Citation
Ecosystem-level responses of carbon and energy from a tropical wet forest in Costa Rica

Material Information

Title:
Ecosystem-level responses of carbon and energy from a tropical wet forest in Costa Rica
Creator:
Loescher, Henry W
Publication Date:
Language:
English
Physical Description:
x, 91 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Aerodynamics ( jstor )
Biological rhythms ( jstor )
Canopy ( jstor )
Carbon ( jstor )
Carbon dioxide ( jstor )
Climate models ( jstor )
Covariance ( jstor )
Ecosystems ( jstor )
Forests ( jstor )
Tropical forests ( jstor )
Dissertations, Academic -- Forest Resources and Conservation -- UF ( lcsh )
Forest Resources and Conservation thesis, Ph.D ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 2002.
Bibliography:
Includes bibliographical references (leaves 80-90).
General Note:
Printout.
General Note:
Vita.
Statement of Responsibility:
by Henry William Loescher.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
029220972 ( ALEPH )
51020748 ( OCLC )

Downloads

This item has the following downloads:


Full Text










ECOSYSTEM-LEVEL RESPONSES OF CARBON AND ENERGY FROM A
TROPICAL WET FOREST IN COSTA RICA













By

HENRY WILLIAM LOESCHER













A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2002














I DEDICATE THIS DISSERTATION TO MY FATHER AND TO ALL THOSE WHO
ACTIVELY ENJOY LIFE















ACKNOWLEDGMENTS

I wish to acknowledge the financial support of many organizations which made

this study possible: U.S. Department of Energy Office of Science contract IBN-9652699

within the interagency Terrestrial Ecology and Global Change, the School of Forest

Resources and Conservation, the Department of Energy as the National Institute for

Global Environmental Change Program (NIGEC), the Organization for Tropical Studies,

and the Biological Sciences Department at Florida International University.

My degree is in part, due to the assistance and dedication of many people. I wish

to thank the faculty and staff of the School of Forest Resources and Conservation for

academic guidance and logistical support. I wish to thank the chairman of my committee,

professor Henry L. Gholz, for academic, moral and creative support throughout my

tenure as a graduate student, and sincerely acknowledge my other supervisory committee

members for being thoughtful through difficult times, Stephan Mulkey, Francis Putz,

Jennifer Jacobs, and Steve Oberbauer. I am indebted to David Clark, Deborah Clark,

Monique Leclerc, Tim Martin, Tilden Meyers, Thara Prabha, and Anand Karipot.

Thanks also go to David Hollinger for assistance with tower site location, Matt 'el tejafio'

Schroeder for data collection, and Steve Moore for programming assistance. I would

specifically like to thank JeffAmthor, and Wayne H. Smith, for additional financial

support and guidance, and to Jennifer Jacobs for stimulating my interest in fluid

dynamics.



iii















TABLE OF CONTENTS

page

ACKNOWLEDGEMENTS.............................................................. .... iii

LIST OF TABLES..................................................................................... vi

LIST OF FIGURES............................................................... vii

ABSTRACT...............................................................................................ix

CHAPTERS

1. INTRODUCTION........................................................................... ... 1

2. ECOSYSTEM-LEVEL CARBON EXCHANGE.........................................5
Materials and Methods....................................................................5
Study Site...................................... .......................................... .5
Meteorological Data.......................................................................7
Net Ecosystem Exchange (NEE) Measurements......................................9
Data Screening.................................................................... ...13
Leaf Area Estimation....................................................................13
R esults................................... ... ............................................ ... 14
Characterizing Canopy-Level Turbulence ........................................... 14
Diurnal Patterns in NEE................................................................15
Environmental Controls on NEE...................................................... 16
Estimating Annual NEE.................................................................17
D iscussion ..................................... ............................................ 18
Characteristics of the La Selva Canopy.............................................. 18
Environmental Controls on NEE......................................................19
Comparisons with Other Tropical Sites...............................................25

3. ENERGY BALANCE AND MODELED EVAPORATION FOR A WET
TROPICAL FOREST IN COSTA RICA...............................................47

M ethods......... .. ........................................................................ .. 47
Meteorological Data................. .............. ................................47
Energy Balance Estimates............................................................ 48
Evapotranspiration Model............................................................ 51
R esults................................... .......... .... .................................. 54


iv















Above-Canopy Environment .................................... .... ........54
Energy Balance ................................................ .......................55
Modeled Conductance and Evaporation ............................... ...........57
D iscussion ............................................................................58
Ecosystem Energy Balance ............................................................58
Conductances and Other Limits to Annual Energy Fluxes .........................59

4. CONCLUSIONS......................................................................................79

LIST OF REFERENCES .........................................................................80

BIOGRAPHICAL SKETCH..................................................................91



































v














LIST OF TABLES
Table page

2-1. Estimates of aerodynamic parameters, zero-plane displacement (d) from eq. 2-1,
aerodynamic roughness length (zo) from eq. 2-2, and u* estimates according to
stability class (L), eq. 2-3, from La Selva Biological Station. zo, d and u* estimates
are median values �95% CI, L values are means �1 SE, and n is number of 30-min
periods........................................................................... .............28

2-2. Parameter estimates and statistics from the light response function, eq. 2-5, across
VPD classes and year ......................................................................29

2-3. Annual and seasonal differences in estimated leaf area index (eLAI) ...............31

2-4. Meteorolgical data for years 1998-2000 from La Selva Biological Station, CR.....32

2-5. Across site comparison of stand attributes from four neotropical eddy covariance
studies................... ........ . ......... .. .... ......... ..................................... 33

2-6. Between year measures of productivity and ecosystem efficiency from La Selva,
Costa Rica and Cuieiras, Brazil...........................................................34

3-1. First order regression parameters for the energy balance from 1998-2000............63

3-2. Annual evaporative fluxes calculated using the Priestly-Taylor equation from La
Selva, Costa Rica.................................................. ... ...................... 64

3-3. Annual fraction of time that the canopy was wet at two heights in the canopy.....65

















vi














LIST OF FIGURES

Figure

2-1. The relationship between the normalized power spectra.............................36

2-2. Diurnal time series of the spectra-based correction factor............................37

2-3. Relationships between u* and daytime NEE............................ ...........38

2-4. Diurnal chracterisitics of A) storage fluxes and line-average temperature and
non-rotated vertical windspeed..................................... .................. 39

2-5. The mean diurnal pattern of above-canopy eddy covariance (EC) ............... 40

2-6. Main effects of environmental variables on daytime NEE.............................41

2-7. NEE as a function of PPFD across a gradient of vapor pressure deficits............42

2-8. The relationship between NEEnight and below-canopy temperature .................45

2-9. Cumulative NEE from La Selva, Costa Rica for 1998-2000...........................46

3-1. Diurnal relationships of friction velocity and Monin-Obukov length ...............66

3-2. Cumulative net radiation for 1998-2000 over and old-growth forest ...............67

3-3. Typical diurnal changes in below-canopy temperature and water vapor.............68

3-4. Diurnal patterns of storage energy fluxes................................................ 69

3-5. Daytime Bowen ratios for each year..................................................... 70

3-6. The relationship between net radiation and estimated energy flux....................71

3-7. Emprical relations to model both aerodynamic and bulk conductance ...............74

3-8. The diurnal relationship of aerodynamic and bulk conductance.......................75

3-9. The relationship between aerodynamic conductance and latent energy flux........76



vii















3-10. Diurnal changes in a decoupling coefficient, Q,.......................................77

3-11. Relationship between empirical and modeled estimates of latent energy..........78













































viii














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

ECOSYSTEM-LEVEL RESPONSES OF CARBON AND ENERGY FROM A
TROPICAL WET FOREST IN COSTA RICA

By

Henry William Loescher

August 2002

Chair: Dr. Henry L. Gholz
Department: School of Forest Resources and Conservation

Whether tropical forests are sources, sinks, or neutral with respect to their carbon

balance with the atmosphere remains unclear. To address this issue, estimates of net

ecosystem exchange of carbon and energy (NEE) were made for 3 years (1998-2000)

using the eddy-covariance technique in a tropical wet forest in Costa Rica. Mean

daytime NEE was ca. -18 tmol CO2 2 m s'(uptake) and mean nighttime NEE 4.6 pmol

CO2 m"2 s"1 (efflux). However, because -80% of the nighttime data in this forest were

collected during laminar flow conditions (< 0.2 m s"'), nighttime NEE was likely

underestimated. Using an alternative analysis, mean nighttime NEE increased to 6.9

p.mol CO2 m"2 s. Incident radiation accounted for ~51% of the variation in the daytime

fluxes, with temperature and vapor pressure deficit together accounting for another

-20%. This forest was a slight negative carbon sink in 1998 (-0.08 to -1.42 t C ha' y'),

a moderate sink in 1999 (-1.65 to -3.21 t C ha' y-), and a strong sink in 2000 (-6.1 to -




ix








8.1 t C ha-' y"). This trend is interpreted as relating to the dissipation of warm-phase El

Nifio effects over the course of this study.

The effects of net radiation (Rn), vapor pressure deficit (VPD), and surface

conductances on energy balance and evapotranspiration (ET) were also determined for

this forest. Sensible (H) and latent heat (XE) fluxes were estimated as the sum of above-

canopy eddy-covariance fluxes and changes in below-canopy heat profiles. Albedo was

-12 % of incident radiation and did not differ seasonally. Rn was significantly different

among years, explaining -79% of the variation in each of the H and XE fluxes. The

effects of VPD did not explain any additional variation in heat fluxes. XE was always

greater than H (when Rn exceeded 40 W mn2). The dimensionless decoupling coefficient,

Q was always > 0.5 and peaked at 0.7, suggesting that ET for this the forest was

generally decoupled from physiological controls. There was better precision in

estimating XE flux using the Priestly-Taylor model rather than using the more

physiologically-based Penman-Monteith model. Annual ET was 54-66 % of bulk

precipitation and utilized - 88-97 % of available energy (Rn).




















x














CHAPTER 1
INTRODUCTION


Inverse model calculations based on atmospheric CO2 concentrations and

13C02/12CO2 and 02/N2 ratios indicate that the terrestrial biosphere is currently a net

carbon sink, partially offsetting the additions of CO2 from fossil fuel combustion and

deforestation (Schimel et al. 2001). Temporal variation in carbon uptake and emissions

by terrestrial ecosystems has an effect on interannual variations in atmospheric CO2

concentrations (Schimel et al. 2001), although the magnitude remains uncertain

(Houghton et al. 1998, Houghton 1996, IPCC 1995). Recent attempts to measure forest-

level carbon exchange with the atmosphere have focused on temperate, boreal, and arctic

ecosystems (Aubinet et al. 2001, Black et al. 2000, Law et al. 2000, Valentini et al. 2000,

Clark et al. 1999, Frolking et al 1996, Goulden et al 1996, Vourlitis and Oechel 1997),

with only few eddy flux studies from tropical forests (Fan et al. 1990, Grace et al. 1996,

Mahli et al. 1998). However, tropical forests account for -35% of global net primary

productivity, >50% of the carbon in aboveground terrestrial biomass and -20% of the soil

carbon (Melillo et al. 1993, Dixon et al. 1994). Old-growth rainforest were historically

thought to be carbon neutral (input = output). This was challenged by Grace et al.

(1995a) and Fan et al. (1990) based on eddy covariance, who suggested that apparently

long-undisturbed Amazon forests sequester carbon. If generally true, the implications for

carbon science and policy making are enormous.




1






2


The energy balances of tropical forests are complex and many dynamic feedback

mechanisms among radiation, cloud formation and precipitation have been identified

(Wielicki et al. 2002, Hartman et al. 2001, Sohn and Smith 1992). This complexity

extends to the potential role of the tropical energy balance in affecting tropical and global

climates and general and anomalous circulations (Kelly and Randall 2001, Timmerman et

al. 1999, Chen and Van den Dool 1999, Fasullo and Webster 1999, Larson et al. 1999).

Much of our understanding of these dynamics has relied on model results, which have

shown large spatial and temporal variability in both sensible and latent energy budgets

(Kelly and Randall 2001, Raman et al. 1998, Hulme and Viner 1998, Shuttleworth 1988).

In situ studies have either scaled leaflevel measurements to whole canopies (Bigelow

2001, Avisshar 1993, Roberts et al. 1993), or have estimated the energy balance

components using eddy covariance over short periods (e.g., 8 d, Shuttleworth et al.

1984). Quantifying the variation of energy balance parameters and their biophysical

controls over longer periods should allow for better predictions of runoff and improved

models of regional and global climate.

Both physical and physiological factors influence forest energy balance, including

incident radiation, albedo, rain, interception, canopy capacitance, and aerodynamic (ga)

and bulk surface (gb) conductances. Incident radiation in the tropics varies less

seasonally than to that at higher latitudes, and values at the surface are more related to

cloudiness than changes in solar zenith angle. General circulation models tend to

underestimate net radiation in the tropics because of uncertainties in estimating surface

albedo and cloud cover (Cramer et al. 1999, Ruimy et al. 1999). Albedos of forests range

from 0.1 -0.2, with annual and seasonal differences affecting the surface energy balance.





3


Large differences in annual rainfall have been observed in the tropics that are thought to

be influenced by el Nifio-Southern Oscillation (ENSO) and other anomalous circulations.

This in turn affects the amounts of water available for evapotranspiration. A general

observation is that -50 % of annual rainfall is re-circulated to the atmosphere through

transpiration and evaporation of intercepted water, with the other 50% as runoff

(Shuttleworth 1989). This implies that local climate is strongly effected by how energy is

partitioned at the surface.

Canopy conductances for tropical forests have been estimated using both

ecophysiological and micrometeorological approaches. Ecophysiologists estimate bulk

surface conductance (gb) by scaling leaf-level or sap-flow measurements to the canopy

(Whitehead 1998, Dolman et al. 1991), while micrometeorologists often model gb in

relation to meteorological parameters (Wright et al. 1996). Aerodynamic conductance is

generally calculated as a function of horizontal windspeed, zeroplane displacement, and

roughness length (Denmead and Bradley 1985). Evapotranspiration from tropical forests

is generally thought to be strongly dependent on aerodynamic conductance, because of

the high rainfall and the significant proportion of the time when the canopy is wet,

reducing the importance of gb in evaportranspiration (Shuttleworth 1989).

Research objectives are: 1) Latent heat is greater than sensible heat all the time

(net radiation >40 W m-2), 2) below-canopy energy balance contributes significantly to

the overall energy balance across temporal scales, 3) with ecosystems that receive greater

precipitation, the more total time the canopy is wet. Because this wet tropical forest

receives - 4000 mm y-1 of precipitation, more relative time is needed to dry the canopy,

so evapotranspiration rates rely more on aerodynamic conductance than physiological





4


controls (bulk canopy conductance), and 4) Evapotranspiration from wet tropical forests

is strongly dependant on net radiation, and accounts for - 50% of the precipitation.














CHAPTER 2
ECOSYSTEM-LEVEL CARBON EXCHANGE

The objective of this chapter was to define the patterns of diurnal and annual net

carbon dioxide exchange and their climatic controls, for a lowland tropical wet forest in

northeastern Costa Rica. At this forest, large interannual fluctuations in aboveground net

primary productivity since 1984 closely paralleled fluctuations in the atmospheric CO2

concentrations and were strongly negatively correlated with average nighttime minimum

temperatures (Clark et al. in review). In this paper, we assess the effects of

environmental variations on the diurnal, seasonal, and interannual patterns of forest-level

carbon exchange for this forest from 1998 through 2000.



Materials and Methods

Study Site

This study was conducted as part of a long term study of tropical forest carbon

cycling, the CARBONO project, at the La Selva Biological Station, Puerto Viejo de

Sarapiqui, Costa Rica (10025' 51"N, 840 00'59"W, elevation 80-150 m.a.s.l.). La Selva is

located in northeastern Costa Rica in the Caribbean lowlands at the base of the central

volcanic chain and was classified as tropical wet forest in the Holdridge life zone system

by Hartshorn and Peralta (1988). This forest has an average 400 trees ha'1 > 10 cm

diameter ha-1 from -100 species (Lieberman et al. 1985), dominated by the mimosoid

legume canopy species, Pentaclethra macroloba (34% of the basal area, Clark and Clark



5






6


2000). Mean tree height is 20-25 m, with emergents exceeding 60 m. Canopy gaps

occupy - 0.01-0.04 ha ha' (Denslow and Hartshorn 1994) making the overall canopy

very aerodynamically rough. Incident mean (1993-1998) daily solar radiation was 14.9

MJ m-2 d-', with a range from 0.4 to 31.3 MJ m-2 d"'. Mean annual temperatures were

24.6 "C (1982-1998, Organization for Tropical Studies, OTS, unpublished weather

records). Mean annual precipitation is 4000 mm (from 1963 to 2000), with a short drier

period from December to the end of May, but with no month receiving less than 100 mm

(Sanford et al., 1994). Soils range from relatively fertile Inceptisols in riverine areas to

low pH, low phosphorus Ultisols in upland areas (Sollins et al. 1994).

Moisture-laden northeast trade winds originating over the Caribbean Sea

dominate surface winds (Hassenrath 1991). During most (85%) daytime hours, the

annual mean surface wind direction is 900. The wetter season (June through November)

and drier season (mid December through May) are controlled by the movement of the

equatorial low-pressure trough, i.e., the eastern Pacific intertropical convergence zone

(ITCZ). During the drier season, the sub-tropical Hadley cell dominates general

circulations, while the tropical cell dominates wet season circulations (Sanford et al.

1994). There are no data showing the exact passing of the ITCZ in Costa Rica. Hence,

seasons are defined here as wet season beginning on May 1 (DOY 121), and dry season

on December 20 (DOY 353). These dates are 10 d after the average date that the ITZC

passes through Barro Colorado Island, Panama (pers. comm. Steve Paton). Other

circulations may influence wet season climate including temporales, polar air masses that

move down the North American continent generating depressions and prolonged rain

events, chiefly occurring in November and December (Schlutz et al. 1998). Veranillos,





7


temporary and often irregular movement of the South Pacific anticyclone northward,

create short dry periods, typically lasting 7 to 10 days in September or October. Sanford

et al. (1994) and Holdridge et al. (1971) provide further site information for La Selva,

and Waylen et al. (1996a) and Hassenrath (1991) provide more details on its climatology.

Because La Selva is located at 100 N latitude, there is little diurnal change in

sunlight over the course of the year, with only a 40-min difference in day length between

solstices. For this study, sunrise and sunset were defined as 0600 and 1800 h, delineating

daytime and nighttime periods.

A 42 m tower (Upright Inc. Selma, CA) was used to access the canopy

environment and to support meteorological instrumentation. The site was a relatively flat

ridgetop in an area of generally rolling topography, with -20-30 m relief between stream

bottoms and ridgetops (OTS unpublished digital elevation model). After accounting for

stability effects, a source area model (Schuepp et al. 1990) was used to estimate that

under stable conditions, 95% of the cumulative flux was derived from within 1.2 km of

the tower (at a mean horizontal windspeed of 3 m s1). The tower was sited to minimize

edge effects, below-canopy advection either to or from the site, and any major directional

differences due to forest composition and structure.



Meteorological Data

Microclimate data were collected continuously at the tower top. Measured

variables included incident radiation (LI-190, LI-Cor Inc., Lincoln, NE), photosynthetic

photon flux density (PPFD, LI-200X, LI-Cor Inc.), aspirated air temperature (Ta, 100 Q

platinum RTD, Omega Engineering, Stamford, CT), and bulk rainfall (TE525 metric,





8


Texas Electronics, Dallas, TX). Atmospheric pressure (PB105, Vaisala, Helsinki,

Finland) was monitored at ground level.

All of the above data were collected at an interval of 5 sec and compiled as 30-

min averages with dataloggers (CR1OX and 21X, Campbell Scientific Inc., Logan, UT).

Instruments were cleaned, leveled as necessary, and recalibrated according to

manufacturers' instructions. At times when the PPFD sensor was inoperational, PPFD

was estimated by linear regression equation relating PPFD to incident radiation (R2 >

0.99). Likewise, when either power outages occurred or aspirated air temperature were

not logged, air temperature was estimated from a regression against an air temperature

sensor CS500 (Campbell Scientific) also mounted at 42 m in a radiation shield (R2 >

0.98).

Long term meteorological data from La Selva were used to examine decadal scale

trends in microclimate (OTS, http://www.ots.duke.edu). PPFD and air temperature have

been measured since 1982 and bulk precipitation since 1961.

To assess zero-plane displacement (d), or the mean level of momentum absorption, four

3-cup anemometers (Model 03103-L, R.M. Young, Traverse City, MI) were mounted

vertically along the tower at 35.5, 31.6, 28.2, and 25 m above the ground. d was then

estimated by determining the intercept (yo) of Eq. 2-1:

z35.5 Eq. 2-1
logd = Jlogu+ yo
Z25

where Z is measurement height above the ground (m), i is the 30-min time average of the

instantaneous measurement of horizontal wind velocity (cm s'1) at each height.

Roughness length (zo) was estimated using d:





9


u* /-1 Eq. 2-2
Z0 = (Zm - d). exp Eq. 2-2
\ Umk)

where Zm is the measurement height (42 m), u* is friction velocity (m s'-), im is the 30-

min time average of the instantaneous measurement of horizontal wind velocity at

measurement height (m s'l), and k is the von Karmen constant (0.41, dimensionless).

The ratio of convective to mechanical production of turbulent kinetic energy (Monin-

Obukov length, L) was used to determine atmospheric stability as in Eq. 2-3:

L = pC_ Tu Eq. 2-3
gkH

where p is the density of air (kg m-3), Cp is the specific heat capacity of air (J kg-' Kl'), Ta

is in Kelvin (K), g is acceleration due to gravity (m s-2), and H is the sensible heat flux

density (J m72 s'1) (Pahlow and Parlange 2000, Monteith and Unsworth 1990, Rosenberg

et al. 1983).



Net Ecosystem Exchange (NEE) Measurements

A closed-path eddy covariance system was used to estimate the portion of NEE of

CO2 contributed by turbulent exchange. Because the below-canopy environment was not

always subject to turbulent transfer (i.e., well-mixed conditions), a profile system was

used to estimate the rate of change of [CO2] below the canopy. NEE was then estimated

as

27 8[C2
NEE = w'CO2'+ 8Z27 Eq. 2-4
f.5 at
0.5 t

where w' and C02' are the deviations of instantaneous values from a running mean of

vertical windspeed (m 's) and molar fraction of CO2 (umol CO2 mor'), respectively, and





10


Zx is measurement height (m). The first term is the 30-min time-averaged eddy

covariance flux. The second term is the storage flux below 42 m. The convention used is

that negative values of NEE correspond to uptake of CO2 by the forest from the

atmosphere.

A 3-D sonic anemometer (K-probe, Applied Technologies Inc., Boulder CO) was

used to measure wind velocities in each polar coordinate (w, v, u) and sonic temperature

(0). The gas sampling inlet was mounted on the sonic anemometer and co-located with

the top transducer in the w-axis. Infrared gas analyzers (IRGA, model LI-6262, LI-Cor

Inc.) were used to measure concentrations of CO2 and H20 vapor, controlled for pressure

and temperature at ground level inside a climate-controlled structure. Flow rates were

maintained by pumps (KNF Neuberger, Trenton, NJ) and mass flow controllers (Model

series 200, 0-10 Ipm, Teledyne Hastings Inc., Los Angeles CA). Sampled air flowed

through -60 m of tubing (4.8 mm ID Teflon tubing) at a rate of 8 1pm resulting in a

lagtime of-14.2 s.

The NOAA flux software program (McMillen 1988), with a 400 second digital

recursive running mean and a fixed lagtime, was used to collect raw eddy covariance data

files. IRGA voltage outputs were digitized by a 12-bit analog-to-digital board.

Covariances, wind and scalar statistics, and coordinate rotations were calculated in real

time at 10 hz. Protocols for accuracy, precision, quality control and assurance were used

as defined by the AmeriFlux Science Plan

(http://cdiac.esd.oml.gov/programs/ameriflux/scif.htm).

The response of all instruments must be as fast as, or faster than, the turbulence

that is carrying the bulk of mass and energy. This process occurs between frequencies of





11


1-10 hz, within the inertial sub-range (Kaimal and Finnigan 1994). While the sonic

anemometer operates at these speeds, the response time for the IRGAs is slower, at-8-9

hz. To account for this frequency loss, fast Fourier transfer (FFT) analyses were applied,

where the portion of attenuation shown by the co-spectral density in the inertial sub-range

was compared to the total spectra. This ratio for w'8' was considered ideal, since

temperature (0) was measured by the sonic anemometer at 10 hz, and then compared to

the spectra ofw'C02' to estimate a spectral correction factor, SCf:

I / 10 Eq. 2-5
SnSwco2. * w' C02-1 nSwco2 * w ' CO2
SC f =0 0..001
i / 10
I nSwo * w'Of' nSw,o * w' '-1
0.1 0.001

where Sw,x, is spectral density ofw' and CO2' or 0', n is the natural frequency, and w'x' is

the mean covariance of w' and C02' or 0' (Meyers and Baldocchi 1991, Baldocchi and

Meyers 1989). Although we assumed that the dissipation of turbulent kinetic energy

occurred in the inertial sub-range for all the scalars, we did not expect that the Kaimal

spectral relationship (a slope of- 2/3 within the inertial sub-range) would hold for every

30-min period because of roughness, differing stabilities, and possible density driven

flows over time.

Profile measurements were used to calculate below-canopy CO2 storage dynamics

(Eq. 2-4). CO2 was collected from 6 inlets at 0.5, 7.3, 11.95, 16.55, 21.2, and 27.6 m. A

datalogger (model 21X, Campbell Scientific Inc.) was used to operate solenoids that

switched the flow (-3 1pm) from each inlet through the IRGA (Li-Cor 6262) for 5 min

during each 30-min period and to record the raw data. Platinum resistance thermometers

(100 0 PRT, m68, Omega Engineering, Stamford, CT) housed in radiation shields were





12


co-located with each inlet. When sampling occurred, the airflow acted to aspirate the

PRTs. Temperature and humidity profiles were used to account for changes in mass flow

due to changes in density (Webb et al. 1980). Below-canopy storage was calculated from

line-averaged profile measurements using Eq. 2-4; it was assumed that this profile was

similar across the flux source area.

Both eddy covariance and profile measurements began in April 1998 and

continued through the end of December 2000. Gaps in the dataset occurred for periods of

2-14 days when either power failure or instrumentation malfunctions occurred. IRGAs

were calibrated every 2-3 days. Improved precision in calibration was achieved starting

in February 1999 by plumbing nitrogen through the IRGA reference cell as a zero

reference. A model was used to relate daytime NEE to PPFD (Ruimy et al. 1995):

NaPax Eq. 2-6
NEEday = RE + mE
a + Pmax

where NEEday was calculated using Eq. 2-4, Pma is maximum ecosystem CO2 uptake rate

(p.mol CO2 m"2 s'1), RE is ecosystem respiration (grmol CO2 m-2 s'1), 0 is PPFD (lmol m-2

s-1), and a is apparent quantum efficiency (aCO2/0o). To describe the effects of

temperature on nighttime NEE, a second model was used:

NEEnight = R exp(b*T) Eq. 2-7

where Ro is the base ecosystem respiration rate (pmol CO2 m-2 ss-) when air temperature

is 0 �C, T is temperature (�C), and b is an empirical coefficient.

A general linear model (SAS v. 8.01, Cary, NC) was used to test first and second

order effects of other variables on NEE, including PPFD, temperature, VPD, season and





13


year. Sigmaplot v. 5.0 (SPSS Inc., Richmond, CA) was used to describe these

relationships.

Data Screening


Eddy covariance data were screened for validity and removed when either i) the

standard deviation ofw', CO2', or 0' was > 1.74 times the mean, ii) rain occurred, iii) 30-

min data collection periods were incomplete, or, iv) signals from either the sonic

anemometer or the IRGA were out-of-range. Profile data were removed when either i)

data were beyond 3 SD from the mean, ii) 30-min data collection periods were

incomplete, or iii) signals from the IRGA were out-of-range.



Leaf Area Estimation

Although the La Selva forest is largely evergreen, seasonal differences in leaf area

index (LAI) were expected because 8% of the tree species are deciduous in the dry

season, and 28 % of tree species produce annual leaf flushes, many at the onset of the wet

season (Frankie et al. 1974). Furthermore, many tropical rain forest tree species are

facultatively deciduous, losing up to half of their leaves during prolonged dry periods

(Richards 1996). Photographic estimates of eLAI (estimated LAI) were derived using the

WINPHOT program (Ter Steege 1996) each year during the wet and dry seasons across

18 randomly stratified 0.5 ha plots (description of statistical design for plot layout in

Clark and Clark 2000). Within each plot, 6 photographs were made at each sampling

date under diffuse light conditions at the same randomly chosen points. Because these

estimates were derived optically with no means of direct calibration, they should be

viewed relatively.





14


Results



Characterizing Canopy-Level Turbulence

There were -2/3 slopes for the normalized spectra of wind velocities in the

inertial sub-range during periods of both stable and unstable atmospheric conditions,

confirming a transfer of energy to the canopy with shear forces dominating (Figure 2-1)

The spectral density decreased during stable conditions (Figure 2-1B), as did the eddy

covariance flux estimates, but the general relationships held. Buoyancy forces produced

measurable vertical wind movement at night, as indicated by the positive 1:1 slopes at

wavenumbers < 0.1. The observed shift in the spectral peak between stable and unstable

conditions was similar to that reported in other studies (Kaimal and Finnigan 1994,

Anderson et al. 1986). The spectral correction factor ranged from 1.18 to 1.08 and

varied with u* (Figure 2-2).

Zero-plane displacement (d) and zo for momentum also varied with stability

(Table 2-1). During unstable periods (L< -50 m), zo increased to -2.4 m, a long

roughness length even for a forest (Hansen 1993), with a mean level of drag (d) of-22

m. Aerodynamic roughness lengths sharply decreased from slightly unstable conditions

(-50 < L < -10) to neutral conditions (-10 < L < 10), indicating the quick formation of

stratified laminar flow and the decoupling of the below-canopy environment.

The relationship between NEEday and u* was linear (Figure 2-3A). There was

decrease in the NEEnight with u*, although no threshold was observed. When NEEight

data were averaged across u* intervals of 0.025 m s-", the relationship was strongly linear

below a u* of 0.45 m s"' (Figure 2-3B). No relationship, however, was found between





15


the residuals from the energy budget and u* under any stability conditions, so that no u*

threshold could be determined and used to filter data, as has been done in most other

studies (e.g., Clark et al. 1999). We assumed that the most accurate estimate of nighttime

turbulent exchange occurred at larger values ofu*(i.e., > 0.4 m s"') and that conversely,

storage occurred at very low u* (i.e., < 0.05 m- s"), in spite of the lack of obvious

thresholds. The consequences of these assumptions are discussed below.



Diurnal Patterns in NEE

The diurnal pattern of the CO2 storage flux was very consistent throughout the

year (Figure 2-4A). The greatest fluxes were observed in early morning hours, when

below-canopy CO2 that was respired during the night and stored in the air column below

the sonic anemometer was vented or re-fixed through photosynthesis. The magnitude of

these morning ejections may be underestimated, because venting may skew the

distribution of wind statistics in the 30-min dataset, and so valid data could have been

inadvertently removed during the screening process.

The maximum average storage flux was -5.6 Imol CO2 m-2 S-1, which occurred at

-0800 when the convective boundary layer was developing, as shown by the increasing

vertical windspeed in Figure 2-4B. Storage fluxes decreased until -1400, after which

only net effluxes from below the canopy were observed. This also coincided with peaks

in below-canopy temperature (Figure 2-4B) and vertical windspeed. Storage efflux

increased to a peak just after sunset. The maximum nighttime (before 0600) storage

efflux was 2.97 p�mol CO2 m2 sr1, with an average of 1.6 +0.13 umol CO2 n"2 s'.

Storage generally decreased throughout the night, along with temperature and vertical






16


windspeed. Vertical windspeed and below-canopy temperatures diverged between 1630

and 0130, which indicates horizontal advection of below-canopy CO2 off the site may

have occurred and that a portion of the flux may have been missed (Mahrt et al. 2000).

The maximum NEEday based on the 3-yr mean for each half-hour was -17.3 +0.3 jtmol

CO2 m2 s-', occurred at - 1130, and closely followed the inverse pattern of PPFD (Figure

2-5). Nighttime eddy covariance flux was positive and fairly constant throughout the

night. At dawn, this flux sharply decreased (uptake into the forest) for -30 minutes with

decreases in Monin-Obukov length (Eq. 2-3).



Environmental Controls on NEE

NEEday. NEEday was negatively correlated with PPFD (Figure 2-6A) and had an

estimated mean maximum of -18 � 9 pmol CO2 m2 s'-. A linear model that included

second-order effects of year, PPFD, and VPD explained ~72% of the total variation in

NEEday. PPFD alone accounted for -51% of the variation (Figure 2-6), with no

significant effect of season. The light response function of Eq. 2-6 had a R2 of 0.51.

Residuals from this function were weakly related to temperature and VPD (Figure 2-6B

and 2-6C).

Since VPD includes temperature as a component, VPD and temperature are

correlated, and since VPD directly affects stomatal conductance (Law et al. 2000,

Landsberg and Gower 1997), we examined the influence of VPD on NEEday further by

separating NEEday into three VPD classes (0-0.5, 0.51-1.00, and > 1.0 kPa) and refitting

Eq. 2-6 for each year. The results (Figure 2-7, Table 2-2) show that the response function

is more linear within a VPD class. P.m and RE in 1998 were significantly lower than





17


those found in 1999, while a in 2000 was greater than in the other two years (Table 2-2).

Light compensation points ranged from 110 to 207 umol CO2 m"2 s'.

NEEnight. The mean NEEnight was 4.82 +0.6 upmol CO2 m2 -1 (mean �1 SD). This may

be an underestimate because >80% of the nighttime turbulent exchange measurements

were made with u* < 0.4 m s'-. Using only data with u* > 0.4 m s-, the estimate

becomes considerably larger, 6.98 pmol CO2 m-2 s-1, with 4.83 �0.21 and 2.15 �0.11

pmol CO2 m"2 s-1 (mean +1 SE) the contributions from eddy covariance and storage

fluxes, respectively. In addition, the averaged daytime RE values (from Eq. 2-6, Table 2-

2) ranged from 5.07-6.42 uLmol CO2 m-2 s1, supporting use of the higher NEEnight values

derived here.

NEEnight was weakly related to temperature over the entire sampling period

(Figure 2-8), with a Qio of 1.79 (p < 0.053).



Estimating Annual NEE

I estimated annual NEE in two ways. In the first case, half-hourly meteorological

data were used to drive Eq. 2-6 for each year and VPD class to derive an annual NEEday.

Then a fixed NEEnight of 6.98 umol CO2 m"2 S-1 was subtracted. I refer to these results as

NEEmodeled. In the second case, measured NEEday and NEEnight (with u* > 0.4 m s-1) were

used, and only the gaps were filled using the procedure above; results are referred to as

NEEgap. There were marked differences in the cumulative NEE among years and method

of estimation (Figure 2-9). Annual NEE was estimated to be higher in 1999 and 2000

than in 1998, with 2000 the strongest sink year.





18


Discussion



Characteristics of the La Selva Canopy

I assume that the eddy covariance measurements from La Selva were robust and

represent the nature of CO2 exchange between the canopy and the atmosphere on the

basis of two requirements generally prescribed for this method: (1) a consistent energy

cascade in the inertial sub-range in the power spectra under both stable and unstable

conditions (Kaimal and Finnigan 1994), and (2) the measurement height was at least 1.8

times d (it was in fact at least 8 times zo above d in unstable conditions and 6 zo above d

under all other stability classes, Schmid et al. 2000, Monteith and Unsworth 1990). I did

not expect to see a well-developed energy cascade during nighttime conditions. It is

likely that this was due to roughness induced turbulence in the nighttime flows.

It is unclear why no u* threshold in the eddy covariance data was observed.

Turbulence structures above a fixed plane (e.g., above a vegetated canopy) from other

studies have generally been described using Monin-Obukov theory (Leclerc et al.

submitted) and measured over uniform canopies with short aerodynamic roughness. As

roughness lengths become greater (> 1 m), the effect on turbulence in the well-mixed

layer and applicability of Monin-Obukov theory become questionable (Nakamura and

Mahrt 2001, Ayotte et al. 1999, Raupach and Finnigan 1997). Under these conditions, u*

becomes homogenized over a broader range of turbulence lengths and potentially has less

explanatory power (Nakamura and Mahrt 2001). At La Selva, there were insufficient

data to assess if the linear relationship between u* and NEEnight existed at values >0.45 m

s-1, or if the relation ultimately reached some asymptote. Data that would have





19


contributed toward developing a u* filter may also have been screened out by other

criteria.



Environmental Controls on NEE

The average-diurnal pattern in NEE (Figure 2-5) is strongly symmetrical around

1130. On average, the storage term contributed 33 % to NEEnight. The power spectra for

nighttime eddy covariance indicated a transfer of mass and energy, but the flux was

relatively small and the source distance quite long (1.5-2.0 km). The nighttime

environment below the canopy is quite different from that at the tower top and subject to

diabatic flows controlled by changes in air density and slope (Mahrt 1992), making the

source area for the storage flux more localized. The greatest variation in nighttime CO2

profiles occurred during periods with the most rapid changes in Ta. The diurnal patterns

in CO2 profiles and respective storage fluxes, however, were relatively constant

throughout our measurement period. After the morning re-assimilation of stored CO2

dissipated, NEEay was clearly dominated by above-canopy fluxes.

A coarse estimate of the amount of recycled CO2 can be obtained as the difference

between integrated NEEnight and the morning storage flux. This represents a value of

32% for the fraction of integrated NEEnight that was recycled below 42 m, similar to

isotopically-derived estimates of re-synthesized CO2 from the Ducke forest near Manaus,

Brazil (39%, Sternberg et al. 1997) and within the range from another neotropical forest,

Barro Colorado Island, Panama (31-38%, Sternberg et al. 1989).

The lack of a relationship between 30-min averages of NEEnight and temperature

was likely due to the combination of different factors. First, there are numerous sources





20


of CO2 for ecosystem respiration, each with their own controlling factors (Davidson et al.

1998). Soil respiration can be influenced by soil type, water status and temperature

(Schwendenmann et al. submitted), and foliar respiration may be influenced by nitrogen

and/or photosynthate availability in the canopy at the time of leaf expansion. Second,

there was only a small range annual nighttime temperature (< 9 0C). Third, the above-

canopy source area is integrated over a larger area than that from the below-canopy

environment (Raupach et al. 1992), subjecting the storage flux to localized biotic effects

or the possibility of below-canopy advection. Fourth, as mentioned above, 80% of flux

estimates were made under conditions with low u* (< 0.4 m s-'), questionable conditions

in this case for eddy covariance. Finally, carbon may have been exported from the

system as CO2, due to large morning ventilation events could not be quantified with

confidence (Wilson et al. 1998, Mahrt 1992).

The relationship between NEEnight and temperature (Figure 2-8) based on longer

averaging intervals is subject to the same potential sources of error as mentioned above

and may also lead to underestimates NEEnight. Schwendenmann et al. (submitted)

reported soil respiration rates at La Selva from upland soils ranged from 3.3 to 4.3 utmol

CO2 m 2 S1. Given that soil respiration accounts for 50% of RE for a wide range of

forests (Amthor 1994, Ryan 1991), this further suggests that our higher value (6.98 upmol

CO2 m"2 s-1) for NEEnight is more likely correct. Moreover, when I recalculated annual

NEEnight using the nighttime respiration function in Figure 2-8, it also resulted in a very

low value (4.3 pmol CO2 s'2 s-), similar to that found using the NEEight data across all

u* values, and again a value completely out of step with measured soil respiration and

our other estimation of NEEnight.





21


The potential exists that below-canopy nighttime advective flow contributes error

in NEEnight. However, for this to occur there must exist specific conditions, such as, a

strong upslope temperature gradient, a breakdown of below-canopy resistance to flow,

and/or a net vertical movement of wind into the forest. At La Selva however, there is

only a small change in temperature over a large upslope area, there is considerable

resistance presented by large below-canopy leaf area and tree stems, and nighttime

above-canopy vertical and below-canopy horizontal windspeeds (data not shown) were

not much different from the expected accuracy of the sonic anemometer (i.e., 0.05 m s-).

This is an unresolved issue and these factors are concerns for many tower flux sites

(Massman and Lee 2001).

In the face of increasing global temperatures (National Academy of Science

2000), there is increased focus on the role of temperature in controlling the carbon

dynamics in the tropics. Kindermann et al. (1996) modeled the effects of increased

temperature on carbon stores and with even small increases in temperature (-0.5 �C),

large effluxes of carbon to the atmosphere are expected. It is hypothesized that most of

this carbon will be from the tropics (Trumbore et al. 1996). At La Selva, large year-to-

year fluctuations over the past 16 years in aboveground biomass increments have been

negatively correlated with both the mean nighttime temperature and variations in annual

fluctuations in atmospheric CO2 concentrations (Clark et al. in review). In this study, I

found only a small temperature influence on NEEay and NEEnight and only when all three

years of data were pooled and hourly averages were used. I cannot dismiss the NEEdy

response to temperature as entirely due to VPD, and may in fact be partially due to

photorespiration. Interestingly, Grace et al. (1996) and Malhi et al. (1998) did not report





22


a nighttime temperature response for Amazonian forests using eddy covariance data. It

may well be that effects of temperature on ecosystem respiration from tropical forests

may only become apparent after many years of observation.

There were large interannual differences in apparent forest-level quantum

efficiency (a) estimated from the NEEday data. This may indicate large adjustments in

forest structure and physiology in response to the climatic variation among our study

years. Waring et al. (1995) concluded that seasonal differences in both LAI and canopy-

level quantum efficiency largely controlled productivity from a deciduous forest in

northeastern US. In our study, the estimated a for 2000 was significantly higher than

those from the preceding two years and approached the theoretical maximum for C3

leaves (Lloyd et al. 1995, Farquhar et al. 1980). However eLAI did not follow this

pattern and increased only 1999 and into the dry season of 2000. That neither changes in

LAI nor a were related to annual NEE suggests that there are interactions between NEE,

canopy leaf dynamics and climate for this complex tropical wet forest.

The relationship between NEEday and VPD could be a result of either

physiological or physical effects. A physiological effect could be stomatal closure in

response to a hydrologic limitation, either high VPDs or indirectly, decreases in soil

moisture availability. The location of the hydraulic limitation in the La Selva forest is not

known. A physical effect could be through though modification of canopy architecture

through premature leaf drop, leaf folding or changes in leaf orientation. Whole forest

canopies do not fully saturate even at full insolation (Ruimy et al. 1995, Wang and

Polglase 1995). Changes in leaf angle or leaf closure in the upper canopy allows

penetration of light to deeper canopy layers, allowing for increased carbon gain in the





23


lower canopy. This offsets the effect of leaf closure or changes in orientation at leaf scale

in terms of light response at the ecosystem level. Only 8% amount of the time, however,

were VPD values > 1 kPa and during 97% of the daytime (when net radiation was > 40 w

m 2), latent heat fluxes were greater than sensible heat fluxes (13 <1, unpublished data).

This strongly suggests that the La Selva canopy had access to abundant soil water. The

only exception was in the 1998 dry season, when 30% of daytime VPDs were > 1 kPa,

precipitation was the lowest ever recorded (68, 38, 126 mm monthly total rainfall in

January, February and March, respectively), and daily mean Ta was -1 �C above the long-

term average.

The 1998 dry season was at the end of the 1997-1998 warm-phase El Nifio

Southern Oscillation (ENSO) and was warmer and drier. During December 1998, a cold-

phase (la Nifia) ENSO brought greater precipitation, cooler temperatures, and lower mean

daily insolation (and PPFD), with several days receiving < 5 MJ d"'. Overall, 1998 was

warmer and drier during the dry season, but had more precipitation, cooler temperatures

and reduced light during the latter part of the year (compared to the other two years, 2-4).

The greater eLAI in 1998 coupled with lower a, a greater portion of time with VPDs > 1

kPa in the dry season, and overall lower mean daily insolation (13.3 MJ d"', 2-4), likely

contributed to the La Selva forest being close to carbon neutral in 1998.

During 1999, the daily insolation was well above the long term trend, but in November

and December the insolation was well below the long-term average due to a prolonged

temporal, suggesting that the effects of increased annual incident radiation outweighed

those of reduced eLAI and a prolonged temporal, making this forest a moderate sink of

carbon.





24


Despite these seasonal variations in climate, I did not find any seasonal effects on

NEE for any year. Even though seasonal displacements of the ITCZ alter Hadley cell

circulations, changes in individual environmental factors do not necessarily occur in

concert as a result. The initial passing of the ITCZ can be intermittent and there can be

multiple 'false starts' (Hastenrath 1991). Moreover, the northern most progression of the

ITCZ is just north of Costa Rica and with erratic movement, prolonged periods of dry

weather can occur during the otherwise wet season (Sanford et al. 1994). Other regional

anomalies can occur, as during 1998 and 1999, when heavy rains persisted from

December into January, even after the ITCZ passed. In 1998, this was brought about by a

cold-phase ENSO event, and in 1999 by a prolonged temporal.

The climatic trends observed during the 97/98 ENSO were typical for this region

(Waylen et al. 1996b, Cavazos and Hastenrath 1990), as were the conditions observed

during temporales (Sanford et al. 1994). The 97/98 ENSO, however, brought the highest

temperatures in the 19-year La Selva record. The apparent increasing frequency and

magnitude of ENSO events (Timmermann et al. 1999, Corti et al. 1999), may have

implications for interpreting the effects of climate change on NEE, as at La Selva, 1998

had the lowest estimated NEE. This supports the findings ofTian et al. (1998) who

concluded in a modelling study that variations in NEE of tropical forests are controlled,

in part, by macro-level changes in climate, which in turn are driven by the timing,

frequency and magnitude of ENSO events.

A possible alternative explanation for the large interannual variation in NEE at La

Selva is that a substantial fraction of forest is in an early successional stage. There is a

high frequency oftreefall gaps in this forest, even though La Selva is below most





25


hurricane pathlines. If I assume that the forest is aggrading carbon during its stand half-

life, which was estimated as 77.3 y (Lieberman et al. 1990), and has gap formation rate of

0.96% area y'1 (Denslow and Hartshorn 1994), then 74% of the land area is under

recovery at any time. Moreover, the mortality rate for 1969-1982 was 2.03% (Lieberman

et al. 1990), but increased to 4.77% in 1997-1998 (Clark et al. unpublished data) in

upland plots associated with gaps, suggesting that gap formation rate also increased under

climatic conditions imposed by both ENSO phases.



Comparisons with Other Tropical Sites

The pattern and magnitude of NEE at La Selva were similar to those estimated by

eddy covariance at three tropical moist forest sites in the Brazilian Amazon (Reserva

Ducke, north central Amazon, Fan et al. 1990; Jari, south central Amazon, Grace et al.

1995b; Cuieiras, north central Amazon, Mahli et al. 1998). Mean maximum daytime

NEE estimated for these three forests ca, -18-to-20 uimol CO2 m"2 s-1 and mean nighttime

NEE -5-7 tmol CO2 m-2 s-1 (Table 2-5). This suggests that these ecosystems may have

similar controlling factors on NEE, even though there are marked differences in stand

characteristics. Mahli et al. (1998) hypothesized that cloudiness on insolation were the

strong determinate of NEEday, which may explain the low NEE in 1998 at La Selva. One

difference is that VPD was found to play a stronger role in regulating carbon gain at the

other sites than at La Selva (e.g., Mahli et al. 1998). These other three sites receive <

2500 mm y'- of annual precipitation and have lower soil water availability (Hodnett et al.

1996, Tomasella and Hodnett 1996), so that there are likely to be greater hydrologic






26


constraints on daytime NEE (Mahli et al. 1998, Williams et al. 1998). The range in

annual quantum yield at La Selva overlaps with the values for the other three sites.

Reduction in leaf carbon gain can occur when leaves are wet, lowering rates of

CO2 diffusion into the leaf by a factor of 104 (Jones 1992), thus reducing photosynthesis

(Ishabashi and Terashima 1995). Smith and McClean (1989) also found that

photosynthesis was significantly reduced in wet leaves that had a wettable cuticle, but

increased on leaves that had non-wettable cuticles. The ratio of species with

wettable:non-wettable leaves is at La Selva not known. However, because eddy

covariance data collected during rain events at La Selva were eliminated, NEEday may

have been over-estimated as a result. La Selva experiences rain for 18% of the time

annually, as compared to only 13% and 8% at Janr and Ducke, respectively (data from

1994-96, http\\.www.abracos.com). For this reason, the likelihood of a sampling bias are

potentially greater at La Selva due to a larger fraction of data removed in screening.

The range of our estimates of annual gross ecosystem production (GEP, Table 2-

6) overlaps the GPP estimate at Cuieiras. High ratios of NEE:GEP are thought to exist

only with forests rapidly accumulating carbon (e.g., 0.02 for an old-growth Pseudotsuga

menziesii forest, compared to 0.29 for rapidly a growing pine stand, Waring and

Schlesinger 1985). The NEE:GEP ratios were 0.18 and 0.194 for La Selva and Cuieiras,

respectively, during 2000 and 1995 indicating that these systems can be very productive

compared to other forests. All of the tropical sites cited here are considered "old-

growth", and assumed in the past to be near steady state with respect to carbon.

However, eddy flux data from none of them support this view and suggest instead that

tropical forests may be net sinks from 0 to much as -6 t C ha-' yv' (averaging ca. -2 t C






27


ha'- y-'). However, the systematic errors discussed earlier collectively would tend to

decrease the amount of accumulated NEE into the forest, since the carbon would be

measured when it enters the system at the top but not when it leaves the forest (e.g.,

downslope drainage of C02, loss of C as unmeasured volatile organic compounds, etc.).

Where within the La Selva ecosystem this carbon is apparently accumulating remains an

open question.





28


Table 2-1. Estimates of aerodynamic parameters, zero-plane displacement (d) from eq.
2-1, aerodynamic roughness length (zo) from eq. 2-2, and u* estimates according to
stability class (L), eq. 2-3, from La Selva Biological Station. zo, d and u* estimates are
median values �95% CI, L values are means �1 SE, and n is number of 30-min periods.

Stability class zo (m) d (m) u* (m s-') L (m) n

Unstable 2.41 �1.04 21.5 �1.80 0.37 �0.46 -716 428 57

Slightly unstable 3.62 �0.95 19.4 �1.91 0.34 �0.13 -22 +1.5 51

Neutral 0.45 +0.01 22.1 �6.73 0.11 +0.05 1 +0.6 84

Stable 0.44 +0.21 23.0 �1.00 0.22 �0.05 814 �553 186







Table 2-2. Parameter estimates and statistics from the light response function, eq. 2-5, across VPD classes and year
(mean �+CI, *p-value). The Pmx coefficients are presented at 2200 pmol mn2 s" in the VPD class 0-0.5 kPa, Because the light
response function has a close to linear relationship. *All parameter estimates were at least significant to p<0.0001 with an a = 0.05,
unless otherwise noted. adenotes a significant difference between 1998 and 1999, b between 1998 and 2000, c between 1999 and 2000.
VPD year Pmax a RE R2 Light compensation

point
0-0.50 1998 -32.72 na <0.002 0.0260 �0.003 4.51 �0.35 0.44 na.

0.51-1.00 -43.60 �4.92 0.0356 �0.007 4.50 �1.13 0.45 na.

>1.0 -32.40 �3.70 0.0317 �0.010 0.0019 5.33 �1.90 0.005 0.35 na.

pooled -40.80 �3.95a 0.0346�0.005b 5.07 �0.60a 0.49 146 �1.6

0-0.50 1999 -29.99 na <0.03 0.0217 �0.0020 5.77 �0.44 0.39 na.

0.51-1.00 -52.40 �5.34 0.0344 �0.0052 5.52 �0.99 0.47 na.

>1.0 -37.80 �1.79 0.0377 �0.0083 5.92 �1.73 0.0006 0.52 na.

pooled -55.01 �5.78C 0.0311 �0.0035C 6.42 �0.66 0.50 207 �2.2

0-0.50 2000 -26.52 na 0.054 �0.009 6.02 �0.94 0.48 na.

0.51-1.00 -39.00 �3.0 0.058 �0.016 7.85 +2.13 0.59 na.







>1.0 -40.10 �7.2 0.039 �0.036 0.067 6.57 �8.80 0.22 0.30

pooled -37.00 �2.07 0.058 �0.009 6.33 �0.83 0.51 110 �1.3






31


Table 2-3. Annual and seasonal differences in estimated leaf area index (eLAI)
m2 m-2 from La Selva, Costa Rica (S.F. Oberbauer unpublished data).
Season Year Mean (median) � 1 SE

Dry 1998 3.85 (3.96) �0.19

Wet 1998 4.85 (4.79) �0.11

Dry 1999 2.71 (2.52) �0.13

Wet 1999 3.76 (3.84) �0.11

Dry 2000 3.48 (3.51) �0.07

Wet 2000 3.43 (3.30) �0.13







Table 2-4. Meteorological data for years 1998-2000 from La Selva Biological Station, CR.
Data are means �1SE. For each year, cumulative meteorological data was used with a general linear model to test for between year
statistical significance, wherein, the slopes of the cumulative meteorological data equal the annual means. For each year the slopes
were highly linear, R2 > 0.98, p < 0.001, a, b, and c denote significant difference (a =0.5, p < 0.0001) for 1998, 1999 and 2000 from the
other two years, respectively
Time period mean air mean minimum mean soil mean daily Total

temperature, 'C temperature, �C temperature, �C insolation, MJ d-1 precipitation

1998 a 24.23 �0.015 a 24.03 �0.015 a 23.43 �0.018 a 13.3 �0.24 3495

1999 23.4 �0.014 23.19 �0.014 21.29 +0.023 b 16.82 �0.28 3475

2000 23.66 �0.016 23.44 �0.016 21.24 �0.022 c 14.97 �0.26 4127

January 1998 25.09 �0.025 24.8 �0.026 23.840.002 12.51 �0.12 68

February 1998 25.48 �0.047 25.23 �0.46 24.71 �0.002 12.7 �0.15 38

March 1998 25.44 �0.055 25.17 �0.053 24.79 �0.003 13.3 �0.17 126

June 1998 26.01 �0.031 25.73 �0.029 23.44 �0.043 12.44 �0.13 400

December 1998 23.7 �0.03 23.47 �0.03 21.22 �0.042 9.72 �0.15 909

November 1999 24.62 �0.048 24.31 �0.047 22.67 �0.003 13.78 �0.17 309

December 1999 22.03 �0.037 22.79 �0.05 20.85 �0.002 8.73 �0.13 524








Table 2-5. Across site comparison of stand attributes from four neotropical eddy covariance studies.
site La Selva Ducke Jaru Cuieiras

Location Lat. 100 26' N, 20 57' S 100 04' S 2035' S

Long. 830 59' W 590 57'W 610 56' W 600 06' W

Length of study period days 1006 12 55 365

Mean annual temperature �C 25 26a 25a 27.8

Annual rainfall mm 4000 2400b 1450a 2400b

Holdridge life zone type Tropical wet Tropical moist Tropical moist Tropical moist

Above ground biomass t C ha' 160.5 (� 18) 300-350 140-180 300-350

*Estimated LAI 2.7-4.9 5-6 4 5-6

Estimated quantum yield umol CO2 tmol photon' 0.022-0.058 0.051 0.025 0.048

Mean Nighttime NEE pImol CO2 m-2 s" 6.89 6.0 6.4 6.5

peak Daytime NEE umol COz m-2 s-1 -18 -19 -15 -17 to -21

Light compensation point pmol m72 s'- 147-208 120 130 80

Annual NEE tC ha' y-' c-0.09 to -6.1 -2.2 -1.0 -5.9







Source This study Fan et al. 1990 Grace et al. 1995b Mahli et al. 1998

adata obtained from ABRACOS webpage, http://yabae.cptec.inpe.br/abracos/avaiable.html, bHodnett et al. 1996, C data from
NEEmodeled. Note, there are no estimates of LAI from tropical forests that have been measured directly through destructive sampling.






35


Table 2-6. Between year measures of productivity and ecosystem efficiency from La
Selva, Costa Rica and Cuieiras, Brazil. Data from La Selva are from eq. 2-5 across VPD
classes and a fixed nighttime NEE, as noted in the text. Units ofGEP and NEE are
expressed as MT C ha' y'.

site La Selva, Costa Rica Cuieiras, Brazil

year 1998 1999 2000 1995

GEP 28.41 30.6 33.9 30.4

NEE -0.09 -1.66 -6.1 -5.9

NEE:GEP 0.004 0.055 0.18 0.194

Note: The Cuieriras dataset is from Mahli et al. 1998, and Malhi and Grace 2000. Data
from La Selva is based on NEEmodeled.







36









A -e T
-v- u
0.001 -



-213 slope





0.007 -
t-





0
0 B

0.01 -

I.� -"2/3 slope




000






inertial sub-range

0.001 0.01 0.1 1 10

wavenumber (frequency'1)










Figure 2-1. The relationship between the normalized power spectra, for the three wind
vectors against the wavenumber. Data are averaged from 73 30-min periods from
January 10 to Februray 28th, 1999. A) Mid-day, unstable conditions, beginning at 1200-
1230 p.m, and B) Mid-night, stable conditions, beginning at 2300-2330 p.m. Error bars
are +/-1 SE.







37








1.16 - 0.5


1.14 - - 0.4
U 0.4

1.12 -




0.31
0
1.10 -

0.2 >
0 0
1.08-

-0.1
a 1.06 - SC
-- U*

1.04 . . .. .. .. 0.0
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

time of day







Figure 2-2. Diurnal time series of the spectra-based correction factor,
for CO2 fluxes from La Selva as calculated from Eq 2-5. Data are 90-min running means
centered on the 30-min interval. Data are from 1998-2000. Error bars are +/- 1 SE.







38








80
A) Daytime (R2 = 0.14)
60 .

40 . . .







2 .- *'* ":. :2. �
O: �
S-80
r

-20

-40 -
* -

E -60

U -80
S 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
E 8





















Friction velocity (u*, m s"1)
Figure 2-3. Relationships between u* and daytime NEE, (A), and between B) nighttimeflux
edycoaine rstrg ~xse0 eddy covariancighttime fluxNEE were averaged
6 -



4 -



2 -



0 -




0.0 0.1 0.2 0.3 0 .4 0.5

Friction velocity (u*, m sa")







Figure 2-3. Relationships between u* and daytime NEE, (A), and between B) nighttime
eddy covariance or storage fluxes, and u* (B). Values of nighttime NEE were averaged
across intervals of 0.025 m s-', except for the righthand most eddy covariance point,
which was averaged from all data between 0.4 < u* < 0.54 m s"'. Error bars are +/- 1SE.







39








4 __-----------------,-------, ----
A
3- T






28 - i w
* 2







0" -3
U-

E

-5
.5


-6 _ _
0 500 1000 1500 20)0
S 30-1 0.12
S29 - T B

. 7 - 0.08


0)-
ca 0.04
24 U
m 232
22 22 T 0.00
0 500 1000 1500 2000

time of day












Figure 2-4. Diurnal characteristics of A) storage fluxes, and B) line-averaged
temperature and non-rotated vertical windspeed from La Selva. All data are from 1998-
2000. A 90-min running average was used with each estimate centered on the 30-min
interval as indicated. Sample size for storage and line-averaged temperature were 5566
and for vertical windspeed 16333. Error bars are +/- 1 SE.







40







10 1400
B

S1200


0 ^ 1000



0
-5 -v storage flux

SPPFD -T\ I "-

LL -10 400

-15 -200


-20 .-. r 0
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

time of day


Figure 2-5. The mean diurnal pattern of above-canopy eddy covariance (EC) and storage
(below canopy) flux, NEE (the sum of the fluxes) and PPFD from La Selva (1998-2000).







41








60
60 -i--------------------------------
A R= 0.51
40 .-
40






U 0 . * " " * S



- -I. * *





0 500 1000 1500 2000

PPFD (imol m"s"1)

B . R2 = 0.18

0








S -60
a 15 20 25 30 35





air temperature (C)
- 80 ----------

S60 R2 =0.21
* -20
-40 "


-60



S-" 60- ' " *' *














VPD (kPa)




Figure 2-6. Main effects of environmental variables on daytime NEE,
where, A) is the light response of NEE for all data (1998-2000), and B) and C) show the
residuals from the light response function in relation to temperature and VPD,
resectively.
VPD (kPa)







42








60
1998, VPD < 0.5 kPa R2 = 0.43
40- 0







40- O O O
oo o







-00
00 0


60
260 - A ------ i -------------(b--0---0--0






1998, 0.5 < VPD < 1.0 kPa R2= 0.42
. 40

E 20 0
-20 -
0

00
S-20 0 0
S-40- 0


go -.-----------------









1998, VPD > 1.0 kPa R2 = 0.41
40

20
-20 E








-60 b

0 500 1000 1500 2000
PPFD (2mol m02 "1)









Figure 2-7. NEE as a function of PPFD across a gradient of vapor pressure deficits
for years 1998, 1999 and 2000 from La Selva, Costa Rica.
-20-o~p



















for years 1998, 1999 and 2000 from La Selva, Costa Rica.







43







60
o o R2 = 0.39
40- 0 0

0 00
20 - O O O





-40 -


-60 -
1999, VPD < 0.5 kPa 0
-80 1i ii
60
S0 R2 = 0.47


0 20 000
E D












4R2- = 0.52
S 20

W -40- 0

-60 -
1999,0.5 VPD 1.0 kPa
-80 500 1000 1500 2000
60o


D (R = 0.52

20

V V


-20

-40 V

-60
1999, VPD > 1.0 kPa

0 500 1000 1500 2000

PPFD (Imol m2 s"1)





Figure 2-7. continued.







44







60
2000, VPD < 0.5 kPa R2= 0.48
40

20-

2 . 0


-20- o

-40 -

-60

-80
60

S2000, 0.5 < VPD < 1.0 kPa R2 = 0.59


9 20.

S 0 -

O -20-
E

m -40
w
z
-60









-eo - -i ------ i ------ i --
-80
60

2000, VPD > 1.0 kPa R2 = 0.30

20

01

-20 vv v

-40- v v

-60-

-80i
0 500 1000 1500 2000

PPFD (�mol m2 s"1)





Figure 2-7. continued.






45






6 -
R2 = 0.17
Q, =1.79





5 3-
E




23 24 25


2 -
23 24 25
line-averaged temperature (C)



Figure 2-8. The relationship between NEEnigt and line-averaged (below-canopy)
temperature. NEEnight estimates are the aggregated averages from each nighttime 30-min
period from the whole study period (Figure 2-4A). Temperature is an average from six
measurements through the canopy profile from 27 to 0.5 m.







46








0-

1 0.09 tC ha-' y-

2 -2

-3 -, -1.66 tC ha*' y'

-4

-5

-6
- -6.1 t C ha"' y-1
-' -7
-- 1998
S8... 1999
-8
w A. NEE modeled --- 2000
Lu -9 -
1 |i _ ___ ____________________________________
Z -





0 -2 -.. -1.43 t C ha y

-3

-4 -3.21 t C ha*' y-'

-5

-6

-7

-8
B. NEE gap -8.03 t C ha- y"
-9 111111i.
0 50 100 150 200 250 300 350

Day of Year







Figure 2-9. Cumulative NEE from La Selva, Costa Rica for 1998-2000.
Estimates were calculated using A) the results from the light response equation across
VPD classes and a fixed estimate for nighttime respiration, and B) direct estimates of
daytime and nighttime NEE with gaps filled using data from NEEmodeled















CHAPTER 3
ENERGY BALANCE AND MODELED EVAPOTRANSPIRATION FOR A WET
TROPICAL FOREST IN COSTA RICA

The objective of this chapter was to examine the annual and seasonal effects of

atmospheric environment (albedo, net radiation, vapor pressure deficit) and the surface

controls on the energy balance of a wet tropical forest in Costa Rica.



Methods



Meteorological Data

All measurements referred to here were collected from September 1997 to

December 2000. Instrumentation for measuring air temperature, relative humidity, bulk

precipitation, and net radiation were mounted at the top of a 42 m tower (Upright, Inc.,

Selma, CA). Prior to March 1 1999, air temperature (Ta) was measured with a CS500

probe (Campbell Scientific, Inc., Logan UT) installed within a radiation shield, and

linearly back-corrected to fit the response of the aspirated temperature sensor (R2 = 0.98).

Ta was measured with platinum resistance temperature detector (100 0 platinum RTD,

Omega Engineering, Stamford, CT) mounted in an aspirated shield. Relative humidity

was also measured with the CS500 probe, and rainfall with a tipping bucket rain gauge

(model TE525, metric, Texas Electronics, Dallas TX). Net radiation, Rn, was measured





47






48


from March 1, 1999 to December 2000, with a closed-cell thermopile-style sensor (NR-

lite, Kipp and Zonen, Delft, the Netherlands). From August 1, 1997 to March 1, 1999 Rn

was measured with a Fritschen-style sensor (model Q-7.1, Radiation Energy Balance

Systems, Seattle, WA). All data collected with the Q-7.1 were linearly back-corrected to

fit the response of the NR-lite (R2 = 0.97) and corrected for advected sensible heat.

Another net radiometer (model CNR. 1, Kipp and Zonen, Delft, the Netherlands) was also

used to estimate albedo during February-April, July, and September 2000. Two

resistance grid type leaf wetness sensors (model 237-L, Campbell Scientific, Inc.) were

mounted at 26 and 2 m, and histograms of relative wetness were compiled. To mimic the

wetting of leaves, the sensors were coated with a layer of flat, off-white latex paint.

Soil heat flux plates (model HFT-3, Radiation Energy Balance Systems) were

installed at a depth of 5 cm, in each of three 1 m x 1 m plots > 20 m distance apart near

the base of the tower. Atmospheric pressure was measured at -3 m (PB105, Vaisala,

Helsinki, Finland). All meteorological data were collected at 5 sec intervals and

compiled as 30-min averages with a datalogger (CR10X, Campbell Scientific Inc.,

Logan, UT). Instruments were cleaned, leveled as necessary, and recalibrated according

to manufacturers' instructions.



Energy Balance Estimates

An ecosystem-level energy balance can be estimated by

Rn =XE+H+G Eq. 3-1

assuming horizontal homogeneity, where Rn is net radiation, XE is the latent energy flux,

H is the sensible heat flux, and G is the soil heat flux (all units are W mn2). Both H and





49


XE were estimated by the summation of both eddy covariance (above-canopy) and below-

canopy fluxes, expressed by,

E = w'q'+ 27 [q]az27 Eq. 3-2
z0.5 at

H= w'O'+ {27 - ]-aZ27 Eq. 3-3
z0.5 at

where w', q', and 0' are the deviations of instantaneous values from a running mean of

vertical windspeed (m s'), specific humidity (mmol mor'), and virtual temperature (�C),

respectively, and Zx is measurement height (m). w' q' and w'O' are the turbulent

exchanges of water vapor and heat as estimated by eddy covariance method, and the

second term in each equation is the storage flux in the air column. The storage fluxes are

noted as XEtor and Hstor, respectively. The convention used here is that negative values

correspond to a flux into the forest from the atmosphere.

The eddy covariance system was comprised of a sonic anemometer (K-probe,

Applied Technologies, Inc., Boulder, CO),an infra-red gas analyzer (IRGA, model Li-

6262, Li-Cor, Linclon, NE), -60 m of tubing (4.8 mm ID Teflon) with the inlet co-

located with the sonic probe, a laptop computer, and a pump to pull air through the tubing

at - 8 1pm. The sonic anemometer measured the wind velocities in three dimensions at

10 Hz, where w is vertical windspeed, and u and v are the two horizontal windspeed

components. The anemometer was also used to estimate temperature, 0s, as a function of

the speed of sound and changes in air density (excluding water vapor, i.e., "virtual

temperature"). The IRGA measured the concentration of water vapor at -8 Hz. The

laptop and a flux software program from McMillen (1988), with a fixed lagtime (14.3 s),

were used to collect raw eddy covariance data files. Covariances, wind and scalar






50


statistics, and coordinate rotations were calculated in real time at 10 Hz. Because

transport of mass and energy by turbulence occurs between frequencies of 1-10 Hz,

within the inertial sub-range (Kaimal and Finnigan 1994), and because the IRGA's

response time is -8 Hz, fast-Fourier analyses were applied to correct for frequency loss.

Protocols for accuracy, precision, quality control and assurance were used as defined by

the AmeriFlux Science Plan (http://cdiac.esd.ornl.gov/programs/ameriflux/scif.htm).

The storage fluxes were estimated by adding together the changes of heat through

the forest profile in both the air column below 42 m and in the foliage (Eq. 3-3). Water

vapor was sampled from 6 inlets at 0.5, 7.3, 11.95, 16.55, 21.2, and 27.6 m on the tower.

Solenoids switched the flow (-3 1pm) from each inlet through a second IRGA for 5 min

during each 30-min period. For H, temperature through the profile, Op, was measured

with platinum RTDs housed in radiation shields and co-located with each inlet; when

sampling occurred, the airflow acted to aspirate the platinum RTDs, approximating true

0.

Because leaves have small thermal inertia but significant amounts ofH20,

changes in leafH were estimated by:

f27 aOp
Hleaf = Sw - Cpw - Larea " Eq. 3-4
.0.5

where H�af is the below-canopy leaf heat flux (W mn2), Sw, is the specific weight of leaf

water (kg H20 m-2 leaf area), Cpw is the heat capacity of water (J kg1 K-'), and Lare is the

one-sided leaf area based on LAI from polyculture plantations at La Selva (4.5 m2 m-2, S.

Bigelow and J.Ewel, pers. comm.). It was assumed that the profiles of Hor and XEsor

were similar throughout the flux source area. G was estimated as a 30-min average of the





51


3 soil heat flux plates. The contribution to G by the change in soil heat storage was

ignored because soil temperature varied little in a 30-min period, soil thermal properties

for these soils were not known, and the change in soil heat storage was expected to be

negligible.



Evapotranspiration Model

Evapotranspiration estimates were partitioned into whole forest transpiration and

the evaporation of intercepted precipitation. XET was calculated using the Penman-

Monteith equation;

ARn + PaCp[es(Ta)-ea(Ta)ga
AE = Eq. 3-5
A[A+ (l + ga)]
gb

where XE is latent energy flux (W m-2), A is the is the rate of increase in saturated water

vapor pressure with temperature (kPa K'), Pa is the density of air (kg m-3), es is the

saturated water vapor pressure at Ta, ea is the ambient water vapor pressure (kPa), ga is

the aerodynamic conductance (mol m"2 s'), X is the latent heat of vaporization (J kg-'), y

is the psychometric constant at 25 oC (0.0665 kPa K'), gb is the bulk canopy conductance

(mol m-2 s'-). To change units of energy to depth, XE was multiplied by a conversion

factor that included molar volume (mol mf3) and weight (kg mol'). Evapotranspiration

depth is noted in the text as ET.

A positive momentum flux into the canopy was assumed hence, aerodynamic

conductance was estimated by,






52


k2u Eq. 3-6

z42-d )+ In Z42-d z
In [ l )) [In z + Tm - T
S z J , Zm Z 42

where k is Von Karmen's constant (0.40), d is the zeroplane displacement (m), Zm is the

aerodynamic roughness length, and Tm and Th are the diabatic correction factors (m) for

momentum and sensible heat, respectively (Yasuda 1988, Arya 1988). Zeroplane

displacement and aerodynamic roughness changed with stability and were empirically

estimated for this study period (Table 2-1, Loescher et al. submitted). Diabatic correction

factors are a function of stability, where in stable conditions,

y, = "H = 6 n(l + ) Eq. 3-7

and in unstable conditions,

1+ (1-16)o0.5 Eq. 3-8
TH = -21n[ ;2yM = 0.6YT

with C is a stability parameter ratio of convective to mechanical turbulent production

z42 -d Eq. 3-9
L

where L is the Monin-Obukov length (Equation 2-3).

Bulk canopy conductance, gb was estimated by,

b ga A_ E Eq. 3-10
g- PaD42
CpAE

where D42 is the specific humidity deficit at measurement height (kg kg'1).

When gaps in the measured ?E occurred, both ga and gb were empirically

modelled by using relationships with horizontal windspeed (for ga), and the vapor

pressure deficit (VPD) and Rn, (for gb, Martin et al. 1997, Wright et al. 1996). gb was

then normalized to unity, and the upper limit to VPD determined (Jarvis 1976, Livingston






53


and Black 1987). This limit function was used to estimate a theoretical maximum, gmx,

by increasing gb as though VPD was not limiting. Then gmax, in turn, was related to Rn,

assuming that maximum conductance would take place with 0 VPD and high Rn. A

dimensionless decoupling coefficient, 0, was used to determine the relative effects ofga

and gb on evapotranspiration (Jarvis and McNaughton 1986);

A' +2
SQ= Y Eq. 3-11
A/i+2+ga
/Y gb

Evaporation of free water in the canopy was modeled using a Rutter-type model

(Calder et al. 1986). Because canopy water storage increases exponentially to a

maximum, with increases in precipitation from a single rain event, Ei was modeled every

30-min by,

4t = 30
Ei= -t C-ET})
t=0

Eq. 3-12

where t is min, C is canopy water depth. When the canopy was wet, ET was estimated

with gb set to 0. C was estimated by,


-k IP
C= Cap(l-e Ccap) Eq. 3-13

where Cap is the maximum canopy capacity, mm, k is a unitless canopy fill coefficient,

YP is the cumulative amount of water that fell during a 30-min interval. An empirical

estimate of 1.53 mm was used for Cap (Loescher et al. 2002). The stemflow component

of interception was ignored because it was assumed to be a small volumetric flux, i.e., <

2% of rain, (Schroth et al. 1999, Neal et al. 1993). A value of 0.28 for k from a broadleaf






54


plantation forest at La Selva was used (Bigelow 2001). If modelled ET was greater than

the remainder of canopy free water, the remaining depth was included as transpiration,

ET. Evapotranspiration was then the sum of Ei and ET.

A second method of estimating XET, the Priestly-Taylor equation, was used to

compare with the Penman-Monteith results. The Priestly-Taylor equation (Eq. 3-14)

simplifies the transfer process that is explicit in Eq. 3-5, and in doing so, is thought to be

appropriate for large-scale, well-watered vegetative canopies, like those typically found

in the wet tropics (Priestly and Taylor 1972).

AE = aRn,[ A Eq. 3-14
A+y

where a is a coefficient estimated by fitting the model results to empirical measures of

kE from Eq. 3-2. Monteith (1981) estimated an average a on a theoretical basis as 1.26,

but values observed over rough canopies have varied greatly (Jones 1992).



Results



Above-Canopy Environment

The Monin-Obukov (M-O) stability length did not differ with year or season. On

a diurnal basis, M-O length was neutral (- 0 m) during the night, and decreased during

the day time until 1400 when the boundary layer became weakly unstable (M-O - -125 m

Figure 3-1). After 1400 h, the M-O length sharply increased, and the boundary-layer

became weakly stable (-100 m) at 1600 h, but then returned to neutral conditions by

nightfall. Friction velocity (u*) was -0.1 m s"' during the night and increased during the

morning hours with convective turbulence. u* decreased after solar noon and continued






55


to decrease throughout the afternoon due to the dissipation of turbulent kinetic energy

(Figure 3-1).

Total daily R, ranged from 1.47 to 27.54 MJ d-' during the measurement period,

and differed among years (using a general linear model with an a = 0.05, p < 0.0001,

Figure 3-2) with mean daily totals for 1998-2000 of 13.31 �0.028, 17.48 �0.050, and

15.33 �0.040 MJ d-1 (mean �1 SE), respectively. R, also varied significantly with season

(p < 0.0001). Mid-day albedo did not change seasonally, and ranged from 0.118-0.135 of

incident short wave radiation.



Energy Balance

Diurnal temperature and water vapor profiles (Figure 3-3) followed trends similar

to those of other forests (Shaw et al. 1988). Heating of the air column during the day was

greater with height, i.e., there was a positive temperature gradient. However, counter

gradients were often observable between 21 and 27 m, where the leaf area was

concentrated. Cooling during the night often produced neutral or slightly negative

gradients, often with warmer temperatures at ground levels. Negative or neutral water

vapor gradients were observed all times, with counter gradients present during non-rain

days between 11 and 21 m height.

Soil heat fluxes followed very similar diurnal patterns throughout the year, and

ranged between � 16 W m-2 at any point in time, with negative flux into the system

during the daytime (Figure 3-4A). At night, ,Esor was - 3-5 W m"2. A larger L.Eor

efflux occurred in the early morning hours, presumably from convective winds mixing

the below-canopy airspace and evaporating free water. LEto flux continued to be





56


positive throughout the afternoon, but was more variable, with mean daytime values

ranging from -6 to 7 W m"2. This flux remained positive throughout the day across all

years. Mean nighttime storage of sensible heat (H.or + Hief) ranged from - 1-3 W m2,

and became negative in the early hours as the air space increased in temperature. The

maximum average Htor + Hieaf was ca. -18 W mn2, which occurred at -0800 h when the

convective boundary layer was developing. Hstor + Hilf increased and became positive at

-1400 h, which coincided with a late afternoon weakly stable/unstable boundary layer

(Figure 3-1). The below-canopy environment continued to lose sensible heat until -1900

h, when neutral canopy conditions prevailed. The total storage flux (Figure 3-4B) was

similar to that of Htor + Hef, but daytime fluxes were ameliorated by release of water to

the atmosphere. During nighttime neutral conditions, the flux was -8-10 W mn2. The

daytime minima was - -5 W m72 and the peak efflux was 17 W m-2, which occurred

during weakly stable conditions (Figure 3-4B).

The average 30-min XE was greater than H for all daytime hours and across

seasons and years (i.e., H/XE = p < 1.0, Figure 3-5A). A linear model that included

second-order effects of year, season, VPD and Rn, explained 79% of the total variation in

H +AE (Table 3-1). Because Rn and VPD are auto-correlated and VPD did not explain

any additional variation, it was removed from the linear model. Rn alone accounted for

69% and 68% of the variation in H and XE, respectively.

I could only close the energy balance (Eq. 3-1) to within 32-50%, with the

exception of periods with a wet canopy during the dry season of 1998. In general, H +

XE + G underestimated Rn, particularly when Rn values were < 400 W m72 (Figure 3-6).

The most likely reason for this was that XE estimates, which also contributed to greater





57


variance in P estimate before 1000 h (Figure 3-5A). For this reason, XE was empirically

estimated by XE = R, - H for model comparisons, rather than the XE estimate in Eq 3-2

(e.g., Twine et al. 2000). Bowen ratio calculated by H(Rn-H) were -50% smaller than

those estimated by IE in Eq 3-2, and the largest contribution of H occurring between 0900

and 1200 h (Figure 3-5B), coinciding closely with the observed diurnal pattern of Ta.



Modelled Conductance and Evaporation

ga, was linearly correlated with horizontal windspeed (Figure 3-7A). Values of

WH and WyM were small and all other parameters were relatively constant. We expected

the upper boundary of normalized gb (Figure 3-7B) to be negatively related to VPD,

because of the negative physiological response of canopies to VPD (Martin et al. 1997),

which was found for VPDs > 0.5 kPa. gmax was hyperbolically related to R& (Figure 3-

7C). Estimated values of gb were always less than ga (Figure 3-8A). At dawn, gb was -1

mol m"2 s-1 until 0800 h, then steadily decreased during the day, approaching 0.1 mol m2

s-' at dusk. The minimum ga was 1.5 mol m-2 s-1, which increased to > 2 mol m-2 s'1

during mid-day (1000-1500 h). Modeled values of ga and gb behaved similarly to those

derived using Eq. 3-2 (Figure 3-8B). Using conductances in the Penman-Monteith

equation during periods when the canopy was dry, ga explained 44% of the variation in

XE (Figure 3-9). 0 ranged from -0.3 during the night to -0.7 by 0830 h. For the

majority of daytime hours (0600-1600), I was > 0.5 (Figure 3-10).

The Penman-Monteith equation explained 68% of the observed variation in Rn -

H, but overestimated Rn - H by -28% (Figure 3-11A). In contrast, the more simple

Priestly-Taylor relationship accounted for 98% of the observed variation (Figure 3-11B).






58


Annual ETpT ranged from 1892 mm in 1998 to 2294 mm in 1999, and from 54% to 66%

of bulk precipitation (Table 3-2). Mean daily ETpT rates were also lower in 1998 and

greatest in 1999 (5.19 to 6.29 mm d-', respectively). Interception loss was greatest in

2000, with an annual total of 708 mm, accounting for 18% of bulk precipitation.



Discussion

Ecosystem Energy Dynamics

Daytime profiles of temperature and water vapor in the upper canopy often

showed counter gradients that were the result of winds that did not fully penetrate the

entire canopy. The higher canopy vegetation, between 21-27 m, was a physical barrier to

transpiration from below. As a result, the apparent XEstor at night was due adiabatic

cooling and consequent condensation. Storage fluxes can contribute substantially to the

overall ecosystem energy balance when R. is small or during the night, but on a diurnal

basis this was a very a small component (-2 %).

Bowen ratios (3) consistently < 1 indicated that water was not limiting XE at any

time during the year. Soils in the wet tropics generally do not exert hydraulic limitations

on XE (De Bruin 1983). This is likely also the case at La Selva, where soils have high

water-holding capacity and high hydraulic conductivity (Weitz et al. 1997, Sollins et al.

1994). Pentaclethra macroloba, the dominant tree (42% of the basal area), closes its

stomata and leaves in the late afternoon (- 1530 h). I could not detect associated changes

in either H or LE due to changes in P. macroloba physiology, suggesting that either

single species control on energy partitioning or cannot be detected at the ecosystem-level

for diverse tropical wet forests, the eddy-covariance technique is insensitive to species-






59


level responses in these forests, or that the time of day that P. macroloba closes its

leaves is inconsequential to energy balance.

The lack of energy balance closure at this site is probably attributable to both

losses of high and low frequency signals. This loss is more apparent when Rn < 400 W

m 2. A combination of the long sampling tube combined with high humidity conditions

may have led to some in-tube mixing. Also, the scale lengths for H and kE were highly

variable and may not have been fully captured using eddy covariance (Mahli 1996).

Similar degrees of non-closure have been found at other tower flux sites with

similarly large aerodynamic roughness (- 2 m), such as an old-growth pine stand in

Oregon (to within 20-30 %, Anthoni et al. 2000), and an old-growth conifer stand in

Washington (to within 10-35 %, M. Falk and J. Chen, pers. comm.). Studies from other

tropical have not reported the degree of energy closure for periods greater than 1 day, but

it is likely that similar results would also be obtained for these also complex natural

forests. A high degree of closure for 30-min intervals seems to require aerodynamic

roughness lengths less than -0.5 m to minimize viscous effects, less than saturated air for

long periods, and smaller sensor separation distances (e.g., Gholz and Clark, in press).



Conductances and Other Limits to Annual Energy Fluxes

The general diurnal patterns found for ga and gb are similar to those for other

tropical forests (Bigelow 2001, Wright et al. 1996, Shuttleworth 1984). Using the same

calculations, our gb estimates were almost identical to those found by Wright et al. (1996)

for a tropical forest in Brazil, suggesting similar physical and physiological controls on

XE from these two neotropical forest canopies. Bigelow (2001) examined gb for three





60


monocultural plantations at La Selva and found that all the species had higher late

afternoon rates (-0.5-0.8 mol m-2 S-1) than found for the old-growth forest. However two

species, Cedrela odorata and Cordia alliodora, had late afternoon gb < 0.2 mol m-2 s1 in

the dry season. The lower afternoon gb that we observed was likely an integrated

response ofP. macroloba closing its stomata and folding its leaves closed by 1630 h or

later. Significant differences were observed in gb between years and seasons, which

followed the same trends as PR. Between year differences in Rn explained much of the

interannual variability observed in measured and modelled XE. Increased rainfall in 2000

increased the absolute amount of interception, but not the fraction of rainfall intercepted.

The direct effects ofgb on XE could not be determined because these variables

could not be independently measured. However, low values of (e.g., < 0.3) during

night and early morning and evening hours indicate that these periods are the only times

when physiological control over XE occurs, likely due to the opening and closing of

stomata and leaves. Values of 0 > 0.5 suggest that mid-day XE is controlled more by Rn

and ga than gb. ga explained 44 % of the variation in UE during times when both

conductances were used to calculate ETpM (i.e., dry canopy conditions). The importance

ofga in controlling ETpM increased further because of the very high precipitation at La

Selva. For example, 32 % of the time, the upper canopy was wet.

The Priestly-Taylor relationship for well-watered conditions described the ET

dynamics of this tropical forest quite well. The fraction of available energy used for ETpr

was similar from year-to-year suggesting a thermodynamic constraint on H that limits

maximum daily virtual temperature (Calder 1986). Wright et al. (1996) and Calder et al.

(1986) also found that accounted for a large fraction of R& (>0.80) from a humid





61


Amazonian and Javanese forest, respectively, as did Bigelow (2001) for the monocultural

plantations at La Selva (0.79-0.90).

In a previous study in the old-growth forest at La Selva, Luvall (1984) determined

that the energy required for evapotranspiration exceeded Rn by 25%. Although this

seems counterintuitive, it may very well be true. This phenomenon has been observed

over crops (Ham and Hielman 1991) as well as other tropical forests (Jones 1992,

Shuttleworth 1989, Calder 1986). The likely explanation is that energy is locally

advected into the flux field. This is certainly possible at La Selva, where mean daytime

wind direction is -90 and the fetch is -2 km, beyond which the landscape is dominated

by pastures, crops and patches of secondary forests extending for - 60 km to the

Caribbean shore. Often, small (< 1 km in width) afternoon convective cells deposit rain

heterogeneously across the landscape. Advection of drier air masses with greater

evaporative demand is possible, particularly during Luvall's study in the early 80's when

much of the land in the Costa Rican coastal plain was being converted from forests to

agriculture. Estimates of ETpr reported here did not exceed available R&, suggesting

advection was not significant, in this case, although we cannot rule out the possibility that

advection contributed additional energy towards the overall balance.

The magnitude of canopy capacitance is in large part a function of physical

surface area of a canopy (Waring and Schleshinger 1985). At La Selva, high epiphytic

loads, bromeliad tanks and arboreal soil mats can contribute capacitance and may not

have been fully accounted for in our estimates. We used a fixed estimate of capacitance

of 1.53 mm. The fraction of rainfall that was intercepted seemed consistent from 1998 to

2000, and also with Luvall's study, which suggests that the canopy surface area at La






62


Selva is saturated for a large portion of time, that the relative annual amount of

throughfall is constant, and that surface area does not change to any appreciable degree

over time.






63


Table 3-1. First order regression parameters for the energy balance closure from 1998-
2000 across seasons and canopy conditions from a wet, tropical, old-growth forest at La
Selva, Costa Rica. y = Yo + b*x, y is UE + H, x is Rn, Yo and b values are mean �1SD, n
is number of30-min averages, and all regressions were significant at the p < 0.0001 level
unless otherwise noted.
year Yo b n R

1998

Dry canopy/dry season 12.0 �3.8 p = 0.0016 0.52 �0.02 429 0.72

Wet canopy/dry season 38.0 �6.2 1.07 �0.05 362 0.59

Dry canopy/wet season 12.7 �2.6 0.66 �0.01 2182 0.73

Wet canopy/ Wet season 7.9 �2.3 p= 0.0006 0.63 �0.11 1867 0.66

1999

Dry canopy/dry season 16.0 �2.3 0.63 �0.01 2372 0.83

Wet canopy/dry season 19.4 �2.5 0.57 �0.01 1362 0.75

Dry canopy/wet season 13.1 �2.3 0.55 �0.01 2356 0.74

Wet canopy/ Wet season 22.6 �5.0 0.50 �0.02 483 0.73

2000

Dry canopy/dry season 7.3 �4 p = 0.07 0.61 �0.01 1220 0.81

Wet canopy/dry season 17.5 �5.5 p = 0.0014 0.6 �0.02 256 0.84

Dry canopy/wet season 17.5 �2.6 0.57 �0.01 1868 0.68

Wet canopy/ Wet season 23.1 �4.1 0.68 �0.2 872 0.71







Table 3-2. Evaporative fluxes for La Selva, Costa Rica calculated using the Priestly-Taylor equation from La Selva, Costa Rica,
compared to results from other studies and forests.
Site Forest type Period Annual Annual Annual Daily ETpT Annual Annual Annual Annual

Rainfall ETpI ETPT/Rn (mm �f1SE) Ei ETp-/rain Ei/ETpT Ei/rain

(mm) (mm) (dimensionless) (mm)

This study old-growth 1998 3495 1892 0.95 5.19 +0.007 564 0.55 0.30 0.17

This study " " 1999 3475 2294 0.88 6.29 +0.009 587 0.66 0.26 0.17

This study " " 2000 4127 2230 0.97 6.10 +0.010 708 0.54 0.32 0.18

aLa Selva, CR old-growth Aug 82-Mar 83 4620 2172 1.25 5.9 760 0.47 0.35 0.17

bLa Selva, CR 3 Dec 94-Nov 95 3156 1318- 0.79-0.90 1.65-4.00 74-375 nd. 0.06-0.25 nd.

plantations 1509

cJanlappa, secondary Aug 80-Jul 81 2892 1481 0.96 nd. 595 0.52 0.41 0.21

Java

dDucke, old-growth 8 d in Sept 83 nd. nd. 0.92 nd. nd. 0.48 0.30 nd.

Brazil

References: aLuvall 1984, b Bigelow 2001, c Calder et al. 1986, d Shuttleworth et al. 1984, 1988. nd denotes no data.






65


Table 3-3. Annual fraction of time that the canopy was wet at two heights in the La Selva
canopy.
year Height of leaf wetness sensor

25m 2m

1998 0.42 0.59

1999 0.17 0.57

2000 0.23 0.48

All years 0.32 0.57







66








0.5 - 150
-A- Friction velocity
----- Monin-Obukov length I 100
0.4
(1)- 50
E

03 0.31



S- -150
> 0.2 o
- C


0.1
-150

A B C D A
0.0, ,, --------- -200
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

Time (hour)










Figure 3-1. Diurnal relationships of friction velocity and Monin-Obukov length
over an old-growth tropical forest. Data are averages using all data from 1998-2000.
Intervals A, B, C, and D are neutral, unstable, weakly unstable, weakly stable boundary
layers, respectively. Error bars are +/- 1 SE.






67







7000

- 1998, mean 13.31 MJ d1
. 6000 ....... 1999, mean 17.48 MJ d1
S--- 2000, mean 15.33 MJ d"
5000


j! 4000


% 3000


3 2000

E
o 1000


0-
0 100 200 300
DOY








Figure 3-2. Cumulative net radiation for 1998-2000 over an old-growth forest
in La Selva, Costa Rica. Daily means were derived using first-order regression. Each
year was significantly different at the a = 0.05 level, p < 0.001, and R2 > 0.99.







68








30
0600 h 1330 h 0 0530 h 1800 h
25 -


20


15 HEATING ' COOLING


I 10-


5


0
DAY NIGHT

22 23 24 25 26 27 28 29 3022 23 24 25 26 27 28 29 30

air temperature, C
30
0600 1430 h 0530h 1800 h
25


20


15 GAIN LOSS


10 -


5



DAY NIGHT

38 39 40 41 42 43 44 45 38 39 40 41 42 43 44 45

water vapor (mmol mol"')






Figure 3-3. Typical diurnal changes in below-canopy temperature and water vapor
profiles from an old-growth forest, La Selva, Costa Rica. Data are median values for all
of 1999. Error bars are +/- 1 SE.







69








20 -
A
15 -

10 -

15

0

-5


E -10

x -15 --0- G
SHo, + Hleaf

E,-20
S 20 700
0k ? Em + G + Ht + H-,- B
" R,, - 600
I- 15
u) 500

10
- 400

5 300

200
0
100

-5
0

-10 -- - - -100
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

Time (hour)







Figure 3-4. Diurnal patterns of storage energy fluxes from an old-growth wet tropical
forest in Costa Rica. Data are mean values from 1998-2000 with error bars +/- 1 SE.







70









1.8- A

1.6

1.4

- 1.2-

S 1.0-

S0.8

a 0.6
4-
o 0.4

0.2

0.0-

-0.2

-0.4


0.8 - 2000 B
--] 1999
1 1998
0.6


S0.4

0
. 0.2
C

0.0


-0.2


-0.4
600 800 1000 1200 1400 1600

Time (hour)



Figure 3-5. Daytime Bowen ratios for each year. Estimates of XE were calculated by A)
equation 3-2, and B) Rn - H, with H determined by equation 3-3. Data are measured
median (+ 1 SE) values with net radiation > 40 W m2.








71








1400 -----------
10 1998 wet canopy/dry season
" dry canopy/dry season R2 = 059
R = 0.72
1000

Boo

800
600 0� O' c

400 -

200 0 0 a 0


S-200

W 1400 -
dry canopy/wet season wet canopy/wet season
0 oop 2 =0.66
1200 R =0.73 R2 0.66

1000 o 0 - o

00 0
6000 0 00


-200
go o o ��o oO









_o o
-200
-200 0 200 400 600 800 1000 1200 -200 0 200 400 600 800 1000 1200

Net radiation (W mn)








Figure 3-6. The relationship between net radiation and estimated energy flux

for 1998-2000. Estimated energy flux includes modelled contribution from the

understory.








72









1400 -
dry canopy/dry season wet canopy/dry season a
1200o R2 0.82 o R = 0.76

1000 6- 0S 00 � 0
oo o o o o� oa
o00 o-o a o


000o o o 0o c 0 00
6001








1400
o0 0 o






000 o o� o o 0 0 o
000 00 oo
dry canopy/wet season owet canopy/wet season


S1200 R o.74 oj o
100- � ��� 00





200Soo- 6 QC 00o
0- 0

8 @s o oo o "





-200

-200 0 200 400 600 800 1000 1200 1400 -200 0 200 400 600 600 1000 1200 1400

Net radiation (W m"2)





Figure 3-6. continued








73









1400
2000 wet canopy/dry season
1200 dry canopy/dry season R = 0.84

1000 = 0.81 oa


0 c o 0 0


400 - 0 0
o o 0 o o0
200- - 00 -00co0 0 -0 0.
o o o o o

0
-200

Uj 1400



1000 o 0 oO o o 0

800 Oco
0 0 0
400 0 0

200 0 a o0 o - 0~ c ~ 000 0-





-200 0 200 400 600 800 1000 1200 1400 -200 0 200 400 600 800 1000 1200 1400

Net radiation (W m )




Figure 3-6. continued
Figure 3-6. continued







74





5T
A.
g= 1.37 + 0.66*u
R2 = 0.91




7E 3



2 -



1 -L-----------------------
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
horizontal windspeed (m s")
1.0 ------7
B.
relative gb = exp6


0.







02


0.0
0.0 0.5 1.0 1.5 2.0 2.5
Vapor Pressure Deficit (kPa)
4C.

gx = 0.362 + 3.23*R/(644.26 + R,)
R2 = 0.92
3- IT










0o
0 200 400 600 800 1000 1200
Net radiation (W m )








Figure 3-7. Emprical relations to model both aerodynamic and bulk conductance.
independant variables were averaged at different intervals, i.e., 0.25 m s",
0.025 kPa, and 50 w m-2, for A, B, and C, respectively. All data were median
values with +/- 1 SE.






75








3 -




2 -















E
-1
o 3
E




2 -
Cu


0



0-






--- bulk conductance, gb
---- aerodynamic conductance, gO

-1
400 600 800 1000 1200 1400 1600 1800 2000

Time (hour)





Figure 3-8. The diurnal relationship of aerodynamic and bulk conductance calculated by
A) collected eddy covariance data, and B) modelled based on Figure 7. Data are median
values with +/- 1 SE.






76







1600
o R2 = 0.44
1400

1200 -
0
1000 - 0 O
o 0 00


S600 -�
wO

0 8
200 - (8000 O

0 - 6)0 Cb
S0
-200 0

-400
0 2 4 6 8

g, (mol m-2 s1)







Figure 3-9. The relationship between aerodynamic conductance and latent energy flux
from La Selva during 1999. Data are empirical estimates from Eq. 3-6 (ga), and from Eq.
3-2 (UE).






77








1.0

0.9

0.8 -

0.7 -

i0.6
a 0.6

8 0.2







0.1 -

0.0 . .
0 250 500 750 1000 1250 1500 1750 2000 2250

time (hour)






Figure 3-10. Diurnal changes in the decoupling coefficient, f, from an old-growth wet
tropical forest, Costa Rica. Data is from 1997-2000. SE are typically < 0.006.







78



3000
y = 1 + 1.28*x A.
0 00 0
E 2500 R2 = 0.68 0 0 �

3 0 0 0 0

I0 8- @ 0 0 d�o
S1500 0 - -6z0 -- 0 - o

E5 1000 2o o o
E o
15 000 qo O












.200~~ ~ ~ '0 Sp --- i -- i -- i ------------ --4
0) - 0 0
o

0 oo 0 0

1400
y = 1 + 1*x B.
E 1200 R2 0.98

1000

2800
6W-




S2000








-0
61 400


1400
Nd


1000

-2000400 600 800 0

Rn - 00H (W m
4-
,r_ 400

0 0.



-200
-200 0 200 400 600 800 1000 1200 1400

Rn - H (W M-2)








Figure 3-11. Relationship between empirical and modeled estimates of the latent energy,
where A) is modeled using a Penman-Monteith equation, and B) using Priestly-Talyor
equation with a dry canopy, and C) using a Priestly-Taylor equation with a wet canopy.
All graphs used data from 1998-2000. An a of 1.24 was found for graphs B and C. All
slopes were significant, p < 0.0001. y-intercepts had p-values of 0.5, 0.04 and 0.04 for A,
B and C, respectively..














CHAPTER 4
CONCLUSIONS

Diurnal patterns of NEE at La Selva followed trends similar to those observed

elsewhere, with turbulent fluxes dominating in the daytime, and large storage fluxes

contributing largely in the early morning. Daytime NEE was a function of both abiotic

(PPFD, VPD and temperature) and biotic (quantum efficiency and eLAI) factors. VPD

limited NEEay when values were above > 1 kPa, but this only occurred over a small

percentage of time. There was a positive correlation between nighttime respiration and

temperature based on diurnal averages of all NEEnight measurements. We used a fixed

NEEnigt value, to estimate annual NEE. There was a big difference between our two

alternative estimates of annual NEE. But regardless of the calculation method used, the

results indicate that there is a large interannual variation in NEE at La Selva, related to

large scale regional climate dynamics.

Energy balance closure of this lowland tropical rainforest was ~ 74%. Storage

fluxes contributed very little (-2%) to the overall daily energy balance. Conductances

followed similar trends to those found of other tropical forests. Daytime XE was almost

always greater than sensible heat, suggesting that the trees in this forest are in contact

with sufficient ground-water reserves to minimize hydraulic stress. R& was a large

determinant for the annual energy flux, suggesting that the Priestly-Taylor model for

evapotranspiration is more appropriate in the tropics than the Penman-Monteith model.

The general rule holds true that Rn X E.



79















LIST OF REFERENCES


Amthor, J.S., 1994. Plant respiration. In: Wilkinson R.E (Ed.), Plant-Environment
Interactions. M. Dekker Publisher, New York. pp. 501-554

Anderson, D.E., Verma, S.B., Clement, R.J., Baldocchi, D.D., Matt, D.R., 1986.
Turbulence spectra of C02, water vapor, temperature and velocity over a deciduous
forest. Agricultural and Forest Meteorology, 38, 81-99.

Anthoni, P.M., Law, B.E., Unsworth, M.H., 1999. Carbon and water vapor exchange of
an open-canopied ponderosa pine ecosystem. Agricultural and Forest Meteorology, 95,
151-168.

Arya, S.P., 1988. Introduction to micrometeorology. Academic Press, London. pp. 307.

Aubinet, M., Chermanne, B., Vandenhaute, M., Longdoz, B., Yernaux, M., Laitat, E.,
2001. Long term carbon dioxide exchange above a mixed forest in the Belgian Ardennes.
Agricultural and Forest Meteorology. 108, 293-315.

Avissar, R., 1993. Observations of leaf stomatal conductance at the canopy scale - an
atmospheric modeling perspective. Boundary-Layer Meteorology. 64, 127-148.

Ayotte, K.W., Finnigan, J.J., Raupach, M.R., 1999. A second-order closure for neutrally
stratified vegetative canopy flows. Boundary-Layer Meteorology. 90, 189-216.

Baldocchi, D.D., Meyers, T.P., 1989. The effects of extreme turbulent events on the
estimation of aerodynamic variables in a deciduous forest canopy. Agricultural and
Forest Meteorology. 48, 117-134.

Bigelow, S., 2001. Evapotranspiration modelled from stands of three broad-leaved
tropical trees in Costa Rica. Hydrological Processes. 15, 2779-2796.

Black, T.A., Chen, W.J., Barr, A.G., Arain, M. A., Chen, Z., Nesic, Z., Hogg, E.H.,
Neumann, H.H., Yang, P.C., 2000. Increased carbon sequestration by a boreal deciduous
forest in years with a warm spring. Geophysical Research Letters. 27, 1271-1274.

Calder, I.R., 1986. What are the limits on forest evaporation?-A further comment.
Journal of Hydrology. 89, 33-36.



80






81


Calder, I.R., Wright, I.R., Murdiyaso, D., 1986. A study of evaporation from a tropical
rain forest-west Java. Journal of Hydrology. 89, 13-31.

Cavazos, T, Hastenrath, S.L., 1990. Convection and rainfall over Mexico and their
modulation by the Southern Oscillation. Journal of Climatology, 10, 377-386.

Chen W.Y., Van den Dool, H.M., 1999. Significant change ofextratropical natural
variability and potential predictability associated with the El Nino/Southem Oscillation.
Tellus A51, 790-802.

Clark D.B., Clark D.A., 2000. Landscape-scale variation in forest structure and biomass
in a tropical rain forest. Forest Ecology and Management. 137, 185-198.

Clark, D.A., Piper, S.C., Keeling, C.D., Clark, D.B. (in review) Tropical forest growth
and atmospheric carbon dynamics linked to annual temperature variation. Science

Clark, K.L., Gholz, H.L., Moncreiff, J.B., Cropley, F., Loescher, H.W., 1999.
Environmental controls over net exchanges of carbon dioxide from contrasting Florida
ecosystems. Ecological Applications. 9, 936-948.

Corti, S., Molteni, F., Palmer, T.N.,1999. Signature of recent climate change in
frequencies of natural atmospheric circulation regimes. Nature, 398, 799-802.

Cramer, W., Kicklighter, D.W., Bondeau, A., Moore, B., Churkina, C., Nemry, B.,
Ruimy, A., Schloss, A.L., 1999. Comparing global models of terrestrial net primary
productivity (NPP): overview and key results. Global Change Biology. 5, 1-15.

Davidson, E.A., Belk, E., Boone, R.D., 1998. Soil water content and temperature as
independent or confounded factors controlling soil respiration in a temperate mixed
hardwood forest. Global Change Biology. 4, 217-227.

De Bruin, H.A.R., 1983. Evapotranspiration in humid tropical regions. In: Hydrology of
humid tropical regions with particular reference to the hydrological effects of agriculture
and forestry practice. Proc. Hamburg Symp. Aug. 1983. IAHS Publ. No. 40.

Denslow J.S., Hartshorn G.S., 1994. Tree-fall gap environments and forest dynamic
processes. In: McDade, L.A., Bawa, K.S., Hespenheide, H.A., Hartshorn, G.S. (Eds), La
Selva: Ecology and Natural History of a Neotropical Rain Forest. University of Chicago
Press, Chicago, IL. pp. 120-127.

Dixon, RK., Brown, S., Houghton, R.A., Solomon, A.M., Trexler, M.C., Wisniewski, J.
1994. Carbon pools and flux of global forest ecosystems. Science, 263, 185-190.

Dolman, A.J., Gash, J.H.C., Roberts, J., Shuttleworth, W.J., 1991. Stomatal and Surface
Conductance of Tropical Rain-Forest. Agricultural and Forest Meteorology. 54, 303-318.






82


Fan, S-M., Wofsy, S.C., Bakwin, P.S., Jacob, D.J., 1990. Atmosphere-biosphere
exchange of CO2 and 03 in the Central Amazon forest. Journal of Geophysical Research.
95, 12851-16864.

Farquhar, G.D., Caemmerer, S.V., Berry, J.A., 1980. A biochemical model of
photosynthetic CO2 assimilation in leaves of C-3 Species. Planta. 149, 78-90.

Fasullo J., Webster, P.J., 1999. Warm pool SST variability in relation to the surface
energy balance. Journal of Climate. 12, 1292-1305.

Frankie, G.W., Baker, H.G., Opler, P.A., 1974. Comparative phenological studies of trees
in tropical wet and dry forests in lowlands of Costa-Rica. Journal of Ecology. 62, 881-
919.

Frolking, S., Goulden, M.L., Wofsy, S.C., Fan, S.-M., Sutton, D.J., Munger, J.W.,
Bazzaz, A.M., Daube, B.C., Crill, P.M., Aber, J.D., Band, L.E., Wang, X., Savage, K.,
Moore, T., Harriss, R.C., 1996. Modeling temporal variability in the carbon balance of a
spruce/moss boreal forest. Global Change Biology. 2, 343-366.

Gholz, H.L., Clark, KIL., (in press). Energy exchange across a chronosequence of slash
pine forests in north Florida. Agricultural and Forest Meteorolog.y

Goulden, M.L, Munge, J.W., Fan, S-M., Daube, B.C., Wofsy, S.F., 1996. Measurements
of carbon sequestration by long-term eddy-covariance: methods and critical evaluation of
accuracy. Global Change Biology. 2, 169-182.

Grace, J., Lloyd, J., McIntyre, J., Miranda, A. C., Meir, P., Miranda, H.S., Nobre, C.,
Moncrieff, J., Massheder, J., Malhi, Y., Wright, I., Gash, J., 1995a. Carbon dioxide
uptake by an undisturbed tropical rainforest in southwest Amazonia, 1992 to 1993.
Science, 270, 779-780.

Grace J, Lloyd J, McIntyre J, Miranda, A., Meir, P., Miranda, H., Moncriefi J.,
Massheder, J., Wright, I., Gash, J. 1995b. Fluxes of carbon dioxide and water vapor over
an undisturbed tropical rainforest in south-west Amazonia. Global Change Biology. 1, 1-
12.
Grace, J., Malhi, Y., Lloyd, J., McIntyre, J., Miranda, A.C., Meir, P., Miranda, H.S.,
1996. The use of eddy covariance to infer net carbon dioxide uptake of Brazilian rain
forest. Global Change Biology. 2, 209-217.

Ham, J.M., Heilman, J.L., 1991. Aerodynamic and surface resistances affecting energy
transport in a sparse crop. Agricultural and Forest Meteorology. 53, 267-284.

Hansen, F.V., 1993. Surface roughness lengths. ARL technical Report, U.S. Army, White
Sands Missile Range, NM 88002-5501.






83


Hartmann, D.L., Moy, L.A., Fu, Q., 2001. Tropical convection and the energy balance at
the top of the atmosphere. Journal of Climate. 14, 4495-4511.

Hartshorn, G.S., Peralta R., 1988. Preliminary description of primary forests along the La
Selva-Volcan Barva altitudinal transect, Costa Rica. In: Alemda, F., Pringle, C. (Eds),
Tropical Rainforests: Diversity and Conservation. California Academy of Science, San
Fransisco. pp. 281-295.

Hastenrath, S.L., 1991. Climate Dynamics of the Tropics. Kluwer Academic, Boston,
MA, 488 pp.

Hodnett, M.G., Tomasella, J., Marques Filho, A.O., Oyama, M.D., 1996. Deep water
uptake by forest and pasture in central Amazonia: Predictions from long-term rainfall
data using simple water balance model. In: Gash, J.H.C., Nobre, C.A., Roberts, J.M.,
Victoria, R.L., (Eds.), Amazonian Deforestation and Climate. John Wiley, New York.
pp.79-99.

Holdridge, L.R., Grenke, W.C., Hatheway, W.H., Liang, T., Tosi, J.A. jr. 1971. Forest
Environments in Tropical Life Zones: a Pilot Study. Pergamon Press, Oxford, 747 pp.

Houghton, R.A., 1996. Terrestrial sources and sinks of carbon inferred from terrestrial
data. Tellus Series B-Chemical and Physical Meteorology. 48, 420-432.

Houghton, R.A., Davidson, E.A., Woodwell, G.M., 1998. Missing sinks, feedbacks, and
understanding the role of terrestrial ecosystems in the global carbon balance. Global
Biogeochemical Cycles. 12, 25-34.

Hulme, M., Viner, D., 1998. A climate change scenario for the tropics. Climate Change.
39, 145-176.

IPCC., 1995. IPCC Second Assessment Report: Climate Change. Geneva, Switzerland,
pp 64.

Ishibashi, M., Terashima, I.,1995. Effects of continuous leaf wetness on photosynthesis:
adverse aspects of rainfall. Plant, Cell and Environment. 18, 431-438.

Jarvis, P.G., 1976. The interpretations of the variations in leaf water potential and
stomatal conductance found in canopies in the field. Philosophical Transactionsform the
Royal Society. B273, 593-610.

Jarvis, P.G., McNaughton, K.G., 1986. Stomatal control of transpiration: scaling up from
leaf to region. Advances in Ecological Research. 15, 1-49.

Jones, H., 1992. Plants and microclimate. Cambridge University Press, NY, 428 pp.






84


Kaimal, J.C., Finnigan, J.S., 1994. Atmospheric Boundary-Layer Flows; Their Structure
and Measurement. Oxford University Press, UK, 453 pp.

Kelly, M.A., Randall, D.A., 2001. A two-box model of a zonal atmospheric circulation in
the tropics. Journal of Climate. 19, 3944-3964.

Kindermann, J., Wurth, G., Kohlmaier, G.H., Badec, F.W., 1996. Interannual variation of
carbon exchange fluxes in terrestrial ecosystems. Global Biogeochemical Cycles, 10,
737-755.

Landsberg, J.J., Gower, S.T., 1997. Applications ofPhysiological Ecology to Forest
Management. Academic Press, NY, 354 pp.

Larson, K., Hartmann, D.L., Klein, S.A., 1999. The role of clouds, water vapor,
circulation, and boundary layer structure in the sensitivity of the tropical climate. Journal
of Climate. 12, 2359-2374.

Law, B.E., Williams, M., Anthoni, P.M., Baldocchi, D.D., Unsworth, M.H., 2000.
Measuring and modelling seasonal variation of carbon dioxide and water vapour
exchange of a Pinus ponderosa forest subject to soil water deficit. Global Change
Biology. 6, 613-630.

Leclerc, M.Y., Karipot, A., Prabha, T., Allwine, G., Lamb, B., Gholz, H.L., (submitted)
Tracer flux footprint validation over a forest canopy and effect of larger-scale advection
on eddy-covariance fluxes. Journal of Geophysical Research.

Lieberman, M., Lieberman, D., Hartshorn, G.S., Peralta, R., 1985. Small-scale altitudinal
variation in lowland wet tropical forest vegetation. Journal ofEcology. 73, 505-516.

Lieberman, D., Hartshorn, G.S., Lieberman, M, Peralta, R., 1990. Forest dynamics at La
Selva Biological Station, 1969-1985. In: Gentry, A.H., (Ed.), Four Neotropical
Rainforests. Yale University Press, New Haven. pp. 509-521.

Livingston, N.J., Black, T.A., 1987. Stomatal characteristics and transpiration of three
species of conifer seedlings planted on a high elevation south-facing clear-cut. Canadian
Journal of Forest Research. 17, 1273-1282.

Lloyd, J., Grace, J., Miranda, A.C., Meir, P., Wong, S. C., Miranda, B. S., Wright, I. R.,
Gash, J. H. C., McIntyre, J,. 1995. A simple calibrated model of Amazon rainforest
productivity based on leaf biochemical properties. Plant, Cell and Environment. 18,
1129-1145.

Loescher, H.W., Powers, J.S., Oberbauer, S.F., 2002. Spatial variation ofthroughfall
volume in an old growth tropical wet forest, Costa Rica. Journal of Tropical Ecology. 18,
397-407





85


Loescher, H.W., Oberbauer S.F., Gholz, H.L., Clark, D.B. (submitted) Environmental
controls on net ecosystem-level carbon exchange and productivity in a Central American
tropical wet forest. Global Change Biology.

Luvall, J.C., 1984. Tropical deforestation and recovery: the effects on evaporation
processes. Ph.D. dissertation, University of Georgia, Athens, Georgia. 146 pp.

Mahli, Y., 1996. The behavior of the roughness length for temperature over
heterogeneous surfaces. Quarterly Journal of the Royal Meteorological Society. 122,
1095-1125.

Mahli, Y., Nobre, A.D., Grace, J., Kruijt, B., Pereira, M.G.P., Culf A., Scott, S., 1998.
Carbon dioxide transfer over a Central Amazonian rain forest. Journal of Geophysical
Research. 103, 31,593-31,612.

Malhi, Y., Grace, J., 2000. Tropical forests and atmospheric carbon dioxide. Trends in
Ecology and Evolution, 15, 332-33.

Mahrt, L., 1992. Momentum balance of gravity flows. Journal ofAtmospheric Sciences.
39, 2701-2711.

Mahrt, L., Lee X.H., Black, A., Neumann, H., Staebler, R.M., 2000. Nocturnal mixing in
a forest subcanopy. Agricultural and Forest Meteorology. 101, 67-78.

Martin, T.A., Brown, K.J., Cermak, J., Ceulemans, R., Kucera, J., Meinzer, F.C.,
Rombold, J.S., Sprugel, D.G., Hinckley, T.M., 1997. Crown conductance and tree and
stand transpiration in a second-growth Abies amabilis Forest. Canadian Journal of Forest
Research. 27, 796-808.

Massman, W.J., Lee, X., 2001. Eddy covariance flux corrections and uncertainties in long
term studies of carbon and energy exchanges. Proceedings from the workshop for
unaccounted flux in long term studies of carbon and energy exchanges. Boulder CO.

McMillen, R.T., 1988. An eddy-correlation technique with extended applicability to non-
simple terrain. Boundary-Layer Meteorology. 43, 231-245.

Melillo, J.M., McGuire, A.D., Kicklighter, D.W., Moore, III. B, Vorosmarty, C.J.,
Schloss, A.L., 1993. Global climate change and terrestrial net primary productivity.
Nature. 363, 234-240.

Meyers, T.P., Baldocchi, D.D., 1991. The budgets of turbulent kinetic-energy and
Reynolds stress within and above a deciduous forest. Agricultural and Forest
Meteorology. 53, 207-222.

Monteith, J.L., 1981. Evaporation and surface temperature. Quarterly Journal of the
Royal Meteorological Society. 107, 1-27.






86



Monteith, J.L., Unsworth, M.H., 1990. Principles of Environmental Physics. Edward
Arnold Publishers, New York. pp. 291.

Nakamura, R., Mahrt, L., 2000. Similarity theory for local and spatially averaged
momentum fluxes. Agricultural and Forest Meteorology. 101, 265-279.

National Academy of Science. 2000. Reconciling observations of global temperature
change. National Academy Press, Washington, D.C., p 85.

Neal, C., Robson, A.J., Bhardwaj, C.L., Conway, T., Jeffery, H.A., Neal, M., Ryland, G.
P., Smith, C.J., Walls, J. 1993. Relationships between precipitation, stemflow and
throughfall for a lowland beech plantation, Black-Wood, Hampshire, Southern England -
Findings on interception at a forest edge and the effects of storm damage. Journal of
Hydrology. 146, 221-233.

Pahlow, M., Parlange, M.B., 2001. On Monin-Obukhov similarity in the stable
atmospheric boundary layer. Boundary-layer meteorology. 99, 225-248.

Priestly, C.H.B., Taylor, R.J., 1972. On the assessment of the surface heat flux and
evaporation using large scale parameters. Monthy Weather Review. 100, 81-92.

Raman, S., Niyogi, D.S., Prabhu, A., Ameenullah, S., Nagaraj, S.T., Kumar, U., Jayanna,
S., 1998. VEBEX: Vegetation and surface energy balance experiment for the tropics.
Proceedings from the Indian Academy of Science. 107, 97-105.

Raupach, M.R., Finnigan, J.J., 1997. The influence of topography on meteorological
variables and surface-atmosphere interactions. Journal of Hydrology. 190, 182-213.

Raupach, M.R., Weng, W.S., Carruthers, D.J., Hunt, J.G.R., 1992. Temperature and
humidity fields and fluxes over low hills. Quarterly Journal Royal Meteorological
Society. 118, 191-225.

Richards, P.W., 1996. The tropical rainforest; an ecological study. Cambridge
University Press, Cambridge UK. 575 pp.

Roberts, J., Cabral, O.M.R., Fisch, G., Molion, L.C.B., Moore, C.J., Shuttleworth, W.J.,
1993. Transpiration from an Amazonian rain-forest calculated from stomatal conductance
measurements. Agricultural and Forest Meteorology. 65, 175-196.

Rosenberg, N.J., Blad, B.L., Verma, S.B., 1983. Microclimate: the biological
environment. Wiley, New York. 495 pp.

Ruimy, A., Jarvis, P.G., Baldocchi, D.D., Saugier, B., 1995. CO2 fluxes over plant
canopies and solar radiation: A review. Advances in Ecological Research. 26, 1-81.





87


Ruimy, A., Kergoat, L., Bondeau, A., 1999. Comparing global models of terrestrial net
primary productivity (NPP): analysis of differences in light absorption and light- use
efficiency. Global Change Biology. 5, 56-64.

Ryan, M.G., 1991. A simple method for estimating gross carbon budgets for vegetation in
forested ecosystems. Tree Physiology. 9, 255-266.

Sanford, R, Paaby, P., Luvall, J.C., Phillips, E., 1994. Climate, geomorphology and
aquatic systems. In: McDade ,L.A., Bawa, K.S., Hespenheide, H.A., Hartshorn, G.S.,
(Eds.), La Selva: Ecology and Natural History of a Neotropical Rain Forest University of
Chicago Press, Chicago, IL. pp. 19-33

Schimel, D.S., House, J.I., Hibbard, K.A., Bousquet, P., Ciais, P., Peylin, P., Braswell,
B.H., Apps, M.J., Baker, D., Bondeau, A., Canadell, J., Churkina, G., Cramer, W.,
Denning, A.S., Field, C.B., Friedlingstein, P., Goodale, C., Heimann, M., Houghton, R.
A., Melillo, J.M., Moore, B., Murdiyarso, D., Noble, I., Pacala, S.W., Prentice, I.C.,
Raupach, M.R., Rayner, P.J., Scholes, R.J., Steffen, W.L., Wirth, C., 2001. Recent
patterns and mechanisms of carbon exchange by terrestrial ecosystems. Nature. 414,
169-172.

Schmid, H.P., Grimmond, S.B., Cropley, F., Offerle, B., Su, H-B., 2000. Measurements
of CO2 and energy fluxes over a mixed hardwood forest in the mid-western United
States. Agricultural and Forest Meteorology. 103, 357-374.

Schroth, G., Silva, L.F.d., Wolf, M.A., Teixeira, W.G., Zech, W. 1999. Distribution of
throughfall and stemflow in multi-strata agroforestry, perennial monoculture, fallow and
primary forest in central Amazonia, Brazil. Hydrological Processes. 13, 1423-1436.

Schuepp, P.H., Leclerc, M.Y., Macpherson, J.I., Desjardins, R.L., 1990. Footprint
prediction of scalar fluxes from analytical solutions of the diffusion equation. Boundary-
Layer Meteorology. 50, 355-373.

Schwendenmann, L., Veldkamp, E., Brenes, T., O'Brien, J.J., Mackensen, J., (submitted)
Spatial and temporal variation in soil CO2 efflux in an old-growth neotropical rain forest,
La Selva, Costa Rica. Biogeochemistry.

Shaw, R.H., Den Hartog, G., Neuman, H.H., 1988. Influence of foliar density and
thermal stability on profiles of Reynolds stress and turbulence intensity in a deciduous
forest. Boundary-Layer Meteorol. 45, 391-409.

Shuttleworth, W.J., Gash, H.C., Lloyd, C.R., Moore, C.J., Roberts, J., Marques Filho,
A.d. 0., Fisch, G., Silva Filho, V de P., Molion, L.C.B., Sa, L.D. de A., Nobre, J.C.A.,
Cabral, O.M.R., Patel, S.R., Moraes, J.C., 1984. Eddy correlation measurements of
energy partition for Amazonian forest. Q. J. R. Meteorol. Soc. 110, 1143-1163.





88


Shuttleworth, W.J., 1988a, Evaporation from Amazonian rainforest. Philosophical
Transactions form the Royal Society London. B233, 321-346.

Shuttleworth, W.J., 1989. Micrometeorology of temperate and tropical forests.
Philosophical Transactions form the Royal Society London. B324, 299-334.

Smith, W.K., McClean, T.M., 1989. Adaptive relationship between leaf water repellency,
stomatal distribution, and gas exchange. American Journal of Botany. 76, 456-469.

Sohn, B.J., Smith, E.A., 1992. Global energy transports and the influence of clouds on
transport requirements - a Satellite Analysis. Journal of Climate. 5, 717-734.

Sollins P, Sancho FM, Mata RC, Sanford RL Jr (1994) Soils and soil process research.
In: McDade, L.A., Bawa, K.S., Hespenheide, H.A., Hartshorn, G.S., (Eds.), La Selva:
Ecology and Natural History of a Neotropical Rain Forest University of Chicago Press,
Chicago, IL. pp. 34-53.

Steinberg, L.S.L., Mulkey, S.S., Wright, S.J., 1989. Ecological interpretation of leaf
carbon isotope ratios: influence of respired carbon dioxide. Ecology. 70, 1317-1324.

Stemberg, L.S.L., Moreira, M.Z., Martinelli, L.A., Victoria, R.L., Barbosa, E.M.,
Bonates, L.C.M., Nepstad, D.C., 1997. Carbon dioxide recycling in two Amazonian
tropical forests. Agricultural and Forest Meteorology. 88, 259-268.

Tian, H., Melillo, J.M., Kicklighter, D.W., McGuire, A.D., Helfich, III. J.V.K., Moore,
III. B., Vorosmarty, C.J., 1998. Effect ofinterannual climate variability on carbon storage
in Amazonian ecosystems. Nature. 396, 664-667.

Ter Steege, H.,1996. Winphot 5: a programme to analyze vegetation indices, light, and
light quality from hemispherical photographs. Tropenbos Guyana Report 95-2,
Tropenbos Guyana Programme, Georgetown, Guyana.

Timmermann, A., Oberhuber, J., Bacher, A., Esch, M. Latif M., Roeckner, E., 1999.
Increasing El Nifio frequency in a climate model forced by future greenhouse warming.
Nature. 398, 694-694.

Tomasella, J., Hodnett, M.G., 1996. Soil properties and van Genuchten parameters for an
oxisol under pasture in central Amazonia. In: Gash, J.H.C., Nobre, C.A., Roberts, J.M.,
Victoria, R.L., (Eds.), Amazonian Deforestation and Climate John Wiley, New York.
pp.101-124.

Trumbore, S.E., Chadwick, O.A., Amundson, R., 1996. Rapid exchange between soil
carbon and atmospheric carbon dioxide driven by temperature change. Science. 272, 393-
396.






89


Twine, T.E., Kustas, W.P., Norman, J.M., Cook, D.R., Houser, P.R., Meyers, T.P.,
Prueger, J.H., Starks, P.J., Wesely, M.L., 2000. Correcting eddy-covariance flux
underestimates over a grassland. Agricultural and Forest Meteorology. 103, 279-300.

Valentini, R., Matteucci, G., Dolman, A.J., Schulze, E. D., Rebmann, C., Moors, E.J.,
Granier, A., Gross, P., Jensen, N.O., Pilegaard, K., Lindroth, A., Grelle, A., Bernhofer,
C., Grunwald, T., Aubinet, M., Ceulemans, R., Kowalski, A.S., Vesala, T., Rannik, U.,
Berbigier, P., Loustau, D., Guomundsson, J., Thorgeirsson, H., Ibrom, A., Morgenstern,
K., Clement, R., Moncrieff, J., Montagnani, L., Minerbi, S., Jarvis, P.G., 2000.
Respiration as the main determinant of carbon balance in European forests. Nature. 404,
861-865.

Vourlitis, G.L., Oechel, W.C., 1997. Landscape-scale C02, H20 vapour and energy flux
of moist-wet coastal tundra ecosystems over two growing seasons. Journal of Ecology.
85, 575-590.

Waring, R.H., Schlesinger, W.H., 1985. Forest ecosystems: concepts and management.
Academic Press, NY, 340 pp.

Waring, R.H., Law, B.E., Goulden, M.L., Bassow, S.C., McCreight, R.W., Wofsy, S.C.,
Bazzaz, F.A., 1995. Scaling gross ecosystem production at Harvard Forest with remote
sensing: a comparison of estimates from a constrained quantum-use efficiency model and
eddy correlation. Plant, Cell, and Environment. 18, 1201-1213.

Waylen, P.R., Caviedes, C.N., Quesada, M.E., 1996a. Interannual variability of monthly
precipitation in Costa Rica. Journal of Climate. 9, 2606-2612.

Waylen, P.R., Quesada, M.E., Caviedes, C.N., 1996b. Temporal and spatial variability of
annual precipitation in Costa Rica and the Southern Oscillation. International Journal of
Climatology. 16, 173-193.

Webb, E.K., Perman, G.I, Luening, R., 1980. Correction of flux measurements for
density effects due to heat and water vapor transfer. Quarterly Journal of the Royal
Meteorological Society, 106, 85-100.

Wielicki, B.A.,Wong T.M., Allan, R.P., Slingo, A., Kiehl, J.T., Soden, B.J., Gordon,
C.T., Miller, A.J., Yang, S.K., Randall, D.A., Robertson, F., Susskind, J., Jacobowitz, H.,
2002. Evidence for large decadal variability in the tropical mean radiative energy budget.
Science. 295, 841-844.

Wietz, A.M., Grauel, W.T., Keller, M., Veldkamp, E., 1997. Calibration of time domain
reflectometry technique using undisturbed soil samples from humid tropical soils of
volcanic origin. Water Resources and Research. 33, 1241-2149.






90


Williams, M., Malhi, Y., Nobre, A.D., Rastetter, E.B., Grace, J., Pereira, M.G.P., 1998.
Seasonal variation in net carbon exchange and evaporation in a Brazilian rain forest: a
modelling analysis. Plant, Cell and Environment. 21, 953-968.

Wilson, J.D., Finnigan, J.J., Raupach, M.R., 1998. A first order closure for disturbed
plant canopy flows and its application to winds in a canopy on a ridge. Quarterly Journal
Royal Meteorological Society. 124, 705-732.

Wright, I.R., Gash, J.H.C., Rocha, H.R.d., Roberts, J.M., 1996. Modelling surface
conductance for Amazonian pasture and forest. In: Gash, J.H.C., Nobre, C.A., Roberts,
J.M., Victoria R.L. (Eds.), Amazonina deforestation and climate. Institute of Hydology.
Pp. 437-467.

Yasuda, N., 1988. Turbulent diffusivity and diurnal variations in the atmospheric
boundary layer. Boundary-Layer Meteorology 43, 209-221.




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EU1VCI3LA_MJ5AM3 INGEST_TIME 2015-03-05T21:31:10Z PACKAGE AA00028848_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES



PAGE 1

ECOSYSTEM-LEVEL RESPONSES OF CARBON AND ENERGY FROM A TROPICAL WET FOREST IN COSTA RICA By HENRY WILLIAM LOESCHER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2002

PAGE 2

I DEDICATE THIS DISSERTATION TO MY FATHER AND TO ALL THOSE WHO ACTIVELY ENJOY LIFE

PAGE 3

ACKNOWLEDGMENTS I wish to acknowledge the financial support of many organizations which made this study possible: U.S. Department of Energy Office of Science contract IBN-9652699 within the interagency Terrestrial Ecology and Global Change, the School of Forest Resources and Conservation, the Department of Energy as the National Institute for Global Environmental Change Program (NIGEC), the Organization for Tropical Studies, and the Biological Sciences Department at Florida International University. My degree is in part, due to the assistance and dedication of many people. I wish to thank the faculty and staff of the School of Forest Resources and Conservation for academic guidance and logistical support. I wish to thank the chairman of my committee, professor Henry L. Gholz, for academic, moral and creative support throughout my tenure as a graduate student, and sincerely acknowledge my other supervisory committee members for being thoughtfiil through difficult times, Stephan Mulkey, Francis Putz, Jennifer Jacobs, and Steve Oberbauer. I am indebted to David Clark, Deborah Clark, Monique Leclerc, Tim Martin, Tilden Meyers, Thara Prabha, and Anand Karipot. Thanks also go to David Hollinger for assistance with tower site location. Matt 'el tejano' Schroeder for data collection, and Steve Moore for programming assistance. I would specifically like to thank Jeff Amthor, and Wayne H. Smith, for additional financial support and guidance, and to Jennifer Jacobs for stimulating my interest in fluid dynamics. iii

PAGE 4

TABLE OF CONTENTS page ACKNOWLEDGEMENTS iii LIST OF TABLES vi LIST OF FIGURES vii ABSTRACT ix CHAPTERS 1. INTRODUCTION 1 2. ECOSYSTEM-LEVEL CARBON EXCHANGE 5 Materials and Methods 5 Study Site 5 Meteorological Data 7 Net Ecosystem Exchange (NEE) Measurements 9 Data Screening 13 Leaf Area Estimation 13 Results 14 Characterizing Canopy-Level Turbulence 14 Diurnal Patterns in NEE 15 Environmental Controls on NEE 16 Estimating Annual NEE 17 Discussion 18 Characteristics of the La Selva Canopy 18 Environmental Controls on NEE 19 Comparisons with Other Tropical Sites 25 3. ENERGY BALANCE AND MODELED EVAPORATION FOR A WET TROPICAL FOREST IN COSTA RICA 47 Methods 47 Meteorological Data 47 Energy Balance Estimates 48 Evapotranspiration Model 51 Results 54 iv

PAGE 5

Above-Canopy Environment 54 Energy Balance Modeled Conductance and Evaporation 57 Discussion 58 Ecosystem Energy Balance 58 Conductances and Other Limits to Annual Energy Fluxes 59 4. CONCLUSIONS 79 LIST OF REFERENCES 80 BIOGRAPHICAL SKETCH 91 V

PAGE 6

LIST OF TABLES Table Eige 2-1. Estimates of aerodynamic parameters, zero-plane displacement (d) from eq. 2-1, aerodynamic roughness length (zo) from eq. 2-2, and u* estimates according to stability class (L), eq. 2-3, from La Selva Biological Station, zo, d and u* estimates are median values ±95% CI, L values are means ±1 SE, and n is number of 30-min periods 2-2. Parameter estimates and statistics from the light response fimction, eq. 2-5, across VPD classes and year 29 2-3. Annual and seasonal differences in estimated leaf area index (eLAI) 31 2-4. Meteorolgical data for years 1998-2000 from La Selva Biological Station, CR 32 2-5. Across site comparison of stand attributes from four neotropical eddy covariance studies 26. Between year measures of productivity and ecosystem eflBciency from La Selva, Costa Rica and Cuieiras, Brazil 34 31 . First order regression parameters for the energy balance from 1998-2000 63 3-2. Annual evaporative fluxes calculated using the Priestly-Taylor equation from La Selva, Costa Rica 64 3-3. Annual fraction of time that the canopy was wet at two heights in the canopy 65 vi

PAGE 7

LIST OF FIGURES Figure Eige 2-1 . The relationship between the normalized power spectra 36 2-2. Diurnal time series of the spectra-based correction factor 37 2-3. Relationships between u* and daytime NEE 38 2-4. Diurnal chracterisitics of A) storage fluxes and line-average temperature and non-rotated vertical windspeed 39 2-5. The mean diurnal pattern of above-canopy eddy co variance (EC) 40 2-6. Main effects of environmental variables on daytime NEE 41 2-7. NEE as a function of PPFD across a gradient of vapor pressure deficits 42 2-8. The relationship between NEEnight and below-canopy temperature 45 29. Cumulative NEE fi-om La Selva, Costa Rica for 1998-2000 46 31 . Diurnal relationships of fiiction velocity and Monin-Obukov length 66 3-2. Cumulative net radiation for 1998-2000 over and old-growth forest 67 3-3. Typical diurnal changes in below-canopy temperature and water vapor 68 3-4. Diurnal patterns of storage energy fluxes 69 3-5. Daytime Bowen ratios for each year 70 3-6. The relationship between net radiation and estimated energy flux 71 3-7. Emprical relations to model both aerodynamic and bulk conductance 74 3-8. The diurnal relationship of aerodynamic and bulk conductance 75 3-9. The relationship between aerodynamic conductance and latent energy flux 76 vii

PAGE 8

3-10. Diurnal changes in a decoupling coefiBcient, Q, 3-1 1 . Relationship between empirical and modeled estimates of latent energy viii

PAGE 9

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ECOSYSTEM-LEVEL RESPONSES OF CARBON AND ENERGY FROM A TROPICAL WET FOREST IN COSTA RICA By Henry William Loescher August 2002 Chair: Dr. Henry L. Gholz Department: School of Forest Resources and Conservation Whether tropical forests are sources, sinks, or neutral with respect to their carbon balance with the atmosphere remains unclear. To address this issue, estimates of net ecosystem exchange of carbon and energy (NEE) were made for 3 years (1998-2000) using the eddy-covariance technique in a tropical wet forest in Costa Rica. Mean daytime NEE was ca. -18 |imol CO2 m'^ s'' (uptake) and mean nighttime NEE 4.6 |amol CO2 m"^ s"' (eflQux). However, because -80% of the nighttime data in this forest were collected during laminar flow conditions (< 0.2 m s"'), nighttime NEE was likely underestimated. Using an alternative analysis, mean nighttime NEE increased to 6.9 nmol CO2 m'^ s"' . Incident radiation accounted for ~5 1 % of the variation in the daytime fluxes, with temperature and vapor pressure deficit together accoimting for another -20%. This forest was a slight negative carbon sink in 1998 (-0.08 to -1.42 t C ha'' y"'), a moderate sink in 1999 (-1.65 to -3.21 t C ha'' y'), and a strong sink in 2000 (-6.1 to ix

PAGE 10

8.1 t C ha'' y"'). This trend is interpreted as relating to the dissipation of warm-phase El Nino effects over the course of this study. The efifects of net radiation (R,,), vapor pressure deficit (VPD), and surface conductances on energy balance and evapotranspiration (ET) were also determined for this forest. Sensible (H) and latent heat (XE) fluxes were estimated as the sum of abovecanopy eddy-covariance fluxes and changes in below-canopy heat profiles. Albedo was -12 % of incident radiation and did not differ seasonally. Rn was significantly different among years, explaining -79% of the variation in each of the H and XE fluxes. The effects of VPD did not explain any additional variation in heat fluxes. X,E was always greater than H (when Rn exceeded 40 W m'^). The dimensionless decoupling coefficient, Q was always > 0.5 and peaked at 0.7, suggesting that ET for this the forest was generally decoupled from physiological controls. There was better precision in estimating X,E flux using the Priestly-Taylor model rather than using the more physiologically-based Penman-Monteith model. Annual ET was 54-66 % of bulk precipitation and utilized 88-97 % of available energy (R„). x

PAGE 11

CHAPTER 1 INTRODUCTION Inverse model calculations based on atmospheric CO2 concentrations and '^C02/^^C02 and O2/N2 ratios indicate that the terrestrial biosphere is currently a net carbon sink, partially offsetting the additions of CO2 from fossil fuel combustion and deforestation (Schimel et al. 2001). Temporal variation in carbon uptake and emissions by terrestrial ecosystems has an effect on interannual variations in atmospheric CO2 concentrations (Schimel et al. 2001), although the magnitude remains uncertain (Houghton et al. 1998, Houghton 1996, IPCC 1995). Recent attempts to measure forestlevel carbon exchange with the atmosphere have focused on temperate, boreal, and arctic ecosystems (Aubinet et al. 2001, Black et al. 2000, Law et al. 2000, Valentini et al. 2000, Clark et al. 1999, Frolking et al 1996, Goulden et al 1996, Vourlitis and Oechel 1997), with only few eddy flux studies from tropical forests (Fan et al. 1990, Grace et al. 1996, Mahli et al. 1998). However, tropical forests account for -35% of global net primary productivity, >50% of the carbon in aboveground terrestrial biomass and -20% of the soil carbon (Melillo et al. 1993, Dixon et al. 1994). Old-growth rainforest were historically thought to be carbon neutral (input = output). This was challenged by Grace et al. (1995a) and Fan et al. (1990) based on eddy co variance, who suggested that apparently long-undisturbed Amazon forests sequester carbon. If generally true, the implications for carbon science and policy making are enormous. 1

PAGE 12

The energy balances of tropical forests are complex and many dynamic feedback mechanisms among radiation, cloud formation and precipitation have been identified (Wielicki et al. 2002, Hartman et al. 2001, Sohn and Smith 1992). This complexity extends to the potential role of the tropical energy balance in affecting tropical and global climates and general and anomalous circulations (Kelly and Randall 2001, Timmerman et al. 1999, Chen and Van den Dool 1999, Fasullo and Webster 1999, Larson et al. 1999). Much of our understanding of these dynamics has relied on model results, which have shown large spatial and temporal variability in both sensible and latent energy budgets (KeUy and RandaU 2001, Raman et al. 1998, Hulme and Viner 1998, Shuttleworth 1988). In situ studies have either scaled leaflevel measurements to whole canopies (Bigelow 2001, Avisshar 1993, Roberts et al. 1993), or have estimated the energy balance components using eddy covariance over short periods (e.g., 8 d, Shuttleworth et al. 1984). Quantifying the variation of energy balance parameters and their biophysical controls over longer periods should allow for better predictions of runoff and improved models of regional and global climate. Both physical and physiological factors influence forest energy balance, including incident radiation, albedo, rain, interception, canopy capacitance, and aerodynamic (ga) and bulk surface (gb) conductances. Incident radiation in the tropics varies less seasonally than to that at higher latitudes, and values at the surface are more related to cloudiness than changes in solar zenith angle. General circulation models tend to underestimate net radiation in the tropics because of uncertainties in estimating surface albedo and cloud cover (Cramer a/. 1 999, Ruimy e/ a/. 1999). Albedos of forests range fi-om 0. 1 -0.2, with annual and seasonal differences affecting the surface energy balance.

PAGE 13

3 Large differences in annual rainfall have been observed in the tropics that are thought to be influenced by el Niiio-Southem Oscillation (ENSO) and other anomalous circulations. This in turn affects the amounts of water available for evapotranspiration. A general observation is that -50 % of annual rainfall is re-circulated to the atmosphere through transpiration and evaporation of intercepted water, with the other 50% as runoff (Shuttleworth 1989). This imphes that local climate is strongly effected by how energy is partitioned at the surface. Canopy conductances for tropical forests have been estimated using both ecophysiological and micrometeorological approaches. Ecophysiologists estimate bulk surface conductance (gb) by scaling leaf-level or sap-flow measurements to the canopy (Whitehead 1998, Dolman et al. 1991), while micrometeorologists often model gb in relation to meteorological parameters (Wright et al. 1996). Aerodjuamic conductance is generally calculated as a function of horizontal windspeed, zeroplane displacement, and roughness length (Derunead and Bradley 1985). Evapotranspiration from tropical forests is generally thought to be strongly dependent on aerodynamic conductance, because of the high rainfall and the significant proportion of the time when the canopy is wet, reducing the importance of gb in evaportranspiration (Shuttleworth 1989). Research objectives are: 1) Latent heat is greater than sensible heat all the time (net radiation >40 W m'^), 2) below-canopy energy balance contributes significantly to the overall energy balance across temporal scales, 3) with ecosystems that receive greater precipitation, the more total time the canopy is wet. Because this wet tropical forest receives ~ 4000 mm y"' of precipitation, more relative time is needed to dry the canopy, so evapotranspiration rates rely more on aerodynamic conductance than physiological

PAGE 14

4 controls (bulk canopy conductance), and 4) Evapotranspiration from wet tropical forests is strongly dependant on net radiation, and accounts for ~ 50% of the precipitation.

PAGE 15

CHAPTER 2 ECOSYSTEM-LEVEL CARBON EXCHANGE The objective of this chapter was to define the patterns of diurnal and annual net carbon dioxide exchange and their climatic controls, for a lowland tropical wet forest in northeastern Costa Rica. At this forest, large interannual fluctuations in aboveground net primary productivity since 1984 closely paralleled fluctuations in the atmospheric CO2 concentrations and were strongly negatively correlated with average nighttime minimum temperatures (Clark et al. in review). In this paper, we assess the effects of environmental variations on the diurnal, seasonal, and interannual patterns of forest-level carbon exchange for this forest from 1998 through 2000. Materials and Methods Study Site This study was conducted as part of a long term study of tropical forest carbon cycling, the CARBONO project, at the La Selva Biological Station, Puerto Viejo de Sarapiqui, Costa Rica (10°25' 51"N, 84° 00'59"W, elevation 80-150 m.a.s.l.). La Selva is located in northeastern Costa Rica in the Caribbean lowlands at the base of the central volcanic chain and was classified as tropical wet forest in the Holdridge life zone system by Hartshorn and Peralta (1988). This forest has an average 400 trees ha"' > 10 cm diameter ha'' from -100 species (Lieberman et al. 1985), dominated by the mimosoid legume canopy species, Pentaclethra macroloba (34% of the basal area, Clark and Clark 5

PAGE 16

6 2000). Mean tree height is 20-25 m, with emergents exceeding 60 m. Canopy gaps occupy ~ 0.01-0.04 ha ha"' (Denslow and Hartshorn 1994) making the overall canopy very aerodynamically rough. Incident mean (1993-1998) daily solar radiation was 14.9 MJ m"^ d'', with a range from 0.4 to 31.3 MJ m'^ d''. Mean annual temperatwes were 24.6 °C (1982-1998, Organization for Tropical Studies, OTS, unpublished weather records). Mean annual precipitation is 4000 mm (from 1963 to 2000), with a short drier period from December to the end of May, but with no month receiving less than 100 mm (Sanford et al, 1994). Soils range from relatively fertile Inceptisols in riverine areas to low pH, low phosphorus Ultisols in upland areas (Sollins et al. 1994). Moisture-laden northeast trade winds originating over the Caribbean Sea dominate surface winds (Hassenrath 1991). During most (85%) daytime hours, the annual mean surface wind direction is 90°. The wetter season (June through November) and drier season (mid December through May) are controlled by the movement of the equatorial low-pressure trough, i.e., the eastern Pacific intertropical convergence zone (ITCZ). During the drier season, the sub-tropical Hadley cell dominates general circulations, while the tropical cell dominates wet season circulations (Sanford et al. 1994). There are no data showing the exact passing of the ITCZ in Costa Rica. Hence, seasons are defined here as wet season beginning on May 1 (DOY 121), and dry season on December 20 (DOY 353). These dates are 10 d after the average date that the ITZC passes through Barro Colorado Island, Panama (pers. comm. Steve Paton). Other circulations may influence wet season climate including temporales, polar air masses that move down the North American continent generating depressions and prolonged rain events, chiefly occurring in November and December (Schlutz et al. 1998). Veranillos,

PAGE 17

7 temporary and often irregular movement of the South Pacific anticyclone northward, create short dry periods, typicaUy lasting 7 to 10 days in September or October. Sanford et al. (1994) and Holdridge et a/. (1971) provide fiirther site information for La Selva, and Waylen et al. (1996a) and Hassenrath (1991) provide more details on its climatology. Because La Selva is located at 10° N latitude, there is little diurnal change in sunlight over the course of the year, with only a 40-min difference in day length betw^een solstices. For this study, sunrise and sunset were defined as 0600 and 1 800 h, delineating daytime and nighttime periods. A 42 m tower (Upright Inc. Selma, CA) was used to access the canopy environment and to support meteorological instrumentation. The site was a relatively flat ridgetop in an area of generally rolling topography, with -20-30 m relief between stream bottoms and ridgetops (OTS unpublished digital elevation model). After accounting for stability effects, a source area model (Schuepp et al. 1 990) was used to estimate that under stable conditions, 95% of the cumulative flux was derived fi-om within 1.2 km of the tower (at a mean horizontal windspeed of 3 m s''). The tower was sited to minimize edge effects, below-canopy advection either to or fi"om the site, and any major directional differences due to forest composition and structure. Meteorological Data Microclimate data were collected continuously at the tower top. Measured variables included incident radiation (LI190, Ll-Cor Inc., Lincoln, NE), photosynthetic photon flux density (PPFD, LI-200X, Ll-Cor Inc.), aspirated air temperature (Ta, 100 Q platinum RTD, Omega Engineering, Stamford, CT), and bulk rainfall (TE525 metric.

PAGE 18

8 Texas Electronics, Dallas, TX). Atmospheric pressure (PB105, Vaisala, Helsinki, Finland) was monitored at ground level. All of the above data were collected at an interval of 5 sec and compiled as 30min averages with dataloggers (CRIOX and 2 IX, Campbell Scientific Inc., Logan, UT). Instruments were cleaned, leveled as necessary, and recalibrated according to manufacturers' instructions. At times when the PPFD sensor was inoperational, PPFD • • • • 2 was estimated by linear regression equation relating PPFD to incident radiation (R > 0.99). Likewise, when either power outages occurred or aspirated air temperature were not logged, air temperature was estimated fi-om a regression against an air temperature sensor CS500 (CampbeU Scientific) also mounted at 42 m in a radiation shield (R^ > Long term meteorological data fi-om La Selva were used to examine decadal scale trends in microclimate (OTS, http://www.ots.duke.edu). PPFD and air temperature have been measured since 1982 and bulk precipitation since 1961. To assess zero-plane displacement (d), or the mean level of momentum absorption, four 3-cup anemometers (Model 03103-L, R.M. Young, Traverse City, MI) were mounted vertically along the tower at 35.5, 3 1 .6, 28.2, and 25 m above the ground, d was then estimated by determining the intercept (yo) of Eq. 2-1: where Z is measurement height above the ground (m), u is the 30-min time average of the 0.98). Eq. 2-1 instantaneous measurement of horizontal wind velocity (cm s" ) at each height. Roughness length (zo) was estimated using d:

PAGE 19

( M * V' Eq. 2-2 exp^= — where Zm is the measurement height (42 m), u* is friction velocity (m s" ), Um is the 30min time average of the instantaneous measurement of horizontal wind velocity at measurement height (m s''), and k is the von Karmen constant (0.41, dimensionless). The ratio of convective to mechanical production of turbulent kinetic energy (MoninObukov length, L) was used to determine atmospheric stability as in Eq. 2-3: ^ ^ -pCpT.u' Eq. 2-3 gkH where p is the density of air (kg m"^), Cp is the specific heat capacity of air (J kg"' K"'), Ta is in Kelvin (K), g is acceleration due to gravity (m s'^), and H is the sensible heat flux density (J m"^ s'') (Pahlow and Parlange 2000, Monteith and Unsworth 1990, Rosenberg etal. 1983). Net Ecosystem Exchange (NEE) Measurements A closed-path eddy co variance system was used to estimate the portion of NEE of CO2 contributed by turbulent exchange. Because the below-canopy environment was not always subject to turbulent transfer (i.e., well-mixed conditions), a profile system was used to estimate the rate of change of [CO2] below the canopy. NEE was then estimated as ^27 NEE = w'C02'+ — ^^27 Eq. 2-4 io.5 where w' and CO2' are the deviations of instantaneous values from a running mean of vertical windspeed (m s"') and molar fraction of CO2 (^mol CO2 mol''), respectively, and d[C02].

PAGE 20

10 Zx is measurement height (m). The lirst term is the 30-min time-averaged eddy covariance flux. The second term is the storage flux below 42 m. The convention used is that negative values of NEE correspond to uptake of CO2 by the forest from the atmosphere. A 3-D sonic anemometer (K-probe, Applied Technologies Inc., Boulder CO) was used to measure wind velocities in each polar coordinate (w, v, u) and sonic temperature (0). The gas sampling inlet was mounted on the sonic anemometer and co-located with the top transducer in the w-axis. Infrared gas analyzers (IRGA, model LI-6262, Ll-Cor Inc.) were used to measure concentrations of CO2 and H2O vapor, controlled for pressure and temperature at ground level inside a climate-controlled structure. Flow rates were maintained by pumps (KNF Neuberger, Trenton, N J) and mass flow controllers (Model series 200, 0-10 1pm, Teledyne Hastings Inc., Los Angeles CA). Sampled air flowed through -60 m of tubing (4.8 mm ID Teflon tubing) at a rate of 8 1pm resulting in a lagtime of -14.2 s. The NOAA flux software program (McMillen 1988), with a 400 second digital recursive running mean and a fixed lagtime, was used to collect raw eddy covariance data files. IRGA voltage outputs were digitized by a 12-bit analog-to-digital board. Covariances, wind and scalar statistics, and coordinate rotations were calculated in real time at 1 0 hz. Protocols for accuracy, precision, quality control and assurance were used as defined by the AmeriPlux Science Plan (http://cdiac.esd.oml.gov/programs/ameriflux/scifhtm). The response of all instruments must be as fast as, or faster than, the turbulence that is carrying the bulk of mass and energy. This process occurs between frequencies of

PAGE 21

11 1-10 hz, within the inertia! sub-range (Kaimal and Finnigan 1994). While the sonic anemometer operates at these speeds, the response time for the IRGAs is slower, at~8-9 hz. To account for this frequency loss, fast Fourier transfer (FFT) analyses were applied, where the portion of attenuation shown by the co-spectral density in the inertial sub-range was compared to the total spectra. This ratio for w'0' was considered ideal, since temperature (0) was measured by the sonic anemometer at 1 0 hz, and then compared to the spectra of w'C02' to estimate a spectral correction factor, SCf: where Swx' is spectral density of w' and CO2' or 9', n is the natural frequency, and w'x' is the mean covariance of w' and CO2' or 9' (Meyers and Baldocchi 1991, Baldocchi and Meyers 1 989). Although we assumed that the dissipation of turbulent kinetic energy occurred in the inertial sub-range for all the scalars, we did not expect that the Kaimal spectral relationship (a slope of 2/3 within the inertial sub-range) would hold for every 30-min period because of roughness, differing stabilities, and possible density driven flows over time. Profile measurements were used to calculate below-canopy CO2 storage dynamics (Eq. 2-4). CO2 was coUected from 6 inlets at 0.5, 7.3, 1 1.95, 16.55, 21.2, and 27.6 m. A datalogger (model 2 IX, Campbell Scientific Inc.) was used to operate solenoids that switched the flow (~3 1pm) from each inlet through the IRGA (Li-Cor 6262) for 5 min during each 30-min period and to record the raw data. Platinum resistance thermometers (100 Q PRT, m68. Omega Engineering, Stamford, CT) housed in radiation shields were Eq. 2-5

PAGE 22

12 co-located with each inlet. When sampling occurred, the airflow acted to aspirate the PRTs. Temperature and humidity profiles were used to account for changes in mass flow due to changes in density (Webb et al. 1980). Below-canopy storage was calculated fi-om line-averaged profile measurements using Eq. 2-4; it was assumed that this profile was similar across the flux source area. Both eddy co variance and profile measurements began in April 1 998 and continued through the end of December 2000. Gaps in the dataset occurred for periods of 2-14 days when either power failure or instrumentation mallunctions occurred. IRGAs were calibrated every 2-3 days. Improved precision in calibration was achieved starting in February 1999 by plumbing nitrogen through the IRGA reference cell as a zero reference. A model was used to relate daytime NEE to PPFD (Ruimy et al. 1995): (^>«^max Eq. 2-6 NEE^^^ =Re + where NEEday was calculated using Eq. 2-4, Pmax is maximum ecosystem CO2 uptake rate (fxmol CO2 m'^ s''), Re is ecosystem respiration (jamol CO2 m"^ s"'), 0 is PPFD (jimol m"^ s"'), and a is apparent quantum efficiency (5CO2/50). To describe the effects of temperature on nighttime NEE, a second model was used: where Ro is the base ecosystem respiration rate (|amol CO2 m'^ s"') when air temperature is 0 °C, T is temperature (°C), and b is an empirical coefiBcient. A general linear model (SAS v. 8.01, Gary, NC) was used to test first and second order effects of other variables on NEE, including PPFD, temperature, VPD, season and

PAGE 23

13 year. Sigmaplot v. 5.0 (SPSS Inc., Richmond, CA) was used to describe these relationships. Data Screening Eddy covariance data were screened for validity and removed when either i) the standard deviation of w', CO2', or 9' was > 1 .74 times the mean, ii) rain occurred, iii) 30min data collection periods were incomplete, or, iv) signals from either the sonic anemometer or the IRGA were out-of-range. Profile data were removed when either i) data were beyond 3 SD from the mean, ii) 30-min data collection periods were incomplete, or iii) signals from the IRGA were out-of-range. Leaf Area Estimation Although the La Selva forest is lai gely evergreen, seasonal differences in leaf area index (LAI) were expected because 8% of the tree species are deciduous in the dry season, and 28 % of tree species produce annual leaf flushes, many at the onset of the wet season (Frankie et al. 1974). Furthermore, many tropical rain forest tree species are facultatively deciduous, losing up to half of their leaves during prolonged dry periods (Richards 1996). Photographic estimates of eLAI (estimated LAI) were derived using the WINPHOT program (Ter Steege 1996) each year during the wet and dry seasons across 18 randomly stratified 0.5 ha plots (description of statistical design for plot layout in Clark and Clark 2000). Within each plot, 6 photographs were made at each sampling date under diSuse light conditions at the same randomly chosen points. Because these estimates were derived optically with no means of direct calibration, they should be viewed relatively.

PAGE 24

14 Results Characterizing Canopy-Level Turbulence There were -2/3 slopes for the normalized spectra of wind velocities in the inertial sub-range during periods of both stable and unstable atmospheric conditions, confirming a transfer of energy to the canopy with shear forces dominating (Figure 2-1) The spectral density decreased during stable conditions (Figure 2IB), as did the eddy covariance flux estimates, but the general relationships held. Buoyancy forces produced measurable vertical wind movement at night, as indicated by the positive 1:1 slopes at wavenumbers < 0. 1 . The observed shift in the spectral peak between stable and imstable conditions was similar to that reported in other studies (Kaimal and Finnigan 1994, Anderson et al. 1 986). The spectral correction factor ranged from 1 . 1 8 to 1 .08 and varied with u* (Figure 2-2). Zero-plane displacement (d) and zo for momentum also varied with stability (Table 2-1). During unstable periods (L< -50 m), zo increased to -2.4 m, a long roughness length even for a forest (Hansen 1993), with a mean level of drag (d) of -22 m. Aerodynamic roughness lengths sharply decreased from slightly unstable conditions (-50 < L < -10) to neutral conditions (-10 < L < 10), indicating the quick formation of stratified laminar flow and the decoupling of the below-canopy environment. The relationship between NEEday and u* was linear (Figure 2-3A). There was decrease in the NEEnight with u*, although no threshold was observed. When NEEnight data were averaged across u* intervals of 0.025 m s"', the relationship was strongly linear below a u* of 0.45 m s'' (Figure 2-3B). No relationship, however, was foimd between

PAGE 25

15 the residuals from the energy budget and u* under any stability conditions, so that no u* threshold could be determined and used to filter data, as has been done in most other studies (e.g., Clark et al. 1999). We assumed that the most accurate estimate of nighttime turbulent exchange occurred at larger values of u*(i.e., > 0.4 m s"') and that conversely, storage occurred at very low u* (i.e., < 0.05 ms''), in spite of the lack of obvious thresholds. The consequences of these assumptions are discussed below. Diurnal Patterns in NEE The diurnal pattern of the CO2 storage flux was very consistent throughout the year (Figure 2-4 A). The greatest fluxes were observed in early morning hours, when below-canopy CO2 that was respired during the night and stored in the air column below the sonic anemometer was vented or re-fixed through photosynthesis. The magnitude of these morning ejections may be underestimated, because venting may skew the distribution of wind statistics in the 30-min dataset, and so valid data could have been inadvertently removed during the screening process. The maximum average storage flux was -5.6 j^mol CO2 m'^ s'', which occurred at -0800 when the convective boundary layer was developing, as shown by the increasing vertical windspeed in Figure 2-4B. Storage fluxes decreased imtil -1400, after which only net effluxes from below the canopy were observed. This also coincided with peaks in below-canopy temperature (Figure 2-4B) and vertical windspeed. Storage eflflux increased to a peak just after sunset. The maximum nighttime (before 0600) storage efflux was 2.97 fimol CO2 m"^ s'', with an average of 1.6 ±0.13 i^mol CO2 m'^ s"'. Storage generally decreased throughout the night, along with temperature and vertical

PAGE 26

16 windspeed. Vertical windspeed and below-canopy temperatures diverged between 1630 and 0130, which indicates horizontal advection of below-canopy CO2 ofifthe site may have occurred and that a portion of the flux may have been missed (Mahrt et al 2000). The maximum NEEday based on the 3-yr mean for each half-hour was -17.3 ±0.3 |imol CO2 m'^ s'', occurred at ~ 1 130, and closely followed the inverse pattern of PPFD (Figure 2-5). Nighttime eddy covariance flux was positive and fairly constant throughout the night. At dawn, this flux sharply decreased (uptake into the forest) for -30 minutes with decreases in Monin-Obukov length (Eq. 2-3). Environmental Controls on NEE NEEdayNEEday was negatively correlated with PPFD (Figure 2-6A) and had an estimated mean maximum of -1 8 ± 9 ^imol CO2 m"^ s '. A linear model that included second-order effects of year, PPFD, and VPD explained ~72% of the total variation in NEEday. PPFD alone accounted for -51% of the variation (Figure 2-6), with no significant effect of season. The light response function of Eq. 2-6 had a of 0.51 . Residuals from this function were weakly related to temperature and VPD (Figvire 2-6B and 2-6C). Since VPD includes temperature as a component, VPD and temperature are correlated, and since VPD directly affects stomatal conductance (Law et al. 2000, Landsberg and Gower 1997), we examined the influence of VPD on NEEday further by separating NEEday into three VPD classes (0-0.5, 0.51-1.00, and > 1.0 kPa) and refitting Eq. 2-6 for each year. The results (Figure 2-7, Table 2-2) show that the response function is more linear within a VPD class. Pmax and Re in 1998 were significantly lower than

PAGE 27

17 those found in 1999, while a in 2000 was greater than in the other two years (Table 2-2). 2 -I Light compensation points ranged from 1 10 to 207 |xmol CO2 m' s' . NEEnighi. The mean NEE„igh, was 4.82 ±0.6 ^imol CO2 m"^ s"' (mean +1 SD). This may be an underestimate because >80% of the nighttime turbulent exchange measurements were made with u* < 0.4 m s"'. Using only data with u* > 0.4 m s"\ the estimate becomes considerably larger, 6.98 ^mol CO2 m'^ s'\ with 4.83 ±0.21 and 2.15 ±0.1 1 |j,mol CO2 m'^ s"' (mean ±1 SE) the contributions from eddy covariance and storage fluxes, respectively. In addition, the averaged daytime Re values (from Eq. 2-6, Table 22) ranged from 5.07-6.42 |amol CO2 m"^ s"', supporting use of the higher NEEnight values derived here. NEEnight was weakly related to temperature over the entire sampling period (Figure 2-8), with a Qio of 1 .79 (p < 0.053). Estimating Annual NEE 1 estimated annual NEE in two ways. In the iirst case, half-hourly meteorological data were used to drive Eq. 2-6 for each year and VPD class to derive an annual NEEday. Then a fibced NEEnight of 6.98 fxmol CO2 m"^ s"' was subtracted. I refer to these results as NEEmodeiedIn the second case, measured NEEday and NEEnight (with u* > 0.4 m s'^) were used, and only the gaps were filled using the procedure above; results are referred to as NEEgap. There were marked differences in the cumulative NEE among years and method of estimation (Figure 2-9). Annual NEE was estimated to be higher in 1999 and 2000 than in 1998, with 2000 the strongest sink year.

PAGE 28

18 Discussion Characteristics of the La Selva Canopy I assume that the eddy covariance measurements from La Selva were robust and represent the nature of CO2 exchange between the canopy and the atmosphere on the basis of two requirements generally prescribed for this method: (1) a consistent energy cascade in the inertial sub-range in the power spectra under both stable and unstable conditions (Kaimal and Finnigan 1994), and (2) the measurement height was at least 1.8 times d (it was in fact at least 8 times zo above d in unstable conditions and 6 zo above d under all other stability classes, Schmid et al. 2000, Monteith and Unsworth 1990). I did not expect to see a well-developed energy cascade during nighttime conditions. It is likely that this was due to roughness induced turbulence in the nighttime flows. It is unclear why no u* threshold in the eddy covariance data was observed. Turbulence structures above a fixed plane (e.g., above a vegetated canopy) from other studies have generally been described using Monin-Obukov theory (Leclerc et al. submitted) and measured over uniform canopies with short aerodynamic roughness. As roughness lengths become greater (> 1 m), the effect on turbulence in the well-mixed layer and applicability of Monin-Obukov theory become questionable (Nakamura and Mahrt 2001, Ayotte et al. 1999, Raupach and Finnigan 1997). Under these conditions, u* becomes homogenized over a broader range of turbulence lengths and potentially has less explanatory power (Nakamura and Mahrt 2001). At La Selva, there were insuflHcient data to assess if the linear relationship between u* and NEEnight existed at values >0.45 m s"', or if the relation ultimately reached some asymptote. Data that would have

PAGE 29

19 contributed toward developing a u* filter may also have been screened out by other criteria. Environmental Controls on NEE The average-diurnal pattern in NEE (Figure 2-5) is strongly symmetrical around 1 130. On average, the storage term contributed 33 % to NEEmght. The power spectra for nighttime eddy covariance indicated a transfer of mass and energy, but the flux was relatively small and the source distance quite long (1.5-2.0 km). The nighttime envirorunent below the canopy is quite different from that at the tower top and subject to diabatic flows controlled by changes in air density and slope (Mahrt 1 992), making the source area for the storage flux more localized. The greatest variation in nighttime CO2 profiles occurred during periods with the most rapid changes in Ta. The diurnal patterns in CO2 profiles and respective storage fluxes, however, were relatively constant throughout our measurement period. After the morning re-assimilation of stored CO2 dissipated, NEEday was clearly dominated by above-canopy fluxes. A coarse estimate of the amount of recycled CO2 can be obtained as the difference between integrated NEEnight and the morning storage flux. This represents a value of 32% for the fraction of integrated NEEnight that was recycled below 42 m, similar to isotopically-derived estimates of re-synthesized CO2 from the Ducke forest near Manaus, Brazil (39%, Sternberg et al. 1997) and within the range from another neotropical forest, Barro Colorado Island, Panama (31-38%), Sternberg et al. 1989). The lack of a relationship between 30-min averages of NEEnight and temperature was likely due to the combination of different factors. First, there are numerous sources

PAGE 30

20 of C02 for ecosystem respiration, each with their own controlling factors (Davidson et al. 1998). Soil respiration can be influenced by soil type, water status and temperature (Schwendenmann et al submitted), and foliar respiration may be influenced by nitrogen and/or photosynthate availability in the canopy at the time of leaf expansion. Second, there was only a small range annual nighttime temperature (< 9 °C). Third, the abovecanopy source area is integrated over a larger area than that from the below-canopy environment (Raupach et al. 1992), subjecting the storage flux to localized biotic effects or the possibility of below-canopy advection. Foiuth, as mentioned above, 80% of flux estimates were made under conditions with low u* (< 0.4 m s''), questionable conditions in this case for eddy covariance. Finally, carbon may have been exported from the system as CO2, due to large morning ventilation events could not be quantified with confidence (Wilson et al. 1998, Mahrt 1992). The relationship between NEEnight and temperature (Figure 2-8) based on longer averaging intervals is subject to the same potential sources of error as mentioned above and may also lead to underestimates NEEnightSchwendenmann et al. (submitted) reported soil respiration rates at La Selva from upland soils ranged from 3.3 to 4.3 |amol CO2 m' s" . Given that soil respiration accounts for 50% of Re for a wide range of forests (Amthor 1994, Ryan 1991), this further suggests that our higher value (6.98 nmol CO2 m' s' ) for NEEnight is more likely correct. Moreover, when I recalculated annual NEEnight using the nighttime respiration flinction in Figure 2-8, it also resulted in a very low value (4.3 ^mol CO2 m"^ s''), similar to that found using the NEEnight data across all u* values, and again a value completely out of step with measured soil respiration and our other estimation of NEEnight-

PAGE 31

21 The potential exists that below-canopy nighttime advective flow contributes error in NEEnightHowever, for this to occur there must exist specific conditions, such as, a strong upslope temperature gradient, a breakdown of below-canopy resistance to flow, and/or a net vertical movement of wind into the forest. At La Selva however, there is only a small change in temperature over a large upslope area, there is considerable resistance presented by large below-canopy leaf area and tree stems, and nighttime above-canopy vertical and below-canopy horizontal windspeeds (data not shown) were not much different from the expected accuracy of the sonic anemometer (i.e., 0.05 m s"'). This is an unresolved issue and these factors are concerns for many tower flux sites (Massman and Lee 2001). In the face of increasing global temperatures (National Academy of Science 2000), there is increased focus on the role of temperature in controlling the carbon dynamics in the tropics. Kindermann et al. (1996) modeled the effects of increased temperature on carbon stores and vsdth even small increases in temperature (~0.5 °C), large effluxes of carbon to the atmosphere are expected. It is hypothesized that most of this carbon will be from the tropics (Trumbore et al. 1996). At La Selva, large year-toyear fluctuations over the past 16 years in aboveground biomass increments have been negatively correlated with both the mean nighttime temperature and variations in annual fluctuations in atmospheric CO2 concentrations (Clark et al. in review). In this study, I found only a small temperature influence on NEEday and NEEnight and only when all three years of data were pooled and hourly averages were used. I cannot dismiss the NEEday response to temperatm-e as entirely due to VPD, and may in fact be partially due to photorespiration. Interestingly, Grace et al. (1996) and Malhi et al. (1998) did not report

PAGE 32

22 a nighttime temperature response for Amazonian forests using eddy covariance data. It may well be that effects of temperature on ecosystem respiration from tropical forests may only become apparent after many years of observation. There were large interannual differences in apparent forest-level quantum efficiency (a) estimated from the NEEday data. This may indicate large adjustments in forest structure and physiology in response to the climatic variation among our study years. Waring et al. (1995) concluded that seasonal differences in both LAI and canopylevel quantum efficiency largely controlled productivity from a deciduous forest in northeastern US. In our study, the estimated a for 2000 was significantly higher than those from the preceding two years and approached the theoretical maximum for C3 leaves (Lloyd et al. 1995, Farquhar et al. 1980). However eLAI did not follow this pattern and increased only 1999 and into the dry season of 2000. That neither changes in LAI nor a were related to annual NEE suggests that there are interactions between NEE, canopy leaf dynamics and climate for this complex tropical wet forest. The relationship between NEEday and VPD could be a result of either physiological or physical effects. A physiological effect could be stomatal closure in response to a hydrologic limitation, either high VPDs or indirectly, decreases in soil moisture availability. The location of the hydraulic limitation in the La Selva forest is not known. A physical effect could be through though modification of canopy architecture through premature leaf drop, leaf folding or changes in leaf orientation. Whole forest canopies do not frilly saturate even at frill insolation (Ruimy et al. 1995, Wang and Polglase 1995). Changes in leaf angle or leaf closure in the upper canopy allows penetration of light to deeper canopy layers, allowing for increased carbon gain in the

PAGE 33

23 lower canopy. This offsets the effect of leaf closure or changes in orientation at leaf scale in terms of light response at the ecosystem level. Only 8% amount of the time, however, were VPD values > 1 kPa and during 97% of the daytime (when net radiation was > 40 w m'^), latent heat fluxes were greater than sensible heat fluxes (P <1, unpublished data). This strongly suggests that the La Selva canopy had access to abundant soil water. The only exception was in the 1998 dry season, when 30% of daytime VPDs were > 1 kPa, precipitation was the lowest ever recorded (68, 38, 126 mm monthly total rainfall in January, February and March, respectively), and daily mean Ta was ~1 °C above the longterm average. The 1998 dry season was at the end of the 1997-1998 warm-phase El Nino Southern Oscillation (ENSO) and was warmer and drier. During December 1 998, a coldphase (la Nina) ENSO brought greater precipitation, cooler temperatures, and lower mean daily insolation (and PPFD), with several days receiving < 5 MJ d"'. Overall, 1998 was warmer and drier during the dry season, but had more precipitation, cooler temperatures and reduced light during the latter part of the year (compared to the other two years, 2-4). The greater eLAI in 1998 coupled with lower a, a greater portion of time with VPDs > 1 kPa in the dry season, and overall lower mean daily insolation (13.3 MJ d"', 2-4), likely contributed to the La Selva forest being close to carbon neutral in 1998. During 1999, the daily insolation was well above the long term trend, but in November and December the insolation was well below the long-term average due to a prolonged temporal, suggesting that the effects of increased annual incident radiation outweighed those of reduced eLAI and a prolonged temporal, making this forest a moderate sink of carbon.

PAGE 34

24 Despite these seasonal variations in climate, I did not find any seasonal effects on NEE for any year. Even though seasonal displacements of the ITCZ alter Hadley cell circulations, changes in individual environmental factors do not necessarily occur in concert as a result. The initial passing of the ITCZ can be intermittent and there can be multiple 'false starts' (Hastenrath 1991). Moreover, the northern most progression of the ITCZ is just north of Costa Rica and with erratic movement, prolonged periods of dry weather can occur during the otherwise wet season (Sanford et al. 1994). Other regional anomalies can occur, as during 1998 and 1999, when heavy rains persisted fi^om December into January, even after the ITCZ passed. In 1998, this was brought about by a cold-phase ENSO event, and in 1999 by a prolonged temporal. The climatic trends observed during the 97/98 ENSO were tjpical for this region (Waylen et al. 1 996b, Cavazos and Hastenrath 1 990), as were the conditions observed during temporales (Sanford et al. 1994). The 97/98 ENSO, however, brought the highest temperatures in the 1 9-year La Selva record. The apparent increasing frequency and magnitude of ENSO events (Timmermann et al. 1999, Corti et al. 1999), may have implications for interpreting the effects of climate change on NEE, as at La Selva, 1 998 had the lowest estimated NEE. This supports the findings of Tian et al. (1998) who concluded in a modelling study that variations in NEE of tropical forests are controlled, in part, by macro-level changes in climate, which in turn are driven by the timing, frequency and magnitude of ENSO events. A possible alternative explanation for the large interannual variation in NEE at La Selva is that a substantial fraction of forest is in an early successional stage. There is a high frequency of treefall gaps in this forest, even though La Selva is below most

PAGE 35

25 hurricane pathlines. If I assume that the forest is aggrading carbon during its stand halflife, which was estimated as 77.3 y (Lieberman et al. 1990), and has gap formation rate of 0.96% area y"' (Denslow and Hartshorn 1994), then 74% of the land area is under recovery at any time. Moreover, the mortality rate for 1969-1982 was 2.03% (Lieberman et al 1990), but increased to 4.77% in 1997-1998 (Clark et al. unpublished data) in upland plots associated with gaps, suggesting that gap formation rate also increased under climatic conditions imposed by both ENSO phases. Comparisons with Other Tropical Sites The pattern and magnitude of NEE at La Selva were similar to those estimated by eddy covariance at three tropical moist forest sites in the Brazilian Amazon (Reserva Ducke, north central Amazon, Fan et al. 1990; Jani, south central Amazon, Grace et al. 1995b; Cuieiras, north central Amazon, Mahli et al. 1998). Mean maximum daytime NEE estimated for these three forests ca, -18-to-20 fxmol CO2 m'^ s"' and mean nighttime NEE -5-7 ^imol CO2 m'^ s'' (Table 2-5). This suggests that these ecosystems may have similar controlling factors on NEE, even though there are marked differences in stand characteristics. Mahli et al. (1998) hypothesized that cloudiness on insolation were the strong determinate of NEEday, which may explain the low NEE in 1998 at La Selva. One difference is that VPD was found to play a stronger role in regulating carbon gain at the other sites than at La Selva (e.g., Mahli et al. 1 998). These other three sites receive < 2500 mm y"' of annual precipitation and have lower soil water availability (Hodnett et al. 1996, Tomasella and Hodnett 1996), so that there are likely to be greater hydrologic

PAGE 36

26 constraints on daytime NEE (Mahli et al. 1 998, Williams et al. 1 998). The range in annual quantum yield at La Selva overlaps with the values for the other three sites. Reduction in leaf carbon gain can occur when leaves are wet, lowering rates of CO2 diffusion into the leaf by a factor of 10"* (Jones 1992), thus reducing photosynthesis (Ishabashi and Terashima 1995). Smith and McClean (1989) also found that photosynthesis was significantly reduced in wet leaves that had a wettable cuticle, but increased on leaves that had non-wettable cuticles. The ratio of species with wettable :nonwettable leaves is at La Selva not known. However, because eddy covariance data collected during rain events at La Selva were eliminated, NEEday may have been over-estimated as a result. La Selva experiences rain for 18% of the time annually, as compared to only 13% and 8% at Jani and Ducke, respectively (data fi-om 1994-96, httpW.www.abracos.com). For this reason, the likelihood of a sampling bias are potentially greater at La Selva due to a larger fi-action of data removed in screening. The range of our estimates of annual gross ecosystem production (GEP, Table 26) overlaps the GPP estimate at Cuieiras. High ratios of NEE:GEP are thought to exist only with forests rapidly accumulating carbon (e.g., 0.02 for an old-growth Pseudotsuga menziesii forest, compared to 0.29 for rapidly a growing pine stand. Waring and Schlesinger 1 985). The NEE:GEP ratios were 0. 1 8 and 0. 1 94 for La Selva and Cuieiras, respectively, during 2000 and 1 995 indicating that these systems can be very productive compared to other forests. All of the tropical sites cited here are considered "oldgrowth", and assumed in the past to be near steady state with respect to carbon. However, eddy flux data fi-om none of them support this view and suggest instead that tropical forests may be net sinks from 0 to much as -6 1 C ha'* y"' (averaging ca. -2 t C

PAGE 37

27 ha'' y''). However, the systematic errors discussed earlier collectively would tend to decrease the amount of accumulated NEE into the forest, since the carbon would be measured when it enters the system at the top but not when it leaves the forest (e.g., downslope drainage of CO2, loss of C as unmeasured volatDe organic compounds, etc.). Where within the La Selva ecosystem this carbon is apparently accumulating remains an open question.

PAGE 38

28 Table 2-1 . Estimates of aerodynamic parameters, zero-plane displacement (d) from eq. 2-1 , aerodynamic roughness length (zo) from eq. 2-2, and u* estimates according to stability class (L), eq. 2-3, from La Selva Biological Station, zo, d and u* estimates are median values ±95% CI, L values are means ±1 SE, and n is number of 30-min periods. Stability class zo(ni) d(m) u* (m s"') L(m) n Unstable 2.41 ±1.04 21.5 ±1.80 0.37 ±0.46 -716 ±428 57 Slightly unstable 3.62 ±0.95 19.4 ±1.91 0.34 ±0.13 -22 ±1.5 51 Neutral 0.45 ±0.01 22.1 ±6.73 0.11 ±0.05 1 ±0.6 84 Stable 0.44 ±0.21 23.0 ±1.00 0.22 ±0.05 814 ±553 186

PAGE 39

29 Z -2 U +i CO 73 Si 3 o a o in m o +1 m o o o +1 o o 00 ON ON o o I o ca C +1 o in ro o o +1 in m o o >n »n o o o o On +! rn in ON o o o o o o +1 m o +1 ON o NO o +1 o in in o o o +1 NO m o 0 m 0 0 0 (N 0 in 0 On 00 V ON On V rn t-^ r-i rn in 0 0 0 0 0. ro NO ON r-NO 0 0 +1 +1 +1 -H (N in ON in in in NO 0
PAGE 40

30 o CO (N d o 00 00 +1 o o d +1 o\ o d +1 O +i o u-1 d CO 00 d +1 m en 0^ O o d +1 00 o o +1 o o r~-'

PAGE 41

31 Table 2-3. Annual and seasonal differences in estimated leaf area index (eLAI) m'^ from La Selv a, Costa Rica (S.F. Oberbauer unpublished data). Season Year Mean (median) ± 1 SE Dry 1998 3.85 (3.96) ±0.19 Wet 1998 4.85 (4.79) ±0.11 Dry 1999 2.71 (2.52) ±0.13 Wet 1999 3.76 (3.84) ±0.11 Dry 2000 3.48 (3.51) ±0.07 Wet 2000 3.43 (3.30) ±0.13

PAGE 42

32 03 C o o 1) 0) t3 (N C i CO .9 5 •'^ B ^ .a 3 00 c/3 O c3 O ^ cS i3 > ^ CO '5b S 2 g o o a 00 '-S ON ^ OS 3 u ^ >^ J3 CO CO O .2 Ph 00 -2 w P (Z3 ON S On rr^ On ON u s ^ CO o g o a "3 3 o aII "eo 4> .Si , V •S ^ u 00 S ON 00-^ .23 a> CO U o H t3 o u o a. H e o 9> B o o B u o _ 3 a. S a E s a. E ON d -H 00 O d -H CO 00 On ON 00 d -H 00 o d -H ON ON ON d -H rON CN 0 q d -H irJ (N 00 On ON JS o in — > d -H 00 O -H (N o d -H CO o d -H CN CN CN q d -H NO CN 00 ON ON o d -H CN 00 ON ON I (U u (U Q m o o d -H NO CN CN 00 q d -H CN NO (N ON ON On I > o CN in d -H m 00 CN o o d -H m 00 d CN m ON NO m CN m in 0 0 0 0 0 d d d d d d -H -H -H -H -H -H m CO ON CN r~in in in CN CN CN (N CN CN CN ri o d -H O CN CN ON ON ON I u o Q

PAGE 43

33 CO •3 3 i > o o >^ -o t3 u -a .2 s o c eg S I T3 cn Cm o a o C/3 I o o c« c/3 O < CO Co o (N O o O 00 (N O o c/3 Co O in u. o 0 O in in in o in in o (N in o OS in X! O o (N ON m o 00 o o in o o o CO DO c o o o o c/3 'a _o 'C D•O +-» c« o 00 c u o CO i-l u I — ,cO CO 3 C CO .S2 'o CO 'a. o .22 '6 CO o 'S, o H .S2 "o CO o o _^ CO o '5, o o N
PAGE 44

00 ON ON ON 5 o ON ON 3 1/1 o .s t GO 1/1 ^ U --^ 2 -o .b o -o u l-l e ^ u -5 > CO ca o x: gj If I-4-> • (U ^ s cd o 1=1 I— ' ^ ° X o o C/2 O o s o .1 CO u o (U u ki o

PAGE 45

35 Table 2-6. Between year measures of productivity and ecosystem efficiency from La Selva, Costa Rica and Cuieiras, Brazil. Data from La Selva are from eq. 2-5 across VPD classes and a fixed nighttime NEE, as noted in the text. Units of GEP and NEE are expressed as MT C ha'' y''. site La Selva, Costa Rica Cuieiras, Brazil year 1998 1999 2000 1995 GEP 28.41 30.6 33.9 30.4 NEE -0.09 -1.66 -6.1 -5.9 NEE:GEP 0.004 0.055 0.18 0.194 Note: The Cuieriras dataset is from Mahli et al. 1998, and Malhi and Grace 2000. Data from La Selva is based on NEEmodeied.

PAGE 46

36 0.001 0.01 0.1 1 10 wavenumber (frequency'^) Figure 2-1 . The relationship between the normalized power spectra, for the three wind vectors against the wavenumber. Data are averaged from 73 30-min periods from January 10 to Februray 28th, 1999. A) Mid-day, unstable conditions, beginning at 12001230 p.m, and B) Mid-night, stable conditions, beginning at 2300-2330 p.m. Error bars are +/1 SE.

PAGE 47

37 Figure 2-2. Diurnal time series of the spectra-based correction factor, for CO2 fluxes from La Selva as calculated from Eq 2-5. Data are 90-min running means centered on the 30-min interval. Data are from 1998-2000. Error bars are +/1 SE.

PAGE 48

38 Figure 2-3. Relationships between u* and daytime NEE, (A), and between B) nighttime eddy covariance or storage fluxes, and u* (B). Values of nighttime NEE were averaged across intervals of 0.025 m s"', except for the righthand most eddy covariance point, which was averaged from all data between 0.4 < u* < 0.54 m s'\ Error bars are +/1 SE.

PAGE 49

39 Figure 2-4. Diurnal characteristics of A) storage fluxes, and B) line-averaged temperature and non-rotated vertical windspeed from La Selva. All data are from 1 9982000. A 90-niin running average was used with each estimate centered on the 30-min interval as indicated. Sample size for storage and line-averaged temperature were 5566 and for vertical windspeed 16333. Error bars are +/1 SE.

PAGE 50

40 time of day Figure 2-5. The mean diumal pattern of above-canopy eddy co variance (EC) and storage (below canopy) flux, NEE (the sum of the fluxes) and PPFD from La Selva (1998-2000).

PAGE 51

41 Figure 2-6. Main effects of environmental variables on daytime NEE, where. A) is the light response of NEE for ail data (1998-2000), and B) and C) show the residuals from the light response function in relation to temperature and VPD, respectively.

PAGE 52

42 Figure 2-7. NEE as a function of PPFD across a gradient of vapor pressure deficits for years 1998, 1999 and 2000 fi-om La Selva, Costa Rica.

PAGE 53

43 Figure 2-7. continued.

PAGE 54

44 Figure 2-7. continued.

PAGE 55

45 (A o o o E UJ lU line-averaged temperature (C) Figure 2-8. The relationship between NEEnight and line-averaged (below-canopy) temperature. NEEnight estimates are the aggregated averages from each nighttime 30-min period from the whole study period (Figure 2-4A). Temperature is an average from six measurements through the canopy profile from 27 to 0.5 m.

PAGE 56

46 0 50 100 150 200 250 300 350 Day of Year Figure 2-9. Cumulative NEE from La Selva, Costa Rica for 1998-2000. Estimates were calculated using A) the results from the light response equation across VPD classes and a fixed estimate for nighttime respiration, and B) direct estimates of daytime and nighttime NEE with gaps filled using data from NEEmodeled

PAGE 57

CHAPTER 3 ENERGY BALANCE AND MODELED EVAPOTRANSPIRATION FOR A WET TROPICAL FOREST IN COSTA RICA The objective of this chapter was to examine the annual and seasonal effects of atmospheric environment (albedo, net radiation, vapor pressure deficit) and the surface controls on the energy balance of a wet tropical forest in Costa Rica. Methods Meteorological Data All measurements referred to here were collected fi-om September 1997 to December 2000. Instrumentation for measuring air temperature, relative humidity, bulk precipitation, and net radiation were mounted at the top of a 42 m tower (Upright, Inc., Selma, CA). Prior to March 1 1999, air temperature (Ta) was measured with a CS500 probe (Campbell Scientific, Inc., Logan UT) installed within a radiation shield, and linearly back-corrected to fit the response of the aspirated temperature sensor (R = 0.98). Ta was measured with platinum resistance temperature detector (100 Q platinum RTD, Omega Engineering, Stamford, CT) mounted in an aspirated shield. Relative humidity was also measured with the CS500 probe, and rainfall with a tipping bucket rain gauge (model TE525, metric, Texas Electronics, Dallas TX). Net radiation, R„, was measured 47

PAGE 58

48 from March 1, 1999 to December 2000, with a closed-ceU thermopile-style sensor (NRUte, Kipp and Zonen, Delft, the Netherlands). From August 1, 1997 to March 1, 1999 R„ was measured with a Fritschen-style sensor (model Q-7.1, Radiation Energy Balance Systems, Seattle, WA). All data collected with the Q-7.1 were linearly back-corrected to fit the response of the NR-Ute (R^ = 0.97) and corrected for advected sensible heat. Another net radiometer (model CNR.l, Kipp and Zonen, Delft, the Netherlands) was also used to estimate albedo during February-April, July, and September 2000. Two resistance grid type leaf wetness sensors (model 237-L, CampbeU Scientific, Inc.) were mounted at 26 and 2 m, and histograms of relative wetness were compiled. To mimic the wetting of leaves, the sensors were coated with a layer of flat, off-white latex paint. Soil heat flux plates (model HFT-3, Radiation Energy Balance Systems) were installed at a depth of 5 cm, in each of three 1 m x 1 m plots > 20 m distance apart near the base of the tower. Atmospheric pressure was measured at ~3 m (PB105, Vaisala, Helsinki, Finland). All meteorological data were collected at 5 sec intervals and compiled as 30-min averages with a datalogger (CRIOX, Campbell Scientific Inc., Logan, UT). Instruments were cleaned, leveled as necessary, and recalibrated according to manufacturers' instructions. Energy Balance Estimates An ecosystem-level energy balance can be estimated by R„=AE + H + G Eq. 3-1 assuming horizontal homogeneity, where Rn is net radiation, XE is the latent energy flux, H is the sensible heat flux, and G is the soil heat flux (all units are W m'^). Both H and

PAGE 59

49 XE were estimated by the summation of both eddy co variance (above-canopy) and belowcanopy fluxes, expressed by, where w', q', and 9' are the deviations of instantaneous values from a running mean of vertical windspeed (m s"'), specific humidity (mmol mol''), and virtual temperature (°C), respectively, and Zx is measurement height (m). w'q' and w'0' are the turbulent exchanges of water vapor and heat as estimated by eddy covariance method, and the second term in each equation is the storage flux in the air column. The storage fluxes are noted as X,Estor and Hstor, respectively. The convention used here is that negative values correspond to a flux into the forest from the atmosphere. The eddy covariance system was comprised of a sonic anemometer (K-probe, Applied Technologies, Inc., Boulder, CO),an infra-red gas analyzer (IRGA, model Li6262, Li-Cor, Linclon, NE), -60 m of tubing (4.8 mm ID Teflon) with the inlet colocated with the sonic probe, a laptop computer, and a pump to pull air through the tubing at ~ 8 1pm. The sonic anemometer measured the wind velocities in three dimensions at 10 Hz, where w is vertical windspeed, and u and v are the two horizontal windspeed components. The anemometer was also used to estimate temperature, 9s, as a function of the speed of sound and changes in air density (excluding water vapor, i.e., "virtual temperature"). The IRGA measured the concentration of water vapor at ~8 Hz. The laptop and a flux software program from McMillen (1988), with a fixed lagtime (14.3 s), were used to collect raw eddy covariance data files. Covariances, wind and scalar Eq. 3-2 Eq. 3-3

PAGE 60

50 statistics, and coordinate rotations were calculated in real time at 10 Hz. Because transport of mass and energy by turbulence occurs between frequencies of 1-10 Hz, within the inertial sub-range (Kaimal and Finnigan 1994), and because the IRGA's response time is ~8 Hz, fast-Fourier analyses were applied to correct for frequency loss. Protocols for accuracy, precision, quality control and assurance were used as defined by the AmeriFlux Science Plan (http://cdiac.esd.oml.gov/programs/ameriflux/scifhtm). The storage fluxes were estimated by adding together the changes of heat through the forest profile in both the air column below 42 m and in the foliage (Eq. 3-3). Water vapor was sampled from 6 inlets at 0.5, 7.3, 1 1.95, 16.55, 21.2, and 27.6 m on the tower. Solenoids switched the flow (~3 1pm) from each inlet through a second IRGA for 5 min during each 30-min period. For H, temperature through the profile. Op, was measured with platinum RTDs housed in radiation shields and co-located with each inlet; when sampling occurred, the airflow acted to aspirate the platinum RTDs, approximating true 8. Because leaves have small thermal inertia but significant amounts of H2O, changes in leaf H were estimated by: where Hieaf is the below-canopy leaf heat flux (W m"^), S^vt is the specific weight of leaf one-sided leaf area based on LAI from polyculture plantations at La Selva (4.5 m m" , S. Bigelow and J.Ewel, pers. comm.). It was assumed that the profiles of Hstor and XEstor were similar throughout the flux source area. G was estimated as a 30-min average of the 0.5 Eq. 3-4 water (kg H2O m'^ leaf area), Cpw is the heat capacity of water (J kg' K' ), and Larea is the

PAGE 61

51 3 soil heat flux plates. The contribution to G by the change in soil heat storage was ignored because soil temperature varied little in a 30-min period, soil thermal properties for these soils were not known, and the change in soil heat storage was expected to be negligible. Evapotranspiration Model Evapotranspiration estimates were partitioned into whole forest transpiration and the evaporation of intercepted precipitation. XEj was calculated using the PenmanMonteith equation; M„+PaC[e,iTa)-ea(T^)]ga XE = Eq. 3-5 /l[A + Kl + ")] 8b where XE is latent energy flux (W m"^), A is the is the rate of increase in saturated water vapor pressure with temperature (kPa K""), pa is the density of air (kg m"^), Cs is the saturated water vapor pressure at Ta, ea is the ambient water vapor pressure (kPa), ga is the aerodynamic conductance (mol m'^ s"'), A, is the latent heat of vaporization (J kg"'), y is the psychometric constant at 25 °C (0.0665 kPa K''), gb is the bulk canopy conductance (mol m"^ s''). To change units of energy to depth, XE was multiplied by a conversion factor that included molar volume (mol m"^) and weight (kg mol''). Evapotranspiration depth is noted in the text as Ej. A positive momentum flux into the canopy was assumed hence, aerodynamic conductance was estimated by,

PAGE 62

8a 52 k^u Eq. 3-6 2 f \ In + ln \ ) . z '42 where k is Von Karmen's constant (0.40), d is the zeroplane displacement (m), Zm is the aerodynamic roughness length, and and Th are the diabatic correction factors (m) for momentum and sensible heat, respectively (Yasuda 1988, Arya 1988). Zeroplane displacement and aerodynamic roughness changed with stability and were empirically estimated for this study period (Table 2-1, Loescher et al. submitted). Diabatic correction factors are a function of stability, where in stable conditions, =^>„ =61n(l+0 ^'l-^-^ and in unstable conditions, ^ -21n[— = 0.6T^ with C| is a stability parameter ratio of convective to mechanical turbulent production , Z42j^ Eq. 3-9 L where L is the Monin-Obukov length (Equation 2-3). Bulk canopy conductance, gb was estimated by, _ gg Ag?Eq. 3-10 CpAE where D42 is the specific humidity deficit at measurement height (kg kg"'). When gaps in the measured XE occurred, both ga and gb were empirically modelled by using relationships with horizontal windspeed (for ga), and the vapor pressure deficit (VPD) and Rn, (for gb, Martin et al. 1997, Wright et al. 1996). gb was then normalized to unity, and the upper limit to VPD determined (Jarvis 1976, Livingston

PAGE 63

53 and Black 1987). This limit function was used to estimate a theoretical maximum, gmax, by increasing gb as though VPD was not limiting. Then gmax, in turn, was related to R„, assuming that maximum conductance would take place with 0 VPD and high R„. A dimensionless decoupling coefficient, Q, was used to determine the relative effects of ga and gb on evapotranspiration (Jarvis and McNaughton 1986); A/ +2 Q = Eq.3-11 /r Sb Evaporation of free water in the canopy was modeled using a Rutter-type model (Calder et al. 1986). Because canopy water storage increases exponentially to a maximum, with increases in precipitation from a single rain event, Ei was modeled every 30-min by, t = 0 Eq. 3-12 where t is min, C is canopy water depth. When the canopy was wet, Ex was estimated with gb set to 0. C was estimated by, C^C,^A\-e '^'''P) Eq.3-13 where Ccap is the maximum canopy capacity, mm, k is a unitless canopy fill coefficient , SP is the cumulative amount of water that fell during a 30-min interval. An empirical estimate of 1 .53 mm was used for Ccap (Loescher et al. 2002). The stemflow component of interception was ignored because it was assumed to be a small volumetric flux, i.e., < 2% of rain, (Schroth et al. 1 999, Neal et al. 1993). A value of 0.28 for k from a broadleaf

PAGE 64

54 plantation forest at La Selva was used (Bigelow 2001). If modeUed Et was greater than the remainder of canopy free water, the remaining depth was included as transpiration, Et. Evapotranspiration was then the sum of Ei and Et. A second method of estimating XEr, the Priestly-Taylor equation, was used to compare with the Penman-Monteith results. The Priestly-Taylor equation (Eq. 3-14) simplifies the transfer process that is explicit in Eq. 3-5, and in doing so, is thought to be appropriate for large-scale, well-watered vegetative canopies, like those typically found in the wet tropics (Priestly and Taylor 1972). Ai:^=aR„[-^] Eq.3-14 where a is a coefiBcient estimated by fitting the model results to empirical measures of A.E from Eq. 3-2. Monteith (1981) estimated an average a on a theoretical basis as 1 .26, but values observed over rough canopies have varied greatly (Jones 1992). Results Above-Canopy Environment The Monin-Obukov (M-0) stability length did not differ with year or season. On a diurnal basis, M-0 length was neutral (~ 0 m) during the night, and decreased during the day time until 1400 when the boundary layer became weakly unstable (M-0 ~ -125 m Figure 3-1). After 1400 h, the M-0 length sharply increased, and the boundary-layer became weakly stable (-100 m) at 1600 h, but then returned to neutral conditions by nightfall. Friction velocity (u*) was -0.1 m s"' during the night and increased during the morning hours with convective turbulence, u* decreased after solar noon and continued

PAGE 65

55 to decrease throughout the afternoon due to the dissipation of turbulent kinetic energy (Figure 3-1). Total daily Rn ranged from 1 .47 to 27.54 MJ d"' during the measurement period, and differed among years (using a general linear model with an a = 0.05, p < 0.0001, Figure 3-2) with mean daily totals for 1998-2000 of 13.31 ±0.028, 17.48 ±0.050, and 15.33 ±0.040 MJ d'' (mean ±1 SE), respectively. Rn also varied significantly with season (p < 0.0001). Mid-day albedo did not change seasonally, and ranged from 0.118-0.135 of incident short wave radiation. Enei^ Balance Diurnal temperature and water vapor profiles (Figure 3-3) followed trends similar to those of other forests (Shaw et al. 1988). Heating of the air column during the day was greater with height, i.e., there was a positive temperature gradient. However, counter gradients were often observable between 21 and 27 m, where the leaf area was concentrated. Cooling during the night often produced neutral or slightly negative gradients, often with warmer temperatures at ground levels. Negative or neutral water vapor gradients were observed all times, with counter gradients present during non-rain days between 1 1 and 21 m height. Soil heat fluxes followed very similar diurnal patterns throughout the year, and ranged between ± 1 6 W m~^ at any point in time, with negative flux into the system during the daytime (Figure 3-4A). At night, XEstor was ~ 3-5 W m'^. A larger A,Estor efflux occurred in the early morning hours, presumably from convective winds mixing the below-canopy airspace and evaporating free water. X,Estor flux continued to be

PAGE 66

56 positive throughout the afternoon, but was more variable, with mean daytime values ranging from ~6 to 7 W m"^ This flux remained positive throughout the day across all years. Mean nighttime storage of sensible heat (Hstor + Hieaf) ranged from ~ 1-3 W m" , and became negative in the early hours as the air space increased in temperature. The maximum average Hstor + Hieaf was ca. -18 W m"^, which occurred at -0800 h when the convective boundary layer was developing. Hstor + Hieaf increased and became positive at -1400 h, which coincided with a late afternoon weakly stable/unstable boundary layer (Figure 3-1). The below-canopy environment continued to lose sensible heat until -1900 h, when neutral canopy conditions prevailed. The total storage flux (Figure 3-4B) was similar to that of Hstor + Hieaf, but daytime fluxes were ameliorated by release of water to the atmosphere. During nighttime neutral conditions, the flux was -8-10 Wm". The daytime minima was -5 W m'^ and the peak efflux was 17 W m"^, which occurred during weakly stable conditions (Figure 3-4B). The average 30-min XE was greater than H for all daytime hours and across seasons and years (i.e., H/X.E = p < 1 .0, Figure 3-5A). A linear model that included second-order effects of year, season, VPD and R,,, explained 79% of the total variation in H +XE (Table 3-1). Because Rn and VPD are auto-correlated and VPD did not explain any additional variation, it was removed from the linear model. Rn alone accounted for 69% and 68% of the variation in H and XE, respectively. I could only close the energy balance (Eq. 3-1) to within 32-50%, with the exception of periods with a wet canopy during the dry season of 1998. In general, H + XE + G underestimated R,,, particularly when Rn values were < 400 W m'^ (Figure 3-6). The most likely reason for this was that XE estimates, which also contributed to greater

PAGE 67

57 variance in p estimate before 1000 h (Figure 3-5 A). For this reason, XE was empirically estimated by >.E = Rn H for model comparisons, rather than the A,E estimate in Eq 3-2 (e.g., Twine et al. 2000). Bowen ratio calculated by H/(Rn-H) were -50% smaller than those estimated by IE in Eq 3-2, and the largest contribution of H occurring between 0900 and 1200 h (Figure 3-5B), coinciding closely with the observed diurnal pattern of Tg. Modelled Conductance and Evaporation ga, was linearly correlated with horizontal windspeed (Figure 3-7A). Values of \|/H and vj/M were small and all other parameters were relatively constant. We expected the upper boundary of normalized gb (Figure 3-7B) to be negatively related to VPD, because of the negative physiological response of canopies to VPD (Martin et al. 1997), which was found for VPDs > 0.5 kPa. gmax was hyper bo lically related to Rn (Figure 37C). Estimated values of gb were always less than ga (Figure 3-8A). At dawn, gb was ~1 mol m"^ s'* imtil 0800 h, then steadily decreased during the day, approaching 0. 1 mol m"^ s"' at dusk. The minimum ga was 1 .5 mol m"^ s'', which increased to > 2 mol m'^ s'' during mid-day (1000-1500 h). Modeled values of ga and gb behaved similarly to those derived using Eq. 3-2 (Figure 3-8B). Using conductances in the Penman-Monteith equation during periods when the canopy was dry, ga explained 44% of the variation in ^E (Figure 3-9). Q ranged from -0.3 during the night to -0.7 by 0830 h. For the majority of daytime hours (0600-1600), Q was > 0.5 (Figure 3-10). The Penman-Monteith equation explained 68% of the observed variation in Rn H, but overestimated Rn H by -28% (Figure 3-1 1 A). In contrast, the more simple Priestly-Taylor relationship accounted for 98% of the observed variation (Figure 3-1 IB).

PAGE 68

58 Annual ETft ranged from 1892 mm in 1998 to 2294 mm in 1999, and from 54% to 66% of bulk precipitation (Table 3-2). Mean daily ETpt rates were also lower in 1998 and greatest in 1999 (5.19 to 6.29 mm d'', respectively). Interception loss was greatest in 2000, with an annual total of 708 mm, accounting for 18% of bulk precipitation. Discussion Ecosystem Enei^ Dynamics Daytime profiles of temperature and water vapor in the upper canopy often showed counter gradients that were the result of winds that did not frilly penetrate the entire canopy. The higher canopy vegetation, between 21-27 m, was a physical barrier to transpiration from below. As a result, the apparent A,Estor at night was due adiabatic cooling and consequent condensation. Storage fluxes can contribute substantially to the overall ecosystem energy balance when R„ is small or during the night, but on a diurnal basis this was a very a small component (~2 %). Bowen ratios (P) consistently < 1 indicated that water was not limiting XE at any time during the year. Soils in the wet tropics generally do not exert hydraulic limitations on XE (De Bruin 1983). This is likely also the case at La Selva, where soils have high water-holding capacity and high hydraulic conductivity (Weitz et al. 1 997, SoUins et al. 1994). Pentaclethra macroloba, the dominant tree (42% of the basal area), closes its stomata and leaves in the late afternoon (~ 1530 h). I could not detect associated changes in either H or XE due to changes in P. macroloba physiology, suggesting that either single species control on energy partitioning or cannot be detected at the ecosystem-level for diverse tropical wet forests, the eddy-covariance technique is insensitive to species-

PAGE 69

59 level responses in these forests , or that the time of day that P. macroloba closes its leaves is inconsequential to energy balance. The lack of energy balance closure at this site is probably attributable to both losses of high and low frequency signals. This loss is more apparent when R,, < 400 W m"l A combination of the long sampling tube combined with high humidity conditions may have led to some in-tube mixing. Also, the scale lengths for H and IE were highly variable and may not have been fully captured using eddy covariance (Mahli 1 996). Similar degrees of non-closure have been found at other tower flux sites with similarly large aerodynamic roughness (~ 2 m), such as an old-grow4h pine stand in Oregon (to within 20-30 %, Anthoni et al. 2000), and an old-growth conifer stand in Washington ( to within 10-35 %, M. Falk and J. Chen, pers. comm.). Studies from other tropical have not reported the degree of energy closure for periods greater than 1 day, but it is likely that similar results would also be obtained for these also complex natural forests. A high degree of closiire for 30-min intervals seems to require aerodynamic roughness lengths less than -0.5 m to minimize viscous effects, less than saturated air for long periods, and smaller sensor separation distances (e.g., Gholz and Clark, in press). Conductances and Other Limits to Annual Energy Fluxes The general diurnal patterns found for ga and gb are similar to those for other tropical forests (Bigelow 2001, Wright et al. 1996, Shuttleworth 1984). Using the same calculations, our gb estimates were almost identical to those found by Wright et al. (1996) for a tropical forest in Brazil, suggesting similar physical and physiological controls on from these two neotropical forest canopies. Bigelow (2001) examined gb for three

PAGE 70

60 monocultural plantations at La Selva and found that all the species had higher late afternoon rates (-0.5-0.8 mol m"^ s"') than found for the old-growth forest. However two 2 1 • species, Cedrela odorata and Cordia alliodora, had late afternoon gb < 0.2 mol m' s" in the dry season. The lower afternoon gb that we observed was likely an integrated response of P. macroloba closing its stomata and folding its leaves closed by 1630 h or later. Significant differences were observed in gb between years and seasons, which followed the same trends as Rn. Between year differences in Rn explained much of the interannual variability observed in measured and modelled XE. Increased rainfall in 2000 increased the absolute amount of interception, but not the fi-action of rainfall intercepted. The direct effects of gb on could not be determined because these variables could not be independently measured. However, low values of Q (e.g., < 0.3) during night and early morning and evening hours indicate that these periods are the only times when physiological control over XE occurs, likely due to the opening and closing of stomata and leaves. Values of Q > 0.5 suggest that mid-day X,E is controlled more by Rn and ga than gb. ga explained 44 % of the variation in XE during times when both conductances were used to calculate ETpm (i.e., dry canopy conditions). The importance of ga in controlling ETpm increased fiirther because of the very high precipitation at La Selva. For example, 32 % of the time, the upper canopy was wet. The Priestly-Taylor relationship for wellwatered conditions described the ET dynamics of this tropical forest quite well. The fi-action of available energy used for ETpj was similar fi-om year-to-year suggesting a thermodynamic constraint on H that limits maximum daily virtual temperature (Calder 1986). Wright et al. (1996) and Calder et al. (1986) also found that accounted for a large fi-action of Rn (>0.80) from a humid

PAGE 71

61 Amazonian and Javanese forest, respectively, as did Bigelow (2001) for the monocultural plantations at La Selva (0.79-0.90). In a previous study in the old-grovvth forest at La Selva, Luvall (1984) determined that the energy required for evapotranspiration exceeded Rn by 25%. Although this seems counterintuitive, it may very well be true. This phenomenon has been observed over crops (Ham and Hiehnan 1991) as well as other tropical forests (Jones 1992, Shuttleworth 1989, Calder 1986). The likely explanation is that energy is locally advected into the flux field. This is certainly possible at La Selva, where mean daytime wind direction is -90 and the fetch is ~2 km, beyond which the landscape is dominated by pastures, crops and patches of secondary forests extending for ~ 60 km to the Caribbean shore. Often, small (< 1 km in width) afternoon convective cells deposit rain heterogeneously across the landscape. Advection of drier air masses with greater evaporative demand is possible, particularly during Luvall's study in the early 80's when much of the land in the Costa Rican coastal plain was being converted from forests to agriculture. Estimates of ETpr reported here did not exceed available Rn, suggesting advection was not significant, in this case, although we cannot rule out the possibility that advection contributed additional energy towards the overall balance. The magnitude of canopy capacitance is in large part a fiinction of physical surface area of a canopy (Waring and Schleshinger 1985). At La Selva, high epiphytic loads, bromeliad tanks and arboreal soil mats can contribute capacitance and may not have been fiilly accoimted for in our estimates. We used a fixed estimate of capacitance of 1.53 mm The fraction of rainfall that was intercepted seemed consistent from 1998 to 2000, and also with Luvall's study, which suggests that the canopy surface area at La

PAGE 72

62 Selva is saturated for a large portion of time, that the relative annual amount of throughfaU is constant, and that surface area does not change to any appreciable degree over time.

PAGE 73

63 Table 3-1. First order regression parameters for the energy balance closure from 19982000 across seasons and canopy conditions from a wet, tropical, old-growth forest at La Selva, Costa Rica, y = Yo + b*x, y is X,E + H, x is Rn, Yo and b values are mean ±1SD, n is number of 30-min averages, and all regressions were significant at the p < 0.0001 level unless otherwise noted. ^_ year Yo b n R 1998 Dry canopy/dry season 12.0 ±3.8 p = 0.0016 0.52 ±0.02 429 0.72 Wet canopy/dry season 38.0 ±6.2 1.07 ±0.05 362 0.59 Dry canopy/wet season 12.7 ±2.6 0.66 ±0.01 2182 0.73 Wet canopy/ Wet season 7.9 ±2.3 p= 0.0006 0.63 ±0.11 1867 0.66 1999 Dry canopy/dry season 16.0 ±2.3 0.63 ±0.01 2372 0.83 Wet canopy/dry season 19.4 ±2.5 0.57 ±0.01 1362 0.75 Dry canopy/wet season 13.1 ±2.3 0.55 ±0.01 2356 0.74 Wet canopy/ Wet season 22.6 ±5.0 0.50 ±0.02 483 0.73 2000 Dry canopy/dry season 7.3 ±4 p = 0.07 0.61 ±0.01 1220 0.81 Wet canopy/dry season 17.5 ±5.5 p = 0.0014 0.6 ±0.02 256 0.84 Dry canopy/wet season 17.5 ±2.6 0.57 ±0.01 1868 0.68 Wet canopy/ Wet season 23.1 ±4.1 0.68 ±0.2 872 0.71

PAGE 74

o s a +-» on O U 13 o o cr ki o CO Oh .I" o o o s o a> ^ (-1 S •~ * W o X) B £3 O H o 13 .2 Oh C/2 O bo .s a. H W JH H 00 o d d o m m o d d in «n d d d 00 in 00 o o o ON O O o o d -H d -H d -H ON ON o «n 0^ en U -a .o in oo rON 00 ON d d d CM o ON ON 00 n m ON > o 5 ON o u Q o I "2 u D CO cS X! Pi u i d 1^ d
PAGE 75

65 Table 3-3. Annual fraction of time that the canopy was wet at two heights in the La Selva canopy. year Height of leaf wetness sensor 25 m 2 m 1998 042 059 1999 0.17 0.57 2000 0.23 0.48 All years 0.32 0.57

PAGE 76

66 Time (hour) Figure 3-1 . Diumal relationships of Mction velocity and Monin-Obukov length over an old-growth tropical forest. Data are averages using all data from 1998-2000. Intervals A, B, C, and D are neutral, unstable, weakly unstable, weakly stable boundary layers, respectively. Error bars are +/1 SE.

PAGE 77

67 Figure 3-2. Cumulative net radiation for 1998-2000 over an old-growth forest in La Selva, Costa Rica. Daily means were derived using first-order regression. Each year was significantly different at the a = 0.05 level, p < 0.001, and > 0.99.

PAGE 78

68 Figure 3-3. Typical diurnal changes in below-canopy temperature and water vapor profiles fi-om an old-growth forest, La Selva, Costa Rica. Data are median values for aU of 1999. Error bars are +/1 SE.

PAGE 79

69 Figure 3-4. Diumal patterns of storage energy fluxes from an old-growth wet tropical forest in Costa Rica. Data are mean values from 1998-2000 with error bars +/1 SE.

PAGE 80

70 UJ £^ X o (0 i_ c a> o m 1.8 1.6 1.4 1.2 ^ 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 o c 9) O m If 2000 1 1 1999 1 1 1998 ..ill Mil Mill B fittl. 1 J 1 1 1 600 800 1000 1200 Time (hour) 1400 1600 Figure 3-5. Daytime Bowen ratios for each year. Estimates of were calculated by A) equation 3-2, and B) Rn H, with H determined by equation 3-3. Data are measured median (+ 1 SE) values with net radiation > 40 W m'^.

PAGE 81

71 wet canopy/dty season 1200 Net radiation (W m'') Figure 3-6. The relationship between net radiation and estimated energy flux for 1998-2000. Estimated energy flux includes modelled contribution from the understory.

PAGE 82

72 Figure 3-6. continued

PAGE 83

73 Figure 3-6. continued

PAGE 84

74 g, = 1.37 + 0.66'u = 0.91 Figure 3-7. Emprical relations to model both aerodynamic and bulk conductance, independant variables were averaged at different intervals, i.e., 0.25 m s' , 0.025 kPa, and 50 w m-2, for A, B, and C, respectively. All data were median values wdth +/1 SE.

PAGE 85

75 Figure 3-8. The diumal relationship of aerodynamic and bulk conductance calculated by A) collected eddy covariance data, and B) modelled based on Figure 7. Data are median values with +/1 SE.

PAGE 86

Figure 3-9. The relationship between aerodynamic conductance and latent energy flux from La Selva during 1999. Data are empirical estimates from Eq. 3-6 (ga), and from Eq. 3-2 (XE).

PAGE 87

77 1.0 -r 0.9 0.8 0.2 0.1 0.0 -'-1 — 1 1 1 1 — — ~i — 1 1 1 1 — 0 250 500 750 1000 1250 1500 1750 2000 2250 time (hour) Figure 3-10. Diurnal changes in the decoupling coeflBcient, Q, from an old-growth wet tropical forest, Costa Rica. Data is from 1997-2000. SE are typically < 0.006.

PAGE 88

78 Figure 3-11. Relationship between empirical and modeled estimates of the latent energy, where A) is modeled using a Penman-Monteith equation, and B) using Priestly-Talyor equation with a dry canopy, and C) using a PriestlyTaylor equation with a wet canopy. All graphs used data from 1998-2000. An a of 1 .24 was found for graphs B and C. All slopes were significant, p < 0.0001 . yintercepts had pvalues of 0.5, 0.04 and 0.04 for A, B and C, respectively..

PAGE 89

CHAPTER 4 CONCLUSIONS Diurnal patterns of NEE at La Selva followed trends similar to those observed elsewhere, with turbulent fluxes dominating in the daytime, and large storage fluxes contributing largely in the early morning. Daytime NEE was a function of both abiotic (PPFD, VPD and temperature) and biotic (quantum efficiency and eLAI) factors. VPD limited NEEday when values were above > 1 kPa, but this only occurred over a small percentage of time. There was a positive correlation between nighttime respiration and temperature based on diurnal averages of all NEEnight measurements. We used a fixed NEEnight value, to estimate annual NEE. There was a big difiference between our two alternative estimates of annual NEE. But regardless of the calculation method used, the results indicate that there is a large interannual variation in NEE at La Selva, related to large scale regional climate dynamics. Energy balance closure of this lowland tropical rainforest was ~ 74%. Storage fluxes contributed very little (-2%) to the overall daily energy balance. Conductances foUowed similar trends to those found of other tropical forests. Daytime XE was almost always greater than sensible heat, suggesting that the trees in this forest are in contact with sufBcient ground-water reserves to minimize hydraulic stress. Rn was a large determinant for the annual energy flux, suggesting that the Priestly-Taylor model for evapotranspiration is more appropriate in the tropics than the Penman-Monteith model. The general rule holds true that Rn ~ XE. 79

PAGE 90

LIST OF REFERENCES Amthor, J.S., 1994. Plant respiration. In: Wilkinson R.E (Ed.), Plant-Environment Interactions. M. Dekker Publisher, New York. pp. 501-554 Anderson, D.E., Verma, S.B., Clement, R.J., Baldocchi, D.D., Matt, D.R., 1986. Turbulence spectra of CO2, water vapor, temperature and velocity over a deciduous forest. Agricultural and Forest Meteorology, 38, 81-99. Anthoni, P.M., Law, B.E., Unsworth, M.H., 1999. Carbon and water vapor exchange of an open-canopied ponderosa pine ecosystem. Agricultural and Forest Meteorology, 95, 151-168. Arya, S.P., 1988. Introduction to micrometeorology. Academic Press, London, pp. 307. Aubinet, M., Chermanne, B., Vandenhaute, M., Longdoz, B., Yemaux, M., Laitat, E., 2001. Long term carbon dioxide exchange above a mixed forest in the Belgian Ardennes. Agricultural and Forest Meteorology. 108, 293-315. Avissar, R., 1993. Observations of leaf stomatal conductance at the canopy scale an atmospheric modeling perspective. Boundary-Layer Meteorology. 64, 127-148. Ayotte, K.W., Finnigan, J. J., Raupach, M.R., 1999. A second-order closure for neutrally stratified vegetative canopy flows. Boundary-Layer Meteorology. 90, 189-216. Baldocchi, D.D., Meyers, T.P., 1989. The effects of extreme turbulent events on the estimation of aerodynamic variables in a deciduous forest canopy. Agricultural and Forest Meteorology. 48, 117-134. Bigelow, S., 2001. Evapotranspiration modelled fi-om stands of three broad-leaved tropical trees in Costa Rica. Hydrological Processes. 15, 2779-2796. Black, T.A., Chen, W.J., Barr, A.G., Arain, M. A., Chen, Z., Nesic, Z., Hogg, E.H., Neumann, H.H., Yang, P.C., 2000. Increased carbon sequestration by a boreal deciduous forest in years with a warm spring. Geophysical Research Letters. 21, 1271-1274. Calder, I.R., 1986. What are the limits on forest evaporation? — ^A further comment. Journal of Hydrology. 89, 33-36. 80

PAGE 91

81 Calder, I.R., Wright, I.R., Murdiyaso, D., 1986. A study of evaporation from a tropical rain forest-west Java. Journal of Hydrology. 89, 13-31. Cavazos, T, Hastenrath, S.L., 1990. Convection and rainfall over Mexico and their modulation by the Southern Oscillation. Journal of Climatology, 10, 377-386. Chen W.Y., Van den Dooi, H.M., 1999. Significant change of extratropical natural variability and potential predictability associated with the El Nino/Southern Oscillation. Tellus A51, 790-802. Clark D.B., Clark D.A., 2000. Landscape-scale variation in forest structure and biomass in a tropical rain forest. Forest Ecology and Management. 137, 185-198. Clark, D.A., Piper, S.C., Keeling, CD., Clark, D.B. (in review) Tropical forest growth and atmospheric carbon dynamics linked to annual temperature variation. Science Clark, K.L., Gholz, H.L., Moncreiff, J.B., Cropley, F., Loescher, H.W., 1999. Environmental controls over net exchanges of carbon dioxide from contrasting Florida ecosystems. Ecological Applications. 9, 936-948. Corti, S., Molteni, F., Palmer, T.N.,1999. Signature of recent climate change in frequencies of natural atmospheric circulation regimes. Nature, 398, 799-802. Cramer, W., Kicklighter, D.W., Bondeau, A., Moore, B., Churkina, C, Nemry, B., Ruimy, A., Schloss, A.L., 1999. Comparing global models of terrestrial net primary productivity (NPP): overview and key results. Global Change Biology. 5, 1-15. Davidson, E.A., Belk, E., Boone, R.D., 1998. Soil water content and temperature as independent or confoimded factors controlling soil respiration in a temperate mixed hardwood forest. Global Change Biology. 4, 217-227. De Bruin, H.A.R., 1983. Evapotranspiration in humid tropical regions. In: Hydrology of humid tropical regions with particular reference to the hydrological effects of agriculture and forestry practice. Proc. Hamburg Symp. Aug. 1983. lAHS Publ. No. 40. Denslow J.S., Hartshorn G.S., 1994. Tree-fall gap environments and forest dynamic processes. In: McDade, L.A., Bawa, K.S., Hespenheide, H.A., Hartshorn, G.S. (Eds), La Selva: Ecology and Natural History of a Neotropical Rain Forest. University of Chicago Press, Chicago, IL. pp. 120-127. Dixon, R.K., Brown, S., Houghton, R.A., Solomon, A.M., Trexler, M.C., Wisniewski, J. 1994. Carbon pools and flux of global forest ecosystems. Science, 263, 185-190. Dolman, A.J., Gash, J.H.C., Roberts, J., Shuttleworth, W.J., 1991. Stomatal and Surfece Conductance of Tropical Rain-Forest. Agricultural and Forest Meteorology. 54, 303-318.

PAGE 92

82 Fan, S-M., Wofsy, S.C., Bakwin, P.S., Jacob, D.J., 1990. Atmosphere-biosphere exchange of CO2 and O3 in the Central Amazon forest. Journal of Geophysical Research. 95, 12851-16864. Farquhar, G.D., Caemmerer, S.V., Berry, J. A., 1980. A biochemical model of photosynthetic CO2 assimilation in leaves of C-3 Species. Planta. 149, 78-90. Fasullo J., Webster, P. J., 1999. Warm pool SST variability in relation to the surface energy balance. Journal of Climate. 12, 1292-1305. Frankie, G.W., Baker, H.G., Opler, P. A., 1974. Comparative phenological studies of trees in tropical wet and dry forests in lowlands of Costa-Rica. Journal of Ecology. 62, 881919. Frolking, S., Goulden, M.L., Wofsy, S.C., Fan, S.-M., Sutton, D.J., Munger, J.W., Bazzaz, A.M., Daube, B.C., Crill, P.M., Aber, J.D., Band, L.E., Wang, X., Savage, K., Moore, T., Harriss, R.C., 1996. Modeling temporal variability in the carbon balance of a spruce/moss boreal forest. Global Change Biology. 2, 343-366. Gholz, H.L., Clark, K.L., (in press). Energy exchange across a chronosequence of slash pine forests in north Florida. Agricultural and Forest Meteorology Goulden, M.L, Munge, J.W., Fan, S-M., Daube, B.C., Wofsy, S.F., 1996. Measurements of carbon sequestration by long-term eddy-co variance: methods and critical evaluation of accuracy. Global Change Biology. 2, 169-182. Grace, J., Lloyd, J., Mclntyre, J., Miranda, A. C, Meir, P., Miranda, H.S., Nobre, C, MoncrieflF, J., Massheder, J., Malhi, Y., Wright, I., Gash, J., 1995a. Carbon dioxide uptake by an undisturbed tropical rainforest in southwest Amazonia, 1992 to 1993. Science, 270, 779-780. Grace J, Lloyd J, Mclntyre J, Miranda, A., Meir, P., Miranda, H., Moncrieff, J., Massheder, J., Wright, I., Gash, J. 1995b. Fluxes of carbon dioxide and water vapor over an undisturbed tropical rainforest in south-west Amazonia. Global Change Biology. 1,112. Grace, J., Malhi, Y., Lloyd, J., Mclntyre, J., Miranda, A.C., Meir, P., Miranda, H.S., 1996. The use of eddy co variance to infer net carbon dioxide uptake of Brazilian rain forest. Global Change Biology. 2, 209-217. Ham, J.M., HeUman, J.L., 1991. Aerodynamic and surface resistances affecting energy transport in a sparse crop. Agricultural and Forest Meteorology. 53, 267-284. Hansen, F.V., 1993. Surface roughness lengths. ARL technical Report, U.S. Army, White Sands Missile Range, NM 88002-5501.

PAGE 93

83 Hartmaim, D.L., Moy, L.A., Fu, Q., 2001. Tropical convection and the energy balance at the top of the atmosphere. Journal of Climate. 14, 4495-451 1. Hartshorn, G.S., Peralta R., 1988. Preliminary description of primary forests along the La Selva-Volcan Barva altitudinal transect, Costa Rica. In: Alemda, F., Pringle, C. (Eds), Tropical Rainforests: Diversity and Conservation. California Academy of Science, San Fransisco. pp. 281-295. Hastenrath, S.L., 1991. Climate Dynamics of the Tropics. Kluwer Academic, Boston, MA, 488 pp. Hodnett, M.G., TomaseUa, J., Marques Filho, A.O., Oyama, M.D., 1996. Deep water uptake by forest and pasture in central Amazonia: Predictions from long-term rainfall data using simple water balance model. In: Gash, J.H.C., Nobre, C.A., Roberts, J.M., Victoria, R.L., (Eds.), Amazonian Deforestation and Climate. John Wiley, New York. pp.79-99. Holdridge, L.R., Grenke, W.C., Hatheway, W.H., Liang, T., Tosi, J. A. jr. 1971. Forest Environments in Tropical Life Zones: a Pilot Study. Pergamon Press, Oxford, 747 pp. Houghton, R.A., 1996. Terrestrial sources and sinks of carbon inferred from terrestrial data. Tellus Series B-Chemical and Physical Meteorology. 48, 420-432. Houghton, R.A., Davidson, E.A., Woodwell, G.M., 1998. Missing sinks, feedbacks, and understanding the role of terrestrial ecosystems in the global carbon balance. Global Biogeochemical Cycles. 12, 25-34. Hulme, M., Viner, D., 1998. A climate change scenario for the tropics. Climate Change. 39, 145-176. IPCC, 1995. IPCC Second Assessment Report: Climate Change. Geneva, Switzerland, pp 64. Ishibashi, M., Terashima, I., 1995. Effects of continuous leaf wetness on photosynthesis: adverse aspects of rainfall. Plant, Cell and Environment. 18, 431-438. Jarvis, P.G., 1976. The interpretations of the variations in leaf water potential and stomatal conductance found in canopies in the field. Philosophical Transactions form the Royal Society. B273, 593-610. Jarvis, P.G., McNaughton, K.G., 1986. Stomatal control of transpiration: scaling up from leaf to region. Advances in Ecological Research. 15, 1-49. Jones, H., 1992. Plants and microclimate. Cambridge University Press, NY, 428 pp.

PAGE 94

84 Kaimal, J.C., Finnigan, J.S., 1994. Atmospheric Boundary-Layer Flows; Their Structure and Measurement Oxford University Press, UK, 453 pp. Kelly, M.A., Randall, D.A., 2001 . A two-box model of a zonal atmospheric circulation in the tropics. Journal of Climate. 19, 3944-3964. Kindermann, J., Wurth, G., Kohlmaier, G.H., Badec, F.W., 1996. Interannual variation of carbon exchange fluxes in terrestrial ecosystems. Global Biogeochemical Cycles, 10, 737-755. Landsberg, J.J., Gower, S.T., 1997. Applications of Physiological Ecology to Forest Management. Academic Press, NY, 354 pp. Larson, K., Hartmann, D.L., Klein, S.A., 1999. The role of clouds, water vapor, circulation, and boundary layer structure in the sensitivity of the tropical climate. Journal of Climate. 12, 2359-2374. Law, B.E., Williams, M., Anthoni, P.M., Baldocchi, D.D., Unsworth, M.H., 2000. Measuring and modelling seasonal variation of carbon dioxide and water vapour exchange of a Pinus ponderosa forest subject to soil water deficit. Global Change Biology. 6, 613-630. Leclerc, M.Y., Karipot, A., Prabha, T., Allwine, G., Lamb, B., Gholz, H.L., (submitted) Tracer flux footprint validation over a forest canopy and effect of larger-scale advection on eddy-covariance fluxes. Journal of Geophysical Research. Lieberman, M., Lieberman, D., Hartshorn, G.S., Peralta, R., 1985. Small-scale altitudinal variation in lowland wet tropical forest vegetation. Journal of Ecology. 73, 505-516. Lieberman, D., Hartshorn, G.S., Lieberman, M, Peralta, R., 1990. Forest dynamics at La Selva Biological Station, 1969-1985. In: Gentry, A.H., (Ed.), Four Neotropical Rainforests. Yale University Press, New Haven, pp. 509-521. Livingston, N.J., Black, T.A., 1987. Stomatal characteristics and transpiration of three species of conifer seedlings planted on a high elevation south-facing clear-cut. Canadian Journal of Forest Research. 17, 1273-1282. Lloyd, J., Grace, J., Miranda, A.C., Meir, P., Wong, S. C, Miranda, B. S., Wright, I. R., Gash, J. H. C, Mclntyre, J,. 1995. A simple calibrated model of Amazon rainforest productivity based on leaf biochemical properties. Plant, Cell and Environment. 18, 1129-1145. Loescher, H.W., Powers, J.S., Oberbauer, S.F., 2002. Spatial variation of throughfell volume in an old growth tropical wet forest, Costa Rica. Journal of Tropical Ecology. 18, 397-407

PAGE 95

85 Loescher, H.W., Oberbauer S.F., Gholz, H.L., Clark, D.B. (submitted) Environmental controls on net ecosystem-level carbon exchange and productivity in a Central American tropical wet forest. Global Change Biology. Luvall, J.C., 1984. Tropical deforestation and recovery: the effects on evaporation processes. Ph.D. dissertation. University of Georgia, Athens, Georgia. 146 pp. Mahli, Y., 1996. The behavior of the roughness length for temperature over heterogeneous surfaces. Quarterly Journal of the Royal Meteorological Society. 122, 1095-1125. Mahli, Y., Nobre, A.D., Grace, J., Kruijt, B., Pereira, M.G.P., Culf, A., Scott, S., 1998. Carbon dioxide transfer over a Central Amazonian rain forest. Journal of Geophysical i?esearc/i. 103,31,593-31,612. Malhi, Y., Grace, J., 2000. Tropical forests and atmospheric carbon dioxide. Trends in Ecology and Evolution, 15,332-33. Mahrt, L., 1992. Momentum balance of gravity flows. Journal of Atmospheric Sciences. 39, 2701-2711. Mahrt, L., Lee X.H., Black, A., Neumann, H., Staebler, R.M., 2000. Nocturnal mbdng in a forest subcanopy. Agricultural and Forest Meteorology. 101, 67-78. Martin, T.A., Brown, K.J., Cermak, J., Ceulemans, R., Kucera, J., Meinzer, F.C., Rombold, J.S., SprugeL, D.G., Hinckley, T.M., 1997. Crown conductance and tree and stand transpiration in a second-growth Abies amabilis Forest. Canadian Journal of Forest Research. 27, 796-808. Massman, W.J., Lee, X., 2001. Eddy covariance flux corrections and uncertainties in long term studies of carbon and energy exchanges. Proceedings from the workshop for unaccounted flux in long term studies of carbon and energy exchanges. Boulder CO. McMillen, R.T., 1988. An eddy-correlation technique with extended applicability to nonsimple terrain. Boundary-Layer Meteorology. 43, 231-245. Melillo, J.M., McGuire, A.D., Kicklighter, D.W., Moore, III. B, Vorosmarty, C.J., Schloss, A.L., 1993. Global climate change and terrestrial net primary productivity. Nature. 363, 234-240. Meyers, T.P., Baldocchi, D.D., 1991. The budgets of turbulent kinetic-energy and Reynolds stress within and above a deciduous forest. Agricultural and Forest Meteorology. 53, 207-222. Monteith, J.L., 1981. Evaporation and surfece temperature. Quarterly Journal of the Royal Meteorological Society. 107, 1-27.

PAGE 96

86 Monteith, J.L., Unsworth, M.H., 1990. Principles of Environmental Physics. Edward Arnold Publishers, New York. pp. 291. Nakamura, R., Mahrt, L., 2000. Similarity theory for local and spatially averaged momentum fluxes. Agricultural and Forest Meteorology. 101, 265-279. National Academy of Science. 2000. Reconciling observations of global temperature change. National Academy Press, Washington, D.C., p 85. Neal, C, Robson, A.J., Bhardwaj, C.L., Conway, T., Jeffery, H.A., Neal, M., Ryland, G. P., Smith, C.J., Walls, J. 1993. Relationships between precipitation, stemflow and throughfall for a lowland beech plantation. BlackWood, Hampshire, Southern England Findings on interception at a forest edge and the effects of storm damage. Journal of Hydrology. 146, 221-233. Pahlow, M., Parlange, M.B., 2001. On Monin-Obukhov similarity in the stable atmospheric boundary layer. Boundary-layer meteorology. 99, 225-248. Priestly, C.H.B., Taylor, R.J., 1972. On the assessment of the surface heat flux and evaporation using large scale parameters. Monthy Weather Review. 100, 81-92. Raman, S., Niyogi, D.S., Prabhu, A., Ameenullah, S., Nagaraj, S.T., Kumar, U., Jayanna, S., 1998. VEBEX: Vegetation and surface energy balance experiment for the tropics. Proceedings from the Indian Academy of Science. 107, 97-105. Raupach, M.R., Finnigan, J.J., 1997. The influence of topography on meteorological variables and s\irface-atmosphere interactions. Journal of Hydrology. 190, 182-213. Raupach, M.R., Weng, W.S., Carruthers, D.J., Hunt, J.G.R., 1992. Temperature and humidity fields and fluxes over low hills. Quarterly Journal Royal Meteorological Society. 118, 191-225. Richards, P.W., 1996. The tropical rain forest; an ecological study. Cambridge University Press, Cambridge UK. 575 pp. Roberts, J., Cabral, O.M.R., Fisch, G., Molion, L.C.B., Moore, C.J., Shuttleworth, W.J., 1993. Transpiration fi-om an Amazonian rain-forest calculated fi-om stomatal conductance measurements. Agricultural and Forest Meteorology. 65, 175-196. Rosenberg, N.J., Blad, B.L., Verma, S.B., 1983. Microclimate: the biological environment. Wiley, New York. 495 pp. Ruimy, A., Jarvis, P.G., Baldocchi, D.D., Saugier, B., 1995. CO2 fluxes over plant canopies and solar radiation: A review. Advances in Ecological Research. 26, 1-81.

PAGE 97

87 Ruimy, A., Kergoat, L., Bondeau, A., 1999. Comparing global models of terrestrial net primary productivity (NPP): analysis of differences in light absorption and lightuse efficiency. Global Change Biology. 5, 56-64. Ryan, M.G., 1991 . A simple method for estimating gross carbon budgets for vegetation in forested ecosystems. Tree Physiology. 9, 255-266. Sanford, R, Paaby, P., Luvall, J.C., Phillips, E., 1994. Climate, geomorphology and aquatic systems. In: McDade ,L.A., Bawa, K.S., Hespenheide, H.A., Hartshorn, G.S., (Eds.), La Selva: Ecology and Natural History of a Neotropical Rain Forest University of Chicago Press, Chicago, IL. pp. 19-33 Schimel, D.S., House, J.I., Hibbard, K.A., Bousquet, P., Ciais, P., Peylin, P., BrasweU, B.H., Apps, M.J., Baker, D., Bondeau, A., Canadell, J., Churkina, G., Cramer, W., Denning, A.S., Field, C.B., Friedlingstein, P., Goodale, C, Heimann, M., Houghton, R. A., Melillo, J.M., Moore, B., Murdiyarso, D., Noble, I., Pacala, S.W., Prentice, I.C., Raupach, M.R., Rayner, P. J., Scholes, R.J., Stefifen, W.L., Wirth, C, 2001. Recent patterns and mechanisms of carbon exchange by terrestrial ecosystems. Nature. 414, 169-172. Schmid, H.P., Grimmond, S.B., Cropley, F., OfiFerle, B., Su, H-B., 2000. Measurements of CO2 and energy fluxes over a mixed hardwood forest in the mid-western United States. Agricultural and Forest Meteorology. 103, 357-374. Schroth, G., Silva, L.F.d., Wolf, M.A., Teixeira, W.G., Zech, W. 1999. Distribution of throughfall and stemflow in multi-strata agroforestry, perennial monoculture, fallow and primary forest in central Amazonia, Brazil. Hydrological Processes. 13, 1423-1436. Schuepp, P.H., Leclerc, M.Y., Macpherson, J.I., Desjardins, R.L., 1990. Footprint prediction of scalar fluxes from analytical solutions of the difiusion equation. BoundaryLayer Meteorology. 50, 355-373. Schwendenmann, L., Veldkamp, E., Brenes, T., O'Brien, J.J., Mackensen, J., (submitted) Spatial and temporal variation in soil CO2 efflux in an old-growth neotropical rain forest, La Selva, Costa Rica. Biogeochemistry. Shaw, R.H., Den Hartog, G., Neuman, H.H., 1988. Influence of foliar density and thermal stability on profiles of Reynolds stress and turbulence intensity in a deciduous forest. Boundary-Layer Meteor ol. 45, 391-409. Shuttleworth, W.J., Gash, H.C., Lloyd, C.R., Moore, C.J., Roberts, J., Marques Filho, A.d. O., Fisch, G., Silva Filho, V de P., Molion, L.C.B., Sa, L.D. de A., Nobre, J.C.A., Cabral, O.M.R., Patel, S.R., Moraes, J.C., 1984. Eddy correlation measurements of energy partition for Amazonian forest. Q. J. R. Meteorol. Soc. 110, 1 14311 63.

PAGE 98

88 Shuttleworth, W.J., 1988a, Evaporation from Amazonian rainforest. Philosophical Transactions form the Royal Society London. B233, 321-346. Shuttleworth, W.J., 1989. Micrometeorology of temperate and tropical forests. Philosophical Transactions form the Royal Society London. B324, 299-334. Smith, W.K., McClean, T.M., 1989. Adaptive relationship between leaf water repellency, stomatal distribution, and gas exchange. American Journal of Botany. 76, 456-469. Sohn, B.J., Smith, E.A., 1992. Global energy transports and the influence of clouds on transport requirements a Satellite Analysis. Journal of Climate. 5, 717-734. Sollins P, Sancho FM, Mata RC, Sanford RL Jr (1994) Soils and soil process research. In: McDade, L.A., Bawa, K.S., Hespenheide, H.A., Hartshorn, G.S., (Eds.), La Selva: Ecology and Natural History of a Neotropical Rain Forest University of Chicago Press, Chicago, IL. pp. 34-53. Sternberg, L.S.L., Mulkey, S.S., Wright, S.J., 1989. Ecological interpretation of leaf carbon isotope ratios: influence of respired carbon dioxide. Ecology. 70, 1317-1324. Sternberg, L.S.L., Moreira, M.Z., Martinelli, L.A., Victoria, R.L., Barbosa, E.M., Bonates, L.C.M., Nepstad, D.C., 1997. Carbon dioxide recycling in two Amazonian tropical forests. Agricultural and Forest Meteorology. 88, 259-268. Tian, H., Melillo, J.M., KickUghter, D.W., McGuire, A.D., Helfich, III. J.V.K., Moore, III. B., Vorosmarty, C.J., 1998. Effect of interannual climate variability on carbon storage in Amazonian ecosystems. Nature. 396, 664-667. Ter Steege, H.,1996. Winphot 5: a programme to analyze vegetation indices, light, and light quality from hemispherical photographs. Tropenbos Guyana Report 95-2, Tropenbos Guyana Programme, Georgetown, Guyana. Timmermann, A., Oberhuber, J., Bacher, A., Esch, M. Latif, M., Roeckner, E., 1999. Increasing El Nino frequency in a climate model forced by future greenhouse warming. Nature. 398, 694-694. Tomasella, J., Hodnett, M.G., 1996. Soil properties and van Genuchten parameters for an oxisol under pastiire in central Amazonia. In: Gash, J.H.C., Nobre, C.A., Roberts, J.M., Victoria, R.L., (Eds.), Amazonian Deforestation and Climate John Wiley, New York. pp.101-124. Trumbore, S.E., Chadwick, O.A., Amimdson, R., 1996. Rapid exchange between soil carbon and atmospheric carbon dioxide driven by temperature change. Science. 272, 393396.

PAGE 99

89 Twine, T.E., Kustas, W.P., Norman, J.M., Cook, D.R., Houser, P.R., Meyers, T.P., Prueger, J.H., Starks, P.J., Wesely, M.L., 2000. Correcting eddy-covariance flux underestimates over a grassland. Agricultural and Forest Meteorology. 103, 279-300. Valentini, R., Matteucci, G., Dolman, A.J., Schulze, E. D., Rebmann, C, Moors, E.J., Granier, A., Gross, P., Jensen, N.O., PUegaard, K., Lindroth, A., GreUe, A., Bemhofer, C, Grunwald, T., Aubinet, M., Ceulemans, R., Kowalski, A.S., Vesala, T., Rannik, U., Be'rbigier, P., Loustau, D., Guomundsson, J., Thorgeirsson, H., Ibrom, A., Morgenstem, K., Clement, R., Moncrieff, J., Montagnani, L., Minerbi, S., Jarvis, P.G., 2000. Respiration as the main determinant of carbon balance in European forests. Nature. 404, 861-865. VourUtis, G.L., Oechel, W.C., 1997. Landscape-scale CO2, H2O vapour and energy flux of moist-wet coastal tundra ecosystems over two growing seasons. Journal of Ecology. 85, 575-590. Waring, R.H., Schlesinger, W.H., 1985. Forest ecosystems: concepts and management. Academic Press, NY, 340 pp. Waring, R.H., Law, B.E., Goulden, M.L., Bassow, S.C., M^Creight, R.W., Wofey, S.C., Bazzaz, F.A., 1995. Scaling gross ecosystem production at Harvard Forest with remote sensing: a comparison of estimates from a constrained quantum-use efficiency model and eddy correlation. Cell, and Environment. 18, 1201-1213. Waylen, P.R., Caviedes, C.N., Quesada, M.E., 1996a. Interannual variability of monthly precipitation in Costa Rica. Journal of Climate. 9, 2606-2612. Waylen, P.R., Quesada, M.E., Caviedes, C.N., 1996b. Temporal and spatial variability of annual precipitation in Costa Rica and the Southern Oscillation. International Journal of Climatology. 16, 173-193. Webb, E.K., Perman, G.L Luening, R., 1980. Correction of flux measurements for density effects due to heat and water vapor transfer. Quarterly Journal of the Royal Meteorological Society, 106, 85-100. Wielicki, B.A.,Wong T.M., Allan, R.P., Slingo, A., Kiehl, J.T., Soden, B.J., Gordon, C.T., Miller, A.J., Yang, S.K., Randall, D.A., Robertson, F., Susskind, J., Jacobowitz, H., 2002. Evidence for large decadal variability in the tropical mean radiative energy budget. Science. 295, 841-844. Wietz, A.M., Grauel, W.T., Keller, M., Veldkamp, E., 1997. Calibration of time domain reflectometry technique using undisturbed soil samples from humid tropical soils of volcanic origin. Water Resources and Research. 33, 1241-2149.

PAGE 100

90 WiUiams, M., Malhi, Y., Nobre, A.D., Rastetter, E.B., Grace, J., Pereira, M.G P., 1998. Seasonal variation in net carbon exchange and evaporation in a BraziHan ram forest: a modelling analysis. Plant, Cell and Environment. 21, 953-968. Wilson, J.D., Finnigan, J.J., Raupach, M.R., 1998. A first order closure for disturbed plant canopy flows and its application to winds in a canopy on a ndge. Quarterly Journal Royal Meteorological Society. 124, 705-732. Wright, I.R., Gash, J.H.C., Rocha, H.R.d., Roberts, J.M., 1996. Modelling surface conductance for Amazonian pasture and forest. In: Gash, J.H.C., Nobre, C.A., Roberts, J.M., Victoria R.L. (Eds.), Amazonina deforestation and climate. Institute of Hydology. Pp. 437-467. Yasuda, N., 1988. Turbulent difiusivity and diurnal variations in the atmospheric boundary layer. Boundary-Layer Meteorology 43, 209-221.

PAGE 101

BIOGRAPHICAL SKETCH Hank's interest in natural history began at a very young age with fishing, camping, and hiking with family and fiiends. Since then his inquisitiveness towards understanding how natural systems work has continued, as evidenced by this dissertation. Hank has a few academic degrees: A.A.S. in Agronomy (SUNY Cobleskill), A.A.S. in Applied Science (SUNY Cobleskill), B.S. in Natural Resource Management (State College of Vermont, Johnson), and a M.S. fi-om School of Forest Resources and Conservation, University of Florida, and has served as a director for a NGO in Costa Rica, a project manager for OXFAM/APSINCA in Nicaragua, and as a cabinetmaker. Hank's non-academic interests are many-fold, and he currently enjoys listening to loud music, birdwatching, setting off explosives, biking, shooting weapons, racquetball, cooking large meals and enjoying them with Mends. But mostly he enjoys life, and sucks his orange dry. 91

PAGE 102

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fiilly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. ^enry L. Qnolz, Cha Profj^^r of Foresi Resources and Conservation I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fiilly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Steven F. Oberbauer, Professor of Biological Sciences, Florida International University I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fiilly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. J^mi^ R^^acobs, Associate Professor of Civil and • Coastal Engineering I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fiilly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. icis E. Putz, Professor^ ^rest Resources and Conservation

PAGE 103

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fiilly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Step: of Botany This dissertation was submitted to the Graduate Faculty of the School of Forest Resources and Conservation in the College of Agricultural and Life Sciences and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August 2002 Director, Forest Resources and Conservation Dean, Graduate School