. The sites disturbed by wet-season fires maintain ed lower mortality than the unburned sites during and after Hurricane Andrew in 1992. A simultaneous or subsequent disturbance may cause emerging effects on ecosystem tr aits, which cannot be predicted from the

PAGE 21

21 effect of each disturbance, by altering the ph ysical or biological en vironment shaped by the other disturbances. Although past research on disturbance addr essed the unidirectional effects of disturbance on ecosystem traits as Sousa (1984) reviewed, Mack and DAntonio (1998) and Corenblit et al. (2008) recently addre ssed the importance of the reciprocal adjustments between biotic and abiotic com ponents (disturbances) in ecosystems. Mack and DAntonio (1998) argued that human management activities or community behaviors may inversely alter the disturbance regimes, which ultimately changes the ecosystem structure and functi on and makes it difficult to pr edict the pattern of future disturbance regimes. For in stance, the introduction of a new species to an ecosystem may modify the existing dist urbance regimes, which ultimately pushes the ecosystem to a new transitional state caused by an extinction of the native species. Corenblit et al. (2008) introduced the recent recognition of biotic-abiotic feedbacks from the geomorphological perspective with examples of ecosystem engineers. The concept of niche construction (Odling-Smee et al., 1996), which implies that biotic components build their own or others physical environm ents, provided a new insight on the bioticabiotic feedbacks contrary to the classic conc ept of ecological niche (Hutchinson, 1957). The niche construction concept indicates t hat disturbance regimes may be altered by species feedback controls in t he evolutionary time scale. Disturbances are characterized by their specific regimes. Similar to the definitions of disturbance regimes provi ded by Sousa (1984) and van der Maarel (1993), Shea et al. (2004) recently characterized them by int ensity, frequency, spatial extent, and duration as follows:

PAGE 22

22 Intensity describes how strong a dist urbance is, and can be defined with various units depending on the distur bance agents, such as 10 m/s wind speed or 40C temperature. The same disturbance intens ity, however, does not necessarily have the same effect on different communities. The response of a community depends not only on the disturbance intensity but also on the vulnerability of the community to the disturbance. Sousa (1984) defined the strength of a disturbance agent and the damage of a disturbed system as intensity and severity, respectively. Frequency addresses how often disturbances occur in a given time. Spatial extent addresses the area affected by a disturbance event. Duration indicates how long an individual disturbance lasts. The frequency and duration determine the temporal char acteristics of a disturbance. Disturbance and Ecosystem Structures The importance of biodiversity has been a core issue among contemporary ecologists (see a review by Hooper et al ., 2005), and many studies in ecology have focused on the mechanisms of the maintenance of species diversity. Some researchers have emphasized productivity as a mechanism for species diversity (see a review by Waide et al., 1999). As Waide et al. addre ssed, recent studies on the relationship between productivity and species diversity increasingly suggested the influence of scale. Kassen et al. (2000), in a study of sp ecies diversity in a laboratory microcosm, demonstrated the unimodal relationship betw een productivity and diversity by niche specialization in a small heterogeneous environment. T hey also suggested that diversity would increase li nearly with productivity on a large scale because of the increased number of different niches and decreased immigration between niches. Dodson et al. (2000) analyzed previous data on 33 lakes to investigate the relationship between primary productivity and species richne ss. They showed that the relationship could be varied depending on species types or scales of the ecosystems under investigation. Although the productivity-richne ss relationships for rotifers, cladocerans,

PAGE 23

23 copepods, and macrophytes were unimodal, t hose for phytoplankton and fish were strongly dependent on lake ar ea. Chase and Leibold (2002) investigated the importance of scales in diversity-productivi ty relationships. From the research on sample pond sites, they concluded that spec ies richness at the local scale is maximum at the intermediate level of productivity, while the relationship is positive monotonic at the regional scale becaus e of the increase of -diversity (species compositional difference among localities). Mittelbach et al. (2001) analyzed 171 published studies on productivity-richness relationships, and f ound that the hump-s haped or positive relationship was most prevalent alt hough the percentage of each pattern may depend on the scale or type of ecosystem. Disturbance has been extensively studied as one of the mechanisms for species diversity or other ecosystem traits across various ecosystem types (e.g., Armesto and Pickett, 1985; Denslow, 1980; Eisenbies et al., 2005; McCabe and Gotelli, 2000; McIntyre and Lavorel, 1994; Resh et al., 1988; Sousa, 1979; Wilson and Keddy, 1986; Woodin, 1981). The interm ediate disturbance hypothesis (IDH) has suggested the relationship between disturbance regime and species diversity (richness) as a humpbacked pattern where the maximum diversity occurs in the middle range of disturbance regimes as shown in Figure 1-1 (Connell, 1978; Gr ime, 1973). From t he perspective of the non-equilibrium hypothesis, Connell (197 8) explained the IDH as one possible model for the maintenance of species diver s ity. He suggested that species diversity may be maintained by the combination of six models, which cons ist of the three nonequilibrium hypotheses (the I DH, the equal chance hypothesis, and the gradual change hypothesis) and the three equilibrium hypotheses (the niche diversification hypothesis,

PAGE 24

24 the circular networks hypothesis, and the compensatory mortality hypothesis), although the importance of each model in the cont ribution to diversity differs depending on temporal or spatial scales. Meanwhile, t he mechanism of the I DH has been explained by a trade-off in species-specific abilities: t hey can resist competit ors or disturbance, but cannot excel at both (Petraitis et al., 1989). A recent st udy by Haddad et al. (2008), however, did not observe the competition-colonization trade-off expl ained in the classic IDH mechanism. Roxburgh et al. (2004) suggested the between-patch and the withinpatch models as advanced theoretical models of the species coexistence mechanisms from the IDH to overcome the weakness of the IDH that requires patchy disturbances. In the between-patch model, t he IDH is explained using a trade-off between competitive ability and dispersal in the spatial context. That is, better competitors can survive in the low disturbance frequency while the high fr equency disturbance opens a new patch for better dispersers. In this context, the intermediate disturbance frequency opens a new patch in the spatial scale for colonizers (better dispersers) while maintaining the better competitors, which results in the high number of coexisting species in the long term. In the within-patch model, all species in a pat ch are under the same disturbance regime, so the patch is spatially homogeneous under disturbance but temporally heterogeneous. Once the patch is destroyed by a disturbance, better dispersers can first dominate within the patch with the late r domination by better competitors. The coexistence of dispersers and competitor s thus can be realized by the repeated disturbance events with the intermediate fr equency. Although the IDH was supported by some empirical studies (e.g., Flder and Sommer, 1999; Sousa, 1979; Townsend et al., 1997), Chesson and Huntly (1997) argued that disturbance alone does not explain

PAGE 25

25 species coexistence mechanisms because disturbance not only lessens species competitions but also lowers growth rates. They emphasized that species coexistence requires species interactions and niche op portunities created by disturbance in the spatial and temporal scales. Bongers et al. (2009) recently investigated the diversity patterns in tropical forests of Ghana with large-scale datasets. They concluded that the IDH could explain the variation of diversity in dry forests but not in wet tropical forests. This conclusion may imply that species di ffer in the response to disturbance regimes and that the IDH alone cannot ex plain species diversity pattern s. In this regard, Mackey and Currie (2001) analyzed 197 cases from the re levant literature on the relationship between disturbance regimes and species trai ts (species richness, diversity, or evenness) to investigate w hether the IDH was supported by those studies. They identified various shapes for the relations hips: positive monotonic, negative monotonic, peaked, U-shaped, and non-significant. Th ey pointed out the prevalence of the non-significant patterns in the previous studies: 35% (richness-disturbance), 28% (diversity-disturbance), and 50% (evenness-di sturbance) out of th e five patterns. From the perspective that species divers ity is not determined only by productivity or disturbance, researchers have recent ly suggested the importance of combined effects on species diversity. For exampl e, Fukami and Morin (2003) demonstrated the importance of community assembly history on the productivity-diversity relationships using aquatic microbial communities. Kondoh (2001) studied how disturbance and productivity influence the species richness using a numerical model. In the model, species richness was highest when productivity and disturbance were well balanced, and was lowest at each extreme. This result may be support ed by the study by Jenkins

PAGE 26

26 et al. (1992) on productivity, disturbance, and food web structures. Jenkins et al. suggested that if productivity is high enough to support a long food chain in an ecosystem, as an extreme case, the chain may be easily destroyed under a disturbance. The importance of interaction between productivity and disturbance for species diversity was also demonstrated by Cardinale et al. (2005) and Worm et al. (2002). Haddad et al. (2008), in an experim ental study, concluded that diversity patterns under certain disturbance regimes and productivity levels may be determined by species traits such as competitive ability and growth rate. Disturbance and Ecosystem Processes Although the recent studies on the relati ons hips among productivity, disturbance, and species diversity regarded productivity and disturbance as two independent variables interacting to affe ct species diversity, the relationships can be more complicated because disturbance may influence productivity, as Haddad et al. (2008) pointed out. Wootton (1998) inve stigated the IDH from a mult itrophic perspective in the modeling study. Previous studies on the IDH focused on a single trophic level for species competition and assum ed the other trophic levels to be static. Wootton argued that the validity of the IDH depends on the trophic level of s pecies so that the IDH needs to be carefully interpreted from the multit rophic perspective. The multitrophic approach in disturbance studies may i ndicate potential significance of ecosystem processes in disturbance theories, by which ecosystem-l evel energy flows under disturbance can be assessed in a holistic view. Sprugel (1985) argued that a num ber of disturbance studies have focused on the relations hip between disturbance and ecosystem structures in part because of the less drasti c effect of disturbance on the ecosystems energetic parameters and because of the diffi culty in their measurements. The less

PAGE 27

27 drastic effect may be caused by energetic tra de-offs in a system. For instance, even if an ecosystem loses half of its species, which is a drastic change in the structure of the system, the remnant keystone species (Power et al., 1996) may maintain or improve the major energy pathways, so the energetic parame ters may not be altered by disturbance. Reiners (1983) and Odum and Odum ( 2000) built models and simulated net production under disastrous disturbances. In the destructive model by Odum and Odum (2000, p.241), assets we re drained by a destructive pul sing event. The resultant pattern of the assets was represented as a gradual recovery after a crash during the pulsing event. The models built by Reiner s (1983) and Odum and O dum (2000), where disturbances are large and destructive, howev er, unrealistically represented only one type of disturbance regime instead of the va rious regimes of disturbances in nature. Meanwhile, Scheffer et al. (2001) proposed a hy pothesis of alternative stable states. The alternative stable states hypothesis argues that the state of an ecosystem abruptly changes to an alternative stable state when a disturbance regime exceeds a certain threshold. They also suggested that the threshold of disturbance, under which a system sustains its stable state, is altered by ex ternal perturbation, which was supported from the microcosm study by Jiang and Patel (2008). Although studies on the effects of dist urbance on ecosystem processes and their mechanisms have not been published as often as those on ecosystem structures, a few empirical studies have investigated how di sturbance influences ecosystem productivity or respiration rate. For exam ple, Houser et al. (2005) inve stigated the effect of upland soil and vegetation disturbance on stream metabolism. In their study, disturbance intensity was defined as the land use (percentage of bare ground on slopes greater than

PAGE 28

28 5% in each catchment). They concluded t hat gross primary productivity (GPP) was not significantly correlated with disturbance inte nsity because of the low GPP in stream ecosystems although the relati onship between ecosystem respiration rate (ER) and disturbance intensity was dependent on the season of the year. Productivity has been a fundamental issue in forest managem ent, and Kimmins (1996) addressed the importance of disturbance affecting the determinants of production, such as soil temperature, decomposition rate, and soil drai nage. In a similar context, Eisenbies et al. (2005) investigated the ef fects of soil physical distur bance by intensive harvesting practice on soil productivity in pine plantat ions. They suggested possible mechanisms of increased productivity under moderate distur bance, such as competition control or increased rate of nitr ogen mineralization. Water Motion Disturbance Water motion is an important driving force for production or control of physiological rates and community structure in aquatic ecosystems (Hurd, 2000). There have been many studies on the effects of water mo tion on the physiology of aquatic organisms (Charters et al., 1973; Clarke et al., 2005; Dodds, 1989; Gordon and Brawley, 2004; Hurd, 2000; Mass et al., 2010; McAlister and Stancyk, 2003; Thomas and Vernet, 1995; Whitford and Schumacher, 1961; Whitford and Schumacher, 1964) or on ecosystems (Gagnon et al., 2003; George and Edwards, 1973; Kemp and Mitsch, 1979; Margalef, 1997; Petersen et al., 1998). Whitford and Schumacher (1964) discovered that flowing water is an important mechanism for the mineral uptake of algae in lotic freshwater ecosystems. Mass et al. (2010), in a recent study on the photosyn thesis of marine benthic autotrophs, emphasized the importance of the flow-driven oxygen efflux for hi gh productivity. Kemp

PAGE 29

29 and Mitsch (1979) modeled species coexistence mechanisms under water motion, and suggested that the highest diversity occurs under turbulence with periodicity close to the turnover time of competing phytoplank ton species and under high kinetic energy. Petersen et al. (1998) investigated the ef fects of water motion on the structures and processes of coastal planktonic ecosystems by mimicking the coastal mixing regimes. But they did not find any significant effe ct of the mixing regi mes on the ecosystem processes, such as productivity and resp iration rate. Gor don and Brawley (2004) studied the effect of water motion on al gal reproduction under the controlled water motion regimes in the laboratory. They demonstrated that water motion regimes influence reproduction of Alaria esculenta and Ulva lactuca In addition, they pointed out that the effects of water motion are different among species and that the distinct responses of species to water motion may be a ttributed to their life hi stories. As Sparks et al. (1990) argued, a disturbance is regarded as a deviation from a normal environmental condition in the long term, whic h implies that not all water motions are regarded as disturbances depending on the life histories of species or ecosystems adapted to the water motions. Ecosystem Theories and Disturbance Self-Organization and Hypotheses on Ecosystem-Level Strategies Odum (1988) argued that interactions of species reinforce the organization of the system by trial-and-error processes in a given environment. He exemplified s elforganization with a balanced aquatic microc osm where the initial seeds with many available species and abiotic components de velop a basic ecosystem by selecting necessary components. In the same contex t, Levin (2005) defined an ecosystem as a complex adaptive system, where a whole selects the parts and their interactions by

PAGE 30

30 feedback, adaptation, and regulation to maintain homeostasis. In the early reasoning on self-organization, Ashby (1962) explaine d organization as the interdependence or conditionality of components, where constr aints exist in the relationship between components. Although an ecosystem is regarded as a whole composed of interdependent components in cy bernetics, it is unknown whether the components of a system ultimately have a common goal and whether the organization benefits all individuals. Researchers have proposed hypotheses on the ecosystem-level strategies or tendencies in self-organizing processe s (Lotka, 1922; Odum, 1969; Odum, 1995; Ulanowicz, 1997). Lotka (1922) proposed the maximum power principle (MPP) hypothesizing that flux of available energy directed into a system by efficient energycapturing devices determines the preval ence of the system. Odum (1983) later articulated the MPP as a strategy of a prevailing ecosystem maximizing useful power, by which the system feeds ba ck and amplifies other energy pat hways to sustain itself. In the MPP, Odum (1983) also pointed out the trade-off between po wer and efficiency in a system. This trade-off implies that di sturbances change the efficiency in powercapturing devices of a system increasing or decreasing power inflows into the system. Cai et al. (2004) argued that the maximized power is a systems rate of acquisition of useful energy. Because part of the acquired energy by the system is consumed as useful energy and the rest of it is dissipated, however, it is difficult to recognize how much useful energy is drawn in by the system. Odum ( 1969) suggested that ecosystems generally maintain higher producti vity during the early successional stage and higher species diversity during the late r successional stage, which may imply that

PAGE 31

31 the potential for increasing diversity and overcoming disturbance is supported by accumulation of energy. In this regard, GPP (the rate of photo synthetic production) indicates the degree of active capture of light by autotrophs pr oviding an ecosystem with directly usable energy resource s for enhancement of structures. Ulanowicz (1997) proposed ascendency as a hypothesis on ecosystem development and defined it as the product of total system throughput (TST) and the network average mutual information (AMI). As systems develop, TST (the total energy flow) and AMI (the average amount of constraint between compartments) increase to some extent. Ulanowicz (1997) proposed that a systems total capacity is composed of ascendency and overhead. Overhead is generat ed by redundancy of inputs, exports, dissipations, and pathways. According to the ascendency theory, ascendency grows at the expense of overhead. Because overhea d plays an important role as a buffer against external disturbances, a sustainable system may not maximize the ascendency but will optimize ascendency and overhead. With the strategy of a system balancing between ascendency and overhead, ascendency varies depending on the disturbance regimes and adaptation of specie s. While a destructive di sturbance may reset a system to a low ascendency, a certain range of di sturbance regimes may increase ascendency by promoting efficiency and reduc ing redundancy of the system. Ecosystem Traits during Succession Tansley (1935) described the successi on of an ecosystem as the normal vegetational change and progress toward integr ation and stability of climax. Connell and Slatyer (1977) later crystallized the me chanisms of succession by three models: facilitation, tolerance, and inhibition m odels. The three models suggested how ecosystem succession proceeds after empty spac e is occupied by pioneer species that

PAGE 32

32 have broad dispersal powers and rapid growing ability. Each model has a distinct point of view on the response of species to envir onmental constraints and biotic modifications of the environment. The facilitation model hypothesizes that the later species can colonize after the environment is modified by the pioneer species, which occurs in primary succession or heterotrophic succession. The tolerance model, though there is little evidence, hypothesizes that later specie s that are tolerant in the low resource availability occupy the spac e with the pioneer species. The third inhibition model suggests that later species cannot invade t he space until the pioneer species release resources by death or damage. Although the climax in the succession has been regarded as the stable dynamic equilibrium a fter Tansley (1935), the current paradigm of succession (Pickett, 1976) and pulsing (Odum et al., 1995) recognized the climax as one of the temporary peaks in the long-te rm fluctuations and emphasized the dynamic responses of ecosystems to environment al constraints (e.g., disturbance). Margalef (1963) argued that mature ecosyst ems tend to have higher complexity, more information, and higher efficiency (l ower production/biomass ratio) than less mature ones. One exception suggested by Ma rgalef (1963) is that progressive change through the succession may be limited by disturbance or any changing environment under which the ecosystem is occupied by highly reproductive and less-specialized species. Beyers (1962), through his microcosm study on the response of an ecosystem to external temperature stre sses, hypothesized that a highly integrated ecosystem with strong interdependence among system components is likely to be less affected by environmental extremes because of mainta ined homeostasis. The arguments by Margalef (1963) and Beyers (1962) are synthesiz ed to suggest that the future direction

PAGE 33

33 of a mature ecosystem depends more on inter nal factors than on external disturbances or environmental changes. In a similar cont ext, Odum (1969) suggested more specific trends of ecosystem attributes between dev elopmental and mature stages. According to Odum, mature stages are characterized by complex, well-organized, low-growth, and high-information structures. He suggest ed the ecosystem-level ratio between production and respiration (P/R) as an indicator of autotrophic or he terotrophic state of an ecosystem, and the P/R approaches 1 as the system becomes mature. Sousa (1980), from the perspective of ecosystem stru cture, studied the response of intertidal algal community to the disturbance (turnov er of boulder) under different successional stages. He demonstrated that early succe ssional communities are easily damaged by disturbance but recover more quickly fr om the damage than late successional communities. He also concluded that t he responses of communities are dependent on the disturbance regimes, and i dentified several thresholds of the responses, which may be attributed to the life histories of spec ies. Halpern (1988) in the study of Pseudotsuga forests in the western Cascade R ange, Oregon, suggested that long-term responses of communities are influenced by the interactions among life history, disturbance intensity, and chance. Resistance and resilience are ecosystem trai ts indicating how well an ecosystem resists and recovers from disturbance events. The resistance and resilience of an ecosystem may change throughout the developmental stages depending on the strength of species interactions and buffering capacity of the system. Peterson and Stevenson (1992) investigat ed how disturbance timing influences resistance and resilience of stream ecosystems from the perspective of community structures.

PAGE 34

34 Although they demonstrated that the resi stance of benthic algal communities was dependent on the developmental stages of communities and disturbance regimes, the resistance and resilience of communities could be affected by complex interactions of environmental conditions. In this regard, resistance and resilience may have a close relationship with how components of an ecos ystem develop networks of energy, matter, and information over time. Researchers have recently discovered scale-free networks where some nodes have a number of linkages, and most nodes have a few connections with other nodes so that the number of links and nodes form a power distribution with a hierarchical structure (see reviews by Albert and Barabsi, 2002; Barabsi and Bonabeau, 2003). The scale-free network has been shown in many relations, such as social, Internet, and biological networks. Ecosystem networks resemble the scale-free networks in that a food web is a hierarchical structure wh ere a few high-trophic-level species control the majority of low-trophiclevel species (Odum, 1988). This kind of network is resistant and resilient to di sturbance unless the high-linkage nodes are severely attacked (Barabsi and Bonabeau, 2003). As succession proceeds, an ecosystem tends to accommodate higher-tr ophic-level species, and how much the system is resistant or re silient under disturbance may depend on how much the species in the higher trophic levels are secure from the damage. Pulsing Paradigm Pulsing has recently been recognized as a general pattern of ecosystems (Odum et al., 1995). In ecosystem theori es, pulsing indicates not only external disturbances or regular abiotic fluctuations but also internal oscillations of populati on or energy storage. Campbell (1984) investigated ener gy filter properties of ecosystems where matching input frequencies of external source to tu rnover properties of internal organisms

PAGE 35

35 determined power-capturing capacity. In a modeling study, Kemp and Mitsch (1979) suggested that species coexistence is maxi mized when the frequency of turbulence is close to the turnover rate of the related species. Odum et al. (1995) also introduced the hypothesis that the matching of inter nal and external pulses may enhance the performance of an ecosystem. Kang (1998) modeled different kinds of paired producer-consumer units in series, and the result of simulation suggested that an internally pulsed system draws more untapped ener gy in the production processes than a steady-state one by providing optimum load to the production function. Yamamoto and Hatta (2004) investigated the effects of pulsed nutrient supply on phytoplankton diversity using a numerical model. The re sults of the study suggested that species diversity is maximized in the intermediat e frequency of nutrient pulses although the frequency regime maximizing the density of species may depend on the physiology or life history of each species. In the model, however, the phytoplankton cell density oscillated in accordance with a frequency of nutrient pulse so that no pulsing of cell density was observed under the c ontinuous nutrient supply, which is different from the simulation models by Kang (1998) or Zwick ( 1985) where the oscillat ion of biomass was generated by internal autocatal ytic processes. Microcosm Studies Microcosm Overview While a naturally generated small eco system has been called a microcosm, a microcosm generally means an isol ated ecosystem in an artifi cial container created by humans (Beyers and Odum, 1993). Many re searchers have constructed ecological microcosms to study toxicology, microbiol ogy, and community theories of aquatic and terrestrial ecosystems (Fraser and Keddy 1997). Both reductionistic and holistic

PAGE 36

36 approaches of mesocosms, which have been used to test relevant problems as Odum (1984) addressed, have also been applied to microcosms. From the perspective of reductionism, microcosms have been used to test individual species physiology (e.g., Boersma, 2000). Studies on toxicology (e .g., Komjarova and Blust, 2009; Martin and Holdich, 1986) tested the response of indi vidual species responses to specific chemicals or treatments. On the other hand, from the perspective of holism, ecosystem-level properties have been studied using microcosms. Beye rs (1963b) studied the ecosystem-level metabolism of the 12 aquatic microcosms sampled from San Marcos River, Texas. His additional study on the ecosystem-level me tabolic patterns of eight microcosms (Beyers, 1963a), which were established under different conditions of temperature, salinity, volume, light intensity, and s pecies composition, demonstrated that net photosynthesis (net primary production) and nighttime respiration were maximal in the first half of the light or dar k period. This ecosystem-level approach in microcosm tests by Beyers (1963a) further addressed the probl em of a life-supporting system where the metabolism of man might be threatened duri ng the second half of the night period because of the rapid oxygen metabolism of the ecosystem during the first half of the night. Several researchers studied ecosyst em-level metabolism using microcosms (Cooper and Copeland, 1973; Ferens and Beyers, 1972; McIntire and Phinney, 1965; Sugiura et al., 1982). Cai et al. (2006) recently used aquatic microcosms to test the maximum power principle, and they meas ured and analyzed the change of pH as an ecosystem-level property occurring by community metabolism.

PAGE 37

37 A microcosm transplanted from a parent ecosystem usually self-organizes to a newly balanced isolated ecosystem The result of self-organization produces variations among microcosm replicates even if the initial i nput condition is similar. Roeselers et al. (2006) demonstrated that the three ident ical biofilms grown under the same environment showed similar growth rates bu t different community compositions and species richness. Although it is difficult to completely replicate microcosms, Beyers and Odum (1993) suggested that the replication of microcos ms can be enhanced by crossseeding among the microcosms. The studies on species-area relations hip (see Lomolino, 2000) suggest that species richness may be restricted in a small microcosm. Ruth et al. (1994) studied the effect of microcosm size on the rate of benthic macroinvertebrate recolonization and community structure, and s howed that the total number of taxa increased with the increase of microcosm size. In this c ontext, Carpenter (1996) argued that the size restriction of a microcosm may exclude the important f eatures of communities and ecosystems. But the basic ecosystem-level tr aits of self-organization from dispersed biotic and abiotic components, such as pr oduction-consumption, material cycles, and homeostasis have been demonstrated in many microcosms (see Beyers and Odum, 1993). In addition, a microcosm study qui ckly generates new ecological insights and hypotheses with which a new field experim ent can be designed and conducted. Hall (1964) argued that there is a tr ade-off between laboratory and field studies in ecological research. Laboratory study hardly accommodat es the real conditions of a parent ecosystem, whereas field study may produce significant errors in observation, and natural variables cannot be easily controlled or manipulated for cause-and-effect

PAGE 38

38 analysis. In a similar context, Benton et al. (2007) addressed the usefulness of microcosms for studying the effects of intractable large scale problems. Heath (1979), using state space analysis, s howed that the response of a system to a given input sequence depends on the initial status of the system. The range of response space thus becomes reliable by multiple experiments with various initial conditions under the same input sequence. In this regard, microcosms transplanted from different ecosystems and treated with the same input sequence can test the range of response of such ecosystem types under a given treatment sequence. Ecosystem-Level Traits and Measurements Ecosystem-level ener gy flux cannot be easily inferred from the physiology of individual species. Through the energy transformation processes, species interact and exchange materials with the su rrounding environment. For ex ample, photosynthetic or respiratory processes of bi otic components can be estima ted by monitoring carbon dioxide or oxygen concentrations within the surrounding space. In aquatic ecosystems, assuming material exchange between organi sms and their environment, ecosystemlevel metabolic processes could be esti mated by measuring the change of carbon dioxide, oxygen, or nutrient concentrations in the water column. Odum (1956) and Copeland and Dorris (1964) measured oxyg en concentration to estimate primary production in flowing waters. In t he oxygen method, the change of oxygen concentration in the water column occurs by gross primary production, respiration, and oxygen movement by diffusion. Appropriate corrections of diffusion provide the primary production or respiration rate of the ecosyst em under investigation. Because diffusion is a significant part of the change of ox ygen concentration in the water column, researchers applied distinct methods using a plastic dome (Copeland and Duffer, 1964)

PAGE 39

39 or sulfur hexafluoride (SF6) (Wanninkhof et al., 1985) to estimate diffusive oxygen change. Heffernan and Cohen (2010) measured nitrate concentration in the flowing water resulting from nitrate metabolism to es timate the primary production of spring-fed Ichetucknee River, Florida. The nitrate method does not need gas exchange correction, although the use of the method is limited to the specific ecosystem type. Carbon dioxide metabolism has been estimated eit her by using the C-14 tracer method (Peterson, 1980) or by monitoring water co lumn pH (Beyers and Gillespie, 1964). A pH Method for Continuous Monitori ng of Ecosy stem-Level Metabolism Beyers and Gillespie (1964) used a CO2 water titration method to obtain a pH-CO2 curve, but the method cannot provi de the relationship between pH and CO2 concentration in a continuous pH monitoring or changing alkalinity environment. In a theoretical study, Skirrow ( 1965) provided a thermodynamic relationship among pH, alkalinity, and total CO2 concentration ([TCO2]) in a carbonate system. The [TCO2] can be estimated by measuring water column pH and alkalinity and by plugging them into the thermodynamic equation. The water column [CO2] change occurring by photosynthesis and respirati on is calculated from [TCO2] by correcting [CO2] change by calcium carbonate precipitation and CO2 gas exchange across the air-water interface, which are the major contributing factors to water column [CO2] change (Smith, 1973). A conceptual diagram of this pH-alkalinity method to estimate ecosystem-level photosynthesis and respiration in a freshwat er aquatic microcosm is represented in Figure 1-2 using ener gy systems language (for hi story and syntax of energy systems language, see Brown, 2004). One of the uncertainties in the pH-alkalinity method fo r detecting photosynthetic and respiratory metabolisms of aquatic ecosyst ems is that a pH change may also be

PAGE 40

40 affected by metabolic processes other t han the inorganic carbon exchange, although typical diel patterns of water column pH show the most significant contributor to the pH change is the inorganic carbon exchange occurring by photosynthesis and respiration. The uncertain processes include nitrogen (NH4 + or NO3 -) assimilation, excess cation or anion flux, and organic acid excretion in undissociated forms (Geider and Osborne, 1992). Regarding the pH-alkalinity method, Ryt her (1956) pointed out that a relatively large metabolic [CO2] change may be necessary to detect the pH change in a freshwater or seawater that has buffering capacity. Even if the resolution of a pH meter is as high as three decimal places, the pH meter may not sensitively detect the pH change occurring by production of oligotrophic aquatic ecosystems in the field where pH may vary because of spatial heterogeneity and temporal pH change. The uncertainty in the estimation of CO2 gas diffusive flux across the air-water interface is mainly attributed to the gas transfer velocity, k, which is dependent on the hydrodynamics and wind speed (MacIntyre et al., 1995). In addition, Bade and Cole (2006) pointed out that CO2 mass transfer can be chemically enhanced in the high pH environment by the reaction of CO2 and OH-. On the other hand, MacIntyre et al. (1995) reported that surface slicks, especially in stagnant water, may reduce the CO2 gas diffusion. Chemical enhancement and surfac e slicks can also create uncertainty of CO2 gas exchange across the air-water interf ace in the laboratory open-top aquatic microcosms in a reduced wind and hydrodynamic environment. Simulation Models of Disturbance Modeling a system with computer simulati ons is another useful me thodology in studying ecological systems because the param eters, which are hardly controllable in the ecosystems, can be easily manipulated in the simulation process. Ecological

PAGE 41

41 systems can be modeled and simulated to suggest possible mechanisms of the observed phenomena, or to fore cast system behaviors under dist inct input conditions. Simulation with energy systems language prov ides information regarding storages or flows of ecological or general open systems under various input conditions (Montague, 2007). Developed by H. T. Odum, the energy systems language and its diagram can illustrate components and their in teractions in a system (Brown, 2004). Once the systems diagram is completed in the correct syntax of the energy systems language, the derivation of the equations governing a model system is fixed. Then the simulation of the model is conducted us ing a computer programming language. Regarding the methodology of ecological modeling, it has been controversial whether the goals of modeling, such as pr ediction of patterns, guides for future experimentation, and knowledge synthesis, are best achieved using a simple or complex model (Aumann, 2007) The simple models aggregat e a systems components into a functional unit, and focus on the essentia l part of the interactions in question. On the other hand, the complex models r epresent details of components and their interactions as specifically as possible. In fact, both the simple and complex models have strengths and weaknesses. While the simple models are easy to communicate and minimize the complexity caused by many parameterizations, the complex models better reflect the real ecosystem by addi ng more pathways and parameters (Van Nes and Scheffer, 2005). Several researchers have studied the effe ct of disturbance on ecosystem traits using simulation models. Odum and Odum (2000, p.241), in the mini model of destructive pulses, designated the roles of disturbance as draining accumulated assets

PAGE 42

42 and recycling materials for regeneration ( Figure 1-3). A lthough these roles are important aspects of disturbance, the model is too simplified in that disturbance regimes in nature are varied in intensity and a certai n disturbance does not necessarily drain all energy or material storages in a system. Kondoh (2001) used a numerical model to simulate the effect of productivity and di sturbance on species richness, and his model was represented by the following equation: )n...,,2,1i(pRpcp)Dm(p1Rpc dt dp1i 1k ikkii i 1k k ii i (1-1) where pi is the proportion of patches occupied by species i, the first term is the colonization, the second is the loss by loca l extinction, and the last is the loss by competitive exclusion. In Equation 1-1, increased produ ctivity enhances the colonization rate of all species by a constant, R, and disturbance increases the extinction rate of all species by a constant, D. Although this model includes productivity and disturbance as major factors determining s pecies richness, it does not consider the influence of disturbance on the productivity of the system. Also it assumes that disturbance increases extinction rates of all species even though the effect of disturbance may depend on the life hi stories or successional ages of species, as Sousa (1980) pointed out. As Kazanci (2007) poi nted out, a differential equation-based simulation may have difficulty in representing important flows and restrictions of energy and materials among system components, which can be shown in a descriptive systems diagram as a holistic view. The theory of al ternative stable state (Scheffer et al., 2001) may recommend a certain threshold of a dist urbance to be included in a model, above which a system abruptly changes to a new state.

PAGE 43

43 Research Plan This study was intended to investigate common patterns of ecosystem proce sses under distinct disturbance regimes. Thus, a di stinct water sample was used to set up each microcosm, and common patterns of dist urbance effects were analyzed across the different microcosms at the expense of replicat es of one sample for the test of statistical significance. To consider the effects of water motion disturbance, all the samples were collected only from one ecosystem type: lake. Eight samples of water and sediment were co llected in five lake s at different times over 16 months, and each sample was used to set up two microcosms, one in Tank A and the other in Tank B. Each microcosm was allowed to self-o rganize under the three consecutive periods: initial stabilization, disturbance, and postdisturbance. Each microcosm was divided into four sections (Sec 1, 2, 3, and 4) replicated during the initial stabilization period and treated with water motion disturbance regimes differing either intensity or frequency. Because response of an ecosystem may be different according to the timing and duration of dist urbance, several distinct durat ions of initial stabilization and disturbance periods were applied to different microcosms. The post-disturbance period lasted 15 days for all microcosms, assu ming a 15-day average turnover time of high-trophic-level species in the microcosms. The microcosms for the intensity test were designed to test how distinct disturbance intensities influence GPP and ER. Figure 1-4 shows the overall s cheme of water motion disturbance regimes applied to t he four sections of a microcosm during a disturbance period with time and disturbance pow er in the x and y axis, respectively. The distinct disturbance powers, or intensities, were applied to four sections of each microcosm with the same frequency and duration in the intensity tests. In each graph of

PAGE 44

44 Figure 1-4, the total area under the bars indic ate s the total disturbance energy, and the thickness of each bar is the duration of a disturbance. The GPP and ER of each section of a microcosm were estimated under the alte rnating light regimes of 12 h light and 12 h darkness throughout the experi ments. The mean GPP (MGPP) and mean ER (MER) values of the disturbance or post-distur bance period among the four sections were compared and plotted to analyze the relationships between MGPP or MER and disturbance intensities. In the analysis of data, MGPP and MER were calculated not only because they were proportional to the tota l amount of production and respiration but also because it was difficult to compare GPP or ER levels among different sections of a microcosm in the oscillating time series of GPP or ER. Eleven microcosms were tested under the different experim ental duration plans. In the frequency test, a distinct regime of distur bance frequency was applied to each section of a microcosm tank during t he disturbance period un der the alternating light regimes of 12 h light and 12 h darkness. It was tested how discrete or continuous disturbances influence GPP or ER levels of microcosms after the disturbance events. The overall scheme of disturbance regimes was to make the frequency different but total energy the same among the f our frequency regimes as shown in Figure 1-4. A continuous disturbance with low intensity was applied to Sec 1. Sec 2, 3, and 4 were applied with the same intensity but different frequencies and durations of disturbance to satisfy the same total energy among sections. The intensity of Sec 1 was adjusted to satisfy the same total energy as the other sections. The MGPP and MER values of the disturbance or post-disturbance period among the f our sections in a microcosm were

PAGE 45

45 analyzed in the same manner as the intensity tests. Fi ve microcosms were tested under the different experiment al duration plans. The effects of disturbance on ecosystem pr ocesses under the different maturities of ecosystems (maturity test) were test ed using the three paired microcosm sets to investigate whether ecosystems GPP and ER are less affected by disturbances as the systems age and how the responses of the syst ems to the disturbances are different under distinct maturities. The underlying premise was that initial components in a microcosm develop their interactions from dispersed individuals to a coherent system over time. The paired microcosms A and B we re initially replicated by cross-seeding, and the same disturbance regimes as in the in tensity tests were applied to microcosms A and B at different times since the initial se tup, one in the early and the other in the later days under the alternating light regime s of 12 h light and 12 h darkness throughout the experiments. The MGPP and MER patterns of the two tan ks were compared to show how systems maturities influence the relationship between disturbance regimes and MGPP or MER. In addition, resistance and resilience of GPP and ER of microcosms were analyzed to discuss the effect of maturity on ecosystem stability under disturbance. Pulsing patterns of GPP and ER were analyzed using microcosm data. Analysis was conducted by selecting data where distur bance period was equal to or greater than 10 days because it was impossible to estimate amplitudes or wavelengths of pulses in a shorter period. Wavelength ( ) and peak amplitude (PA) of the disturbance and postdisturbance periods were calculated from the transformed GPP time series to study how disturbance influences patterns of and PA over time.

PAGE 46

46 Finally, conceptual models of the microc osms with water motion disturbances were built using the energy systems language, and then simulated using the computer programming language to suggest possible me chanisms of the relationships between MGPP and disturbance regimes resulting from the microcosms. To investigate the energetic mechanisms of disturbance effects in the holistic view, biotic or abiotic components were aggregated into several functional compartments: producers, consumers, microbes, and nutrient s. Although the microbes were also part of the consumer group in an ecosystem, they were separated from the consumers to model their decomposing function. The produc ers and consumers were regarded as the actively reproducing entities by autocatalyt ic feedbacks. Because the purpose of the simulation model was to suggest possi ble hypotheses on the mechanisms of disturbance, a basic model was built based on the steady-state of the state variables over time. In the steady-sta te, the influence of disturbance is easily recognizable by filtering the other factors generating trends or short-term oscillations. Disturbance factors were further applied to the basic steady-state model under the following two hypotheses: Disturbances change the efficiencies of ener gy flow pathways in an ecosystem. That is, disturbances alter the energy pat hways related to the reproduction and death so that the new energy pathways altered by disturbances increase or decrease the reproduction and death rates with given resources. Although disturbances always increase the death rates of both producers and consumers, they increase, decrease, or do not affe ct the reproduction rates of the producers and consumers depending on the traits of micr ocosm samples, such as species combinations, interactions, and life histories. The changes of the efficiencies in energy flow pathways are dependent on the threshold of a disturbance power, which is determined by the vulnerability of a system to the disturbance. The effici encies of the energy pathways of producers and consumers change temporarily during the disturbance events below the threshold, whereas they change perm anently above the threshold.

PAGE 47

47 Figure 1-1. Diversity-disturbance relationship in the intermediate disturbance hypothesis. (Connell, 1978) High Low Diversity Frequent Disturbances Infrequent Large Disturbance small Soon After a disturbance long after

PAGE 48

48 Figure 1-2. Conceptual diagram of the pH-alka linity method in a freshwater aquatic microcosm. Light Autotrophs Heterotrophs Microbes [CA] pH Nutrient CO2 TCO2 CO2 HCO3 CO3 2 Ca 2 + CaCO3 H3O+ Freshwater aquatic microcosm CO2 Energy or material

PAGE 49

49 A B X Figure 1-3. Model of destructive pulses. A) Energy systems diagram. B) Simulation Result. (Odum and Odum, 2000, p.241) Source R I Materials Recycle Production Assets Destructive pulses Pulse applied Assets Materials Time

PAGE 50

50 A B Figure 1-4. Schematic plan of disturbance regime s. A) Intensity test. B) Frequency test. Sec 1 Sec 2 Sec 3 Sec 4 Sec 2 Sec 1 Sec 3 Sec 4 P P P P P P P P t t t t t t t t

PAGE 51

51 2 CHAPTER 2 METHODS The microcosm experiments were conduct ed over a period of 16 months to discover patterns of ecosystem processes responding to the controlled water motion disturbanc e regimes. Two aquatic microcosms were set up in Tanks A and B with water and sediment samples collected from a lake, and they were tested in a laboratory for 22 to 46 days according to a designated exper imental duration plan composed of initial stabilization, disturbance, and post-disturbance periods ( Table 2-1). This experimental practice was serially repeated eight times by collecting water and sediment samples from the five lakes in F lorida at different times over the 16 months ( Table 2-2). As a result, a total of 16 microcosms were tested i n the laboratory. Ea ch lake had the unique chemical or biological properties of wate r, such as nutrient concentrations, pH, alkalinity, turbidity (Secchi depth), and ch lorophyll concentration as described in Table 2-3. Each microcosm was used to test either the effects of disturbance intens ities or the effects of disturbance frequencies. For the microcosms S1, S2, and S6, a set of the two microcosms in Tanks A and B was used to test the disturbance effects under a distinct age (maturity) of a microcosm since the initial setup ( Table 2-1). D ata from the microcosms were analyzed separately and fi nally synthesized to uncover the common patterns and variations of ecosystem processes under different disturbance regimes. Computer simulation models followed the mi crocosm experiments. Once the data of microcosm tests were analyzed, simula tion models were built to suggest possible mechanisms of the resultant patterns. Afte r a basic steady-state model for a microcosm was established, the hypothesized disturbance factors were applied to the basic steadystate model to further dev elop the disturbance intensity and frequency models.

PAGE 52

52 Microcosm Experiments System Design Two identic al 60 cm 60 cm 38 cm (length width height) open-top glass microcosm tanks (A and B) were constructed. Figure 2-1 depicts the structure of a microcosm tank and the use of the two mi crocosm tanks in the tests. For the application of different distur bance regimes, the interior of each tank was divided into four equal sections (Sec 1, Sec 2, Sec 3, and Sec 4) by sealing acrylic panels on the bottom and sides of the tank with 100% silico ne. A 14 cm 20 cm rectangular hole was made in the middle of each acrylic panel to facilitate cross-seedi ng between sections, and a rubber-rimmed plastic lid was prepared to tightly close the hole and block material transport between the sections before the distur bance period. The exterior of each tank was enclosed by aluminum foil to maximize t he capture of light energy in the tank. A preliminary measurement of light revealed that the tank enclosed by aluminum foil exposes about 20% more available light to the bottom than the bare glass tank. Eight 20 W, 60 cm cool-white fluorescent bulbs were fixed 23 cm above the water surface of each microcosm. A wire was set up across the top of each tank section to hold a pH electrode submersed in the wa ter for long-term monitoring. A pump generating a water motion was attached on one side in each section of a tank. The attachment point of a pump was centered ve rtically and offset from the center horizontally about 7 cm to generate a horizont al circular motion. A pH electrode (Oakton double-junction jellfilled pH electrode, pH 0) submersed 10 cm from the water surface in each section, was connected to a pH meter (Artisan PH2000 from APT Instruments, resolution 0.01 pH, accuracy .02 pH), and the pH readings were transported to the comput er and logged every minute by computer software.

PAGE 53

53 Sampling Each water and sediment sample was colle c ted from one of the five lakes within a 30-minute driving distance from the microcosm laboratory at the Univ ersity of Florida ( Table 2-2). Water was collected in 20 L plas ti c buckets or 20 L plastic bottles within 5 m littoral zone from the s horeline of the sampling poin t in each lake. Water and sediment samples for a set of microcosms (A and B) were collected from one location in a lake. Sediments were collected from t he top 10 cm layer to minimize the anaerobic metabolism of microcosms. The collected sediments were filtered through a 2 mm mesh to maximize the homogeneity of sediment s across the sections of a microcosm. Once the water and sediment samples were delivered to the l aboratory, the water sample (120 L for each tank) was first care fully poured into the microcosm Tanks A and B, and then the same amount of sediment (1 L for each secti on of a tank) was carefully stabilized on the bottom of each section, while minimizing suspension of the sediment. Each tank was filled with microcosm sample water to a height of 33 cm from the bottom. Maintenance and Measurements Microcosms were set up in a dark room of the laboratory. T he fluorescent light was turned on and off by automatic digital time rs with the alternating light regimes of 12 h light and 12 h darkness (light 6:00:00, dark 18:00next day 6:00). Fluorescent bulbs were periodic ally replaced with new ones to maintain the photosynthetically active radiation (PAR: measured by the LI-190 quantum sensor and LI-1400 data logger from LI-COR) at 150 molm-2s-1 on top of each tank. Because water in each section of a microcosm tank evaporated at a rate of approximately 1 L a week, the water level was maintained by refilling microcosm tan ks with deionized water filtered through the

PAGE 54

54 Barnstead NANOpure Infinity Water Purification System every week. The room temperature was maintained at 24C by a thermostat. The eight pH meters were generally calibra ted every week at the three pH points (pH 4.01, 7.00, and 10.01 standar d solutions from Hach), but were calibrated more frequently when pH electrodes were submers ed in a productive water that possibly affects the sensitivity of the electrodes. The eight pH electrodes were rinsed by deionized water in a squeeze bottle once a day to minimize the organic coating on the pH sensor glass. Total alkalinity ([TA]) was measured once a week using the end point titration method at pH 4.5 (Palmer, 1992). A digital titrat or (1.25 L/digit, Hach) and 0.2 N H2SO4 titrant were used for the alkalinity measurements. For a measurement of alkalinity, either 100 ml or 50 ml water sa mple was taken from each microcosm section according to the alkalinity level (100 ml for an alkalinity level lower than 50 mg/L CaCO3 and 50 ml for an alkalinity level higher than 50 mg/L CaCO3). In each microcosm test, pH and alkalinit y were monitored through the successive periods of initial stabilizati on, disturbance, and post-distur bance. The rectangular holes on the acrylic panels were open during the init ial stabilization period to maximize the homogeneous condition across the four sections in each tank. The holes were blocked by the watertight rubber-rimmed lids before t he disturbance period, and they remained in place until the end of the post-disturbance period. During the initial stabilization period, microcosms were additionally cross-seeded several times by moving approximately 300 ml water betw een sections each time. During the disturbance period, the pumps were connected to the automatic timers and operated in accordance with the

PAGE 55

55 designated disturbance regimes. After the disturbance period, microcosms were not altered or manipulated except to take measurements and to add make-up water. After the initial construction of the two gl ass microcosm tanks, they were filled with tap water to minimize the odor or chemicals from silicone. The tap water was replaced several times for two weeks before the first microcosm te sts. Once each set of microcosms (A and B) was terminated at t he end of the post-dist urbance period, the microcosm tanks were emptied, cleaned with tap water, and dried for the setup of the next microcosms. The sides and bottom of each tank were carefully cleaned with an aquarium scrub. The pumps and plastic lid s were also cleaned with tap water and a brush. Disturbance Regimes The pump (Aquarium Systems Mini-Jet 404) generated a horizontal circular water motion in the water column by ejecting the absorbed water from the surrounding water column. The dischar ge rate of the pump was mechanically controllable at six levels. The power required by a pump can be calcul ated from the following equation (Simon, 1986): HQP (2-1) where is the unit weight of the fluid, Q is the discharge rate, and H is the total dynamic head. When disturbance power is defined as t he power of the discharged water directly influencing the water motion, it is proportional to the discharge rate regardless of the pressure exerted on the pump head. The dist urbance power, or intensity, was defined as a relative value by omitting the unit (ml/s ) from a discharge rate of a pump. Because the six discharge rates were slightly inconsistent between pumps with 1 ml/s

PAGE 56

56 differences, they were measured before t he experiments. The average disturbance powers (intensities) of the six pumping le vels were 8, 28, 40, 50, 56, and 62. For the intensity tests, the four disturbance intensities, 0, 28, 50, and 62, were applied to Sec 1, 2, 3, and 4 of a tank, resp ectively. As an except ion, the microcosms S2-A and S2-B were applied with intensitie s 0, 40, 62, and 113 by combining two pumps in each section to test a microc osms response to the higher disturbance intensities. A pump was also installed in Sec 1 (intensity 0), but was not operated. Sec 1 was used as an undisturbed reference system in the intensity tests. Pumps were operated either for an hour ( 11:00:00) a day for 10 cons ecutive days or two hours (11:00:00) a day for five consecutiv e days depending on the duration plans. Table 2-4 shows the disturbance regi mes applied to the microcosms. For the frequency tests, different frequen cy regimes wer e applied to the four sections of a tank for five consecutiv e days, while maintaining the same total disturbance energy among sections. In Sec 1, a pump was continuously operated for 105 hours with an intensity of 8 (regime 1). Sec 2 was applied with an intensity of 56 for 30 minutes every four hours (2:00:30, 6:00:30, 10:00:30, 14:00:30, 18:00 18:30, and 22:00:30) (regime 2). Sec 3 was also applied with an intensity of 56 for an hour every eight hours (2:00:00, 10:00: 00, and 18:00:00) (regime 3). The intensity of 56 was applied to Sec 4 for three hours a day (10: 00:00) (regime 4). The total disturbance energy was calculated by multiplying intensity (power) and total duration (hour) of disturbances as a unitless relative value. The purpose of the maturity test was to investigate how the responses of GPP and ER of microcosms to disturbance regimes are different according to the distinct age of a

PAGE 57

57 system. The premise of the maturity test was that initial components in a microcosm self-organize and develop their interactions from dispersed individuals to a coherent system over time. For the maturity tests, two replicated microcosms A and B were used as a pair in S1, S2, and S6. The same distur bance regimes as the intensity tests, for an hour (11:00:00) a day for 10 consecutive days, were applied to both microcosms, but at the different periods since the initial setup: Day 6 for microcosm A, and Day 21 for microcosm B. Data Processing and Analysis Data Processing: From Raw Data to Gross Primar y Productivity (GPP) and Ecosystem Respiration Rate (ER) Once pH and alkalinity data were obtained, they were arranged in a spreadsheet in five-minute intervals. To tal alkalinity levels ([TA]) between two measurement points were estimated by linear interpolation. U nder the assumption that alkalinity mainly results from carbonate and bicarbonate ions in the freshwater condition, the measured [TA] (mg/L CaCO3) was converted into carbonate alkalinity ([CA]) (meq) using the following Equations 2-2 and 2-3: ]H[]OH[]CO[2]HCO[]TA[2 3 3 (2-2) ]CO[2]HCO[]CA[2 3 3 (2-3) Because [CA] is defined as Equation 2-3, [C A] was calculated from the measured [TA] and pH using Equation 2-2. The alkalinity unit of mg/L CaCO3 was converted into meq by dividing 50. The thermodynamic relationship among pH, alkalinity, and total CO2 (Skirrow, 1965) was used to estimate the total CO2 concentration ([TCO2]) in the water column as the following equation:

PAGE 58

58 H 2 L1 H H 2 2a 2K 1 K a a K 1 [CA]] [TCO (2-4) where [CA] is carbonate alkalinity, aH is hydrogen ion activity, K'L1 is Lymans first apparent dissociation constant, and K'2 is second apparent dissociation constant. In Equation 2-4, aH is typically substituted by 10-pH assuming the proximity to the ideal behavior of hydrogen ions in the water column. The theoretical pH-[TCO2] relationship at 25C and zero salinity is represented in Figure 2-2. [TCO2] is the sum of the three carbon compounds as the following equation: ]CO[]CO[]HCO[]TCO[2 2 3 3 2 (2-5) In the water column, [TCO2] changes are driven mainly by metabolic processes of photosynthesis and respiration (PCO2), CaCO3 precipitation or dissolution (CCO2), and CO2 gas exchange across the air-water interface (GCO2) (Smith, 1973). Thus, [PCO2] can be calculated by correcting [CCO2] and [GCO2] from [TCO2] by the following equation: ]GCO[]CCO[]PCO[]TCO[2 2 2 2 (2-6) [CCO2] is dictated by one of the following Equations 2-7, 2-8, 2-9, and 2-10 related to carbonate or bicarbonate ions (Smith and Key, 1975): 2 3 2 3COCa)s(CaCO (2-7) 23 22 3)HCO(CaOHCO)s(CaCO (2-8) )s(CaCO COCa3 2 3 2 (2-9) H)s(CaCO HCO Ca3 3 2 (2-10)

PAGE 59

59 In any reaction of Equations 2-7, 2-8, 2-9, and 2-10, wa ter column gains or loses [CCO2] by half of increased or decr eased total alkalinity. Thus, [CCO2] was calculated by the following equation: ]TA[ 2 1 ]CCO[2 (2-11) The four time series of water column [PCO2] of the four sections of each microcosm were obtained by correcting [CCO2] from [TCO2] using Equation 2-6. But [GCO2] was not corrected because of the uncertainty of the CO2 gas transfer velocity. Instead, the influence of [GCO2] on the sensitivity of the resultant patterns according to various gas transfer velocities was analyzed separately. GPP was defined as a rate of ecosyst em-level photosynthetic gain of CO2-C, and ER was defined as a rate of ecosystem (pla nt, animal, and microbial) respiratory loss of CO2-C (Chapin et al., 2006). The [PCO2] generally showed diel fl uctuations: increasing during a dark period and decreasing dur ing a light period as depicted in Figure 2-3. Thus, positive slopes indicate the nighttime respiration rates (ERn), while negative slopes show the difference between daytime respiration rate and gross primary productivity (ERd-GPP). The slopes of [PCO2] were calculated by dividing the difference between adjacent maximum and minimum peak values by time duration of each photoperiod (12 h). The peaks o ccurred around 6 am or 6 pm but not exactly at those times, so that the maximum and minimum [PCO2] values were selected within h range of 6 am and 6 pm, respectively. Bec ause the nighttime and daytime ER values may not be identical with each other in any two adjacent [PCO2] slopes, the daytime ER on Day i+1 (ERi+1,d) was estimated by taking an average of the two nighttime ER values on Day i and Day i+1 (ERi,n and ERi+1,n) as the following Equations 2-12 and 2-13:

PAGE 60

60 2 ERER ERn,1ini, d1,i (2-12) 2 )ER3ER( ER ERERn,1i n,i n,1id,1i1i (2-13) })ERGPP({ 2 )ERER( })ERGPP({ER GPPd,1i n,1in,i d,1i d,1i1i (2-14) The daily GPP for each section of a microcosm was calculated us ing Equation 2-14. The GPP and ER were expressed as the unit of mM-CO2/day. Although a calculated GPP has a unit of mM-CO2 per 12 hours, it was expressed as mM-CO2/day because the nighttime GPP was assumed to be zero. The time series of GPP, ER, and GPP/ER were plotted by calculating GPP and ER for each day. Ranges of pH and Alkalinity The pH and alkalinity pairs in 50-minute intervals from the eight microcosm sets were plotted with pH in x-axis and [CA] in y-axis to show the range of pH and alkalinity in each microcosm. An equilibrium line of pH and alkalinity satisfying the dynamic equilibrium of CO2 gas exchange across the air-water interface was also plotted on the pH-alkalinity graph. Considering that CO2 is a slightly soluble gas in water where the transfer rate is determined, CO2 gas diffusive flux, F, acro ss the air-water interface is represented by the following equa tion (MacIntyre et al., 1995): )CC(kFwa (2-15) where k is gas transfer velocity (cm/h), is solubility coefficient (molL-1atm-1), Ca is gas concentration in the air (ppm), and Cw is gas concentration in the water (mmol/L). The solubility coefficient is dependent on temperature and salinity (Weiss, 1974). The Ca

PAGE 61

61 in the microcosm room was approximately 440 ppm. The Cw of CO2 gas was calculated using the following thermody namic equation (Skirrow, 1965): H 2 1L H wa K2 1K a ]CA[C (2-16) When gas exchange across the air-water inte rface is predicted using Equation 2-15, dynamic equilibrium of CO2 gas exchange between the air and water is established if CO2 gas diffusive flux, F, equal s zero. In the equilibrium condition under the constant Ca, pH-alkalinity relationship was estab lished using the following equation: H H 2 1L aa a K2 1K C]CA[ (2-17) The pH-alkalinity data pairs of the eight microcosm sets and the equilibrium line showed the direction of CO2 gas diffusion for each microcosm. Analysis of Disturbance Intensity and Frequency Effects The mean GPP (MGPP) and mean ER (MER) during the disturbance and postdisturbance periods were calcul ated from the GPP time seri es of each section of each microcosm, and the subsequent relationship between MGPP and disturbance regime was plotted. This procedure is exemplified in Figure 2-4. The relationship between MGPP and disturbance regime could change acco rding to the time range where the MGPP values were calculated. The MGPP during a time range from t1 to t2 was calculated using the following equation: n )i(GPP Mn ti ]tt[GPP1 21 (2-18)

PAGE 62

62 where n is the number of days during a time range of t1t2, and GPP(i) is a GPP value at time i. Likewise, MER could be defined during a time range t1t2, if necessary. The time series of MGPP[t1t2] under the fixed t1 were plotted to determine whether the relationship between MGPP and disturbance regime was maintained throughout the disturbance and post-disturbance periods. In the time series graphs of MGPP[t1t2] (Appendix A), t1 was set as the beginning of eit her the disturbance or post-disturbance period. If the time series of MGPP[t1t2] between two sections do not intersect, then one section maintains constantly higher MGPP or MER than the other during the designated period. Based on the MGPP[t1t2] time series plots (Appendix A), all possible relationships between MGPP and disturbance regime were plotted according to the range of time [t1t2] in which MGPP values were calculated. Analysis of GPP and ER Pulsing Pattern The pulsing patterns of GPP and ER we re analyzed by detrending the GPP and ER time series of each microcosm. The modifying procedure is illustrated in Figure 2-5. Because a GPP time series generally in cluded a trend during the disturbance and postdisturbance periods, a GPP at each time (GPPt) was first transformed by subtracting the value on the trend line (regressi on line) at that time (GPPt L). Then, the difference between the GPPt and GPPt L was divided by GPPt L to eliminate the e ffect of GPP level on the pulsing amplitude under the assumpti on that the amplit ude of short-term oscillation becomes higher as the level of a trend line increases. That is, GPP values in each time series of a section were transformed into GPPP using the following equation: 1 GPP GPP GPPL t t P t (2-19)

PAGE 63

63 where GPPt P is a transformed GPP point at time t for the GPP pulsing analysis. The same method was applied to ERt P. The pulsing patterns of GPPP and ERP were analyzed in terms of wavelength ( ) and peak amplitude (PA), wh ich are illustrated in Figure 2-6. The was defined as a time difference between adjacent two maxima or two minima in t he pulsing pattern. The d (disturbance period ) and p (post-disturbance ) were calculated by averaging all values identified in the disturbance and postdisturbance period, respectively. The PA was defined as a difference between adjac ent maximum and minimum peak values. The PAd (disturbance period PA) and PAp (post-disturbance PA) were obtained by subtracting the average of all minimum peak values from the average of all maximum peak values within the disturbance and postdisturbance period, respectively. Analysis of Resistance and Resilience Resistance and resilience of each disturbed microcosm section were calculated in terms of GPP and ER using t he microcosm data obtained from the intensity tests. Microcosm data obtained fr om the frequency tests were not analyzed because there was no reference section where no distur bance was applied. Although several researchers used different indices for quantifying resistance and resilience (for comparison, see Orwin and Wardle, 2004), new indices were defined to calculate resistance and resilience of the GPP or ER time series of microcosms in this study. Resistance was used as an indicator of how much GPP or ER of a disturbed section (Sec 2, 3, or 4) deviates less from that of the reference system, Sec 1, during the disturbance period in each microcosm. Resistance (RS) of Se c 2, 3, or 4 was calculated using the follo wing equation for GPP:

PAGE 64

64 m 1 GPP GPP 1 RSm i )1Sec(i )xSec(i )xSec( (2-20) where GPPi(Sec x) is GPP at time i in Sec x (x=2, 3, or 4), and m is the number of days in the disturbance period. The maximum RS(Sec x) is 1 when GPP(Sec x) does not deviate from GPP(Sec 1), and RS(Sec x) decreases as the GPP(Sec x) deviates from GPP(Sec 1). RS(Sec x) for ER was calculated using the same method. Resilience was used as an indicator of how much GPP or ER, once deviated from that of reference section during the disturbance period, re turns close to the reference section during the post-distur bance period. Resilience (RL) of Sec 2, 3, or 4 was calculated using the follo wing equation for GPP: )xSec( n j )1Sec(j )xSec(j )xSec(RS1 n 1 GPP GPP 1 RL (2-21) where GPPj(Sec x) is GPP at time j in Sec x (x=2, 3, or 4), and n is the number of days during the post-disturbance period. The maximum RL(Sec x) is 1 when post-disturbance GPP(Sec x) returns to the reference GPP(Sec 1), and RL(Sec x) decreases as the GPP(Sec x) deviates from GPP(Sec 1) during the post-dist urbance period. RL(Sec x) for ER was calculated using the same method. Analysis of Uncertainty in the Resultant Patterns The uncertainty of the MGPP-disturbance relationships resulting from [GCO2] correction was analyzed by comparing MGPP-disturbance relationships at different k values. The CO2 gas transfer velocity, k (cm/h), was estimated from previous studies at low wind speed conditions. At low wind speed (U10) less than 3 m/s, k generally ranged

PAGE 65

65 0 to 4 cm/h (Borges et al., 2004; Cole and Caraco, 1998; Crusius and Wanninkhof, 2003; MacIntyre et al., 1995). The sensitivity of MGPP-disturbance relationships for the full post-disturbance period was tested with the k (cm/h) values at 0, 1, 2, and 4 for all microcosms. When k equals zero, [GCO2] was also zero. Additionally, to examine the cause of the changes in the MGPP-disturbance relationships under the [GCO2] corrections, the sensitivity of [PCO2] and GPP time series of Sec 1 under the different k values was tested by selecting two representative microcosms: one having the most significant change in the MGPP-disturbance relationship by k, and the other showing the least change. The uncertainty of the MGPP-disturbance relationships resulting from potential instrumental errors was also analyzed. To determine potential errors occurring from pH measurements, the relationships between pH and the theoretical [TCO2] change by 0.01 pH error ([TCO2]/( 0.01 pH)) were plotted under several different alkalinity levels. From these relationships between pH and [TCO2]/( 0.01 pH), the possible maximum and minimum [TCO2]/( 0.01 pH) values were determined within the pH and alkalinity range in each microcos m. The maximum and minimum MGPP differences between sections in each microcosm we re calculated to evaluate how much MGPPdisturbance relationships are reliable under t he potential instrumental errors across the microcosms. Simulation Models Models of microcosms and disturbances were built using the energy systems language, and simulated using the programming package, R (available at the R Project for Statistical Computing, www.r-project.org). The steady-state of state variables was assumed in the calibration of the basic m odel of a microcosm without disturbance, and

PAGE 66

66 this basic model was modified by addition of disturbance factors to test the effects of intensity and frequency of dist urbance. The models were simulated in five-minute intervals for 50 days. Model of a Microcosm The basic steady-state model of a mi crocosm was repres ented with aggregated components and their interactions. T he model was composed of producers, consumers, decomposers, and available nutrient and light energy source as shown in Figure 2-7. The light energy was represented as a step func tion with the alternating 12 h light and 12 h darkness. Producers (P ro) and consumers (Con) were assumed to reproduce by first-order aut ocatalytic processes (J3 and J5), and their death rates (J6 and J7) were proportional to their total bioma ss, Pro and Con (mg-C). It was assumed that the microcosm was phosph orus-limited. The nutrient (phosphorus) was recycled by microbial decomposition of dead organic matters (ProD and ConD) and stored in the water column (Nut) for further primary production. In the m odel, the total amount of the nutrient (TN) was conserved, and part of it was st ored either in the organic matters (Pro, Con, ProD, ConD) or water column (N ut). The reproduction flow of producers, J2, indicates the GPP of the microcosm model. The storages, flows, and some parameters, such as P/C (phosphorus /carbon) for producers (fP) and P/C for consumers (fC), were estimated from the microcosm re sults, literature, or by gue ss. The coefficient involved in each flow was obtained by back-calculating in the equations using the numbers from storages and flows (Odum and Odum, 2000). Simulation of Disturbance Once the basic microcosm model was bu ilt and the steady-sta te of the state variables was identified in the computer simulation, disturbance factors were embedded

PAGE 67

67 in the model. Following the two hypotheses pr ovided in the Research Plan, all possible changes of coefficients (ki) of the reproduction or deat h flows of producers and consumers were simulated. The changes of the coefficients were calculated by the following equation: id ik)fp1(k (2-22) where k'i is a new coefficient after disturbance, p is the power (i ntensity) of the disturbance, fd is the disturbance factor determined by the vulnerability of a system to a disturbance and the threshold of p, and ki is the coefficient involved in Ji. Equation 2-22 implies that the effect of disturbance is represented as a percent increas e or decrease of each coefficient. If pfd is positive under a disturban ce, the disturbance increases the coefficient, ki, by (pfd)%. As hypothesized in the Research Plan, the signs of pfd for k6 and k7, coefficients for death rates of producers (J6) and consumers (J7), were set as positive. But the coefficients related to the reproducti on rates of producers and consumers, k2-k3 and k4-k5, could have nine combinations of the pfd signs made of positive (+), negative (-), or no effect (0). Figure 2-8 illustrates how the fd changes under the diff erent disturbance powers (intensities) above or below a threshold over time. The fd was assumed to be constant only during the disturbance events when the disturbance power was less than a threshold. The fd was assumed to increase during the disturbance ev ents, and the increased fd values remained in the rest of t he experimental per iod unless another disturbance event occurred when the disturbance power wa s equal to or greater than the threshold. Thus, a unit value of t he disturbance factor for five minutes (fd-5) was assigned in each simulation. The fd-5 was the same as fd when the disturbance power

PAGE 68

68 was less than a threshold. But fd was calculated by adding fd-5 every five minutes when the disturbance power (p) was equal to or great er than the threshold. Two distinct fd-5 values, one for a p level above the threshold and one for a p level below the threshold, were applied to the reproduction and death rates in each simulation. In the intensity tests, five power levels (intensities) of 0, 1, 2, 3, and 4 were simulated with two different fd-5 values for the p levels above and below the assigned power threshold. The disturbances were applied for an hour ( 11:00:00) during 10 consecutive days as in the microcosm tests. In the frequency tests, the same disturbance frequency and duration regimes as in the micr ocosm tests were applied with the power level 1 for regime 1, and the power level 7 for regimes 2, 3, and 4. For comparison, the reference (power level 0) wa s also simulated. The disturbance period was during Day 20 in the simulation of intensity tests and Day 20 in the simulation of frequency tests. The postdisturbance period was 15 days after the disturbance period. Validation of Models The sign combinations of the pfd for k2-k3 and k4-k5, and the values of fd-5 and power threshold were manipulated in the simulations to generate possible MGPP-disturbance relationships. The results of simulations were compared with the resultant MGPP-disturbance relationships of the 16 microcosms. Possible mechanisms of resultant relationships based upon the hypotheses of the simulation models were suggested using the param eters, such as pfd signs, fd-5 values, and power threshold.

PAGE 69

69 Table 2-1. Experimental durat ion plans and test types of microcosms. Microcosm Tank Experimental duration plan (days) (initial stabilizationdisturbance-post disturbance) Test type S1 A B 5-10-15 20-10-15 Intensity Intensity maturity S2 A B 5-10-15 20-10-15 Intensity Intensity maturity S3 A B 1-5-15 1-5-15 Intensity Frequency S4 A B 1-5-15 1-5-15 Intensity Frequency S5 A B 5-10-15 5-5-15 Intensity Frequency S6 A B 5-10-15 20-10-15 Intensity Intensity maturity S7 A B 5-5-15 5-5-15 Intensity Frequency S8 A B 5-5-15 5-5-15 Intensity Frequency Table 2-2. Sampling site information. Sample Site (Lake) Location Sampling date Temperature (C) (Air) Microcosm 1 Newnan +29'11", -82'25" April 30, 2009 28 S1 2 Alice +29'36", -82'45" August 17, 2009 28 S2 3 Santa Fe +29'49", -82'48" October 5, 2009 27 S3 4 Wauberg +29'06", -82'13" February 20, 2010 22 S4 5 Orange +29'16", -82'59" March 16, 2010 15 S5 6 Alice +29'36", -82'45" April 17, 2010 24 S6 7 Alice +29'36", -82'45" June 20, 2010 29 S7 8 Newnan +29'11", -82'25" July 17, 2010 28 S8

PAGE 70

70 Table 2-3. Chemical and biological properties of the sample lakes. (Florida LAKEWATCH, 2005) Lake Properties Alice Newnan Orange Santa Fe Wauberg Total phosphorus concentration ( g/L) 454 121 68 11 120 Total nitrogen concentration ( g/L) 627 3362 1698 449 1805 pH 7.5 6.9 7.1 5.9 7.6 Alkalinity (mg/L CaCO3) 85.0 12.4 19.4 1.8 19.7 Secchi depth (m) 1.5 0.3 0.8 2.3 0.6 Chlorophyll concentration ( g/L) 13 208 51 7 92 Table 2-4. Disturbance regimes applied to microcosms. Tank section Intensity Frequency Duration Total energy For intensity tests 1 0 N/A N/A 0 2 28 1 day 1 hour (10-day disturbance) 2 hours (5-day disturbance) 280 3 50 1 day 1 hour (10-day disturbance) 2 hours (5-day disturbance) 500 4 62 1 day 1 hour (10-day disturbance) 2 hours (5-day disturbance) 620 For frequency tests 1 8 continuous 105 hours 840 2 56 4 hours 30 minutes (5-day disturbance) 840 3 56 8 hours 1 hour (5-day disturbance) 840 4 56 24 hours 3 hours (5-day disturbance) 840 S2-A and S2-B were applied with intensities 0, 40, 62, and 113, so the total energy was 0, 400, 620, and 1130.

PAGE 71

71 60cm 60cm 38cm water level 33cm 18cm Enclosed by aluminum foil to pH meter & datalogging by computer software pump S1-AS1-BS2-AS2-BS3-AS3-BS4-AS4-B S5-AS5-BS6-AS6-BS7-AS7-BS8-AS8-BApril 30, 2009 June 15, 2009 August 17, 2009 October 1, 2009 October 5, 2009 October 27, 2009 February 20, 2010 March 14, 2010 March 16, 2010 April 16, 2010 April 17, 2010 June 2, 2010 June 20, 2010 July 16, 2010 July 17, 2010 August 12, 2010 6mm glass 6mm acrylic panel *A pump and a pH electrode were installed in each section 20W, 60cm cool-white fluorescent bulb 20cm 14cm Figure 2-1. Microcosm tank design and plan for each microcosm pair. Sec 1 Sec 2 Sec 3 Sec 4

PAGE 72

72 0 1 2 3 4 5 5678910 pH[TCO2] (mM) [CA] 0.4 meq [CA] 0.8 meq [CA] 1.2 meq [CA] 1.6 meq [CA] 2.0 meq Figure 2-2. Theoretical relationship between pH and [TCO2] (25C, zero salinity). Figure 2-3. [PCO2] diel pattern and calculation of GPP and ER. Dark Dark Dark Light Light Light ERi,n ERi+1,n-(GPP-ER)i+1,d -(GPP-ER)i,d 6am 6pm 6am 6pm 6am Day i Day i+1 Time (Day) [PCO2] (mM)

PAGE 73

73 0.00 0.10 0.20 0 5 10 15 20 25 30Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 0.03 0.04 0.05 0.06 0.07 691215Time (Day)MGPP[6-t] (mM-CO2/day) 0.03 0.05 0.07 0.09 1619222528Time (Day)MGPP[16-t] (mM-CO2/day) 0.03 0.06 0.09 0204060Disturbance intensity MGPP (mM-CO2/day) [6-15] [16-17] [16-22] [16-30] Figure 2-4. Representation of MGPP-disturbance relationship from GPP time series. Intensity (0) (28) (50) (62) Disturbance Post-disturbance average Time range

PAGE 74

74 Disturbance Post-disturbance Time (Day) Regression line (GPPL) Transformation to residual plot = (GPPtGPPL t) To GPP pulsing (GPPP) = (GPPtGPPL t)/(GPPL t) 0 0 0 Figure 2-5. Procedure to tr ansform GPP to GPPP. GPP (mM-CO2/day) GPP-GPP L (mM-CO2/day) GPP P

PAGE 75

75 Wavelength ( 1) 2 3 GPPPTime (Day) 0Maximum Minimum M1M2M3m1m2 nn i i b m a M PAb i i a i i; where n is the number of ; where a is the number of M and b is the number of m Figure 2-6. Wavelength ( ) and peak amplitu de (PA) of GPPP.

PAGE 76

76 Figure 2-7. Energy systems diagram and equations of the basic microcosm model. Light L R J1 J2 J3 J4 J5 J6 J7 J8 J9 Nut Pro Con ConD ProD Equations d(Pro) = J2 J3 J4 J6 d(Con) = J4 J5 J7 d(ProD) = J6 J8 d(ConD) = J7 J9 TN = Nut + fP(Pro + ProD) + fC(Con + ConD)

PAGE 77

77 Figure 2-8. Change of fd over the experimental peri ods according to power (p). Experiental periods Initial stabilization Disturbance Post-disturbance Disturbances (p < threshold) (p threshold) fd Time (Day) 0 0

PAGE 78

78 3 CHAPTER 3 RESULTS Structures and Processes of Microcosms Microcosm Features During the initial microcosm setup, part of the sediment was mixed with water and suspended in the water column for several days. It took three to five days for the sediments suspended in the water column to st abilize since the initia l microcosm setup. Once the sediments stabilized on the botto m of each tank, the bottom was clearly visible from the top of each tank. The bottom of the microcosms, S1 and S8, which were set up using water samples from Lake Newnan, however, was not visible after the sediments stabilized because the in itial algal density was high. The highest-trophic-level s pecies observed in the microcosms were snails or freshwater shrimps with body lengths of less than 1 cm at the end of the experiments. The number of individuals of the highesttrophic-level species in each section was generally three to five, but more than 20 freshwater shri mps were observed in each section of S2. Exceptionally, a fish with a body length of less than 2 cm was seen in Sec 4 of S4-A at the end of the experiment. Tw o or three duckweed individuals were observed on the surface of the water in eac h section of S3 and S4 but they did not outgrow and reproduce more than five individuals. Algae generally grew on the bottom, wall, water column, and water surface of t he microcosms. The microcosms, initially constructed with water and sediment sample s, generally culminated in the combination of algal clusters (green, yellow, or bla ck) and heterotrophic spec ies at the end of the experiments.

PAGE 79

79 Ranges of pH and Alkalinity Figure 3-1 shows the distribution of the pairs of pH and al kalinity data collected with 50-minute intervals during the experimen tal periods for the eight microcosm sets (S1S8). The pH and alkalinity of the eight microcosm sets were widely distributed in the ranges of pH from 5.1 to 10.0 and alkalinity fr om 0.06 to 1.84 meq. The ranges of pH and alkalinity of the microcosms were di stinct according to the water samples and disturbance regimes. During the experimental periods, alkalinity was not constant in all microcosms. The alkalinity in S1 and S2 asymptotically increased while it linearly increased in S8. The alkalinity change in S1-B was most significant among microcosms with the range of 0.46.60 meq for 45 days. In S4, S5, and S7 alkalinity linearly decreased. The alkalinity of S3 was constant with a r ange of fluctuations (0 .072.004 meq). The alkalinity of S6 fluctuated, and the fluctuat ing patterns over time were different among the microcosm sections that were under distinct disturbance regimes. The range of pH also differed among microcosms ( Table 3-1). S3, sampled from Lake Santa Fe, had the pH range of 5.3.8 in S3-A and 5.1.7 in S3-B. The pH temporarily increased in diel fluctuations up to the range of 9 in S1, S6, S7, and S8 that were sampled either from Lake Alic e or Lake Newnan. Each section of each microcosm had a distinct pH tr ajectory over time under a di stinct disturbance regime. For example, Sec 2, 3, and 4 of S8-A had diel pH fluctuations of up to 1.0, 1.5, and 1.6 difference between the daily maximum and mi nimum pH values, respectively, after Day 6, which was the first day that the microcosm was disturbed by water motions, whereas Sec 1 had the maximum daily pH difference of 0.6 as shown in Figure 3-2.

PAGE 80

80 GPP and ER of Microcosms The time series of GPP and GPP/ER of the 16 microcosms are presented in Appendix A. The gross primar y productivity (GPP) and ecosyst em respiration rate (ER) oscillated during the experimental periods wi th the range of average wavelengths of 2.8.0 days in the microcosm sections. The va rious amplitudes of diel fluctuations of [PCO2] generated distinct GPP values in each mi crocosm. The extr eme GPP values of minimum 0.01 and maximum 0.92 mM-CO2/day were observed in S6-B for 45 days as shown in Table 3-1. The GPP ranges were diffe rent even among the microcosms sampled at the same location if the sampling times were different. For instance, S2-B and S6-B, sampled from the same location in Lake Alice at different times, had the maximum GPP values of 0.07 and 0.29 mM-CO2/day in the undisturbed Sec 1 for 45 days (Figures A-4 and A-12). The correlation between GPP and ER was quantified by calcul ating Pearsons correlation coefficient (r) between GPP and ER for each section of each microcosm during the disturbance and post-disturbance periods as shown in Table 3-2. The r values of the post-disturbance period in mi crocosm sections ranged from 0.60 to 1.00 except those in Sec 3 (r=0.22) and Sec 4 (r=0.46) of S3-A. T he r values of the disturbance period were not consistently hi gh across the sections of the microcosms, especially in S1-A, S2-A, S2-B, S3-B, S7-B, and S8-B, although S1-B, S4-A, S5-A, S6-B, S7-A, and S8-A showed strong correla tions between GPP and ER with the range of r from 0.84 to 1.00. In S1, S2, and S6, where the sa me disturbance regimes were applied to the two replicated microcosms A and B during Day 6 and Day 21, respectively, the correlation between GPP and ER of microcosm B was higher than that of microcosm A in each microcosm se ction during the di sturbance period.

PAGE 81

81 The ecosystem-level ratio of GPP and ER (GPP/ER) has been used as an indicator of the autotrophic or heterotrophic character of an ecosystem (Odum, 1969). In the microcosm tests, the GPP/ER tended to converge to 1 over time regardless of the autotrophic or heterotrophic status of systems during the disturbance period as shown in Table 3-3. That is, the GPP/ ER of the autotrophic system s (A) during the disturbance period decreased during the pos t-disturbance period, while that of the heterotrophic systems (H) increased. As an exception, the GPP/ER values of S2-A deviated from 1 over time. Because the GPP/ER values of S8-A and S8-B were close to 1 (0.95 to 1.00) during the disturbance period as well as the initial stabilization period, no significant convergence of GPP/ER to 1 was observed in each section during the postdisturbance period. The average deviations of GPP/ER of t he three disturbed Sec 2, 3, and 4 from the reference Sec 1 were calculat ed for the microcosms that were used for the intensity tests to study the effect of disturbance on t he GPP/ER. But no remarkable increase or decrease of average GPP/ER de viation was observed during either the disturbance or post-disturbance period in each microcosm. Effect of Disturbance Intensity GPP and Disturbance Intensity In the intensity tests, the four distinct experimental dur ation plans were applied to the 11 microcosms: 1-5-15 (S3-A and S4-A), 5-5-15 (S7-A and S8-A ), 5-10-15 (S1-A, S2-A, S5-A, and S6-A), and 20-10-15 (S1-B, S2-B, and S6-B). As a result, the 11 microcosms showed five distinct relationships between mean GPP (MGPP) and disturbance intensities: positive monotonic, negative monotonic, U-shaped, peaked, and non-significant. The relationship between MGPP and disturbance intensities during the disturbance and post-disturbance periods are represented in Figure 3-3 and Table 3-4.

PAGE 82

82 Because the rank of MGPP among microcosm sections changed over the experimental period depending on the time frame where G PP values were averaged, two to four MGPP-intensity relationships in different time ranges were selectively plotted considering the change of relationships: an MGPP-intensity relationship for the whole disturbance period and one to three MGPP-intensity relationships for part of the pos t-disturbance period. The time series of MGPP of the 16 microcosms under the changing time frames from the beginning of each disturbance and post-disturbanc e period are presented in Appendix A. The differences of the GPP levels among th e four sections of a microcosm in the GPP time series were not always discernib le because the rank of GPP values among the sections changed over time under the fluct uating GPP time series. For example, in the GPP time series of S1-A ( Figure A-1), no section of the microcosm consistently shows higher GPP values than the others during the post-disturbance period. In some microcosms, however, the GPP time series clearly showed the differences of the GPP levels among sections over time. For exampl e, the GPP values of Sec 2, 3, and 4 were always higher than those of Sec 1 in S8-A ( Figure A-15). In S8-A, the maximum MGPP difference was 0.132 mM-CO2/day between Sec 3 and 1, and the minimum difference was 0.054 mM-CO2/day between Sec 3 and 4 duri ng the 15-day post-disturbance period. Among the 11 microcosms, only S5-A maintained the single MGPP-intensity relationship (U-shaped) throughout t he experimental period. The MGPP-intensity relationships changed in the other microcos ms depending on the time frame where the MGPP values were calculated. Although each microcosm showed distinct MGPP-intensity

PAGE 83

83 relationship, not all microcosms represented cl ear relationships. Fo r example, in S1-A, the MGPP[16-30] difference between the int ensities 62 and 50 was 0.0222 mM-CO2/day, while it was as low as 0.0001 mM-CO2/day between intensities 28 and 0 ( Figure 3-3). S2-A and S2-B were disturbed by water moti on intensities of 0, 40, 62, and 113, which were different from those of the other microcosms. The MGPP-intensity relationship was non-significant in both S2-A and S2-B, al though the relationships were U-shaped and peaked in the intensity range of 0 in S2-A and S2-B, respectively. The GPP values at Day 45 were missing in S2-B. In S3-A and S4-A, the GPP values on Day 1 among sections did not agree with one another, with the average GPP difference between two sections of 0.052 and 0.025 mM-CO2/day, respectively (Figures A-5 and A-7). In S3-A, the GPP of Sec 1 was -0.034 mM-CO2/day on Day 1, while the ER values of the four sections on Day 1 were all negative. The negative values of GPP and ER in S3-A resulted from the incomplete diel patterns of [PCO2] on Day 1. S6-A and S6-B showed significant MGPP differences between the disturbed and undisturbed sections (Figures A-11 and A-12). In S6-A, the undisturbed Sec 1 had a GPP of 0.23 mM-CO2/day on Day 30, compared with t he GPP values of 0.08, 0.05, 0.03 mM-CO2/day of the distur bed Sec 2, 3, and 4. In S6 -B, Sec 2, 3, and 4 had high GPP values of 0.92, 0.92, and 0.81 mM-CO2/day, while the undi sturbed Sec 1 had the GPP of 0.15 mM-CO2/day on Day 45. In the microc osms S6-A and S6-B, the disturbed sections showed remarkable differences in turbidity from the undisturbed reference sections, and the difference was more signific ant in S6-A than in S6-B as shown in Figure 3-4.

PAGE 84

84 Pulsing Patterns of GPP and ER under Disturbance The four GPP time series and the four ER time series of each microcosm were transformed into the time series of GPP pulsing (GPPP) and ER pulsing (ERP) using the method described in Figure 2-5. Because the five-day disturbance period was too short to estimate the wavelength ( ) or peak amplitude (PA), the microcosms having either the 5-10-15 or 20-10-15 duration plan (S1-A, S2-A, S5 -A, S6-A, S1-B, S2-B, and S6-B) were analyzed to study how water motion disturbance influences the and PA of the GPPP and ERP time series. The exponential regre ssion model by least squares fit was used for the transformation of the GPP (r2=0.82) and ER (r2=0.77) time series for Sec 1 of S6-A during the post-distur bance period considering the ex ponential increase of GPP and ER (Figure A-11). The other time series of GPP and ER were transformed using the linear regression model by least squares fit. The values of the GPPP or ERP time series ranged from 2.8 to 9.0 days during the disturbance period and from 2.3 to 7.0 days during the post-disturbance period across the investigated microcosms. The PA values of the GPPP or ERP time series ranged from 0.09 to 0.67 during the disturbance period and fr om 0.15 to 0.63 during the post-disturbance period. The values were not correlated with disturbance intensities in a microcosm. For example, the GPPP time series of the four sections in S6-A, which were applied with different dist urbance intensities, had the same d ( of the disturbance period) of 4.0 days, while those in S5-A had the d values of 6.0, 3.3, 2.8, and 3.0 days at intensities of 0, 28, 50, and 63, respectively. Neither (disturbance or postdisturbance) nor p/ d (ratio of values between the post-disturbance and disturbance periods) showed a consistent correlation wi th disturbance intens ities in the seven

PAGE 85

85 microcosms. Likewise, neither PA (d isturbance or post-disturbance) nor PAp/PAd was correlated with disturbance intens ities in the microcosms. Table 3-5 shows the p/ d and PAp/PAd of the GPPP and ERP time series in each section of the seven microcosms. The p/ d and PAp/PAd values of both GPPP and ERP time series were higher in the disturbed Se c 2, 3, and 4 than in the undisturbed Sec 1 for microcosms S1-A, S2-A, S5 -A, and S6-A, except for the f our underlined cases. The p/ d values of the ERP time series in Sec 2 and 4 of S2-A were slightly lower than in Sec 1 by 0.04 and 0.01. In S1-A, S2-A, S5-A, and S6-A, p/ d and PAp/PAd of Sec 1 were less than 1, whic h indicated that the and PA of GPPP and ERP time series decreased over time in an undisturbed environment. But p/ d and PAp/PAd under disturbance were as high as 1.82 for GPPP in Sec 2 of S5-A and 2.31 for ERP in Sec 2 of S6-A, respectively. In S1-B, S2-B, and S6-B, which were tested under the 20-1015 duration plan, no common pattern of disturbance effect on and PA was discovered. In S1-B, the PAp/PAd values of Sec 2, 3, and 4 were high er than those of Sec 1 for both the GPPP and ERP time series. Whereas S2-B had no consistent effect of disturbance on p/ d and PAp/PAd, S6-B had lower p/ d and PAp/PAd in the disturbed Sec 2, 3, and 4 than in the undisturbed Sec 1 for both the GPPP and ERP time series. Effect of Disturbance Frequency In the frequency tests, the experimental duration plan of 1-5-15 was applied to S3-B and S4-B, and the plan of 5-5-15 was applied to S5-B, S7-B, and S8-B. The frequency regimes (regimes 1, 2, 3, and 4) were applied to the four sections of each microcosm, Sec 1, 2, 3, and 4, res pectively. The relationships between MGPP and disturbance frequency regime s are represented in Figure 3-5 and Table 3-6. In Figure

PAGE 86

86 3-5, bar graphs were used to show the MGPP-frequency relationships where the frequency regimes were categorical data. In S3-B, the rank of MGPP[7-t] among the four sections remained during the postdisturbance period ( Figure A-6). Regardless of the time point t, MGPP[2-6] and MGPP[7-t] were highest under regime 4 in S3-B ( Figure 3-5). The MGPP[7-21] difference between regimes 4 and 2 in S3-B was 0.1083 mM-CO2/day, while it was 0.0258 mM-CO2/day between regimes 1 and 2. In S4-B, MGPP[2-6] values were higher in regimes 2, 3, and 4 than in regime 1 with the maximum MGPP[2-6] difference of 0.024 mM-CO2/day between regimes 2 and 1 ( Figure 3-5). In S4-B, the MGPP gaps between sections diminished during the post-dist urbance period, and MGPP was highest in regime 1 (continuous disturbance) in the MGPP[7-21]-frequency relationship. The difference of MGPP[7-21] values between regimes 1 and 4 was 0.0074 mM-CO2/day, while it was 0.0032 mM-CO2/day between regimes 1 and 2. W hen the disturbance and post-d isturbance periods were aggregated to calculate MGPP values in S4-B, the MGPP values in regimes 2 and 3 were higher than that in regi me 1 by 0.0036 and 0.0016 mM-CO2/day, while the MGPP in regime 4 was lower than that in regime 1 by 0.0018 mM-CO2/day. In S5-B, S7-B, and S8-B, the MGPP[11-t] time series between sections in each microcosm did not intersect during the pos t-disturbance period e xcept for a one-point turnover between regimes 2 and 4 at Day 19 (S5-B), 12 (S7-B), and 25 (S8-B) (Figures A-10, A-14, and A-16). In S7-B, Sec 1 maintained the lowest MGPP[11-t], compared with other sections after Day 14. S5-B, S7-B, and S8-B commonly showed higher MGPP values during the 15-day post-distur bance period under r egimes 2, 3, and 4 (discrete disturbance) than under regime 1 (cont inuous disturbance) ( Figure 3-5).

PAGE 87

87 Among regimes 2, 3, and 4, regime 3 had the lowest MGPP[11-25] during the postdisturbance period. The maximum MGPP[11-25] differences between two sections in the three microcosms were 0.0363 (regime s 2 and 1), 0.0120 (regimes 4 and 1), and 0.1224 (regimes 2 and 1) mM-CO2/day in S5-B, S7-B, and S8-B, respectively. The minimum MGPP[11-25] differences between regimes 3 and 1 were 0.0164, 0.0045, and 0.0681 mM-CO2/day in S5-B, S7-B, and S8-B, respectively. The MGPP under discrete disturbances could be higher than that under continuous disturbances, even if the total energy of the discrete disturbances is lower. In S8-A and S8-B, which were established by the same water and sediment source and tested under the same 5-5-15 dur ation plan, the MGPP[11-25] values of Sec 2, 3, and 4 of S8-A were 0.113, 0.185, and 0.131 mM-CO2/day, while the MGPP[11-25] value of Sec 1 of S8-B was 0.072 mM-CO2/day. It should be noted that the microcosm sect ions for the intensity tests were disturbed discretely. The total disturbance energy exerted on Sec 2, 3, and 4 of S8-A was 280, 500, and 620, while that on Sec 1 of S8-B was 840. Effect of Disturbance under Different Ecosystem Maturities S1, S2, and S6 microcosm sets were desig ned to test the effe cts of disturbance under different ecosystem maturities usi ng the replicated mi crocosms A and B. Microcosms A and B in S1, S2, and S6 were treated with the same disturbance regimes as in the intensity tests (10-day disturbance) during Day 6 and Day 21, respectively. Three ecosystem-level charac teristics were considered in the analysis of data: MGPP-disturbance relationship, resistance and resilience, and internal pulsing pattern.

PAGE 88

88 Relationship between MGPP and Disturbance Regime The microcosms S1-A, S2-A, and S6-A, which were disturbed during Day 615, tended to maintain the MGPP-intensity relationships from the disturbance to the postdisturbance period ( Figure 3-3). In S1-A, the positive monotonic MGPP-intensity relationship during the disturbance period was maintained during the post-disturbance period with an ex ception of MGPP[16-18] at intensity 62, where the MGPP[16-18] was lower than that at intens ity 50 by 0.006 mM-CO2/day. Although the MGPP-intensity relationships in S2-A were nonsignificant, the tendency of low MGPP values at intensities 40 and 113, and that of high MGPP values at intensities 0 and 62 during the disturbance period, remained during the pos t-disturbance period. In S6-A, the MGPP[6-15]-intensity relationship was non-significant but close to negative monotonic, and the negative monotonic relationship remain ed during the post-disturbance period, Day 16. The microcosms S1-B, S2-B, and S6-B, which were distur bed during Day 21, did not consistently maintain the MGPP-intensity relationships fr om the disturbance to the post-disturbance period ( Figure 3-3). S1-B and S6-B showed different MGPP-intensity relationships between the disturbance and post-disturbance period. The positive monotonic MGPP[21-30]-intensity relationship in S1-B did not remain during the postdisturbance period, where the MGPP[31-45]-intensity relationship was U-shaped with the minimum MGPP at intensity 50. The negative monotonic MGPP-intensity relationship in S6-B during the disturbance per iod did not remain during the post-disturbance period, and gradually changed to a peaked pattern after Day 35. The non-significant pattern of the MGPP[21-30]-intensity relationship during the dist urbance period in S2-B, however, was

PAGE 89

89 repeated in the MGPP[31-44]-intensity relationship with the transitional peaked pattern of the MGPP[31-33]-intensity relationship. The microcosms A and B in S1, S2, and S6 showed opposite effects of disturbance on MGPP during the post-disturbance peri od when the microcosm sections were categorized into the disturbed and undisturbed groups ( Figure 3-3). During the post-disturbance period, MGPP values at intensities 28, 50, and 62 were higher than that at intensity 0 in any time range of S1-A, w hereas they were lower in S1-B with the exception of MGPP[31-35] where MGPP at intensity 62 was higher t han that at intensity 0 by 0.002 mM-CO2/day. The MGPP[31-35] difference between intensities 62 and 50 was 0.014 mM-CO2/day in S1-B. S2-A showed lower MGPP values at intensities 40, 62, and 113 than that at intensity 0 dur ing the post-disturbanc e period with the exception of the two higher MGPP values at intensity 62 than that at intensity 0 for MGPP[16-18] and MGPP[16-30]. The MGPP differences between intens ities 62 and 0 were 0.0019 and 0.0016 mM-CO2/day for MGPP[16-18] and MGPP[16-30] in S2-A. The maximum MGPP differences were 0.0078 (intensities 62 and 40) and 0.0102 (intensities 62 and 113) mM-CO2/day for MGPP[16-18] and MGPP[16-30] in S2-A. In S2-B, the MGPP values were higher at intensities 40, 62, and 113 than that at intensity 0 in any time range. The MGPP[16-30]-intensity relationship in S6-A and MGPP[31-45]-intensity relationship in S6-B clearly showed the opposite effects of disturbance on MGPP: decreases by disturbances in S6-A with the MGPP[16-30] difference of 0.022 mM-CO2/day between intensities 0 and 28, and increases by disturbances in S6-B with the MGPP[31-45] difference of 0.252 mM-CO2/day between intensities 62 and 0.

PAGE 90

90 Resistance and Resilience of GPP and ER Resistance and resilience were calculated using Equations 2-20 and 2-21. The resistance (RS) and resil ience (RL) of GPP and ER for the disturbed sections in S1-A, S1-B, S2-A, S2-B, S6-A, and S6-B are represented in Table 3-7. The RS values were higher in S1-B than in S1-A for GPP and ER with differences (RS difference between S1-A and S1-B at each intensity level) ranging from 0.04 to 0.24, while RL values were higher in S1-A than in S1-B with differences ranging from 0.08 to 1.16. S2-A and S2-B did not show a consistent high or low tendency in resistance or resilience of GPP or ER. While the RS values of S6-B were slightly higher than those of S6-A by 0.01.02 with the exception of the lower RS value of S6-B for ER by 0.06 at intensity 50, the RL values of S6-B were lower t han those of S6-A by 6.90.87. GPP and ER Pulsing Comparisons of internal pulsing variables between microcosms A and B in S1, S2, and S6 from Table 3-5 revealed that the p/ d and PAp/PAd of the GPPP or ERP time series were higher in the disturbed Sec 2, 3, and 4 than in the undisturbed Sec 1 with several exceptions, when the microcosm sections were disturbed during Day 6 in S1-A, S2-A, and S6-A. Bu t microcosms S1-B, S2-B and S6-B, disturbed during Day 21, showed different tendencies of p/ d and PAp/PAd among microcosms under disturbance. In S6-B, the p/ d and PAp/PAd of both the GPPP and ERP time series were lower in Sec 2, 3, and 4 than in Sec 1. In S1-B, only the PAp/PAd values of GPPP and ERP were higher in Sec 2, 3, and 4 than in Sec 1. Although the p/ d values of the GPPP time series were lower in the distur bed sections than in the undisturbed one in S1-B, those of the ERP time series were not. No c onsistent pattern was discovered regarding p/ d or PAp/PAd between disturbed and undisturbed sections in S2-B. In

PAGE 91

91 terms of the GPPP time series of S2-B, p/ d and PAp/PAd were higher in Sec 3 and 4 but lower in Sec 2, compar ed with those in Sec 1. In S1, S2, and S6, microcosm B generally had higher p/ d and PAp/PAd than microcosm A in the undisturbed Sec 1 (Table 3-5). S1-A, S2-A, and S6-A had p/ d values of 0.56, 0. 49, and 0.95 for GPPP and 0.74, 0.51, and 0.50 for ERP in Sec 1, while S1-B, S2-B, and S6-B had p/ d values of 1.00, 0. 72, and 1.52 for GPPP and 0.58, 1.77, and 1.10 for ERP. S1-A, S2-A, and S6-A had PAp/PAd values of 0.99, 0.27, and 0.67 for GPPP and 0.90, 0.31, and 0.80 for ERP in Sec 1, while S1 -B, S2-B, and S6-B had PAp/PAd values of 1.02, 1. 22, and 1.84 for GPPP and 0.86, 2.48, and 1.16 for ERP. Uncertainty of Resultant Patterns Uncertainty from [GCO2] Correction The theoretical pH-alkalinity relationship at the equilibrium of CO2 gas exchange across the air-water interface was plotted using the black line in Figure 3-1. The equilibrium line divides the area into two distinct realms: CO2 gas is transferred from the water to the air for any state of a system on the left side, while CO2 diffusion occurs from the air to the water on the right. Figure 3-1 shows that CO2 gas transfer is unidirectional throughout the ex perimental period in S3. CO2 gas transfer occurs in both directions for the other microcosms in which pH-alka linity pairs are distributed on both sides of the equilibrium line. The [GCO2] change occurring by CO2 gas exchange across the air-water interface was not corrected for the calculation of [PCO2] because the gas transfer velocity, k, was uncertain. To test whether the MGPP-disturbance relationships are affected by [GCO2] correction, the s ensitivities of MGPP values and MGPP-disturbance relationships were examined under the four diffe rent k (cm/h) values of 0, 1, 2, and 4. The [GCO2] was

PAGE 92

92 calculated using Equation 2-15. Figures 3-6 and 3-7 represent the MGPP-disturbance relationships for all microcosms including [GCO2] corrections with the four k values during the full 15-day pos t-disturbance period. When k equals 0, no CO2 gas diffusion was assumed. Among the microcosms for the intensity tests, the MGPP-intensity relationship changed by [GCO2] corrections only in S3-A ( Figure 3-6). As k changed from 0 to 4 cm/h, the MGPP-intensity relationship in S3-A changed from non-significant to U-shaped, and the MGPP values increased at the four intensities with a maximum increase from 0.136 to 0.356 mM-CO2/day at intensity 0. As k changed from 0 to 4 cm/h, MGPP values also increased at the four intensities of S1-B with a maximum increase from 0.097 to 0.113 mM-CO2/day at intensit y 50, but the MGPP-intensity relationship did not change. Among the mi crocosms for the frequency tests, the MGPP-frequency relationship only changed in S4-B ( Figure 3-7). In S4-B, the MGPP increased in frequency regimes 2, 3, and 4 with the increasing k values, and the maximum increase was observed in regime 2 by 0.005 mM-CO2/day, while the MGPP decreased in regime 1 by 0.003 mM-CO2/day between the k of 0 and 4 cm/h. Although the MGPP-frequency relationship did not change in S3-B and S5-B, the MGPP values increased with the increase of k (0 cm/h) by a maximum of 0.301 (regime 4 of S3-B) and 0.021 (regime 2 of S5-B) mM-CO2/day, respectively. The four time series of [PCO2] and GPP including [GCO2] corrections under the four different k values were plotted for Sec 1 of the microc osms S6-B and S3-A (Figures 3-8 and 3-9) to analyze what caused the change of the MGPP-disturbance relationships under the different k values. As shown in Figure 3-6, the MGPP-intensity relationship and MGPP values remained regardless of k in S6-B, while the MGPP-intensity relationship and

PAGE 93

93 MGPP values changed in S3-A with the incr easing k. In S6-B, the constant MGPP values and the fixed MGPP-disturbance relationship under the different k values were attributed to the agreement of the short-term [PCO2] patterns among different k values ( Figure 3-8). Although t he long-term [PCO2] differed up to 0.65 mM at Day 22 under the k values between 0 and 4 cm/h, diel [PCO2] patterns were similar under the four k values, 0, 1, 2, and 4 cm/h in each day. As a result the four time series of GPP under the four k values agreed with one another in S6-B. Bu t the time series pattern and values of GPP at the k value of 0 were di fferent from those at the k values of 1, 2, and 4 in S3-A ( Figure 3-9), which was attributed to t he discrepancy among the short-term [PCO2] patterns under the four k values The diel patterns of [PCO2], which determined the daily GPP values, disappeared when CO2 gas exchange occurred (k>0) in S3-A as shown in Figure 3-10. Uncertainty from pH Measurement Error It was examined how much [TCO2] potentially changes by pH measurement error. The theoretical [TCO2] per 0.01 pH change ([TCO2]/( 0.01 pH)) was affected by pH and alkalinity. The [TCO2]/( 0.01 pH) values at the four different [CA] levels were plotted in the pH range of 5 using t he thermodynamic relationship among pH, alkalinity, and [TCO2] ( Figure 3-11). The [TCO2]/( 0.01 pH) became lower as [CA] decreased and pH converged to 8.26. At pH 8.26, the minimum [TCO2]/( 0.01 pH) values were 3.40-5, 2.27-4, 5.67-4, and 1.02-3 mM at the [CA] levels of 0.06, 0.40, 1.00, and 1.80 meq, respectively. All mi crocosms, except S3-A, S3-B, and S5-A, had the pH of 8.26 in their pH ranges. The maximum and minimum [TCO2]/( 0.01 pH) values were calculated using the [TCO2]/( 0.01 pH) graphs in Figure 3-11 and the pH-alkalin ity data of the post-

PAGE 94

94 disturbance periods of the 16 microcosms. For the 16 microcosms, the maximum and minimum differences between MGPP values for the full post-disturbance period were also calculated from Figures 3-3 and 3-5. The maximum and minimum MGPP differences and [TCO2]/( 0.01 pH) values of the 16 mi crocosms are presented in Table 3-8. For the calculation of the maximum and minimum differences between MGPP values, only significant sections for the MGPP-disturbance relationships were considered. For example, if the MGPP-disturbance relationship is U-s haped in an intensity test, the difference of MGPP values between the highest and lo west intensities does not provide much information on the significance of the shape. The important point in making the U-shaped relationship is that MGPP values in the intermediate intensities are lower than those in the highest and lowest intensities. If pH measurement errors occur by 0.01 at peak points of pH diel patterns, the resultant GPP error ( GPP) becomes a twofold of [TCO2] because GPP is calculated by the difference between two adjacent slopes in a [PCO2] graph. The potential unreliability of results from S3 -A and S3-B was identified in Table 3-8, where the maximum -2 [TCO2]/( 0.01 pH) values were gr eater than the minimum MGPP differences. The -2 [TCO2]/( 0.01 pH) of S3-A ranged from 0.0128 to 0.0322 mM, while the maximum MGPP difference between two sections was 0.0139 mM-CO2/day. Although the MGPP in regime 2 was less than that in regime 1 in S3-B, their difference was 0.0365 mM-CO2/day, while the maximum -2 [TCO2]/( 0.01 pH) was 0.0416 mM. Also, the difference of MGPP values between regimes 3 and 1 was 0.0258 mM-CO2/day, which was less than the maximum -2 [TCO2]/( 0.01 pH) of 0.0416 mM in S3-B. In S1-A, the minimum MGPP difference was as low as 0.0001 mM-CO2/day between

PAGE 95

95 intensities 28 and 0, while the -2 [TCO2]/( 0.01 pH) ranged from 0.0012 to 0.0030 mM. Thus, the positive monotonic MGPP-intensity relationship of S1-A can be misleading, although the MGPP difference between intensit ies 62 and 0 was 0.0222 mM-CO2/day, which was high enough to explain the incr ease of GPP by int ensity 62 even under 0.1 pH measurement error. Simulation Models Model Equations and Calibrations The flows, storages, and calibrations of the basic steady-stat e microcosm model are presented in Table 3-9. Production functions of interacting components were assumed to be multiplicative. From the lit erature and results of microcosm experiments, storages and flows were estimated as shown in Figure 3-12, while satisfying the steady-state of the state variables. A two-day period and a 15-day period were assigned for the turnover rate s of the producer and consumer storages, respectively. The flows in Figure 3-12, which were on a one-day basis, were substituted by a five-minute basis for the computer simulation as represented in Table 3-9. Because J2 and J3 in Figure 2-7 were assumed to operate only dur ing the light periods, they were regarded as operating on a 12-hour basis. Simulation Results Tables B-1, B-2, and B-3 present the R codes for the computer simulations of the basic steady-state, dist urbance int ensity, and dist urbance frequency models of a microcosm, respectively. The simulation period lasted 50 days. The time interval of simulation (DT) was initially set at 1 (5 minutes) in the program, and it was small enough so that the graphs of the state variables did not change with smaller DTs.

PAGE 96

96 The state variables of the basic steady -state model approached constant values after several days with daily fluctuations caused by the alternat ing light regimes ( Figure 3-13). A daily GPP value was calculated by averaging 114 GPP values of each day, and it approached 0.0999 mM-C/day around Day 10. The st orage of producers (Pro) maintained the steady-state af ter Day 10. While the consum er storage (Con) leveled off after Day 10, biomass of Con decreased by 2 mg-C from 270 to 268 mg-C during Day 1. The water column nutrient concentration (Nut) approached 3.9 mg-P after Day 10. The dead organic matter from producers (ProD) approached 72.3 mg-C around Day 40, while it slightly increased from Day 20 to 40 by 0.4 mg-C. The dead organic matter from consumers (ConD) decreased from Day 1 to 50 by 0.5 mg-C. Figure 3-14 depicts the general patterns of the GPP time series under the distinct sign combinations of pfd from the simulation of in tensity and frequency tests. Regardless of discrete or continuous disturbance, the GPP values of the models affected by disturbance eventually appr oached the reference GPP during the post-disturbance period when t he reproduction coefficients of producers and consumers were temporarily changed durin g the disturbance period (p
PAGE 97

97 and frequency models, the pattern of the GPP time series changed according to the combination of the signs of pfd between k2-k3 and k4-k5, but only the GPP level changed under the different magnitudes of the pfd. Distinct MGPP-disturbance relationships were te mporarily generated under different fd-5 values, even if the sign combinations of the pfd involved in the coefficients, k2-k3 and k4-k5, were the same. For example, under the positive signs of pfd for k2-k3 and k4-k5, the MGPP-disturbance relationship during the post-distur bance period changed from a negative monotonic to non-significant when the fd-5 values changed from 0.3 to 0.5 in the condition of p less than the threshold, and from 0.0002 to 0.0001 in the condition of p equal to or greater than the threshold, as shown in Figure 3-15. But the non-significant MGPP-disturbance relationship in D of Figure 3-15 will return to a negative monotonic pattern in the long term as s hown in B (after Day 45) because GPP under the low p condition approaches t he reference GPP line over time. The ratio of water column nutrient to total nutrient (Nu t/TN) affected the patterns and levels of GPP time series ( Figure 3-16). Although most sign combinations of pfd for k2-k3 and k4-k5 did not change the pattern of GPP time series under different Nut/TN values, the GPP time series were significant ly altered in the combination of (+,0) when the Nut/TN changed from 0.24 to 0.39 as shown in Figure 3-16. In the initial model, the storage Nut was 3.85 mg-P and TN was 16 mg-P (Nut/TN=0.24). To st udy the effect of Nut/TN on the pattern of GPP time series, th e other condition was established with Nut as 7.85 mg-P and TN as 20 mg-P (Nut/TN=0.39). The increased Nut/TN mimicked a nutrient enrichment or an incr ease of Nut by increased turnover or decomposition rates of organic matters. Under t he sign combination of (+,0), the increase of the Nut/TN

PAGE 98

98 raised the GPP levels at p level 0. W hen Nut/TN changed from 0.24 to 0.39 (A and B in Figure 3-16), the GPP values und er the condition of p less than the threshold deviated more from the reference GPP during t he disturbance period, and the decreasing patterns of GPP values under the condition of p equal to or greater than the threshold changed to increasing patterns duri ng the post-disturbance period. Validation of the Models The resultant MGPP-disturbance relationships from the simulation models were compared with the microcosm results to suggest possible mechanisms of the disturbance effects on the MGPP-disturbance relationships of the microcosms. The distinct MGPP-disturbance relationships were determined by a threshold p, fd-5 in the p less than the threshold, fd-5 in the p equals to or greater than the threshold, and the sign combinations of the pfd for k2-k3 and k4-k5. These factors were manipulated to produce several relationships between MGPP and disturbance regimes. From the simulations, possible MGPP-disturbance relationships we re generated under the distinct combinations of the factors, and each relati onship represented an MGPP-disturbance relationship observed in the microcosm results as shown in Figures 3-17 and 3-18. Although the power threshold may change du ring the experimental period by either structural improvement or deterioration of a microcos m under disturbanc e, it was assumed that the power threshold remained constant throughout the experimental period in the simulation models. Bec ause the sign combinations of the pfd for k2-k3 and k4-k5 may also be different between the power values above and below a threshold, all possible combinations of the si gns were simulated to find the MGPP-disturbance relationships observed in the microcosm tests.

PAGE 99

99 In the simulation of the intensity model ( Figure 3-17), similar MGPP-intensity relationships as the microcos m results were generated by various sign combinations of pfd for k2-k3 and k4-k5 between the power values above and below a threshold. While most MGPP-intensity relationships were generated by the same sign combinations of either (+,+), (+,-), (0,-), (+,0), or (-,-) between the power values above and below a threshold, C (S2-A) and K (S8-A) were gener ated by different sign combinations: (+,+) for a p less than the threshold and (0,+) for a p equal to or greater than the threshold in C, and (0,-) for a p less than the threshold a nd (+,-) for a p equal to or greater than the threshold in K. Even under the same sign combination of pfd for k2-k3 and k4-k5, different MGPP-intensity relationships were generated depending on fd-5 values and a threshold of p. For exam ple, E, G, and H had non-sign ificant, U-shaped, and negative monotonic relationships during the post-disturbance period, and the distinct relationships were caused by different fd-5 values and power thresholds under the same sign combination of (+,+). Although the steady-state models could not represent the development of microcosms, where system configuration changes over time, the models for the microcosms S1-A, S1-B, S2-A, S2-B, S6 -A, and S6-B showed how each microcosm pair, A and B, in S1, S2, and S6 responded to the same disturbance regimes in the maturity tests. In the model s for S1-A and S1-B (A and B of Figure 3-17), the sign combinations were (0,-) and (+,-), respecti vely. While disturbances always reduced the reproduction rates of consumers in both mi crocosms A and B of S1, they increased the reproduction rate of producers in microcosm B. In the models fo r S2-A and S2-B (C and D of Figure 3-17), the sign combinations were (+ ,+) for a p less than the threshold

PAGE 100

100 and (0,+) for a p equal to or greater than the threshold for the S2-A model, and (0,-) for the S2-B model. Whereas the disturbance e ffect was positive on the reproduction rate of consumers in microcosm A of S2, it was negative in microcosm B. The positive effect of disturbance on the reproduction rate of producers in microcosm A of S2 disappeared in microcosm B. In the model s for S6-A and S6-B (H and I of Figure 3-17), the sign combinations were (+,+) and (-,-), respec tively. While it was suggested from the simulation model that disturbances r educed reproduction ra tes of producers and consumers in microcosm B of S6 in contrast to microcosm A, the MGPP values were increased by the disturbances as shown in the result of the microcosm S6-B. In the simulation of the frequency model ( Figure 3-18), the x-axis indicates the frequency regimes applied in the frequency te sts of the microcosms. Thus, the continuous low power (p=1) of disturbance was applied in regime 1, while the high power (p=7) was applied in regimes 2, 3, and 4 with the distinct frequencies and durations used in the mi crocosm tests. The MGPP of the undisturbed reference section was represented as regime 0. The threshol d of p was set at 2, so the temporary fd change occurred only under the frequency regime 1. The fd was permanently changed in the other regimes, 2, 3, and 4. The MGPP-frequency relationship of the microcosm S3-B was generated by the sign combination of (-,-) (A of Figure 3-18). The MGPPfrequency relationship of S4-B was possible when different sign combinations were applied above and below the power threshold: (-,+) for a p less than the threshold and (+,0) for a p equal to or great er than the threshold (B of Figure 3-18). The sign combination of (0,-) was applied to generate the MGPP-frequency relationships of S5-B, S7-B, and S8-B (C of Figure 3-18). The MGPP values in frequency regimes 2, 3, and 4

PAGE 101

101 were higher than that in regime 1, which was observed in the microcosms S5-B, S7-B, and S8-B. The frequency model, how ever, failed to generate the MGPP-frequency relationship among frequency regimes 2, 3, and 4 observed in the microcosms S5-B, S7-B, and S8-B. In the simu lation model for the microc osms S5-B, S7-B, and S8-B, MGPP values among frequency regimes 2, 3, and 4 monotonically increased from regime 2 to 4 during the post-disturbance pe riod. In the microcosms S5-B, S7-B, and S8-B, the MGPP in frequency regime 3 was lower than those in regimes 2 and 4 during the 15-day post-disturbance period. The general patterns of GPP time se ries during the disturbance and postdisturbance periods, represented in Figure 3-14, show that GPP always increases by disturbance when the sign combination of pfd for k2-k3 and k4-k5 is (0,-) regardless of a power threshold in both the intensity and frequency models. T hat is, GPP increased when disturbances reduced the reproduction rate of consumers but did not affect that of producers. In A and D of Figure 3-17, where the sign combination of pfd for k2-k3 and k4-k5 was (0,-), MGPP values were higher at power leve ls 1, 2, 3, an d 4 than at power level 0, although MGPP-intensity relationships were different depending on the values of fd-5 and the power threshold. In C of Figure 3-18, where the sign co mbination was also (0,-), the MGPP values of the disturbed systems (regimes 1, 2, 3, and 4) were higher than that of the undisturbe d system (regime 0). An incr eased GPP by disturbances also could be generated by different sign combinations of (+,-) for a p equal to or greater than a threshold and (0,-) for a p less than a threshold during bot h the disturbance and post-disturbance periods as shown in Figure 3-14. In K of Figure 3-17, where sign combinations were (0,-) for a p less than the threshold and (+,-) for a p equal to or

PAGE 102

102 greater than the threshold, GPP increased by disturbances during both the disturbance and post-disturbance periods. The GPP values always decreased by di sturbances during both the disturbance and post-disturbance periods only when the sign combination was (+,+) at any power and only in the simulation of the frequency model as shown in Figure 3-14. The GPP could decrease by disturbances during both the disturbance and post-disturbance periods in the simulation of the intens ity model, however, when the values of fd-5 and power threshold were well matched with the sign combination of (+,+) for any power, or the sign combinations of (+,+) for a p less t han a threshold and either (-,+) or (0,+) for a p equal to or greater than a threshol d as shown in C, E, G, and H of Figure 3-17.

PAGE 103

103 Table 3-1. Ranges of pH, alkalinity and GPP in the 16 microcosms. Microcosm Sample lake Tank pH range Alkalinity range (meq) GPP range (mM-CO2/day) S1 Newnan A B 7.2.6 7.4.6 0.48.28 0.46.60 0.07.32 0.07.32 S2 Alice A B 7.6.9 7.6.9 1.46.82 1.46.84 0.02.06 0.02.07 S3 Santa Fe A B 5.3.8 5.1.7 0.06.08 0.06.08 0.04.21 0.05.53 S4 Wauberg A B 7.0.3 7.1.6 0.32.40 0.32.40 0.01.11 0.04.07 S5 Orange A B 6.5.0 6.5.9 0.34.44 0.38.42 0.05.30 0.05.25 S6 Alice A B 7.7.2 7.7.0 1.50.62 1.52.66 0.02.23 0.01.92 S7 Alice A B 7.8.1 7.8.2 1.24.40 1.26.38 0.02.09 0.02.12 S8 Newnan A B 7.9.7 7.6.9 0.56.80 0.58.84 0.02.42 0.02.50 Table 3-2. Pearsons correlation coefficient (r) between GPP and ER in the four sections of each microcosm during the dist urbance and post-disturbance periods. Disturbance period Post -disturbance period Microcosm Sec 1 Sec 2 Sec 3Sec 4Sec 1Sec 2 Sec 3 Sec 4 S1-A S1-B 0.90 0.95 -0.61 0.84 0.38 0.90 0.60 0.99 0.95 0.96 0.97 0.83 0.78 0.85 0.98 0.94 S2-A S2-B 0.26 0.44 -0.07 0.48 -0.25 0.74 -0.09 0.58 0.60 0.72 0.86 0.72 0.87 0.64 0.72 0.75 S3-A S3-B 0.62 -0.36 0.89 -0.58 0.69 0.95 0.94 -0.06 0.68 0.83 0.60 0.87 0.22 0.81 0.46 0.88 S4-A S4-B 0.93 0.76 0.89 1.00 0.98 0.98 0.98 0.99 0.96 0.95 0.97 0.95 0.91 0.89 0.95 0.97 S5-A S5-B 0.93 0.91 0.97 0.99 0.98 1.00 0.92 0.77 0.99 0.89 0.96 0.95 0.94 0.97 0.86 0.95 S6-A S6-B 0.96 0.99 0.55 0.99 0.80 0.98 0.86 0.98 1.00 0.97 0.99 1.00 0.98 1.00 0.98 1.00 S7-A S7-B 0.90 0.91 0.99 0.33 0.99 -0.26 0.94 0.85 0.90 0.99 0.82 0.99 0.98 0.99 0.86 0.99 S8A S8-B 1.00 0.21 0.99 0.99 0.99 0.99 0.87 0.97 0.99 0.99 1.00 0.99 1.00 0.98 0.99 0.99

PAGE 104

104 Table 3-3. Average GPP/ER ratio in the four sections of each microcosm during the disturbance and post-disturbance peri ods. ((A) and (H) denote autotrophic and heterotrophic state of each microcosm determi ned by GPP/ER ratio during the disturbance period.) Disturbance period Post -disturbance period Microcosm Sec 1 Sec 2 Sec 3 Sec 4 Sec 1 Sec 2 Sec 3 Sec 4 S1-A (H) S1-B (H) 0.86 0.90 0.89 0.92 0.90 0.90 0.90 0.91 0.94 0.93 0.93 0.92 0.91 0.91 0.94 0.92 S2-A (H) S2-B (H) 1.01 0.88 0.91 0.98 0.95 0.87 0.93 0.94 0.85 0.96 0.87 0.99 0.86 1.00 0.86 1.02 S3-A (A) S3-B (A) 1.28 1.64 1.48 1.40 1.40 1.45 1.80 1.37 1.07 1.03 1.11 0.99 1.15 1.04 1.05 1.05 S4-A (A) S4-B (A) 1.10 1.06 1.21 1.09 1.15 1.09 1.13 1.14 1.06 1.13 1.20 1.02 1.00 1.01 1.04 1.05 S5-A (A) S5-B (A) 1.06 1.51 1.14 1.50 1.21 1.63 1.16 1.48 1.01 1.01 1.00 0.99 1.00 0.99 1.01 0.99 S6-A (H) S6-B (A) 0.91 1.06 0.89 1.06 0.87 1.06 0.89 1.06 1.01 1.04 0.99 1.04 1.02 1.03 0.95 1.03 S7-A (A) S7-B (A) 1.37 1.22 1.28 1.18 1.25 1.14 1.27 1.24 1.01 1.11 0.98 1.06 1.06 1.09 1.04 1.03 S 8-A (H) S8-B (H) 0.96 1.00 0.97 0.99 0.96 0.98 0.97 0.95 0.90 0.90 0.94 0.97 0.98 0.96 0.97 0.99 Table 3-4. MGPP-intensity relationship of the 11 micr ocosms in the intensity tests. (Numbers in each bracket indica te the time (Day) range of t1t2 where MGPP was calculated. NM: negative monotoni c, NS: non-significant, P: peaked, PM: positive monotonic, U: U-shaped.) Microcosm Duration plan Disturbance period Post -disturbance period S1-A 5-10-15 PM [6] P [16], PM [16] S1-B 20-10-15 PM [21] U [31] S2-A 5-10-15 NS [6] U (intensity 0)a NS [16] U (intensity 0)a S2-B 20-10-15 NS [21] P (intensity 0)a P [31], NS [31] P [31] (intensity 0)a S3-A 1-5-15 P [2] NS [7] S4-A 1-5-15 NS [2] U [7 -8], NM [7], NS [7] S5-A 5-10-15 U [6] U [16] S6-A 5-10-15 NS [6] PM [16], P [16], NM [16] S6-B 20-10-15 U [21] NM [31], P [31] S7-A 5-5-15 P [6] NM [11], U [11] S8-A 5-5-15 PM [6] P [11] aMGPP-intensity relationship in the intensity range of 0 was provided in S2-A and S2-B for comparison with other microcosms. The full intensity range was 0 in S2-A and S2-B.

PAGE 105

105 Table 3-5. Ratio of post-disturbance to disturbance in wavelength ( ) and peak amplitude (PA) of the pulsing GPP (GPPP) and pulsing ER (ERP) of microcosms. (The subscripts p and d denote the post-disturbance and disturbance periods.) GPPP ERP Micro cosm Variable Sec 1 Sec 2Sec 3Sec 4Sec 1Sec 2 Sec 3Sec 4 S1-A p/ d PAp/PAd 0.56 0.99 0.57 2.21 0.76 1.31 1.33 2.17 0.74 0.90 1.18 2.27 0.88 1.39 1.08 1.52 S2-A p/ d PAp/PAd 0.49 0.27 0.51 0.46 0.71 0.85 0.61 0.58 0.51 0.31 0.47 0.52 0.60 0.78 0.50 0.48 S5-A p/ d PAp/PAd 0.40 0.38 1.82 0.56 1.32 0.52 0.77 0.84 0.44 0.38 1.82 0.51 1.32 0.49 0.62 0.82 S6-A p/ d PAp/PAd 0.95 0.67 1.05 0.96 1.68 0.52 1.43 1.16 0.50 0.80 1.20 2.31 0.50 0.52 0.71 0.94 S1-B p/ d PAp/PAd 1.00 1.02 0.91 1.21 0.70 1.40 0.55 1.70 0.58 0.86 1.21 1.56 0.60 1.64 0.51 1.86 S2-B p/ d PAp/PAd 0.72 1.22 0.62 0.48 1.00 1.56 1.20 1.87 1.77 2.48 1.27 3.81 2.12 4.43 1.22 0.99 S6-B p/ d PAp/PAd 1.52 1.84 0.69 0.43 0.41 0.54 0.43 0.55 1.10 1.16 0.60 0.52 0.40 0.82 0.37 0.90 The underlined numbers indicate lower p/ d or PAp/PAd of disturbed sections than that of Sec 1 in S1-A, S2-A, S5-A, and S6-A. Table 3-6. Rank of MGPP among the four sections in each microcosm under disturbance frequency regimes. (Numbers in each bracket indicate the time (Day) range of t1t2 where MGPP was calculated. Ranked in the order of high to low MGPP.) Microcosm Duration plan Disturbance period Post-disturbance period S3-B 1-5-15 4-2-1-3 [2] 43-2-1 [7], 43-1-2 [7] S4-B 1-5-15 2-3-4-1 [2] 12-4-3 [7], 1-2-3-4 [7] S5-B 5-5-15 4-2-3-1 [6] 2-4-3-1 [11], 4-2-3-1 [11], 2-4-3-1 [11] S7-B 5-5-15 2-3-4-1 [6] 42-1-3 [11], 4-2-3-1 [11] S8-B 5-5-15 3-2-4-1 [6] 2-4-3-1 [11]

PAGE 106

106 Table 3-7. Resistance (RS) and resilience (RL) of GPP and ER in the three microcosm sets S1, S2, and S6. (Section num bers are substituted by disturbance intensities in the parentheses.) Microcosm Duration plan GPP (Disturbance intensity) ER (Disturbance intensity) RS S1-A S1-B 5-10-15 20-10-15 (28) 0.81 0.91 (50) 0.65 0.89 (62) 0.57 0.81 (28) 0.89 0.93 (50) 0.75 0.90 (62) 0.66 0.83 S2-A S2-B 5-10-15 20-10-15 (40) 0.81 0.55 (62) 0.88 0.91 (113) 0.86 0.70 (40) 0.85 0.69 (62) 0.84 0.89 (113) 0.91 0.76 S6-A S6-B 5-10-15 20-10-15 (28) 0.80 0.81 (50) 0.73 0.74 (62) 0.76 0.76 (28) 0.80 0.82 (50) 0.80 0.74 (62) 0.75 0.77 RL S1-A S1-B 5-10-15 20-10-15 (28) 0.38 0.03 (50) 0.52 -0.64 (62) 0.47 0.29 (28) -0.20 -0.29 (50) 0.38 -0.61 (62) 0.33 0.25 S2-A S2-B 5-10-15 20-10-15 (40) 0.38 0.33 (62) 0.05 -0.20 (113) -0.50 0.49 (40) 0.24 0.17 (62) 0.36 -0.16 (113) -1.41 0.36 S6-A S6-B 5-10-15 20-10-15 (28) -0.33 -13.86 (50) 0.08 -8.76 (62) -1.00 -7.90 (28) -0.31 -16.18 (50) -0.14 -9.63 (62) -0.83 -9.03

PAGE 107

107 Table 3-8. MGPP differences in the MGPP-disturbance relationships and potential [TCO2] changes by 0.01 pH error in the 16 microcosms. (The values of MGPP difference were calculated from the 15-day post-disturbance period.) MGPP difference (mM-CO2/day) [TCO2]/( 0.01 pH) (mM) Microcosm Maximum(-4) Minimum(-4) Maximum(-4) Minimum(-4) S1-A 222 1 15 6 S1-B 208 107 23 7 S2-A 102 21 15 10 S2-B 134 22 17 10 S3-A 139 37 161 64 S3-B 1083 258 208 80 S4-A 63 38 14 2 S4-B 74 32 14 2 S5-A 512 129 37 3 S5-B 363 164 36 2 S6-A 510 68 32 9 S6-B 3782 448 69 9 S7-A 150 23 19 7 S7-B 120 45 21 9 S8-A 1324 540 28 4 S8-B 1224 681 31 4

PAGE 108

108 Table 3-9. Equations and cali brations of the basic steady-state model of a microcosm. (Numbers are based on one sect ion (30 L) of a microc osm with five-minute simulation intervals.) Variable Equation Unit Steady-state value Source Source Light L mmolm-2(5min)-1 0 (6pmam) 45 (6ampm) Microcosm Remaining light (R) 10% of L mmolm-2(5min)-1 0 (6pmam) 4.5 (6ampm) Guess Storages Producer Pro mg-C 72 Wetzel (2001)a Consumer Con mg-C 270 Wetzel (2001)b Dead producer ProD mg-C 72 Rodgers and DePinto (1983)c Dead consumer ConD mg-C 72 Guessc Nutrient (Nut) TN-fP(Pro+ProD) -fC(Con+ConD) mg-P 3.85 Florida LAKEWATCH (2005)d Flows J1 k1RNutPro mmolm-2(5min)-1 40.5 Guess J2 k2RNutPro mg-C(5min)-1 0.25 Microcosm S1 J3 k3RNutPro mg-C(5min)-1 0.075 Guess from Lindeman(1942)e J4 k4ProCon mg-C(5min)-1 0.0625 Guessf J5 k5ProCon mg-C(5min)-1 0.0375 Guess from Lindeman(1942)g J6 k6Pro mg-C(5min)-1 0.025 Steady-state J7 k7Con mg-C(5min)-1 0.025 Steady-state J8 k8ProD mg-C(5min)-1 0.025 Steady-state J9 k9ConD mg-C(5min)-1 0.025 Steady-state Coefficients k1 (mg-Pmg-C)-1 3.25E-2 k2 m2(mmolmg-P)-1 2.00E-4 k3 m2(mmolmg-P)-1 6.01E-5 k4 (mg-Cmin)-1 3.21E-6 k5 (mg-Cmin)-1 1.93E-6 k6 (5min)-1 3.47E-4 k7 (5min)-1 9.26E-5 k8 (5min)-1 3.47E-4 k9 (5min)-1 3.47E-4 fP 0.025 Redfield ratioh fC 0.025 Guess aThe storage of producers was calculated based on the 2-day turnover rate. bAverage 15-day turnover rate for consumers. c10-day turnover rate for decomposition. dWater column P was assumed to be 128 g/L (Lake Newnan). eJ3 was assumed to be 30% of J2 (considering a new environment for species, a higher percentage th an the one Lindeman suggested was assumed). fHalf of GPP. gJ5 was assumed to be 60% of J4 (a higher percentage was assumed in the same reason for e). hMass ratio of C/P=41 from Redfield ratio of C:N:P=106:16:1.

PAGE 109

109 0.0 0.5 1.0 1.5 2.0 567891 0 pH[CA] (meq) Figure 3-1. Ranges of pH-alkalinity pairs of the eight microcosm sets. S2 S5 S6 S1 Dynamic equilibrium of CO2 diffusion across the air-water interface S7 S8 S3 S4

PAGE 110

110 Sec17.5 8.0 8.5 9.0 9.5 10.0 0510152025 Time (Day) pH Sec27.5 8.0 8.5 9.0 9.5 10.0 0510152025 Time (Day) pH Sec37.5 8.0 8.5 9.0 9.5 10.0 0510152025 Time (Day) pH Sec47.5 8.0 8.5 9.0 9.5 10.0 0510152025 Time (Day) pHFigure 3-2. Time series of pH in the four sections of S8-A.

PAGE 111

111 0.07 0.09 0.11 0.13 0204060 [6-15] [16-18] [16-22] [16-30] 0.07 0.09 0.11 0.13 0204060 [21-30] [31-33] [31-35] [31-45] 0.02 0.03 0.04 0.05 04080120 [6-15] [16-18] [16-25] [16-30] 0.03 0.04 0.05 0.06 04080120 [21-30] [31-33] [31-44] 0.08 0.12 0.16 0.20 0204060 [2-6] [7-11] [7-16] [7-21] 0.03 0.05 0.07 0.09 0204060 [2-6] [7-8] [7-17] [7-21] 0.07 0.12 0.17 0.22 0204060 [6-15] [16-30] 0.03 0.05 0.07 0.09 0204060 [6-15] [16-17] [16-22] [16-30] 0.00 0.20 0.40 0.60 0204060 [21-30] [31-32] [31-35] [31-45] 0.02 0.04 0.06 0.08 0204060 [6-10] [11-11] [11-18] [11-25] 0.00 0.10 0.20 0.30 0204060 [6-10] [11-25] x-axis: Disturbance intensity y-axis: MGPP (mM-CO2/day) Figure 3-3. MGPP-intensity relationships of the 11 micr ocosms for the intensity tests. (Numbers in each bracket indica te the time (Day) range of t1t2 where MGPP was calculated.) S4A S3A S1-A S1-B S2A S2-B S5-A S6A S6-B S7-A S8A

PAGE 112

112 Figure 3-4. Top view of S6 microcosms on Day 45. A) S6-A. B) S6-B. A B Sec 1 Sec 2 Sec 3 Sec 4 Sec 1 Sec 2 Sec 3 Sec 4

PAGE 113

113 0.00 0.10 0.20 0.30 1234 [2-6] [7-8] [7-21] 0.00 0.03 0.06 0.09 1234 [2-6] [7-13] [7-21] 0.00 0.05 0.10 0.15 1234 [6-10] [11-12] [11-19] [11-25] 0.00 0.02 0.04 0.06 1234 [6-10] [11-13] [11-25] 0.00 0.10 0.20 0.30 1234 [6-10] [11-25] x-axis: Disturbance frequency regime y-axis: MGPP (mM-CO2/day) Figure 3-5. MGPP-frequency relationships of the five microcosms for the frequency tests. (Numbers in each bracket indica te the time (Day) range of t1t2 where MGPP was calculated.) S8-B S7-B S5-B S3-B S4-B

PAGE 114

114 0.09 0.11 0.13 0204060 k=0 k=1 k=2 k=4 0.09 0.11 0.13 0204060 0.02 0.03 0.04 0.05 04080120 0.04 0.05 0.06 0.07 04080120 0.00 0.15 0.30 0.45 0204060 0.030 0.035 0.040 0.045 0204060 0.06 0.10 0.14 0.18 0204060 0.02 0.05 0.08 0.11 0204060 0.00 0.20 0.40 0.60 0204060 0.02 0.03 0.04 0.05 0204060 0.05 0.10 0.15 0.20 0204060 x-axis: Disturbance intensity y-axis: MGPP (mM-CO2/day) Figure 3-6. Sensitivity of MGPP-intensity relationships of the 15-day post-disturbance period under the different k (cm/h) values in the 11 microcosms for the intensity tests. S4A S3A S1-A S1-B S2A S2-B S5-A S6A S6-B S7-A S8A

PAGE 115

115 0.00 0.20 0.40 0.60 1234 k=0 k=1 k=2 k=4 0.040 0.045 0.050 0.055 1234 0.12 0.15 0.18 0.21 1234 0.04 0.05 0.06 0.07 1234 0.05 0.10 0.15 0.20 1234 x-axis: Frequency regime y-axis: MGPP (mM-CO2/day) Figure 3-7. Sensitivity of MGPP-frequency relationships of the 15-day post-disturbance period under the different k (cm/h) values in the five microcosms for the frequency tests. S8-B S3-B S4-BS5-B S7-B

PAGE 116

116 A 1.3 1.6 1.9 2.2 051015202530354045Time (Day)[PCO2] (mM) k=0 k=1 k=2 k=4B 0.0 0.1 0.2 0.3 051015202530354045Time (Day)GPP (mM-CO2/day)Figure 3-8. Time series of [PCO2] and GPP in Sec 1 of S6-B under the different k (cm/h) values. A) [PCO2]. B) GPP.

PAGE 117

117 A 0 10 20 30 40 05101520Time (Day)[PCO2] (mM) k=0 k=1 k=2 k=4B 0.0 0.1 0.2 0.3 0.4 0.5 05101520Time (Day)GPP (mM-CO2/day)Figure 3-9. Time series of [PCO2] and GPP in Sec 1 of S3-A under the different k (cm/h) values. A) [PCO2]. B) GPP.

PAGE 118

118 A 0.5 0.6 0.7 0.8 101112131415Time (Day)[PCO2] (mM) B 4.5 5.5 6.5 7.5 101112131415Time (Day)[PCO2] (mM) C 9 11 13 15 101112131415Time (Day)[PCO2] (mM) D 18 21 24 27 101112131415Time (Day)[PCO2] (mM) Figure 3-10. [PCO2] patterns during Day 10 in Sec 1 of S3-A under the different k (cm/h) values. A) k=0. B) k=1. C) k=2. D) k=4.

PAGE 119

119 0.000 0.005 0.010 0.015 0.020 5678910pH[TCO2]/( 0.01pH) (mM) [CA] = 0.06 meq [CA] = 0.40 meq [CA] = 1.00 meq [CA] = 1.80 meq Figure 3-11. Change of [TCO2] per 0.01 pH error in the di fferent pH and [CA] levels.

PAGE 120

120 Figure 3-12. Flows and storages with numbers in the basic steady-state model of a microcosm. Light (L) 6.48 (J1) 5.83 0.65 (R) Nut 3.85 Pro72 Con270 10.8 36 7.2 7.2 7.2 7.2 72 72 18 10.8 ConD ProD Light energy (L, R, J1): mol/m 2 /day Nut storage: mg-P Main flows: mg-C/day Main storages: mg-C

PAGE 121

121 A GPP (mM-C/day) 01020304050 0.098 0.099 0.100 0.101 0.102 B Pro (mg-C) 01020304050 40 50 60 70 80 90 100 C Con (mg-C) 01020304050 265 266 267 268 269 270 271 D Nut (mg-P) 01020304050 3.0 3.5 4.0 4.5 5.0 E ProD (mg-C) 01020304050 69 70 71 72 73 74 F ConD (mg-C) 01020304050 69 70 71 72 73 74 Time (Day) Time (Day) Figure 3-13. Simulation results of the variables from the basic steady-state model of a microcosm. (GPP is an average of all GPP values in each day. Each state variable is based on a 30 L microcosm tank section.)

PAGE 122

122 Signs of pfd(k2-k3) (k4-k5) + + + -+ -p < threshold p threshold Intensity test + + + -+ --Frequency test disturbance Post-disturbance + 0 0 0 -0 0 + 0 0 0 -0 0 + + 0 0 Figure 3-14. Patterns of GPP time series from the simulation of the disturbance intensity and frequency models under the differ ent combinations of pfd signs between k2-k3 and k4-k5. (The horizontal lines indicate the reference GPP.)

PAGE 123

123 A GPP (mM-C/day) 01 02 03 04 05 0 0.0990 0.0995 0.1000 0.1005 p = 0 p = 1 p = 2 p = 3 p = 4 B GPP (mM-C/day) 01 02 03 04 05 0 0.0990 0.0995 0.1000 0.1005 Time (Day) C MGPP (mM-C/day) 01234 0.0990 0.0995 0.1000 0.1005D MGPP (mM-C/day) 01234 0.0990 0.0995 0.1000 0.1005 Figure 3-15. Effect of fd-5 values on GPP and MGPP patterns from the simulation of the disturbance intensity model. (T he sign combination of pfd for k2-k3 and k4-k5 was (+,+). The p threshold was 3. The disturbances were applied from Day 20 to 29 for an hour each day.) A) GPP at the fd-5 values of 0.3 (p
PAGE 124

124 A GPP (mM-C/day) 01 02 03 04 05 0 0.0990 0.0995 0.1000 0.1005 0.1010 p = 0 p = 1 p = 2 p = 3 p = 4 B GPP (mM-C/day) 01 02 03 04 05 0 0.0990 0.0995 0.1000 0.1005 0.1010 Time (Day) C MGPP (mM-C/day) 01234 0.0990 0.0995 0.1000 0.1005 0.1010D MGPP (mM-C/day) 01234 0.0990 0.0995 0.1000 0.1005 0.1010 Figure 3-16. Effect of water column nutrient percentage on GPP and MGPP patterns from the simulation of the disturbance intensity model. (The sign combination of pfd for k2-k3 and k4-k5 was (+,0). The p threshold was 3. The disturbances were applied from Day 20 to 29 for an hour each day. The fd-5 values were 0.4 (p
PAGE 125

125 A 01234 0.0998 0.0999 0.1000 0.1001 0.1002B 01234 0.0997 0.0998 0.0999 0.1000 0.1001 C 01234 0.0996 0.0997 0.0998 0.0999 0.1000 0.1001 0.1002D 01234 0.0998 0.0999 0.1000 0.1001 0.1002E 01234 0.0994 0.0996 0.0998 0.1000 0.1002 F 01234 0.0994 0.0996 0.0998 0.1000 0.1002 0.1004G 01234 0.0992 0.0994 0.0996 0.0998 0.1000 0.1002 0.1004 x-axis: Disturbance power (intensity) (p) y-axis: MGPP (mM-C/day) Figure 3-17. Possible MGPP-intensity relationships from the simulation models for the intensity tests. (Dashed line: MGPP during the 10-day disturbance period, solid line: MGPP during the 15-day post-disturbance period. Low p: p less than a threshold, high p: p equal to or greater than a threshold.) Model: S3-A Threshold p=3 Sign(pfd) for (k2-k3,k4-k5)=(+,+) fd-5(low p)=0.4 / fd-5(high p)=0.00008 Model: S1-B Threshold p=4 Sign(pfd) for (k2-k3,k4-k5)=(+,-) fd-5(low p)=0.1 / fd-5(high p)=0.00001 Model: S2-B Threshold p=3 Sign(pfd) for (k2-k3,k4-k5)=(0,-) fd-5(low p)=0.3 / fd-5(high p)=0.00002 Model: S1-A Threshold p=4 Sign(pfd) for (k2-k3,k4-k5)=(0,-) fd-5(low p)=0.1 / fd-5(high p)=0.0001 Model: S5-A Threshold p=4 Sign(pfd) for (k2-k3,k4-k5)=(+,+) fd-5(low p)=0.4 / fd-5(high p)=0.0001 Model: S4-A Threshold p=3 Sign(pfd) for (k2-k3,k4-k5)=(+,0) fd-5(low p)=0.4 / fd-5(high p)=0.0002 Model: S2-A Threshold p=3 Sign(pfd) for (k2-k3,k4-k5) Low p=(+,+)/high p=(0,+) fd-5(low p)=0.2 / fd-5(high p)=0.00005

PAGE 126

126 H 01234 0.0994 0.0996 0.0998 0.1000 0.1002 I 01234 0.0996 0.0998 0.1000 0.1002 0.1004 0.1006 0.1008J 01234 0.0996 0.0998 0.1000 0.1002 0.1004 K 01234 0.0998 0.0999 0.1000 0.1001 0.1002Figure 3-17. Continued. Model: S7-A Threshold p=3 Sign(pfd) for (k2-k3,k4-k5)=(+,-) fd-5(low p)=0.3 / fd-5(high p)=0.00005 Model: S8-A Threshold p=4 Sign(pfd) for (k2-k3,k4-k5) Low p=(0,-)/high p=(+,-) fd-5(low p)=0.3 / fd-5(high p)=0.00015 Model: S6-B Threshold p=3 Sign(pfd) for (k2-k3,k4-k5)=(-,-) fd-5(low p)=0.4 / fd-5(high p)=0.0001 Model: S6-A Threshold p=3 Sign(pfd) for (k2-k3,k4-k5)=(+,+) fd-5(low p)=0.3 / fd-5(high p)=0.00012

PAGE 127

127 A 01234 0.0998 0.1000 0.1002 0.1004 0.1006B 01234 0.0994 0.0996 0.0998 0.1000 0.1002 0.1004C 01234 0.0998 0.1000 0.1002 0.1004 0.1006 x-axis: Disturbance frequency regime (0: reference) y-axis: MGPP (mM-C/day) Figure 3-18. Possible MGPP-frequency relationships from the simulation models for the frequency tests. (Dashed line: MGPP during the five-day disturbance period, solid line: MGPP during the 15-day post-disturbance period. Low p: p less than a threshold, high p: p equal to or greater than a threshold.) Model: S4-B Threshold p=2 Sign(pfd) for (k2-k3,k4-k5) Low p=(-,+)/high p=(+,0) fd-5(low p)=0.02 / fd-5(high p)=0.0001 Model: S3-B Threshold p=2 Sign(pfd) for (k2-k3,k4-k5)=(-,-) fd-5(low p)=0.1 / fd-5(high p)=0.00008 Model: S5-B, S7-B, S8-B Threshold p=2 Sign(pfd) for (k2-k3,k4-k5)=(0,-) fd-5(low p)=0.02 / fd-5(high p)=0.0001

PAGE 128

128 4 CHAPTER 4 DISCUSSION Conclusions This study investigated t he effects of disturbance on ecosystem processes, gross primary productivity (GPP) and ecosystem re spiration rate (ER), using microcosms and simulation models. The effects of distur bance were studied from the perspective of three major topics in ecology: the interm ediate disturbance hypothesis, ecosystem maturity, and pulsing paradigm. The analys es of the microcosm experiments and subsequent simulation models were ultimate ly synthesized to draw the following conclusions: The relationship between ecosystem GPP and disturbance intensity regimes can be positive monotonic, negative monotonic, U-shaped, peaked, or non-significant. Although the relationship between G PP and disturbance frequency regimes can be various, GPP is generally higher under discrete dist urbances than under continuous disturbances when the total ener gy is the same between the discrete and continuous disturbances. The relationships between G PP and disturbance regimes may be determined by changes of e fficiencies in energy flow pathways of a system under disturbances and by a disturbance threshold above which the efficiencies are permanently altered. The effects of disturbance on ecosystem processes depend on the maturity of the ecosystem at the time of the disturbance. If an ecosystem reinforces interactions among biotic and abiotic components over time, a less mature ecosystem is generally more resilient but less resistant to external disturbances than a mature counterpart. Whereas internal processes of a less mature ecosystem tend to be significantly affected by external distur bances, those of a ma ture counterpart tend be more influenced by internal trajecto ry of succession than by the external disturbances. The pulsing pattern of GPP or ER is in fluenced by disturbances. The wavelengths and peak amplitudes of GPP and ER pul sing patterns are amplified under disturbances, compared with a reference stat e. This tendency is more obvious in a less mature ecosystem than in a mature counterpart.

PAGE 129

129 Discussion Credibility of MGPP-Disturbance Relationships in the Microcosms The credibility of the rela tionships between mean GPP (MGPP) and disturbance regimes could be affected by pH m easurement error. As shown in Figure 3-11, the value of [TCO2]/( 0.01 pH) increased as pH devia ted from 8.26 or alkalinity increased. The U-shaped relationship between [TCO2]/( 0.01 pH) and pH in Figure 3-11 was attributed to the low [TCO2]/( 0.01 pH) values within the pH range of 7.5 9.0 in the pH-[TCO2] graphs as shown in Figure 2-2. S3-A and S3-B had pH values in the range of 5.1.8, whereas pH values of the other microcosms were distributed in the range of 6.5.0. Thus, S3-A and S3-B could have the greatest [TCO2] error due to pH measurement error. Among the microcosms, the MGPP-disturbance relationships of S1-B, S5-A, S5-B, S6-B, S8-A, and S8-B were most re liable and not likely to change by pH measurement error, as shown in Table 3-8. When the credibi lity of the MGPP-disturbance relationship is considered, Table 3-8 must be carefully inte rpreted. Because MGPP for a post-disturbance period was calculated by averaging 15-day GPP values, the resultant MGPP-disturbance relationship was likely to be affected by pH measuremen t error when the pH error occurred every day at peak pH points by which the daily GPP is calculated or when large pH errors occurred over a couple of days at peak pH points. In addition, the maximum or minimum [TCO2]/( 0.01 pH) in each microcosm presented in Table 3-8 is the extreme that rarely occurs. Thus, a simp le comparison between a minimum MGPP difference and a maximum -2 [TCO2]/( 0.01 pH) in each microcosm does not guarantee the credibility of a resultant MGPP-disturbance relationship, but it provides a guideline for determining the significance of the relationship.

PAGE 130

130 The credibility of MGPP-disturbance relationships could also be affected by a gas exchange coefficient, k. The c hange of k did not influence the MGPP-disturbance relationships in most microcosms, but it significantly changed the values of MGPP or MGPP-disturbance relationships in S1-B, S3-A, S3-B, and S4-B. Because a daily GPP of each section of a microcosm was determined by peak values in a [PCO2] time series, whether the daily [PCO2] patterns remained under different k values was critical for constant MGPP values or MGPP-disturbance relationships, even if the [GCO2] corrections changed the overall level of [PCO2] in the long term. The change of [PCO2] patterns by different k values in Figure 3-10 indicates the significant contribution of CO2 gas exchange across the air-water interface in S3-A. The k values of 1, 2, and 4 cm/h not only eliminated diel [PCO2] fluctuations but also raised the overall [PCO2] levels in S3-A because the theoretical [GCO2] was much greater than [TCO2] in the low pH range of 5.1.8. Inversely, the loss of [PCO2] diel patterns by [GCO2] corrections in S3-A, as shown in Figure 3-10, implies that a gas transfer ve locity (k) greater than 1 cm/h may not be reasonable for the [GCO2] corrections. If a microc osm is sustainable, production and consumption must be balanced, which is r epresented by diel fl uctuations of [PCO2]. Although the loss of diel fluctuations in a [PCO2] time series may be observed in an extremely heterotrophic system for several days, the heterotrophy cannot be sustained for a long time without significant primary pr oduction or external food supplies. Thus, a reasonable k value may be between 0 and 1 cm/h in the laboratory environment. Then the [GCO2] correction will have negligible effects on the patterns of GPP time series or MGPP-disturbance relationships of the microcosms.

PAGE 131

131 Water motions during the disturbance period might influence MGPP-disturbance relationships. Borges et al. (2004) estimated k values contributed by water current in an estuary. According to their estimation, k was about 4.5 cm/h at water velocity of 77 cm/s with an average water depth of 11 m in a low wind condition. In the microcosms, the average surface water velocity during disturbance was less than 5 cm/s at intensity 62. Although k values may be different among sections under distinct water motion intensities during di sturbance events, the difference influences GPP or ER only during the di sturbance period. In additi on, the difference could be minor because water motions were applied onl y 1 hours a day in the intensity tests and maximum 3 hours a day in the frequency te sts. The surface water motion was negligible at intensity 8, which was cont inuously applied for 105 hours to Sec 1 of the microcosms for the frequency test. The method to calculate dayti me respiration rate (ERd) might have affected the accuracy of GPP and ER estimation. Maloney et al. (2008), in the estimation of stream metabolism from the change of dissolved o xygen (DO) concentration, calculated ERd by averaging the nighttime respiration rate s measured one hour bef ore dawn and one hour after dusk. But the calculati on from a short-term change of O2 or CO2 concentration, where the number of dat a is small, may generate inaccura te GPP or ER values if the concentration change includes va riations from a straight line or smooth curve. The [PCO2] curves obtained from the microcosm tests showed interm ittent variations from straight lines or smooth curves, although the overall [PCO2] for a 12-hour period showed clear patterns of lines or curves in most microcosms. The occasional deviation of [PCO2] from a normal curve seemed to be caus ed by irregular homogenization of pH

PAGE 132

132 in the water column or by discrete metabolic processes over time. As Beyers (1963a) pointed out, the daytime or nighttime [PCO2] change formed a curve, where the magnitude of the slope gradually decreased, in certain microcosms. The 12-hour average slope of [PCO2], which could be estimated by averaging tangent lines, may be close to the slope between two peak values. The accuracy of ERi+1,d calculated by averaging the two ERn values, ERi,n and ERi+1,n, would be decreased if a nighttime respiration rate is significantly different from a daytime respiration rate because the calculation of a daytime respiration ra te was based on the assumption of linear relationship among ERi,n, ERi+1,d, and ERi+1,n. The alkalinity levels of microcosms we re measured once a week to minimize the sampling of water, and their values between two measurement points were estimated by linear interpolation. Although alkalini ty levels may not change linearly between two measurement points, alkalinity values am ong the four sections of each microcosm showed similar levels and trajectories over time. Thus, the comparison of the relative GPP or ER levels among the sections in a microcosm was not likely to be affected by the alkalinity measurement intervals. The credibility of MGPP-disturbance relationships could also be affected by several problems identified during the experiments. The disagreement of GPP values among the four sections during the initial stabilization period was observed in S3, S4, and S5. In S3, the one-day initial stabilization per iod was too short for the microcosms to represent diel fluctuations of [PCO2]. The incomplete diel fluctuations of [PCO2] during the initial stabilization period in S3-A and S3-B produced negative GPP and ER values on Day 1. In S5-A and S5-B, the initia l errors in pH readings resulted in the

PAGE 133

133 disagreement of GPP values among microcosm sections, although the problem was fixed during the disturbance period. The long-term submersion of the pH electrodes in a microcosm for 22 days sometimes altered the responses of electrodes. After one microcosm set was terminated, pH elec trodes were cleaned and stored in a pH 4.01 standard solution or 4 M KCl solution until t he next microcosm test. But pH readings sometimes disagreed with one an other during the initial st abilization period of a new microcosm even after initial calibrations so that the pH readings were carefully monitored and the pH meters were recalibrated whenever necessary. The maintenance practice of microcosms har dly affected the test results. Each section of a tank was refilled with 1 L of deionized water to maintain the water level. The pH did not deviate from the normal traj ectory by refilling because the amount of added water was small, compared with the large vo lume (30 L) of water in each section. Even if the pH deviated by t he added water, it quickly retur ned to the normal trajectory within several minutes. Even in the microcosms S3 where alkalinity level was as low as 0.06 meq, the pH did not deviate from the normal trajectory by the added water. The daily cleaning of the pH electrodes by deioni zed water also did not affect the pH readings. The pH readings returned to their nor mal trajectory within a couple of minutes after being washed by deionized water in a squeeze bottle. No maintenance was done during hour at 6 am and 6 pm to mini mize the influence of maintenance on the calculation of GPP and ER. The pH meters were calibrated once every week, and pH values usually drifted less than 0.05 a week The disagreement of pH values before and after calibration might have affected GPP values of the day the pH meters were calibrated.

PAGE 134

134 Self-Organization Self-organization was a key concept in t he microcosm studies. The microcosms set up by sampled water and sediment sources went through the self-organizing processes. The increase, dec rease, or fluctuation of GPP and ER during the initial stabilization period in each micr ocosm may reflect a systems selection or elimination of biotic and abiotic elements as well as fine-t uning of their interactions. The initial increase or decrease of state variables in the steady-state simulation model resembled the initial self-organization of the microcosms in that the state variables significantly changed until all flows of energy and materials are balanced in the system as a whole. The self-organization of microcosms s eemed to balance internal production and consumption, which was infe rred from the GPP/ER pattern s. The GPP/ER patterns were not correlated with disturbance regimes, but they were likely to be a function of time. In the microcosms, the GPP/ER generally converged to 1 over time, which implied that the systems self -organized to balance the pr oduction and consumption for sustainability ( Table 3-3 and Appendix A). The microcosms also seemed to change the in ternal organization over time so that the responses of systems to disturbance regi mes were different depending on the time of the disturbances since t he initial microcosm setup. Without self-organization, external forces would shape the internal structures and proce sses of the system. In the maturity tests, however, t he two replicated microcosms of A and B, applied with the same external disturbance regi mes but at different times, showed very different results in GPP, pulsing patterns, resistance, and res ilience. While different responses of GPP and ER between A and B microcosms were cl early observed in S1 and S6, they were obscure in S2. It was conjectured that S2-A and S2-B did not show clear differences in

PAGE 135

135 the responses of GPP or ER to distur bances because they already had an optimal condition of components and interactions in the given environment during the initial setup. In this regard, ecosystems may self-organize to reduce the gap between the current status of a system and the pref erred goal in a given environment. Effect of Disturbance Intensity In disturbance studies, it is uncertain how long the effects of a disturbance last during the post-disturbance peri od in a system. Because environmental or internal factors other than the dist urbance influence the structur es and processes of the ecosystem, it would become obscure whether a certain change of traits in the system is caused by the disturbance or another factor unless it is ex amined in a short period after the disturbance. In the mi crocosm studies, each microcosm was monitored for 15 days after disturbance because not only it was an assumed average turnover time of zooplankton, a major high-trophic-level species group in the microcosms, but also it was the shortest time to observe the response of one generation of the high-trophic-level species. In the microcosms, the effect of disturbance on GPP wa s represented by a deviating pattern of GPP of a disturbed syst em from a referenc e GPP of an undisturbed system continuously from the beginning of the disturbance until the end of the post-disturbance period. Neverthel ess, this study showed the MGPP-disturbance relationships of the disturbance and pos t-disturbance periods separately. The separation of periods was useful for analyzi ng how disturbance effects were different between the disturbance and pos t-disturbance periods. In most microcosms, the MGPPdisturbance relationship of the combined period of disturbance and post-disturbance was close to that of the 15-day pos t-disturbance period. Thus, the MGPP-disturbance

PAGE 136

136 relationship of the combined period of dist urbance and post-disturbance can be inferred from that of the 15-day pos t-disturbance period. From the results of the in tensity tests, no common MGPP-intensity relationship was discovered ( Figure 3-3). Even in S5-A and S8-A, where the MGPP-intensity relationships remained throughout the post-di sturbance period, the rela tionships were U-shaped and peaked, respectively. Except for S3-A, where the maximum MGPP difference was within the 0.01 pH error range, the other 10 microcosms showed various MGPP-intensity relationships of positive monotonic, negative monotonic, Ushaped, peaked, or nonsignificant. Because the intermediat e disturbance hypothesis (IDH) has been addressed to explain the relationship between disturbance r egimes and ecosystem diversity, it may be unreasonable to apply t he mechanism of the IDH (the competitioncolonization trade-off) to the MGPP-disturbance relationship. Nevertheless, the peaked relationship in the IDH could be addre ssed to determine whether intermediate disturbance regimes are optimal conditions fo r the highest level of ecosystem traits, such as GPP or ER. S1-B, S2-A, S4-A, and S5-A never showed a peaked MGPPintensity relationship during the disturbance or post-distur bance period. The IDH cannot be completely negated, however, even if it does not explain all MGPP-intensity relationships of the microcosms. It is in triguing that the peaked relationships were shown only in S6-B and S8-A during t he 15-day post-disturbance period but their MGPP-intensity relationships were quite reliable even under the [GCO2] corrections and potential pH measurement errors. The various MGPP-intensity relationships among the microcosms imply that different mechanism s operate in each micr ocosm. As Connell (1978) pointed out, it may be more reasonabl e to regard the IDH not as the only

PAGE 137

137 explanation for the disturbanc e-trait relationships but as one component among several mechanisms. The resultant relationshi p between an ecosystem trait and disturbance regimes may be peaked if the ef fect of the IDH is dominant in the ecosystem. There was no clue that the MGPP-disturbance relationship was affected by the experimental duration plan. For exampl e, S1-A and S6-A showed positive monotonic and negative monotonic relationships durin g the 15-day post-disturbance period, respectively, even though they were test ed under the same 5-1015 duration plan. Although the insi gnificance of MGPP[16-30] differences among intensities 0, 28, and 50 in S1-A may change the MGPP-intensity relationship, the MGPP-intensity relationship is unlikely to be negative monotonic because of the significant difference of MGPP[16-30] at intensity 62 from the others. MGPP-intensity relationships may differ accord ing to the initial biotic and abiotic components of the microcosms. It seemed that samples co llected from the same lake but at different times had distinct compos ition of species and inorganic matters. For example, S2-A and S6-A were constructed with Lake Alice water collected at different times. While S2-A showed U-shaped MGPP-intensity relationship, S6-A showed positive monotonic, peaked, or negative monotonic re lationships during the post-disturbance period in the intensity range of 0. Lake Alice, managed by the University of Florida, has shown several algal blooms every year, which the university manages by spraying chemicals. Thus, the microcosms S2-A S6-A, and S7-A might have had distinct combinations of species caused by the management practice or seasonal change. It was demonstrated that the MGPP-disturbance relationships were not simply formed by internal factors or variation of microcosms other than water motion

PAGE 138

138 disturbance. S6-A and S6-B clearly showed the difference of primary production between disturbed and undist urbed sections ( Figure 3-4). In most microcosms, however, primary production was not discernib le by the naked eye so that it could be recognized by analyzing pH and alkalinity data. Effect of Disturbance Frequency The GPP and ER of S4-B, sampled from Lake Wauberg, declined throughout the experimental period. The continuous decline of the GPP over time may indicate that the components of the microcosm, S4-B, had difficu lty in adapting to the new environment of the laboratory apart fr om their original habi tat. In S4-B, the MGPP values of the discretely disturbed sections were higher t han that of the cont inuously disturbed one during the disturbance per iod, whereas the MGPP of the continuously disturbed section was highest during the post-disturbance period. S4-B did not follow the patterns observed in S5-B, S7-B, and S8-B, where MGPP values under discrete disturbances were greater than that under continuous o ne. But it should be noted that MGPPfrequency relationship of S4-B may change by 0.02.03 pH error. In each microcosm pair A and B of S3, S4, S7, and S8, where each microcosm pair was established by the same wate r and sediment source and tested under the same duration plan, the MGPP value of Sec 1 of microcosm B (continuously disturbed section in the frequency test) were always hi gher than that of Se c 1 of microcosm A (undisturbed section in the intensity test) during the 15-day post-disturbance period. This may imply that a low-intensit y continuous disturbance increases MGPP, although a high-intensity discrete di sturbance either increases or decreases the MGPP depending on the microcosms and disturbance regimes as discussed in the result s of the intensity

PAGE 139

139 tests. The comparison of the two microcosms A and B in S3, S4, S7, and S8 must be careful, however, because microcosms A and B were not cross-seeded each other. The IDH was not also supported in the frequency tests. The IDH hypothesized that the level of an ecosystem trait is maximized in the middle range of disturbance frequency. In S5-B, S7-B, and S8-B, the MGPP value under discrete disturbances was minimal in regime 3, where the disturbance frequency was between those of regimes 2 and 4, during the post-disturbance period ( Figure 3-5). Even if t he middle range of frequency regimes, where MGPP is maximized, was out of the frequency range, regimes 1, the MGPP-frequency relationships among the regimes 1, 2, and 3 should have been positive monotonic or negative monot onic to support the IDH. Bu t more importantly, it is necessary to precisely define frequency in the IDH whether t he different frequency levels are based on the same disturbance power or on the sa me disturbance energy. Disturbance and Ecosystem Maturity The premise under the maturity tests was that a microcosm develops from dispersed individuals to an interconnected whol e. But the development of a freshwater aquatic microcosm did not always follow this premise. While a certain microcosm may start from unassociated co mponents and gradually develop the interaction, the other may initially have interconnected individuals and lose the connections under the new container environment over time. In S2-A for example, the G PP/ER deviated from 1 over time in all sections in contrast to the general tendency of GPP/ER approaching 1 ( Table 3-3). If S1-A, S2-A, and S6 -A are defined as less ma ture systems, and S1-B, S2-B, and S6-B as mature syst ems, resistance was generally higher in the mature systems whereas resilience was higher in the less mature systems in terms of GPP or

PAGE 140

140 ER. But this general pattern of resistanc e and resilience was not clear in S2-A and S2-B ( Table 3-7). Recalling the argument by Margalef (1963) that less mature ecosystems are easily affected by external forces and that mature counterparts are more affected by internal factors, the MGPP-intensity relationships of the post-disturbance periods were similar to those of the di sturbance periods in the less mature systems, whereas the MGPP-intensity relationships of the post-dist urbance periods were shaped regardless of the patterns during the disturbance periods in the mature systems. But such a tendency was not observed in S2-A and S2-B. Al though mature systems may better resist disturbances than less mature counterparts, t hey may not easily return to a reference state once the complex internal ne tworks are destroyed by disturbances. The timing of a disturbance was a critical but difficult component of treatments in the microcosm tests. If a disturbance is app lied too early after t he initial microcosm setup, the influence of di sturbance on GPP or ER may not be easily observed because the system can be highly resilient. On the contrary, the change of GPP or ER may be strongly influenced by a systems internal fact ors if the system is disturbed too late. Although the appropriate timi ng of disturbance to observe the clear effects of disturbance must be between those extremes, each microcosm sample may have distinct timing where the effects of disturbance are clearly shown. Disturbance and Pulsing The words pulsing and disturbance are often used interchangeably in ecological studies. The two terms are simila r in that they generally describe discrete or periodic events exerted on ec osystems. But pulsing seems to have a more comprehensive meaning than disturbance. Although disturbance implies any discrete

PAGE 141

141 events destroying or modifying a ll or part of a system configurat ion, the word pulsing is used for all kinds of periodic or discrete ev ents regardless of its e ffects on the system. The internal oscillation of population or of GPP is a kind of pul sing. The repeating photoperiod of sunlight is also a kind of pulsing. Any kind of natural or anthropogenic disturbance may be called pulsing. It was not simple to calculate and PA from the GPPP because the pulsing pattern was sometimes obscure. In some cases, a maximum GPPP was slightly below the zero line, while a minimum GPPP was above the zero line. A temporary deviation of a pattern within a large pulsing, which look ed like a pulsing, was not counted as an internal pulsing. An incomplete pulsing cycle was not counted as a pulsing unless the or PA of the incomplete pulsing was greater than that of the ot her complete pulsing patterns. The higher p/ d and PAp/PAd in the disturbed sections than the undisturbed one in S1-A, S2-A, S5-A, and S6-A indicated that disturbed microcosms generally amplified the wavelengths and peak amplitudes of GPPP and ERP time series more than the undisturbed microcosms did w hen they are disturbed early si nce the initial setup. Several exceptions indica ted with the underlines in Table 3-5 were minor. The amplified and PA by disturbances, howev er, do not imply that the or PA of disturbed sections simply increased from the disturbance period to the post-dis turbance period. In S1-A, S2-A, S5-A, and S6-A, the and PA of the GPPP time series decreased over time in the undisturbed Sec 1, where p/ d and PAp/PAd were lower than 1. The p/ d and PAp/PAd of the disturbed Sec 2, 3, and 4 were relative ly higher than those of the reference Sec 1,

PAGE 142

142 even if the p/ d and PAp/PAd were less than 1. Thus, it means that and PA were amplified under disturbances compared with their refe rence trajectories. It was not discovered why microcosms amplified and PA under disturbances in the early stage of the te st period. Because and PA were amplifie d, the product of them, PA, was also amplified under disturbances. It c an be conjectured that the expedited material recycling by disturbance may increase the total production, which is represented by the amplified PA, in the microcosms S1-A S2-A, S5-A, and S6-A. But this conjecture was incorrect because the disturbance decreased MGPP in S5-A and S6-A despite of the amplified and PA. The mechanisms of the internal GPP pulsing under disturbance may be more than the mate rial recycling. Also, it is unknown whether the amplified and PA are passive changes caused by disturbances or whether they are a strategy of a system for survival against the disturbances. Possible Mechanisms of Disturban ce Effects from Simulation Models Although the simulation models were us ed to find possible mechanisms of the patterns discovered in the microcosm tests, it was challenging to embed some features of ecosystem development in the models. In the simulation models employed in this study, the configurations of the systems were assumed to be static over time. In reality, an ecosystem or microcosm self-organizes by accommodating, eliminating, or modifying components and interactions over time. The self-organization process is critical to understand how a system develops in the temporal or spatial scales. In this regard, it was difficult to represent the maturity tests in the simulation models because first it must be determined how self-organization or su ccession generally develops the systems configuration. The use of simulation model s was limited to the identification of the

PAGE 143

143 mechanisms for the MGPP-disturbance relationships discovered from the intensity and frequency tests of t he microcosms. The simulation models suggested some possible mechanisms for generating the MGPP-disturbance relationships. All MGPP-disturbance relationships observed in the microcosm studies were represented by the models when t he hypothesis on the disturbance power and its threshold was applied to the simulation models. The U-shaped, peaked, or non-significant MGPP-disturbance relationship was supported by the hypothesis that the effect of dist urbance on the reproduction coefficients is temporary or permanent depending on the disturbance power (p) and its threshold. As the absolute value of the pfd increased, a time series of GPP tended to be more distant from the reference GPP line during the post-distur bance period regardless of the threshold of p. Thus, if there is no threshold of p determi ning the temporary or permanent coefficient change, the MGPP-disturbance relationship will be positive or negative monotonic in the result of simulation model. Although the MGPP-disturbance relationships of the post-dist urbance period resulting from the microcosms were easily generated by varying sign combinations of pfd for k2-k3 and k4-k5 in the simulation models, the p threshold, pfd sign, and fd-5 had to be narrowed to satisfy the MGPP-disturbance relationships for both the disturbance and post-dist urbance periods at the same time. In the simulation models, p and fd were important distur bance parameters forming the distinct MGPP-disturbance relationships. While the power level (p) represented the intensity of a disturbance the disturbance factor (fd) indicated how vulnerable an ecosystem is under disturbances. The combination of p and fd showed the reality of

PAGE 144

144 disturbance in nature where the results of disturbance events are determined by interactions between the dist urbances and biotic component s of the ecosystem. The various sign combinations of pfd were critical in the models to represent the effects of disturbance on specific energy pat hways, contrary to the prev ious studies on the effects of disturbance (e.g., Kondoh, 2001; Odum and Odum, 2000, p.241; Roxburgh et al., 2004), where disturbances had only destructive effects on the struct ures and processes of an ecosystem. There were several discrepancies betw een the results of microcosms and simulation models. In the model of the microcosm S1-B (B of Figure 3-17), GPP was maximum at power level 3 during the di sturbance period and the model could not represent the positive monotonic relationship shown in the microcosm. But it is possible that intensity 62 of the microcosm S1-B is lower than power level 4 of the simulation model so that the MGPP-intensity relationship may be peaked during the disturbance period with a maximum MGPP at intensity 62. The model of the microcosm S2-A (C of Figure 3-17) did not have the lowest MGPP value at power level 4 during the postdisturbance period, whereas the microcosm result had the lowest value at intensity 113. This might be caused by the shortage of intensit y levels in the simulation model. If the simulation model had higher power levels, lower MGPP values would be observed during the post-disturbance period. The model of the microcosm S6-B (I of Figure 3-17) did not show the peaked MGPP-intensity relationship of the microcosm because of the lower MGPP at power level 3 than that at power level 4. The MGPP-intensity relationships, however, will be the same between the model and the microcosm if in tensity 62 of the microcosm is equivalent to power level 3 of the simulation model.

PAGE 145

145 The simulation models for MGPP-disturbance relationships of the microcosms could be improved by adding more details to the hypotheses. The assumption of the simulation models that the power threshol d remains throughout the experimental period may be unrealistic in certain microcosms If repeating dist urbances alter the reproduction or death rates of a microcosm by structural change, they may influence the systems power threshold on the edge of whic h the response of the microcosm to the disturbances dramatically changes during the experimental period, as argued by Scheffer et al. (2001) in t he alternative stable state hypothesis. Thus, a new assumption or hypothesis needs to be established to apply the dynamic power threshold in the models. The dynamic pow er threshold may generate more diverse and concrete MGPP-disturbance relationships depending on the timing of the threshold change. In the simulation models, GPP changed by less than 0.001 mM-C/day, which is 1 or 2 orders of magnitude less than the microcosm results. Because the simulation models were built based on one steady-state mi crocosm, they could not represent the same GPP levels observed in the microcos m tests. Instead, the simulation models were intended to generate and represent the general patterns observed in the microcosms. To match the GPP values as well as those of ot her state variables, different parameters of t he models should be used. Trade-Offs in the Study The results of scientific studies are confirmed by repeated expe riments. On the one hand, the repetition im plies the replication of a sample by which the significance of treatment effects in the sample is eval uated in a statistical method. Thus, the replication of the sample and its statistical analysis disti nguish the treatment effects

PAGE 146

146 from any undetectable instrumental or expe rimental error. On the other hand, the repetition implies multiple experiments under the same treatment sequence but different samples with distinct initial conditions. The multiple experiments expand the range of possible results from various initial condition s. In this regard, both replication and multiple experiments are critical to increas e the credibility of the experimental results and they minimize a bias occurring from the observation of a single case. The availability of a scientific investigation, however, is constrained by the trade-offs between temporal, spatial, and economic aspec ts. For example, if the size of a microcosm increases, it can accommodat e higher-trophic-level species, which eventually make the microcosm resemble a r eal ecosystem. But it requires more space and eventually decreases the available numbe r of replicates. Although increasing the number of replicates increases the credibility of experim ental results, replication increases economic burden and management load. In this study, multiple experiments were employed at the exp ense of replication under the constraints of time, space, and money. The eight microcosm sets, where each set was composed of two microcosms, were sampled from different lakes at different times. Considering that the major role of a microcosm is to generate hypotheses that can be further tested in t he field, the multiple experiments without replication quickly provide new hypotheses on the problem under inve stigation or test research questions as a prelimin ary study. In this study, si gnificance of the results was discussed using the possible instrumental error ranges and uncertainty analyses of parameters to complement the absence of replication. T he multiple experiments from different samples were critical in this study because each microcosm represented the

PAGE 147

147 distinct MGPP-disturbance relationship. If only one or two samples had been studied, the conclusion might have been much different: s upporting either the I DH or one specific MGPP-disturbance relationship. Only lakes were selected as sampling sites for the microcosms because one type of disturbance may have different effects on different ecosystem types with distinct mechanisms. One challenge in any disturbance study is that it is hard to examine the responses of an ecosystem under an extensive range of disturbance regimes. Although it is possible to apply many disturbance regimes in the microcosm expe riments, they are bound by the constraints of space, time, and m oney. If more sections in the microcosm tank were employed, the resultant MGPP-intensity relationships in the intensity tests of the microcosms might have been further suppor ted or rejected. In all microcosms except for S2-A and S2 -B, the maximum intensity of di sturbance was 62. Because the MGPP values are unknown regarding the intensities above 62 or th e intensities inbetween the tested levels, the MGPP-intensity relationships must be carefully interpreted. Future Work The results of this study suggested possi ble research topics to be further studied using microcosms or simulation models, and provided hypotheses that could be tested in the field. The suggested topics are as follows: As prevalent in nature, the combined dist urbance types occurring at the same time or in a sequence may distinctly influence the ecosystem processes, compared with the effects of one disturbanc e type. For example, t he two disturbance types of nutrient and water motion may show distinct MGPP-disturbance relationships depending on the timings and thei r regimes. A factoria l experiment would be employed to test the effe cts of the combined di sturbances with various combinations of timings and regimes of the disturbances. T he experiment could be done by making more sections in a micr ocosm tank. Simulation models could provide possible mechanisms of resultant MGPP-disturbance relationships based on a hypothesis on the mechanisms of distin ct effects of each disturbance type. A dynamic power threshold would be included in the simulation models.

PAGE 148

148 Disturbance events in nature do not occur regularly like the micr ocosm tests. The effects of the randomly generated disturbances on GPP or ER could be studied using the same aquatic microcosm systems. The pumps generating water motion disturbances could be controlled to operat e with random intens ities, frequencies, and durations. In the ex periments, the randomly generated disturbance regimes could be analyzed using indices such as total energy, average power, average frequency, variation of power or frequency, and so on. This study proposed energetic mechanisms of disturbance effects using simulation models. Modifying the details of model s and disturbance fa ctors would improve the results and suggest virt ual disturbance models for ecosystems. First, adding compartments of higher-trophic-level s pecies would make the models more realistic. Second, paramet ers for the dynamic power threshold could be estimated from resistance and resilience data of t he ecosystems and include d in the models. Third, the change of efficiencies or confi gurations of energy flow pathways could be diversified. For example, the effici encies of energy flow pathways may change in other pathways such as microbial dec ompositions. The configuration of the energy flow pathways would change under disturbance depend ing on the power level. For example, disturbances may pr ompt a quadratic drai n of biomass from the producers or consumers. Simulation models may be useful to understand the mechanisms of internal GPP or ER pulsing. Feedback controls would be included in the m odels to generate the oscillating patterns of GPP. Once a model generating internal GPP pulsing is established, disturbance fa ctors would be included. Parameters or pathway configurations would be varied to discove r possible mechanisms of the amplified pulsing wavelength and amp litude under disturbance. The results of the microcosms and simulation models suggested that the relationship between MGPP and disturbance regime can vary depending on disturbance regimes and the responses of an ecosystem to the regimes. This result could be tested in the field. Fo r example, change of nutrient concentration and its timing by irregular nutrient inflow s are detectable dist urbance regimes. Because disturbances occur sequentially in the field, BACI (Before-After/ControlImpact) design (e.g., Conquest, 2000; Mont efalcone et al., 2008) should be used.

PAGE 149

149 5 APPENDIX A GPP, MGPP, AND GPP/ER OF MICROCOSMS A 0.00 0.10 0.20 0.30 051015202530Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.07 0.09 0.11 0.13 691215Time (Day)MGPP[6-t] (mM-CO2/day)C 0.07 0.09 0.11 0.13 1619222528Time (Day)MGPP[16-t] (mM-CO2/day)D 0.6 1.0 051015202530Time (Day)GPP/ER Figure A-1. Time series of S1-A. A) GPP. B) MGPP[6-t]. C) MGPP[16-t]. D) GPP/ER.

PAGE 150

150 A 0.00 0.10 0.20 0.30 051015202530354045Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.07 0.09 0.11 0.13 21242730Time (Day)MGPP[21-t] (mM-CO2/day)C 0.07 0.09 0.11 0.13 3134374043Time (Day)MGPP[31-t] (mM-CO2/day)D 0.7 1.0 051015202530354045Time (Day)GPP/ER Figure A-2. Time series of S1-B. A) GPP. B) MGPP[21-t]. C) MGPP[31-t]. D) GPP/ER.

PAGE 151

151 A 0.00 0.03 0.06 0.09 051015202530Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.02 0.03 0.04 0.05 691215Time (Day)MGPP[6-t] (mM-CO2/day)C 0.02 0.03 0.04 0.05 1619222528Time (Day)MGPP[16-t] (mM-CO2/day)D 0.5 1.0 0 5 10 15 20 25 30Time (Day)GPP/ER Figure A-3. Time series of S2-A. A) GPP. B) MGPP[6-t]. C) MGPP[16-t]. D) GPP/ER.

PAGE 152

152 A 0.00 0.03 0.06 0.09 051015202530354045Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.02 0.04 0.06 21242730Time (Day)MGPP[21-t] (mM-CO2/day)C 0.02 0.04 0.06 3134374043Time (Day)MGPP[31-t] (mM-CO2/day)D 0.5 1.0 051015202530354045Time (Day)GPP/ER Figure A-4. Time series of S2-B. A) GPP. B) MGPP[21-t]. C) MGPP[31-t]. D) GPP/ER.

PAGE 153

153 A 0.00 0.05 0.10 0.15 0.20 05101520Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.08 0.12 0.16 0.20 23456Time (Day)MGPP[2-t] (mM-CO2/day)C 0.08 0.12 0.16 0.20 710131619Time (Day)MGPP[7-t] (mM-CO2/day)D 0.5 1.0 0 5 10 15 20Time (Day)GPP/ER Figure A-5. Time series of S3-A. A) GPP. B) MGPP[2-t]. C) MGPP[7-t]. D) GPP/ER.

PAGE 154

154 A 0.00 0.20 0.40 0.60 05101520Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.06 0.12 0.18 0.24 0.30 23456Time (Day)MGPP[2-t] (mM-CO2/day)C 0.06 0.12 0.18 0.24 0.30 710131619Time (Day)MGPP[7-t] (mM-CO2/day)D 0.5 1.0 05101520Time (Day)GPP/ER Figure A-6. Time series of S3-B. A) GPP. B) MGPP[2-t]. C) MGPP[7-t]. D) GPP/ER.

PAGE 155

155 A 0.00 0.05 0.10 0.15 05101520Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.05 0.08 0.11 0.14 23456Time (Day)MGPP[2-t] (mM-CO2/day)C 0.03 0.04 0.05 0.06 0.07 710131619Time (Day)MGPP[7-t] (mM-CO2/day)D 0.5 1.0 05101520Time (Day)GPP/ER Figure A-7. Time series of S4-A. A) GPP. B) MGPP[2-t]. C) MGPP[7-t]. D) GPP/ER.

PAGE 156

156 A 0.00 0.05 0.10 0.15 05101520Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.05 0.07 0.09 0.11 23456Time (Day)MGPP[2-t] (mM-CO2/day)C 0.04 0.05 0.06 0.07 0.08 710131619Time (Day)MGPP[7-t] (mM-CO2/day)D 0.5 1.0 05101520Time (Day)GPP/ER Figure A-8. Time series of S4-B. A) GPP. B) MGPP[2-t]. C) MGPP[7-t]. D) GPP/ER.

PAGE 157

157 A 0.00 0.10 0.20 0.30 051015202530Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.08 0.11 0.14 0.17 691215Time (Day)MGPP[6-t] (mM-CO2/day)C 0.08 0.11 0.14 0.17 1619222528Time (Day)MGPP[16-t] (mM-CO2/day)D 0.5 1.0 0 5 10 15 20 25 30Time (Day)GPP/ER Figure A-9. Time series of S5-A. A) GPP. B) MGPP[6-t]. C) MGPP[16-t]. D) GPP/ER.

PAGE 158

158 A 0.00 0.10 0.20 0.30 0510152025Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.06 0.08 0.10 0.12 0.14 678910Time (Day)MGPP[6-t] (mM-CO2/day)C 0.12 0.14 0.16 0.18 0.20 1114172023Time (Day)MGPP[11-t] (mM-CO2/day)D 0.5 1.0 0510152025Time (Day)GPP/ER Figure A-10. Time series of S5-B. A) GPP. B) MGPP[6-t]. C) MGPP[11-t]. D) GPP/ER.

PAGE 159

159 A 0.00 0.10 0.20 051015202530Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.03 0.04 0.05 0.06 0.07 691215Time (Day)MGPP[6-t] (mM-CO2/day)C 0.03 0.05 0.07 0.09 1619222528Time (Day)MGPP[16-t] (mM-CO2/day)D 0.5 1.0 0 5 10 15 20 25 30Time (Day)GPP/ER Figure A-11. Time series of S6-A. A) GPP. B) MGPP[6-t]. C) MGPP[16-t]. D) GPP/ER.

PAGE 160

160 A 0.00 0.30 0.60 0.90 051015202530354045Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.03 0.05 0.07 0.09 21242730Time (Day)MGPP[21-t] (mM-CO2/day)C 0.00 0.20 0.40 0.60 3134374043Time (Day)MGPP[31-t] (mM-CO2/day)D 0.5 1.0 051015202530354045Time (Day)GPP/ER Figure A-12. Time series of S6-B. A) GPP. B) MGPP[21-t]. C) MGPP[31-t]. D) GPP/ER.

PAGE 161

161 A 0.00 0.03 0.06 0.09 0510152025Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.02 0.04 0.06 678910Time (Day)MGPP[6-t] (mM-CO2/day)C 0.02 0.04 0.06 1114172023Time (Day)MGPP[11-t] (mM-CO2/day)D 0.5 1.0 0510152025Time (Day)GPP/ER Figure A-13. Time series of S7-A. A) GPP. B) MGPP[6-t]. C) MGPP[11-t]. D) GPP/ER.

PAGE 162

162 A 0.00 0.04 0.08 0.12 0 5 10 15 20 25Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.03 0.05 0.07 678910Time (Day)MGPP[6-t] (mM-CO2/day)C 0.03 0.05 0.07 1114172023Time (Day)MGPP[11-t] (mM-CO2/day)D 0.5 1.0 0510152025Time (Day)GPP/ER Figure A-14. Time series of S7-B. A) GPP. B) MGPP[6-t]. C) MGPP[11-t]. D) GPP/ER.

PAGE 163

163 A 0.00 0.10 0.20 0.30 0.40 0510152025Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.00 0.10 0.20 0.30 678910Time (Day)MGPP[6-t] (mM-CO2/day)C 0.00 0.20 0.40 1114172023Time (Day)MGPP[11-t] (mM-CO2/day)D 0.5 1.0 0510152025Time (Day)GPP/ER Figure A-15. Time series of S8-A. A) GPP. B) MGPP[6-t]. C) MGPP[11-t]. D) GPP/ER.

PAGE 164

164 A 0.00 0.10 0.20 0.30 0.40 0.50 0510152025Time (Day)GPP (mM-CO2/day) Sec1 Sec2 Sec3 Sec4 B 0.00 0.10 0.20 0.30 678910Time (Day)MGPP[6-t] (mM-CO2/day)C 0.00 0.20 0.40 1114172023Time (Day)MGPP[11-t] (mM-CO2/day)D 0.5 1.0 0510152025Time (Day)GPP/ER Figure A-16. Time series of S8-B. A) GPP. B) MGPP[6-t]. C) MGPP[11-t]. D) GPP/ER.

PAGE 165

165 6 APPENDIX B R CODES FOR SIMULATION MODELS Table B-1. R code for a basic steady-state model of a microcosm Basic steady-state model DT=1 k=c(0.0325,2.00E-4,6.01E-5,3.21E-6,1.93E-6,3. 47E-4,9.26E-5,3.47E-4,3.47E-4,0.025,0.025) tseq=seq(0,14500,by=DT) #DT=1 --> 5-min interval t=c(0) #5-min interval t1=c() #1-day interval GPPa=c() #average GPP for each day TN=16 R=c(4.5) GPP=c(0) Pro=c(72) Con=c(270) ProD=c(72) ConD=c(72) Nut=c(3.85) for(i in c(1:(length(tseq)-1))) { t[i+1]=i/288*DT #t[1]=time zero if((i%%(288/DT))>=1&&(i%%(288/DT))<=(144/DT)) {L=0} else {L=45} #light energy DPro=(k[2]-k[3])*Nut[i]*Pro[i] *R[i]-k[4]*Pro[i]*Con[i]-k[6]*Pro[i] DCon=(k[4]-k[5])*Pro[i]*Con[i]-k[7]*Con[i] DProD=k[6]*Pro[i]-k[8]*ProD[i] DConD=k[7]*Con[i]-k[9]*ConD[i] Pro[i+1]=Pro[i]+DPro*DT Con[i+1]=Con[i]+DCon*DT ProD[i+1]=ProD[i]+DProD*DT ConD[i+1]=ConD[i]+DConD*DT Nut[i+1]=TN-k[10]*(Pro[i+1]+P roD[i+1])-k[11]*(Con[i+1]+ConD[i+1]) R[i+1]=L/(k[1]*Nut[i+1]*Pro[i+1]+1) GPP[i+1]=k[2]*Nut[i+1]*Pro[i+1]*R[i+1] } for (j in c(1:50)) { t1[j]=j GPPa[j]=mean(GPP[(146+(j-1)*288):(289+(j-1)*288)])/ 2.5 #2.5-->unit from mg-C/5-min to mM-C/day }

PAGE 166

166 Table B-2. R code for di sturbance intensity tests Intensity test DT=1 k=c(0.0325,2.00E-4,6.01E-5,3.21E-6,1.93E-6,3. 47E-4,9.26E-5,3.47E-4,3.47E-4,0.025,0.025) tseq=seq(0,14500,by=DT) linecolor=c("green","orange","blue","red") pointch=c(1,16,17,15) t=c(0) t1=c() GPPa=c() MGPP_d=c() #MGPP during the full disturbance period MGPP_p=c() #MGPP during the full post-disturbance period TN=16 R=c(4.5) GPP=c(0) Pro=c(72) Con=c(270) ProD=c(72) ConD=c(72) Nut=c(3.85) fd=c(0) #disturbance factor (fd) dis1=20 #beginning of disturbance(day) dis2=30 #end of disturbance(day) thre=3 #p threshold m=0.3 #fd-5 for a p below the threshold n=0.0005 #fd-5 for a p above the threshold for(p in c(0:4)){ #p: power for(i in c(1:(length(tseq)-1))) { t[i+1]=i/288*DT #t[1]=time zero if((i%%(288/DT))>=1&&(i%%(288/DT))<= (144/DT)) {L=0} else {L=45} if(p=dis1&&t[i+1]=205&&i%%(288/DT)<217) #disturbance period & 11am-12pm {fd[i+1]=m} else {fd[i+1]=0}} else {if(t[i+1]>=dis1&&t[i+1 ]=205&&i%%(288/DT)<217) {fd[i+1]=fd[i]+n} else {fd[i+1]=fd[i]}} DPro=(1+sign_a*p*fd[i+1])*(k[2]-k[3])*Nut[i]*Pr o[i]*R[i]-(1+sign_b*p*fd[i+ 1])*k[4]*Pro[i]*Con[i] -(1+p*fd[i+1])*k[6]*Pro[i] DCon=(1+sign_b*p*fd[i+1])*(k[4]-k[5 ])*Pro[i]*Con[i]-(1+p*f d[i+1])*k[7]*Con[i] DProD=(1+p*fd[i+1])* k[6]*Pro[i]-k[8]*ProD[i] DConD=(1+p*fd[i+1]) *k[7]*Con[i]-k[9]*ConD[i] Pro[i+1]=Pro[i]+DPro*DT Con[i+1]=Con[i]+DCon*DT ProD[i+1]=ProD[i]+DProD*DT ConD[i+1]=ConD[i]+DConD*DT Nut[i+1]=TN-k[10]*(Pro[i+1]+P roD[i+1])-k[11]*(Con[i+1]+ConD[i+1])

PAGE 167

167 Table B-2. Continued R[i+1]=L/(k[1]*Nut[i+1]*Pro[i+1]+1) GPP[i+1]=k[2]*Nut[i+1]*Pro[i+1]*R[i+1] } for (j in c(1:50)) { t1[j]=j GPPa[j]=mean(GPP[(146+(j-1)*288):(289+(j-1)*288)])/2.5 } MGPP_d[p+1]=mean(GPPa[(dis1):(dis2-1)]) MGPP_p[p+1]=mean(GPPa[(dis2):(dis2+14)]) if(p==0) {plot(t1,GPPa,type="o",xlab="Time (D ay)",ylab="GPP (mM-C/day)",xlim=c(0,50))} else {lines(t1,GPPa,type="o",pc h=pointch[p],col=linecolor[p])} } #graph of MGPP pattern p=c(0,1,2,3,4) x11() plot(p,MGPP_d,type="o",lty=2) lines(p,MGPP_p,type="o")

PAGE 168

168 Table B-3. R code for disturbance frequency tests Frequency test DT=1 k=c(0.0325,2.00E-4,6.01E-5,3.21E-6,1.93E-6,3.47E -4,9.26E-5,3.47E-4,3.47E-4,0.025,0.025) tseq=seq(0,14500,by=DT) linecolor=c("black","orange","blue","red") pointch=c(21,16,17,15) t=c(0) t1=c() GPPa=c() MGPP_d=c() MGPP_p=c() TN=16 R=c(4.5) GPP=c(0) Pro=c(72) Con=c(270) ProD=c(72) ConD=c(72) Nut=c(3.85) fd=c(0) dis1=20 dis2=25 m=0.3 n=0.0005 for(r in c(0:4)){ if(r==0) {p=0} #Reference(undisturbed) system else if(r==1) {p=1} #regime 1 else {p=7} #regime 2~4 for(i in c(1:(length(tseq)-1))) { t[i+1]=i/288*DT if(r<=1) #below p threshold {sign_a=1 sign_b=1} else #above p threshold {sign_a=1 sign_b=1} if((i%%(288/DT))>=1&&(i%%(288/DT))<=(144/DT)) {L=0} else {L=45} if(r==0) {fd[i+1]=0} else if(r==1) #frequency regime 1 {if(t[i+1]>=dis1&&t[i+1]=dis1&&t[i+1 ]=1&&i%%(48/DT)<6) {fd[i+1]=fd[i]+n} else {fd[i+1]=fd[i]}} else if(r==3) #frequency regime 3 {if(t[i+1]>=dis1&&t[i+1 ]=1&&i%%(96/DT)<12) {fd[i+1]=fd[i]+n} else {fd[i+1]=fd[i]}}

PAGE 169

169 Table B-3. Continued else #frequency regime 4 {if(t[i+1]>=dis1&&t[i+1]=192&&i%%(288/DT)<228) {fd[i+1]=fd[i]+n} else {fd[i+1]=fd[i]}} DPro=(1+sign_a*p*fd[i+1])*(k[2]-k[3])*Nut[i]*Pr o[i]*R[i]-(1+sign_b*p*fd[i+ 1])*k[4]*Pro[i]*Con[i] -(1+p*fd[i+1])*k[6]*Pro[i] DCon=(1+sign_b*p*fd[i+1])*(k[4]-k[5 ])*Pro[i]*Con[i]-(1+p*f d[i+1])*k[7]*Con[i] DProD=(1+p*fd[i+1])* k[6]*Pro[i]-k[8]*ProD[i] DConD=(1+p*fd[i+1]) *k[7]*Con[i]-k[9]*ConD[i] Pro[i+1]=Pro[i]+DPro*DT Con[i+1]=Con[i]+DCon*DT ProD[i+1]=ProD[i]+DProD*DT ConD[i+1]=ConD[i]+DConD*DT Nut[i+1]=TN-k[10]*(Pro[i+1]+P roD[i+1])-k[11]*(Con[i+1]+ConD[i+1]) R[i+1]=L/(k[1]*Nut[i+1]*Pro[i+1]+1) GPP[i+1]=k[2]*Nut[i+1]*Pro[i+1]*R[i+1] } for (j in c(1:50)) { t1[j]=j GPPa[j]=mean(GPP[(146+(j-1)*288):(289+(j-1)*288)])/2.5 } MGPP_d[r+1]=mean(GPPa[(dis1):(dis2-1)]) MGPP_p[r+1]=mean(GPPa[(dis2):(dis2+14)]) if(r==0) {plot(t1,GPPa,type="l",col="grey",xlab="Time (Day)",ylab="GPP (mM-C/day)",xlim=c(0,50)} else {lines(t1,GPPa,type="o",pc h=pointch[r],col=linecolor[r])} } #graph of MGPP pattern r=c(0,1,2,3,4) x11() plot(r,MGPP_d,type="o",lty=2) lines(r,MGPP_p,type="o")

PAGE 170

170 LIST OF REFERENCES Albert, R., Barabsi, A., 2002. Statistical me chanics of complex networks. Reviews of Modern Physics. 74, 47. Armesto, J.J., Pickett, S.T. A., 1985. Experiments on dist urbance in old-field plant communities: Impact on species richness and abundance. Ecology. 66, 230. Ashby, W.R., 1962. Principles of the self-organizing system. In: Foerster, H.V. and Zopf, G.W. (eds.), Principles of Self-Organization : Transactions of the University of Illinois Symposium. Pergamon Pre ss, London, UK, pp. 255. Aumann, C.A., 2007. A methodology for dev eloping simulation models of complex systems. Ecological M odelling. 202, 385. Bade, D.L., Cole, J.J., 2006. Impact of c hemically enhanced diffusion on dissolved inorganic carbon stable isotopes in a fertiliz ed lake. Journal of Geophysical Research. 111, C01014. Barabsi, A., Bonabeau, E., 2003. Scale-free networks. Scient ific American. 288, 50. Benton, T.G., Solan, M., Travis, J.M.J., Sait, S.M., 2007. Microcosm experiments can inform global ecological problems. Trends in Ecology and Evolution. 22, 516. Beyers, R.J., 1962. Relationship betw een temperature and t he metabolism of experimental ecosystems. Science. 136, 980. Beyers, R.J., 1963a. Balanced a quatic microcosms: Their imp lications for space travel. The American Biology Teacher. 25, 422. Beyers, R.J., 1963b. The metabolism of tw elve aquatic laboratory microecosystems. Ecological Monographs. 33, 281. Beyers, R.J., Gillespie, B., 1964. Measuri ng the carbon dioxide metabolism of aquatic organisms. The American Biol ogy Teacher. 26, 499. Beyers, R.J., Odum, H.T., 1993. Ecological Microcosms. Springer-Verlag, New York, NY. Boerner, R.E.J., 1982. Fire and nutrient cycling in temperat e ecosystems. BioScience. 32, 187. Boersma, M., 2000. The nutritional quality of P-limited algae for Daphnia. Limnology and Oceanography. 45, 1157.

PAGE 171

171 Bond-Lamberty, B., Peckham, S.D., Ahl, D.E., Gower, S.T., 2007. Fire as the dominant driver of central canadian boreal fore st carbon balance. Nature. 450, 89. Bongers, F., Poorter, L., Hawthorne, W.D., Sheil, D., 2009. The intermediate disturbance hypothesis applies to tropical fore sts, but disturbance contributes little to tree diversity. Ecology Letters. 12, 798. Borges, A.V., Vanderborght, J., Schiettecatte, L., Gazeau, F ., Ferron-Smith, S., Delille, B., Frankignoulle, M., 2004. Variability of the gas transfer velocity of CO2 in a macrotidal estuary (the Scheldt). Estuaries. 27, 593. Bornette, G., Amoros C., 1996. Disturbance regimes and vegetation dynamics: Role of floods in riverine wetlands. Journal of Vegetation Science. 7, 615. Brown, M.T., 2004. A picture is worth a thousand words: Energy systems language and simulation. Ecological Modelling. 178, 83. Cai, T.T., Montague, C.L., Davis, J.S., 2006. The maximum power principle: An empirical investigation. Ecol ogical Modelling. 190, 317. Cai, T.T., Olsen, T.W., Campbell, D. E., 2004. Maximum (em) power: A foundational principle linking man and nature. Ecological M odelling. 178, 115. Campbell, D.E., 1984. Energy Filter Properties of Ecosyst ems. Ph.D. Dissertation, University of Florida, Gainesville, FL, USA. Cardinale, B.J., Palmer, M.A., Ives, A.R. Brooks, S.S., 2005. Di versity-productivity relationships in streams vary as a function of the natural disturbance regime. Ecology. 86, 716. Carpenter, S.R., 1996. Microc osm experiments have limited relevance for community and ecosystem ecology. Ecology. 77, 677. Chapin, F.S., Woodwell, G.M., Randerson, J.T., Rastetter, E.B., Lovett, G.M., Baldocchi, D.D., Clark, D.A., Harmon, M.E., Schimel, D.S., Valentini, R., Wirth, C., Aber, J.D., Cole, J.J., Goulden, M.L., Harden, J. W., Heimann, M., Howarth, R. W., Matson, P.A., McGuire, A.D., Melillo, J.M., Mooney, H. A., Neff, J.C., H oughton, R.A., Pace, M.L., Ryan, M.G., Running, S.W., Sala, O.E., Schlesinger, W.H., Schulze, E. D., 2006. Reconciling carboncycle concepts, terminology, and methods. Ecosystems. 9, 1041. Charters, A.C., Neushul, M., Coon, D., 1973. The effect of water motion on algal spore adhesion. Limnology and Oceanography. 18, 884. Chase, J.M., Leibold, M.A., 2002. Spatial scale dictates the productivity-biodiversity relationship. Nature. 416, 427.

PAGE 172

172 Chesson, P., Huntly, N., 1997. The roles of harsh and fluc tuating conditions in the dynamics of ecological communities. The American Natura list. 150, 519. Clarke, R.D., Buskey, E.J., Marsden, K.C ., 2005. Effects of wa ter motion and prey behavior on zooplankton capture by two coral reef fishes. Marine Biology. 146, 1145 1155. Cole, J.J., Caraco, N.F., 1998. Atmospheric exchange of carbon dioxide in a low-wind oligotrophic lake measur ed by the addition of SF6. Limnology and Oceanography. 43, 647. Connell, J.H., 1978. Diversity in tropical rain forests and coral reefs: High diversity of trees and corals is maintained only in a nonequilibrium state. Science. 199, 1302. Connell, J.H., Slatyer, R.O., 1977. Mechanism s of succession in natural communities and their role in community stability and organization. The American Naturalist. 111, 1119. Conquest, L.L., 2000. Analysis and interpretation of ecological field data using BACI designs: Discussion. Journal of Agricultural, Biological, and Environm ental Statistics. 5, 293. Cooper, D.C., Copeland, B.J., 1973. Responses of c ontinuous-series estuarine microecosystems to point-source input va riations. Ecological Monograph. 43, 213. Copeland, B.J., Dorris, T.C., 1964. Community metabolism in ecosystems receiving oil refinery effluents. Limnology and Oceanography. 9, 431. Copeland, B.J., Duffer, W.R., 1964. Use of a clear plastic dome to measure gaseous diffusion rates in natural waters. Limnology and Oceanogr aphy. 9, 494. Corenblit, D., Gurnell, A.M. Steiger, J., Tabacchi, E., 2008. Reciprocal adjustments between landforms and living organisms: Extended geomorphic evolutionary insights. CATENA. 73, 261. Crusius, J., Wanninkhof, R., 2003. Gas transf er velocities measured at low wind speed over a lake. Limnology and Oceanography. 48, 1010. Denslow, J.S., 1980. Pattern s of plant species divers ity during succession under different disturbance regime s. Oecologia. 46, 18. Dodds, W.K., 1989. Photosynthes is of two morphologies of Nostoc parmelioides (cyanobacteria) as related to current velo cities and diffusion patterns. Journal of Phycology. 25, 258.

PAGE 173

173 Dodson, S.I., Arnott, S.E., Cottingham, K.L., 2000. The relationship in lake communities between primary productivity and specie s richness. Ecology. 81, 2662. Eisenbies, M.H., Burger, J.A., Aust, W.M., Patterson, S.C., 2005. Soil physical disturbance and logging residue effects on changes in soil productivity in five-year-old pine plantations. Soil Science Societ y of America Journal. 69, 1833. Ferens, M.C., Beyers, R.J., 1972. Studies of a simple laboratory microecosystem: Effects of stress. Ecology. 53, 709. Flder, S., Sommer, U., 1999. Diversity in pl anktonic communities: An experimental test of the intermediate disturbance hypot hesis. Limnology and Oceanography. 44, 1114 1119. Florida LAKEWATCH, 2005. Florida LAKE WATCH Annual Data Summaries 2004. Department of Fisheries and Aquatic Sciences, University of Flori da/Institute of Food and Agricultural Sciences Library, Gainesville, FL. Fraser, L.H., Keddy, P., 1997. The role of experimental microcosms in ecological research. Trends in Ecology and Evolution. 12, 478. Fukami, T., Morin, P.J., 2003. Productivity-biodiversity relationships depend on the history of community assembly. Nature. 424, 423. Gagnon, P., Wagner, G., Himmelm an, J.H., 2003. Use of a wa ve tank to study the effects of water motion and algal movement on the displa cement of the sea star Asterias vulgaris towards its prey. Marine Ecology Progress Series. 258, 125. Gates, R.D., Baghdasarian, G., Muscatine, L., 1992. Temperature stress causes host cell detachment in symbiotic Cnidarians: Implications for coral bleaching. Biological Bulletin. 182, 324. Geider, R.J., Osborne, B., 1992. Algal Photosynthesis: The Measurement of Algal Gas Exchange. Chapman and Hall, New York, NY, pp. 11. George, D.G., Edwards, R.W., 1973. Daphnia distribution within langmuir circulations. Limnology and Oceanography. 18, 798. Gordon, R., Brawley, S.H., 2004. Effects of water motion on propagule release from algae with complex life historie s. Marine Biology. 145, 21. Grime, J.P., 1973. Competit ive exclusion in herbaceous vegetation. Nature. 242, 344 347. Grumbine, R.E., 1994. What is ecosystem management? Conv ervation Biology. 8, 27 38.

PAGE 174

174 Haddad, N.M., Holyoak, M., Mata, T.M., Davies, K.F., Melbourne, B.A., Preston, K., 2008. Species' traits predict the effects of disturbance and productivity on diversity. Ecology Letters. 11, 348. Hall, D.J., 1964. An experimen tal approach to the dynamics of a natural population of Daphnia galeata mendotae. Ecology. 45, 94. Halpern, C.B., 1988. Early successional pat hways and the resistance and resilience of forest communities. Ecology. 69, 1703. Heath, R.T., 1979. Holistic study of an aquatic microcosm: Theoretical and practical implications. International Journal of Environmental Studies. 13, 87. Heffernan, J.B., Cohen, M.J., 2010. Direct and indirect coupling of primary production and diel nitrate dynamics in a subtropical spring-fed river. Limnology and Oceanography. 55, 677. Hooper, D.U., Chapin, F.S., Ewel, J.J., Hect or, A., Inchausti, P., Lavorel, S., Lawton, J.H., Lodge, D.M., Loreau, M., Naeem, S., Schmid, B., Set l, H., Symstad, A.J., Vandermeer, J., Wardle, D.A., 2005. Effects of biodiversity on ecosystem functioning: A consensus of current knowledge. Ecological Monographs. 75, 3. Houser, J.N., Mulholland, P.J., Mal oney, K.O., 2005. Catc hment disturbance and stream metabolism: Patterns in ecosystem respiration and gro ss primary production along a gradient of upland soil and vegetat ion disturbance. Jour nal of the North American Benthological Society. 24, 538. Hurd, C.L., 2000. Water moti on, marine macroalgal physiol ogy, and production. Journal of Phycology. 36, 453. Hutchinson, G.E., 1957. Concluding remark s. Cold Spring Harbor Symposia on Quantitative Biolog y. 22, 415. Jenkins, B., Kitching, R.L., Pimm, S.L. 1992. Productivity, disturbance and food web structure at a local spatial scale in experim ental container habitats Oikos. 65, 249. Jiang, L., Patel, S.N., 2008. Community asse mbly in the presence of disturbance: A microcosm experiment. Ecology. 89, 1931. Kang, D.S., 1998. Pulsing and Self-Organizati on. Ph.D. Dissertation, University of Florida, Gainesville, FL, USA. Kassen, R., Buckling, A., Bell, G., Rainey P.B., 2000. Diversity peaks at intermediate productivity in a laboratory microcosm. Nature. 406, 508.

PAGE 175

175 Kazanci, C., 2007. EcoNet: A new software for ecological modeling, simulation and network analysis. Ecological Modelling. 208, 3. Kemp, W.M., Mitsch, W.J., 1979. Turbulenc e and phytoplankton di versity: A general model of the "paradox of plankton" Ecological Modelling. 7, 201. Kimmins, J.P., 1996. Importance of soil and role of ecosystem disturbance for sustained productivity of cool temperate and boreal fo rests. Soil Science Society of America Journal. 60, 1643. Komjarova, I., Blust, R., 2009. Effect of Na, Ca and pH on simult aneous uptake of Cd, Cu, Ni, Pb, and Zn in the water flea Daphnia magna measured using stable isotopes. Aquatic Toxicology. 94, 81. Kondoh, M., 2001. Unifying the relationships of species richness to productivity and disturbance. Proceedings of the Royal Societ y of London. 268, 269. Larsson, S., Wirn, A., Lundgren, L., Ericsson, T., 1986. Effects of light and nutrient stress on leaf phenolic chemistry in Salix dasyclados and susceptibility to Galerucella lineola (Coleoptera). Oikos. 47, 205. Laska, G., 2001. The disturbance and vegetati on dynamics: A review and an alternative framework. Plant Ecology. 157, 77. Levin, S.A., 2005. Self-organization and the emergence of complexity in ecological systems. BioScience. 55, 1075. Levin, S.A., Paine, R.T., 1974. Disturbance, patch formation, and community structure. Proceedings of the Na tional Academy of Sciences of the USA. 71, 2744. Lindeman, R.L., 1942. The trophic-dynamic as pect of ecology. Ecology. 23, 399. Lomolino, M.V., 2000. Ecology 's most general, yet protean pattern: The species-area relationship. Journal of Biogeography. 27, 17. Lotka, A.J., 1922. Contributi on to the energetics of evol ution. Proceedings of the National Academy of Sciences of the USA. 8, 147. MacIntyre, S., Wanninkhof, R., Chanton, J.P., 1995. Trac e gas exchange across the airwater interface in freshwater and coasta l marine environment. In: Matson, P.A. and Harriss, R.C. (eds.), Biogenic Trace Gases: Measuring Emissions from Soil and Water (Methods in Ecology). Wiley-Blac kwell, Cambridge, MA, pp. 52. Mack, M.C., DAntonio, C.M. 1998. Impacts of biological invasions on disturbance regimes. Trends in Ecology and Evolution. 13, 195.

PAGE 176

176 Mackey, R.L., Currie, D.J., 2001. The diversity-disturbance re lationship: Is it generally strong and peaked? Ecology. 82, 3479. Maloney, K.O., Feminella, J.W., Mitchell, R.M. Miller, S.A., Mulholland, P.J., Houser, J.N., 2008. Landuse legacies and small streams: Identifying relationships between historical land use and contem porary stream conditions. J ournal of the North American Benthological Society. 27, 280. Margalef, R., 1963. On certain unifying prin ciples in ecology. The American Naturalist. 97, 357. Margalef, R., 1997. Turbulence and marine life. Scientia Marina. 61, 109. Martin, T.R., Holdich, D.M., 1986. The acute lethal toxicity of heavy metals to peracarid crustaceans (with particular reference to freshwater Asellids and Gammarids). Water Research. 20, 1137. Mass, T., Genin, A., Shavi t, U., Grinstein, M., Tcher nov, D., 2010. Flow enhances photosynthesis in marine benthic autotrophs by increasing the efflux of oxygen from the organism to the water. Proceedings of the Na tional Academy of Sciences of the USA. 107, 2527. McAlister, J.S., Stancyk, S. E., 2003. Effects of variable water motion on regeneration of Hemipholis elongata (Echinodermata, Ophiuroidea). In vertebrate Biology 122, 166. McCabe, D.J., Gotelli, N.J., 2000. Effects of disturbance frequency, intens ity, and area on assemblages of stream macroinv ertebrates. Oecologia. 124, 270. McIntire, C.D., Phinney, H.K., 1965. Labor atory studies of per iphyton production and community metabolism in lotic environment s. Ecological M onographs. 35, 237. McIntyre, S., Lavorel, S., 1994. How environ mental and disturbance factors influence species composition in temperate Australian grasslands. Journal of Vegetation Science. 5, 373. Mittelbach, G.G., Steiner, C. F., Scheiner, S.M., Gross, K. L., Reynolds, H.L., Waide, R.B., Willig, M.R., Dodson, S .I., Gough, L., 2001. What is the observed relationship between species richness and productivity? Ecology. 82, 2381. Montague, C.L., 2007. Dynamic simulation wi th energy systems language. In: Fishwick, P.A. (ed.), Handb ook of Dynamic System Modeling. Chapman & Hall/CRC, Boca Raton, FL, pp. 28.1.32. Montefalcone, M., Chiantore, M., Lanzone, A., Morri, C., Al bertelli, G., Bianchi, C.N., 2008. BACI design reveals the decline of the seagrass Posidonia oceanica induced by anchoring. Marine Polluti on Bulletin. 56, 1637.

PAGE 177

177 Nowacki, G.J., Kramer, M. G., 1998. The Effe cts of Wind Disturbance on Temperate Rain Forest Structure and Dynamics of Sout heast Alaska. U.S. Dept. of Agriculture, Forest Service, Pacific Northwest Research Station, Portland, OR. Odling-Smee, F.J., Laland, K.N., Feldman, M.W., 1996. Niche construction. The American Naturalist. 147, 641. Odum, E.P., 1969. The strategy of eco system development. Science. 164, 262. Odum, E.P., 1984. The mesocosm. BioScience. 34, 558. Odum, H.T., 1983. Maximum power and efficien cy: A rebuttal. Ecological Modelling. 20, 71. Odum, H.T., 1956. Primary pr oduction in flowing waters. Limnology and Oceanography. 1, 102. Odum, H.T., 1988. Self-organization, transform ity, and information. Science. 242, 1132 1139. Odum, H.T., 1995. Energy systems concept s and self-organization: A rebuttal. Oecologia. 104, 518. Odum, H.T., Odum, E.C., 2000. Modeling for All Scales: An Introduction to System Simulation. Academic Press, San Diego, CA. Odum, W.E., Odum, E.P., Odum, H.T., 1995. Nature's pulsing paradigm. Estuaries. 18, 547. Orwin, K.H., Wardle, D.A., 2004. New in dices for quantifying the resistance and resilience of soil biota to exogenous dist urbances. Soil Biology & Biochemistry. 36, 1907. Paine, R.T., Tegner, M.J. Johnson, E.A., 1998. Compounded perturbations yield ecological surprises. Ecosystems. 1, 535. Palmer, M., 1992. Standard Operating Procedure for GLNPO Tota l Alkalinity Titration. U.S. EPA Great Lakes National Pr ogram Office, Chicago, IL. Petersen, J.E., Sanford, L.P., Kemp, W. M., 1998. Coastal plankton responses to turbulent mixing in experimental ecosystem s. Marine Ecology-Progress series. 171, 23 41. Peterson, B.J., 1980. Aquatic primary productivity and the 14C-CO2 method: A history of the productivity problem. Annual Review of Ecology and Systematics. 11, 359.

PAGE 178

178 Peterson, C.G., Stevenson, R.J., 1992. Resistance and re silience of lotic algal communities: Importance of disturbance timing and current. Ecology. 73, 1445. Petraitis, P.S., Latham, R.E., Niesenbaum, R.A., 1989. The maintenance of species diversity by disturbance. The Quarte rly Review of Biology. 64, 393. Pickett, S.T.A., White, P.S ., 1985. The Ecology of Natu ral Disturbance and Patch Dynamics. Academic Pre ss, San Diego, CA. Pickett, S.T.A., 1976. Succession: An evol utionary interpretation. The American Naturalist. 110, 107. Pickett, S.T.A., Kolasa, J., Armesto, J.J., Collin s, S.L., 1989. The ec ological concept of disturbance and its expression at various hi erarchical levels. Oikos. 54, 129. Platt, W.J., Beckage, B., Doren, R.F., Slater H.H., 2002. Interactions of large-scale disturbances: Prior fire regimes and hurricane mortality of savanna pines. Ecology. 83, 1566. Power, M.E., Tilman, D., Este s, J.A., Menge, B.A., Bond, W.J., Mills, L.S., Daily, G., Castilla, J.C., Lubchenco, J., Paine, R.T. 1996. Challenges in the quest for keystones. BioScience. 46, 609. Randerson, J.T., Liu, H., Flanner M.G., Chambers, S.D., Jin, Y., Hess, P.G., Pfister, G., Mack, M.C., Treseder, K.K., We lp, L.R., Chapin, F.S., Ha rden, J.W., G oulden, M.L., Lyons, E., Neff, J.C., Schuur, E.A.G., Zender, C.S., 2006. The impact of boreal forest fire on climate warming. Science. 314, 1130. Reiners, W.A., 1983. Disturbance and basic properties of ecosystem energetics. In: Mooney, H.A. and Godron, M. (eds.), Dist urbance and Ecosystems. Springer-Verlag, Heidelberg, Germany, pp. 83. Resh, V.H., Brown, A.V., Covich, A.P., Gurt z, M.E., Li, H.W., Mi nshall, G.W., Reice, S.R., Sheldon, A.L., Wallace, J.B., Wissmar R.C., 1988. The role of disturbance in stream ecology. Journal of the North Amer ican Benthological Society. 7, 433. Rodgers, P.W., DePinto, J., 1983. Estima tion of phytoplankton decomposition rates using two-stage continuous flow st udies. Water Research. 17, 761. Roeselers, G., Zippel, B., Staal, M., van Loosdrecht, M. Muyzer, G., 2006. On the reproducibility of microcosm experimentsdifferent communi ty composition in parallel phototrophic biofilm microcosms. Microbial Ecology. 58, 169. Roxburgh, S.H., Shea, K., W ilson, B., 2004. The intermedi ate disturbance hypothesis: Patch dynamics and mechanisms of specie s coexistence. Ecology. 85, 359.

PAGE 179

179 Ruth, B.F., Flemer, D.A., Bundr ick, C.M., 1994. Recolonizati on of esturine sediments by macroinvertebrates: Does microcosm si ze matter? Estuaries. 17, 606. Ryther, J.H., 1956. The measurement of primary production. Limnology and Oceanography. 1, 72. Scheffer, M., Carpenter, S., Foley, J.A., Folk e, C., Walker, B., 2001. Catastrophic shifts in ecosystems. Nature. 413, 591. Shea, K., Roxburgh, S.H., Rauschert, E.S.J. 2004. Moving from pattern to process: Coexistence mechanisms under intermediate di sturbance regimes. Ecology Letters. 7, 491. Skirrow, G., 1965. The dissolved gascarbon dioxide. In: Riley, J.P. and Skirrow, G. (eds.), Chemical Oceanography. Academic Press, New York, NY, pp. 227. Smith, S.V., 1973. Carbon dioxide dynamics: A record of organic carbon production, respiration, and calcificati on in the eniwetok reef flat community. Limnology and Oceanography. 18, 106. Smith, S.V., Key, G.S., 1975. Carbon dioxide and metabolism in marine environments. Limnology and Oceanography. 20, 493. Sousa, W.P., 1979. Disturbance in marine inte rtidal boulder fields: The nonequilibrium maintenance of species dive rsity. Ecology. 60, 1225. Sousa, W.P., 1980. The responses of a community to disturbance: The importance of successional age and species' life hi stories. Oecologia. 45, 72. Sousa, W.P., 1984. The role of disturbance in natural communities. Annual Review of Ecology and Systematics. 15, 353. Sparks, R.E., Bayley, P.B., K ohler, S.L., Osborne, L.L., 1990. Disturbance and recovery of large floodplain rivers. Envi ronmental Management. 14, 699. Sprugel, D.G., 1985. Natural di sturbance and ecosystem energet ics. In: Pickett, S.T.A. and White, P.S. (eds.), The Ecology of Natural Disturbance and Patch Dynamics. Academic Press, San Diego, CA, pp. 335. Sugiura, K., Goto, M., Kuri hara, Y., 1982. Effect of Cu2+ stress on an aquatic microcosm: A holistic study. Envi ronmental Research. 27, 307. Tansley, A.G., 1935. The us e and abuse of vegetational concepts and terms. Ecology. 16, 284.

PAGE 180

180 Thomas, W.H., Vernet, M., 1995. Effects of small-scale turbulence on photosynthesis, pigmentation, cell division, and cell size in the marine dinoflagellate Gonyaulax polyedra (dinophyceae). Journal of Phycology. 31, 50. Townsend, C.R., Scarsbrook, M.R., Dolede c, S., 1997. The intermediate disturbance hypothesis, refugia, and biodiversity in streams. Limnology and Oceanography. 42, 938. Turner, M.G., Baker, W.L. Peterson, C.J., Peet, R. K., 1998. Factors influencing succession: Lessons from large, infrequent natural disturbances. Ecosystems. 1, 511 523. Ulanowicz, R.E., 1997. Ecology, the Ascendent Perspective. Columbia University Press, New York, NY. van der Maarel, E., 1993. So me remarks on disturbance and it s relations to diversity and stability. Journal of Vegetation Science. 4, 733. Van Nes, E.H., Scheffer, M., 2005. A strategy to improve the contribution of complex simulation models to ecological theor y. Ecological Modelling. 185, 153. Vandermeer, J., Cerda, I.G.d.l., Boucher, D., Perfecto, I., Ruiz, J., 2000. Hurricane disturbance and tropical tree species diversity. Science. 290, 788. Waide, R.B., Willig, M.R., Stei ner, C.F., Mittelbach, G., Gough, L., Dodson, S.I., Juday, G.P., Parmenter, R., 1999. The relationship between productivity and species richness. Annual Review of Ecology and Systematics. 30, 257. Wanninkhof, R., Ledwell, J.R., Broecker, W.S., 1985. Gas exchange-wind speed relation measured with sulfur hexafluoride on a lake. Science. 227, 1224. Ward, J.V., 1998. Riverine landscapes: Biodiversity patterns, disturbance regimes, and aquatic conservation. Biologica l Conservation. 83, 269. Weiss, R.F., 1974. Carbon dioxide in water and seawater: The solubility of a non-ideal gas. Marine Chemistry. 2, 203. Wetzel, R.G., 2001. Limnology: Lake and Ri ver Ecosystems. Academic Press, San Diego, CA, pp. 468. Whitford, L.A., Schumacher, G.J., 1961. E ffect of current on mineral uptake and respiration by a fresh-water alga. Limnology and Oceanography. 6, 423. Whitford, L.A., Schumacher, G. J., 1964. Effect of a current on respiration and mineral uptake in Spirogyra and Oedogonium. Ecology. 45, 168.

PAGE 181

181 Wilson, S.D., Keddy, P.A., 1986. Species com petitive ability and position along a natural stress/disturbance gradien t. Ecology. 67, 1236. Woodin, S.A., 1981. Disturbance and community structure in a shallow water sand flat. Ecology. 62, 1052. Wootton, J.T., 1998. Effect s of disturbance on species diversity: A multitrophic perspective. The American Naturalist. 152, 803. Worm, B., Lotze, H.K., Hillebrand, H., Sommer, U., 2002. C onsumer versus resource control of species diversity and eco system functioning. Nature. 417, 848. Yamamoto, T., Hatta, G., 2004. Pulsed nutrient supply as a factor inducing phytoplankton diversity. Ecological Modelling. 171, 247. Zwick, P.D., 1985. Energy Systems and Iner tial Oscillators. Ph.D. Dissertation, University of Florida, Gainesville, FL, USA.

PAGE 182

182 BIOGRAPHICAL SKETCH Seungj un Lee graduated from Daegu Science High School and earned a Bachelor of Science in chemical engineering from Pohang University of Science and Technology (POSTECH) in South Korea. While at POSTE CH, he spent half a year at the University of Birmingham in England as an exch ange student. In 2006, he joined Systems Ecology group at the University of Florida, where he has ex plored the world of ecology and contemporary environmental issues.