SPATIAL VARIABILITY IN THE INTERCEPTION OF INCLINED RAINFALL BY A TROPICAL RAINFOREST CANOPY

12 Pages·1992·1 MB·

by  Herwitz, Stanley R

1992] GARCiA-FRANCO & RICO-GRAY: TILLANDSIA GALLS 61
(Laessle, 1961; Benzing, 1990). It is likely that GARDNER, C. S. 1986. Inferences about pollination
the association between epiphytic bromeliads, in Tillandsia (Brome1iaceae). Selbyana 9: 76-87.
gall-forming insects, and parasitoids are far more GARciA-FRANCO, J. G., V. RICO-GRAY, AND O. ZAYAS.
common than previously documented, and that 1991. Seed and seedling predation of Bromelia
the small number of reports reflects only that the pinquin L. by the land crab Gecarcinus lateralis
Frem. in Veracruz, Mexico. Biotropica 23: 96-97.
association has passed unnoticed in most habi
---AND ---. (in press) Biologia reproductiva de
tats.
Tillandsia deppeana Steudel (Bromeliaceae) en
Veracruz, Mexico. Brenesia 35.
ACKNOWLEDGMENTS HAWKINS, B. A. AND R. J. GAGNE. 1989. Determi
nants of assemblage size for the parasitoids of Ce
We acknowledge V. Hernandez (Instituto de cidomyiidae (Diptera). Oecologia (Berlin) 81: 75-
Ecologia), R. J. Gagne (USDA) and M. E. Schauff 88.
(USDA) for their help in insect determination, JANZEN, D. H. 1974. Epiphytic myrmecophytes in
and E. Saavedra for the illustration of T. ionan Sarawak: mutualism through the feeding of plants
by ants. Biotropica 6: 237-259.
tha. Field work was partially supported by CON
LAESSLE, A. M. 1961. A micro-limnological study of
ACYT grant No. 903579 to VRG.
Jamaican bromeliads. Ecology 42: 499-517.
LALoNDE, R. G. AND J. D. SHORTHOUSE. 1985. Growth
LITERATURE CITED and development of larvae and galls of Urophora
cardui (Diptera, Tephritidae) on Cirsium arvense
ACKERMAN, J. D. 1986. Coping with the epiphytic (Compositae). Oecologia (Berlin) 65: 161-165.
existence: pollination strategies. Selbyana 9: 52- MORENO, P., E. VAN DER MAAREL, S. CASTILLO, M. L.
60. HUESCA, AND I. PISANTY. 1982. Ecologia de la
BENZING, D. H. 1970. An investigation of two bro vegetaci6n de dunas costeras: estructura y com
meliad myrmecophytes: Tillandsia butzii Mez, T. posicion en el Morro de la Mancha, Ver. Biotica
caput-medusae E. Morren and their ants. Bull. (Mexico) 7: 491-526.
Torrey Bot. Club 97: 109-115. NEIL, W. 1951. A bromeliad herpetofauna in Florida.
---. 1990. Vascular epiphytes. Cambridge Uni Ecology 32: 140-143.
versity Press, Cambridge. 354 pp. NOVELO, A. 1978. La vegetaci6n de la Estacion Biolo
BEUTELSPACHER, B. C. R. 1971. Una Bromeliaceae gica El Morro de la Mancha, Veracruz, Biotica
como un ecosistema. Biologia (Mexico) 2: 82-88. (Mexico) 3: 9-23.
---. 1972. Fauna de Tillandsia caput-medusae E. RICO-GRAY, V. AND A. LoT-HELGUERAS. 1983. Pro
Morren, 1880 (Bromeliaceae). An. Inst. BioI., ducci6n de hojarasca del manglar de la Laguna de
U.N.A.M., Ser. Zool. 43: 25-30. la Mancha, Veracruz, Mexico. Biotica (Mexico) 8:
BORROR, D. J., D. M. DE LoNG, AND C. A. TRIPLEHORN. 295-301.
1981. An introduction to the study of insects, 5th SALAS, D. S. 1973. Una bromeliaceae costarricense
ed. Saunders College Publishing, Philadelphia. 827 polinizada por murcielagos. Brenesia 2: 5-10.
pp. UTLEy,J.F. 1983. A revision oft he Middle American
BURT-UTLEY, K. AND J. F. UTLEY. 1975. Calcium45 Thecophylloid Vrieseas (Bromeliaceae). Tulane
translocation in Tillandsia balbisiana Schult. Stud. Zool. Bot. 24: 1-81.
(Bromeliaceae). Brenesia 5: 51-65.
Selbyana l3: 62-71
SPATIAL VARIABILITY IN THE INTERCEPTION OF
INCLINED RAINFALL BY A TROPICAL
RAINFOREST CANOPY
STANLEY R. HERWITZ
Graduate School of Geography, Clark University,
Worcester, Massachusetts 01610 U.S.A.
ROBERT E. SLYE
Ecosystem Science and Technology Branch, NASAlA mes Research Center,
Moffett Field, California 94035 U.S.A.
ABSTRACT. Many epiphytes in forest canopies are drought-adapted due to the limited water-holding
capacity of the aboveground woody surfaces that support them. The objective of our study was to evaluate
whether neighboring tropical rainforest canopy trees differentially intercept inclined rainfall, and discuss
the possible effects on epiphyte distribution patterns. The study involved: (1) developing a computer model
representing the 3-D geometry of a tropical rainforest canopy surface in northeast Queensland, Australia,
using photogrammetric crown elevation data and a geographic information system; (2) generating a shaded
canopy relief image to compute the effective rainfall-intercepting crown areas of 50 selected canopy trees
during a single precipitation event; and (3) analyzing the relationships between the selected trees' effective
crown areas and the net rainfall totals measured beneath their crowns. A significant correlation between
the effective crown areas and net rainfall totals indicated that the inclined rainfall was differentially inter
cepted, with the more prominent canopy trees creating rainshadows on less prominent neighboring canopy
trees. It has been generally assumed that the vertically projected crown areas of canopy trees intercept the
same depth equivalent of precipitation. The results of our study suggest that rainforest canopy trees in the
cyclone-prone tropics do not receive the same depth equivalents of mean annual rainfall. We conclude that
the differential interception of rainfall may influence the spatial pattern of water availability and thus the
distribution of epiphytes in a forest canopy.
La geometria tri-dimensional y microhabitats de las copas del bosque pluvioso tropical.
REsUMEN. Muchas epifitas que crecen en las bovedas arboreas de los bosques tropicales humedos estim
adaptadas a la sequia debido a que las cortezas que las sostienen por encima del suelo ofrecen muy baja
capacidad de retencion de agua. La meta de nuestro estudio consistio en analizar si en estos bosques, las
bovedas de los arboles vecinos interceptaban en forma diferencial el agua de lluvia que cae en lIngulo para,
enseguida, evaluar los efectos que esto pudiese tener en las pautas de distribucion de las epifitas. En el
estudio se incluyolo siguiente: (1) e1 desarrollo de un modelo informatico en el que figura la representacion
tridimensional de la superficie de una boveda arbOrea en un bosque tropical humedo del noreste de
Queensland, Australia. Esto se hizo en base a datos sobre la elevacion de la corona arborea obtenidos
mediante fotogrametria, junto con un sistema de informacion geografica; (2) se genero la imagen del relieve
de una boveda ensombrecida a fin de ca1cular las areas efectivas de intercepcion de precipitacion pluvial
de las coronas de 50 arboles seleccionados en un evento de precipitacion; y (3) se llevo a cabo el analisis
de las relaciones existentes entre las superficies efectivas de las coronas de los arboles seleccionados y los
totales netos de precipitacion pluvial medidos bajo dichas coronas. Al establecerse la correlacion entre las
superficies efectivas de las coronas y los totales netos de precipitacion pluvial se observo que la lluvia que
caia en forma oblicua era interceptada en forma diferenciltl, dandose el caso que los arboles de bovedas
mas prominentes obraban como cortinas, impidiendo asi que la lluvia cayese sobre las cupulas de arboles
vecinos menos prominentes. Por cuanto hace a las bovedas arbOreas, en general, se supone que las superficies
de las coronas que se proyectan verticalmente, interceptan la misma profundidad de precipitacion equi
valente. Los resultados de nuestro estudio sugieren que las bovedas arboreas de los bosques tropica1es
hUmedos expuestos a huracanes no reciben en promedio anual el mismo equivalente de profundidad de
lluvia. Por 10 tanto, llegamos a la conclusion de que la intercepcion diferencial de precipitacion pluvial
puede influenciar las pautas espaciales del agua disponible y, con ello, la distribucion de las epifitas en la
boveda del bosque.
1952; Madison, 1977; Nadkarni, 1984; Gentry
INTRODUCTION
& Dodson, 1987). Factors affecting the distri
The substantial epiphyte loads that character bution and abundance of epiphytes in forest can
ize many tropical rainforest canopies contribute opies include differences in canopy tree archi
significantly to the ecosystem's biotic diversity tecture, microclimate, bark structure, and the
and many ecosystem-level processes (Richards, accumulation ofh umus on aboveground surfaces
62
1992] HERWITZ & SLYE: RAINFALL INTERCEPTION 63
(Sugden & Robins, 1979; Benzing, 1983; Nad been clear-cut, making State Forest No. 452 es
kami, 1986; Lawton, 1991). Wateris considered sentially a forest island. A rainfall interception
to be the resource in shortest supply for many plot occupying an area of 3 ha was established
epiphytes (Benzing, 1990), and so available within this stand. The plot was selected because
moisture may be a key factor that detennines the ofi ts accessibility and its occurrence on relatively
suitability of a canopy microsite for epiphytes. level ground. The elevation of the plot is about
Thus, factors affecting the interception and re 707 m.
tention of rainfall in tree crowns could affect epi Mean annual rainfall is > 1,400 mm, with more
phyte distribution patterns. than 60% of the annual total occurring during
Most tropical rainforests are characterized by the wet season months (January-March). FIGURE
a rough canopy surface due to the varied crown 2 shows the annual rainfall totals over the period
heights and structural differences of neighboring 1926-1990 at an Australian rain station located
canopy tress (Richards, 1952; Odum, 1970; Ash in Yungaburra, less than 2 km from the study
ton, 1978; Halle et al., 1978; Herwitz, 1982; Lloyd plot. Tropical depressions associated with the
& Marques, 1988). In cyclone-prone tropical southward shift of the intertropical convergence
regions such as the Caribbean, northeastern Cen zone provide an influx of moist, unstable tropical
tral America, the Philippines, and northeast air. These depressions often intensify into well
Queensland, Australia, rainforests are subject to developed low pressure cells over the Coral Sea,
a relatively high frequency of non-cyclonic (as subjecting northeast Queensland (including the
well as cyclonic) rainfall events associated with Atherton Tableland) to a high frequency of rain
high horizontal windspeeds (e.g., >400 km d-1). fall events associated with strong winds (Gentilli,
High windspeeds are responsible not only for a 1971). The prevailing wind direction during the
high incidence of treefalls, which increases the wet season in northeast Queensland is between
roughness of the overall canopy surface, but also azimuths of 45° and 180°, with extreme gusts on
for causing the rain to fall at angles inclined from the Atherton Tableland as high as 120 km hr-1
vertical. Depending on the position oft ree crowns (Gordon, 1971).
in relation to neighboring crowns, and the azi Forest structure in the plot bears a greater re
muth and angle ofthe inclined rain, canopy trees semblance to lowland tropical rainforests than
could differentially intercept inclined rainfall. to montane rainforests. The height of the upper
In regions characterized by wind-driven rain canopy is 25-35 m, and the mean trunk diameter
fall where the rain often does not fall vertically, at breast height (DBH) is > 35 cm. Both the den
canopy rainshadows would be expected among sity (620 trees ha-1) and the basal area (65 m2
neighboring canopy trees having different crown ha-1) of trees > 10 cm DBH are comparable to
elevations. Our study is a test of the hypothesis lowland rainforests in the Old and New World
that moisture availability varies significantly tropics (Herwitz, 1981). In a representative 0.5
among neighboring canopy trees in a cyclone ha sample area in the plot, 45 tree species were
prone tropical rainforest. We specifically exam represented.
ined the question of whether the vertically pro
jected crown area of a canopy tree is represen
tative ofi ts effective intercepting crown area under MATERIALS AND METHODS
inclined, wind-driven rainfall conditions. Interception and Net Rainfall
The architecture and geometry of tree crowns
and forest canopies are usually considered in re The standard method of measuring the inter
lation to the functional processes of light inter ception of rainfall in a forested ecosystem is by
ception, gas exchange, productivity, and heat and comparing the gross rainfall incident on the can
energy transfer (Oldeman, 1978; Miller & Stoner, opy with the net rainfall reaching the forest floor
1979; Hutchison et al., 1986; Baldocchi & Mey as throughfall and stemflow drainage (Leonard,
ers, 1988; Pinker & Holland, 1988; Shawet al., 1961; Jackson & Aldridge, 1973; Calder et al.,
1988). Few studies have considered the 3-D ge 1986; Clarke, 1987; Lloyd et al., 1988).
ometry of a forest canopy in relation to rainfall Throughfall and stemflow were collected from
interception processes. 50 canopy trees representing five species in the
plot. The five tree species were: Aleurites moluc
cana (L.) WiUd. (N = 10), Argyrodendron pera
STUDY AREA
latum (F. M. Bailey) H. L. Edlin ex I. H. Boas
The study site was located in a 258 ha stand (N = 10), Castanospermum australe A. Cunn. &
of tropical rainforest (Queensland State Forest C. Fraser ex Hook. (N = 10), Dysoxylum petti
No. 452) on the Atherton Tableland in northeast grewianum F. M. Bailey (N = 10), and Toona
Queensland, Australia (17°16'S, 145°35'E) australis (F. Muell.) Harms (N = 10). These five
(FIGURE 1). Much of the Atherton Tableland has species were selected because each species was
64 SELBYANA [Volume 13
o 5 10 km 25
Coral Sea
....:t
Atherton•
Yungaburra
Atherton Tableland
Interception Plot
FIGURE i. Location of the interception piot on the Atherton Tableland in northeast Queensland, Australia.
represented in the plot by at least ten canopy 0.53 m2 and its height above the ground was
trees having crowns fully exposed in the upper about 1 m. The gauges and troughs were posi
canopy. The DBH of the 50 selected trees ranged tioned beneath the 50 canopy trees in a pattern
between 42 and 122 em, with crown heights rang that extended outward 1 m from the base of each
ing between 24 and 39 m. tree to the perimeter of its crown. Stemflow was
Throughfall was collected from the 50 trees collected from the trunk of each tree using a
using 100 throughfall gauges and 20 throughfall stemflow collar consisting ofa high-quality hose
troughs. The orifice area of each gauge was 0.05 that was slit longitudinally and sealed to the trunk
m2 and its height above the ground was about with U-shaped staples and silicon sealant. The
0.5 m. The collection area of each trough was standard equation for calculating net rainfall from
1992] HERWITZ & SLYE: RAINFALL INTERCEPTION 65
3000r---.-,------.-,-----.,-------.,-------.-,------r-,----~1~
r
-
-
-E
E 2000r -
-
(ij
c
·cu
.... -
(ij
:J
cC 1000 '- -
<x::
-
1930 1940 1950 1960 1970 1980 1990
Year
FIGURE 2. Annual rainfall totals over the period 1926-1990, recorded at the Yungaburra rainfall station on
the Atherton Tableland.
individual trees is Rn = (TJ T.) + (SJc), where volume of intercepted water detained by each
Tv is the throughfall volume, T a is the collection sapling and LA is its total leaf area. The inter
area of the throughfall gauges, Sv is the stemflow ception storage per unit bark volume was deter
volume, and C is the tree's vertically projected mined by: (1) weighing bark samples from the
crown area (Horton, 1919). experimental trees, (2) measuring their volume
Gross rainfall was measured using a cluster of by rapid displacement of an aqueous solution,
15 rain gauges each with an orifice area of 0.05 and (3) saturating the bark samples in distilled
m2 positioned in a cleared area 280 m from the water.
edge of State Forest No. 452. Daily windspeed The total leaf surface area and bark volume of
totals and the prevailing daily wind directions each selected tree in the interception plot was
were recorded at the CSIRO Division of Forest calculated on the basis of: (I) field measurements
Research Meteorological Station in Atherton, 8 of trunk diameter, tree height, and leaflitterfall,
km west of the plot. (2) ground-based photography of crown coverage
and image analysis of obstructed sky, (3) field
Interception Storage Capacities counts of primary branches radiating from trees
of known trunk diameter, height, and crown cov
The rainfall interception storage capacities of erage, and (4) destructive sampling of the leaf
the 50 selected trees were determined by a com and woody surface area and the bark of fallen
bination of laboratory and field measurements primary branches.
described in detail by Herwitz (1982, 1985), and The total interception storage capacity of each
summarized briefly here. tree was the sum of its leaf and bark interception
Laboratory measurements established the in storages expressed as depth equivalents of water
terception storage per unit leaf area and per unit relative to the tree's projected crown area (Her
bark volume for each of the five species. The witz, 1985).
storage per unit leaf area was determined by: (1)
suspending saplings of the selected species on a
Crown Elevations
calibrated spring balance, (2) saturating their
aboveground leaf surfaces using a rainfall sim The crown elevation of each canopy tree (N >
ulator, (3) defoliating the experimental saplings, 500) in the plot was determined photogram
(4) measuring their total leafa rea using a LI- 3000 metrically. Color aerial photos of the plot were
Area Meter (LI-COR, Lincoln, Nebraska), and obtained from a Cessna 182 aircraft at a flight
(5) applying the equation IJLA where Iv is the height of670 m using a Hasselblad 70 mm cam-
66 SELBYANA [Volume 13
era, a 15 cm lens, and Kodak aerocolor negative boring trees. Increased effective crown areas were
film (Myers & Benson, 1981; Myers, 1982). En determined for individual trees by adding: (1)
larged prints of these negatives were used to pro the shaded crown areas of neighboring trees lo
duce a stereomodel of the forest canopy on a cated on the tree's leeward side, and (2) the area
Zeiss D2 Planimat. Horizontal and vertical con of its crown that was exposed on its windward
trols were established using maps prepared by side above less prominent neighboring crowns.
the Queensland Survey Office. After orienting,
levelling, and scaling the aerial photos on the Inclination Angle and Azimuth of
Planimat, the elevation of each canopy tree crown Incident Rainfall
in the stereomodel of the plot was measured in
meters above sea level. The result was a detailed The relationships between rainfall intensity,
crown elevation map. median raindrop size, terminal fall velocity of
raindrops, and horizontal windspeed have been
successfully used to calculate rainfall inclination
Three-Dimensional Model of
angles (Sharon, 1980). On the selected rainday
Canopy Surface
during the 1981 wet season (27 February 1981),
A three-dimensional model ofthe interception when the gross rainfall total was 32.5 mm and
plot canopy surface was created using ARC-INFO the wind run total was 645.3 km, the rainfall
software. ARC-INFO is a geographic informa inclination angle was calculated using the equa
tion system that can generate computer models tions presented below.
of 3-D surfaces by means of a triangulated ir Assuming a normal distribution of raindrop
regular network (TIN) data structure. TIN or sizes, the median droplet size in the population
ganizes x, y and z coordinates into a topological of raindrops was calculated on the basis of rain
network composed of triangles and interpolates fall intensity using the best-fit equation D = 2.23
a continuous 3-D surface that fits through the X pa.102, where D is the raindrop diameter in
vertex points of each triangle. The crown ele mm and P is the rainfall intensity in inch hr-I
vation map of the plot was used as input for the (Laws & Parsons, 1943). The terminal fall ve
TIN data base. The z values representing the locity of the rain was computed on the basis of
crown elevation points were entered manually. the median droplet size using a best-fit equation
The x and y coordinates corresponding to each derived from experimental data reported by Gunn
z value were entered using a digitizing tablet. and Kinzer (1949). The equation is in the form
Outline drawings of the perimeters of all the y = (3.378 x [In(x)]) + 4.213, where y is the
canopy tree crowns visible in the aerial photos terminal velocity in m secl and x is the droplet
of the plot were made on a transparency. These diameter in mm. Rainfall inclination was cal
crown perimeters were incorporated into the culated on the basis of the terminal fall velocity
ARC-INFO data base using an image scanning of the raindrops and the horizontal windspeed
device and the computer program SCITEX. The using the equation tan(b) = W/U where b is the
y,
crown perimeter and canopy surface images were angle of rainfall inclination in degrees from the
registered using ARC-INFO after identifYing ten vertical, W is the windspeed in m sec-1 and Uy
control points. is the terminal fall velocity in m sec-I. The az
imuth of the inclined rainfall was determined on
Effective Rainfall Intercepting the basis of the prevailing wind direction re
Crown Areas corded on the selected rainday.
The effective rainfall intercepting crown areas
RESULTS
of the 50 canopy trees and those canopy trees on
the leeward side of them were determined using The rainfall inclination angle at the plot on 27
a computer-based simulation procedure. We de February 1981 was 51.60 ± 100 from the vertical,
veloped a ray-tracing program that analyzed the with an azimuth of 180°. Net rainfall totals from
3-D model of the canopy surface in relation to the 50 selected canopy trees ranged between 24.2
the azimuth and angle of inclined rainfall on a and 61.8 mm (TABLE 1), or 74 to 190% of the
selected rainday characterized by high wind gross rainfall measured in the open. The coeffi
speeds. The 3-D model of the canopy was in the cient of variation of the net rainfall totals was
form of a raster image (cell size resolution = 21.7%. The coefficient of variation of the gross
0.3588 m2). The ray-tracing program generated rainfall measured in the open was 4.4%.
rainshadows (equivalent to sun shadows) over Spatial variation in net rainfall has been widely
the registered, contoured canopy surface and noted in forested ecosystems (e.g., Stout & Mc
computed the proportion of vertically projected Mahon, 1961; Czarnowski & Olszewski, 1970;
crown areas that were in the rainshadow of neigh- Kimmins, 1973; Calder et al., 1986; Lloyd &
1992] HERWITZ & SLYE: RAINFALL INTERCEPTION 67
TABLE 1. Net rainfall inputs from the 50 canopy trees in the study plot and their effective intercepting crown
areas on 27 Feb 81.
Standard
Mean deviation Maximum
Throughfall (mm) 33.3 7.16 60.9
Stemflow (mm) 0.32 0.22 0.99
Net rainfall, Rn (mm) 33.6 7.27 61.8
Interception storage + R" (mm) 35.2 7.18 62.7
Projected crown area, C (m2) 94.7 60.4 307.9
Effective crown area, C, (m2) 97.8 63.7 307.9
CjC ratio 1.03 0.18 1.68
Marques, 1988) and is often attributed to differ corrected gross rainfall, suggesting that the can
ences in the water-storage capacities of the opy trees did not intercept the same depth equiv
aboveground vegetative surfaces and the rate of alent of rainfall. We hypothesized that inclined
evaporation from these surfaces (e.g., Jackson, rainfall is differentially intercepted by neighbor
1975; Hancock & Crowther, 1979; Herwitz, 1987; ing canopy tree crowns.
Pook et aI., 1991; Teklehaimanot et aI., 1991). FIGURE 3 illustrates how the effective rainfall
Theoretically, some of the variance in net rainfall intercepting crown area (Ce) of an individual can
could be explained if one could correct for the opy tree may not equal its vertically projected
interception storage capacities of the trees from crown area (C). A taller, more prominent canopy
which the throughfall and stemfiow inputs were tree crown may have an effective intercepting
collected. crown area that exceeds its vertically projected
For the selected rainday, which followed sev crown area (C > C). Canopy trees in the rain
e
eral days of rainfall, the leaf interception storage shadow of (but not overgrown by) taller neigh
capacities were used to approximate the quantity boring trees may have effective crown areas that
of incident rainfall retained by the aboveground are less than their projected crown areas (C <
e
surfaces ofe ach tree. Adding the leaf interception C).
storage capacity of each tree to its net rainfall The differential interception of inclined rain
total yielded depth equivalents of water between fall by neighboring canopy trees is distinct from
24.8 and 62.7 mm (TABLE 1), with a coefficient
of variation of 20.5%. This minor reduction in
the coefficient of variation indicated that the in
terception storage capacities of the 50 sampled
trees do not explain all of the inter-tree differ
ences in net rainfall.
The finding that 27 of the 50 sampled trees
had net rainfall totals exceeding the gross rainfall
measured in the open cannot be explained by
differences in their aboveground interception
storage capacities. This finding may be attributed
to: (1) an underestimate of the incident gross
rainfall measured in the cleared area, and/or (2)
canopy rainshadows resulting in the differential
interception of the inclined rainfall.
If one assumes that the effective rainfall col
lecting area of a rain gauge is always equal to the
vertically projected area of its orifice, then gross
rainfall inputs may be underestimated. The de
crease in the effective orifice area of a rain gauge
as a function of rainfall inclination angle can be
corrected by a basic trigonometric model (Shar
on, 1980). Following this correction, the depth
FIGURE 3. Illustration of the differential intercep
equivalent of gross rainfall in the open was re tion of inclined rainfall by two neighboring canopy
calculated to have been 41.3 mm. However, even trees (tree A and tree B) showing how their effective
after employing this correction, six of the 50 sam rainfall intercepting crown areas (Co) do not equal to
pled trees had net rainfall totals exceeding the their vertically projected crown areas (C).
68 SELBYANA [Volume 13
TABLE 2. Effective rainfall intercepting crown areas of two neighboring canopy trees in the study plot shown
in relation to their projected crown areas, demonstrating how canopy trees may intercept different depth
equivalents of rainwater from the same rainfall event.
Tree A' Tree Bb
Gross rainfall, P (mm) 41.3 41.3
g
Vertically projected crown area, C (m2) 158 67
Effective crown area, C (m2) 139 86
e
Volume of intercepted rainwater, Pg x Ce (liter) 5,741 3,552
Depth equivalent of intercepted water, P x C/C (mm) 36.3 53.0
g
'Tree #47, Aleurites moluccana, positioned on the leeward side of tree #29.
Tree #29, Argyrodendron peralatum.
b
the vertical obstruction of subcanopy trees. Ifthe boring crowns in determining the C/C ratio and
rain fell vertically, then the exposed canopy tree the extent of differential interception. A canopy
crowns would all intercept the same depth equiv tree may be tall, but if it is surrounded by tree
alent of water. The differential interception of crowns of comparable elevation, its effective
inclined rainfall means that neighboring canopy rainfall intercepting crown area (Ce) will not ex
trees intercept different depth equivalents from ceed its vertically projected crown area (C).
the same spatially-isotropic rainfall event. The
reason for this inter-tree difference is because the
DISCUSSION
depth equivalent intercepted by an individual
tree is calculated as the volume of intercepted Inclined rainfall has been previously ex
water divided by the tree's projected crown area. amined in temperate environments. Hayes and
Eleven of the experimental canopy trees had Kittredge (1949) and Leonard and Reinhart
shaded crowns (C < C), 17 had increased effec (1963) evaluated the accuracy of rain gauges un
e
tive crown areas (C > C), and 22 were unaffected der conditions of inclined rainfall in various lo
e
by neighboring canopy tree crowns (Ce = C). cations in the USA. Aldridge (1975) used an ar
Multiplying the corrected gross rainfall value of ray of tilted rain gauges in New Zealand and
41.3 mm by the C/C ratio of each of the 50 found that inclined rainfall contributed> 90% of
selected trees resulted in depth equivalents of the total rainfall over a three-year period, with
intercepted rainfall as high as 69.4 mm. TABLE inclination angles exceeding 50% from the ver
2 is an example contrasting the depth equivalents tical. In an earlier study, Aldridge and Jackson
of rainwater intercepted by two neighboring can (1973) noted that stemflow quantities in a forest
opy trees, with the more prominent tree crown stand of Notho/agus truncata in the same part of
having intercepted 16.7 mm more rainfall than New Zealand were greater when the rain events
the tree crown on its leeward side. were accompanied by strong wind, suggesting that
The C/C ratios computed using the ray-trac stemflow inputs may be related to the intercep
ing program and the ARC-INFO generated mod tion of inclined rainfall.
el of the forest canopy did not explain all of the Ford and Deans (1978) were more precise in
inter-tree variation in net rainfall. Other factors their description of the interception of inclined
that could have contributed to the remaining rainfall by a 14-year-old Sitka spruce plantation
unexplained variation include inter-tree differ in Scotland. They reasoned that when rain falls
ences in intercepted rainwater retention under at an angle, the leading shoots at the top of a
turbulent wind conditions and in rates of evap canopy tree present a greater intercepting area
oration. Further refinements are currently being than its vertically projected crown area. The dif
made in the computer model of the canopy sur ferential interception of inclined rainfall thus may
face. Nevertheless, the statistical analysis re contribute to the spatial patterning of throughf all
vealed a significant correlation (P < 0.02) be inputs.
tween the observed net rainfall totals and the None of these studies, however, provided a
corresponding C/C ratios. The correlation coef means of quantifying the differential interception
ficient increased slightly when the C/C ratios of inclined rainfall by individual canopy trees.
were considered in relation to the sum of each The interception ofw ind-driven rainfall by forest
tree's net rainfall input and its interception stor canopies is a complex process that is difficult to
age capacity. The lack of a significant correlation measure under field conditions. Aboveground
between the net rainfall totals and the elevations vegetative surfaces form a complicated lower
of the tree crowns (P > 0.1) demonstrated the boundary for atmospheric flow that involves
importance of crown position relative to neigh- aerodynamic drag and turbulent eddies (Tajch-
1992] HERWITZ & SLY E: RAINFALL INTERCEPTION 69
FIGURE 4. Computer-generated image showing the three-dimensional geometry of the interception plot
canopy surface.
man, 1981; Baldocchi & Meyers, 1988; Shaw et in canopy rainshadows supported by tree crowns
al., 1988). However, our finding of a significant that lack significant accumulations of humus.
correlation between field measurements of net We hypothesize that in regions where wind is
rainfall and a computer-based approximation of often associated with rainfall events and wind
effective rainfall intercepting crown areas (Co) in direction is relatively consistent (as in northeast
dicates that inclined rainfall was differentially Queensland), quantification ofthe differential in
intercepted by the rough canopy surface of the terception of the inclined rainfall could assist in
interception plot (FIGURE 4). identifying and distinguishing: (1) prominent
When high windspeeds generate inclined rain canopy trees that lack epiphytes because large
fall, it is not reasonable to assume: (1) that the quantities of water are transmitted down their
effective rainfall intercepting area of an individ branches and trunks, hindering the establish
ual canopy tree crown is equal to its vertically ment of epiphyte propagules; (2) epiphytes that
projected crown area, and (2) that all neighboring can withstand differentially intercepted, high-in
canopy tree crowns receive the same depth tensity rainfall that drains rapidly from the host
equivalent of gross rainfalL Our study demon tree's woody surfaces; (3) relatively xerophytic
strated the importance of recognizing that neigh epiphytes that are located in canopy rainshadows
boring canopy tree crowns may not receive the and are adapted to drier (and perhaps more oli
same mean annual depth equivalent of precipi gotrophic) conditions; and (4) more me sophyt ic
tation, especially in the cyclone-prone tropics. epiphytes located in the interior parts of prom
The differential interception of rainfall could inent canopy tree crowns. We recommend that
have several possible effects on the distribution studies of the distribution and ecology of epi
of epiphytes in a forest canopy. Increased mois phytes in tropical rainforests direct attention to
ture availability in the crown of a prominent the influence of canopy roughness on rainfall in
canopy tree could result in microsites particu terception processes and moisture availability in
larly favorable for the germination and growth the crowns of canopy trees.
of epiphytes. On the other hand, if prominent
canopy trees were subject to frequent high-in
ACKNOWLEDGMENTS
tensity rainfall, differential interception could re
sult in poor opportunities for rapid colonization, Financial support for this study was provided
with rapid drainage from their branches and by the National Science Foundation (Grant No.
trunks, which could dislodge epiphyte propa SES-90 13329), the Australian-American Edu
gules before they become anchored (Benzing, cational Foundation, and The Australian Na
1990). In canopy rain shadows where the rainfall tional University. We acknowledge the assis
totals are lower and the availability of moisture tance of D. Peterson at the Ecosystem Science
(and possibly nutrients) may be more limited, and Technology Branch of the NASNA mes Re
the epiphytes that thrive may be more xerophytic search Center, Moffett Field, California; G. The
and may have a higher nutrient-use efficiency. lin, D. Aitken, R. Heimes, and R. Lugo of the
Xerophytic epiphytes characterized by a high nu U.S. Geological Survey, Menlo Park, California;
trient-use efficiency would most likely be found B. Myers of the CSIRO Division of Forest Re-
70 SELBYANA [Volume 13
search in Canberra, ACT, Australia; G. Stocker, HERWITZ, S. R. 1981. Regeneration of selected trop
and T. Irvine of the CSIRO Division of Forest ical tree species in Corcovado National Park, Cos
Research in Atherton, Queensland, Australia; and ta Rica. University of California Publications in
Geography, Volume 24. University of Caiifornia
J. Chappell, J. Neale, D. Walker, and J. Jennings
Press, Berkeley, California.
of The Australian National University, and W.
---. 1982. Tropical rainforest influences on rain
Wassermann and G. Schmid.
water flux. Ph.D. dissertation, The Australian Na
tional University, Canberra, Australia.
---. 1985. Interception storage capacities oftrop
LITERATURE CITED
ical rainforest canopy trees. J. Hydro!. 77: 237-
ALDRIDGE, R. 1975. The resultant direction and in 252.
clination of rainfall at Arahura, Wairarapa, New ---. 1987. Raindrop impact and water flow on
Zealand. J. Hydrol. (N.Z.) 14: 55-63. the vegetative surfaces of trees and the effects on
---ANDR.J.JACKSON. 1973. Interceptionofrain stemflow and throughfall generation. Earth Surf.
fall by hard beech (Nothofagus truncata) at Taita, Process. Landforms 12: 425-432.
New Zealand, N.Z. J. Sci. 16: 185-198. HORTON, R. 1919. Rainfall interception. Monthly
AsHTON, P. S. 1978. Crown characteristics of tropical Weather Review 47: 603-623.
trees. Pp. 591-615 in P. B. TOMLINSON AND M. H. HUTClllSON, B. A., D. R. MATT, R. T. McMILLEN, L.
ZIMMERMANN, eds., Tropical trees as living sys J. GROSS, S. J. TAICHMAN, AND J. M. NORMAN.
tems, Cambridge University Press, Cambridge. 1986. The architecture of a deciduous forest can
BALDOCClll, D. D. AND T. P. MEYERS. 1988. Tur opy in eastern Tennessee, U.S.A. J. Ecol. 74: 635-
bulence structure in a deciduous forest. Boundary 646.
Layer Meteorol. 43: 345-364. JACKSON, I. J. 1975. Relationships between rainfall
BENZING, D. H. 1983. Bark surfaces and the origin parameters and interception by tropical forest. J.
and maintenance of diversity among angiosperm Hydrol. 24: 215-238.
epiphytes: an hypothesis. Selbyana 5: 248-255. JACKSON, R. J. AND R. ALDRIDGE. 1973. Interception
--'- . 1990. Vascular epiphytes: general biology and of rainfall by kamahi (Weinmannia racemosa) at
related biota. Cambridge University Press, Cam Taita, New Zealand. N.Z. J. Sci. 16: 573-590.
bridge. 354 pp. KIMMINS,J. P. 1973. Some statistical aspects of sam
CALDER, I. R., I. R. WRIGHT, AND D. MURDIYARSO. pIing throughfall precipitation in nutrient cycling
1986. A study of evaporation from tropical rain studies in British Columbian coastal forests. Ecol
forest-West Java. J. Hydrol. 89: 13-31. ogy 54: 1008-1019.
CLARKE, R. T. 1987. The interception process in trop LAWS, J. O. AND D. A. PARSONS. 1943. The relation
ical rain forests: a literature review and critique. ofr aindrop-size to intensity. Trans. Am. Geophys.
Acta Amazonica 16/17: 225-238. Un. 2: 452-460.
CZARNOWSKI, M. S. AND J. L. OLSZEWSKI. 1970. LAWTON, R. O. 1992. Facultative epiphytism in a
Number and spacing of rainfall-gauges in a decid neotropical cloud forest and the evolution of the
uous forest stand. Oikos 21: 48-51. epiphytic habit. Se1byana 13: 156-157.
FORD, E. D. AND J. D. DEANS. 1978. The effects of LEONARD, R. E. 1961. Net precipitation in a northern
canopy structure on stemflow, throughfall and in hardwood forest. J. Geophys. Res. 66: 2417-2421.
terception loss in a young Sitka spruce plantation. ---AND K. G. REINHART. 1963. Some observa
J. Appl. Ecol. 15: 905-917. tions on precipitation measurement on forested
GENTILU, J. 1971. Dynamics of the Australian tro experimental watersheds. U.S. Forest Service Re
posphere. pp. 53-117 in Climates of Australia and search Note NE-6.
New Zealand, World Survey Of Climatology, Vol LLOYD, C. R. AND A. DE O. MARQUES. 1988. Spatial
ume 13. Elsevier, New York. variability of throughfall and stemflow measure
GENTRY, A. H. AND C. H. DODSON. 1987. Diversity ments in Amazonian rainforest. Agric. For. Me
and biogeography of neotropical vascular epi teorol. 42: 63-73.
phytes. Ann. Missouri Bot. Gard. 74: 205-233. ---, J. H. C. GASH, W. J. SHUTTLEWORTH, AND A.
GoRDON, B. 1971. Climatic Survey Northern Region DE O. MARQUES. 1988. The measurement and
16-Queensland. Bureau of Meteorology, De modelling of rainfall interception by Amazonian
partment of the Interior, Commonwealth of Aus rain forest. Agric. For. Meteorol. 43: 277-294.
tralia. MADISON, M. 1977. Vascular epiphytes: their system
GUNN, R. AND G. D. KINZER. 1949. The terminal atic occurrence and salient features. Selbyana 2:
velocity of fall for water droplets in stagnant air. 1-13.
J. Meteorol. 6: 243-248. MILLER, P. C. AND W. A. STONER. 1979. Canopy
HALLE, F., R. A. A. OLDEMAN, AND P. B. TOMLINSON. structure and environmental interactions. Pp. 428-
1978. Tropical trees and forests: an architectural 458 in O. T. SOLBRIG, S. JAIN, G. B. JOHNSON, AND
analysis. Springer-Verlag, New York. 441 pp. P. H. RAVEN, eds., Topics in plant population bi
HANcocK,N. H. ANDJ. M. CROWTHER. 1979. A tech ology. Columbia University Press, New York.
nique for the direct measurement of water storage MYERS, B. J. 1982. Large-scale color aerial photo
on a forest canopy. J. Hydrol. 41: 105-122. graphs-a useful tool for tropical biologists. Bio
HAYES, G. L. AND J. KITTREDGE. 1949. Comparative tropica 14: 156-157.
rain measurements and rain-gage performances on ---AND M. L. BENSON. 1981. Rainforest species
a steep slope adjacent to a pine stand. Trans. Am. on large-scale color photos. Photogramm. Eng.
Geophys. Un. 30: 295-301. Remote Sens. 47: 505-513.