Trees Lose Their Leaves Later in Agroforestry Systems
ISP paper
1. How Does Canopy Coverage Influence the
Composition and Distribution of Understory
Vegetation?
Julianna Kurpis
SIT Tanzania: Wildlife Conservation and Political Ecology
Fall 2014
3. 3
Acknowledgements
I would like to thank Sokoine University for allowing us to stay at Mazumbai and do
research. Also thank you to David and Richard, you guys are amazing cooks, and your banana
bread is out of this world.
I would also like to thank my guide, Babu Francis. Thank you for chopping through all
that thick vegetation and identifying all the plants. I could not have done this project without
you.
Thank you Baba Jack for helping me with the statistics and explaining ANOVA to me. I
would have been so lost otherwise!
I would also like to thank my parents for funding my trip to Tanzania.
4. 4
Abstract
This study was conducted in Mazumbai Forest Reserve from November 8 to November
23. I investigated how different levels of canopy cover (0-33%, 33-67%, and 67-100%) affect the
composition and distribution of understory vegetation. Percent vegetation coverage for grasses,
herbs, shrubs and trees were recorded, as well as the number and types of species, and the
number of trees and their height within each plot. I predicted that plant covered by 33-67%
canopy would have the greatest amount of species and vegetation coverage, and plant
communities covered by 67-100% canopy would have the least amount of species and vegetation
coverage. I found that plots within 0-33% canopy had the most variety of species whereas plots
within 33-67% canopy had the least amount of species. There was no significant difference
among the vegetation cover for grasses, herbs, and shrubs for all three categories of canopy
cover, although there was a significant difference with trees. The greatest percentage of trees was
found under the greatest amount of canopy (67-100%), whereas the smallest percentage of trees
was found under the least amount of canopy (0-33%). Overall I found that canopy coverage
affects the composition of understory vegetation, but it does not affect plant community structure
or vegetation distribution.
5. 5
Introduction
A tropical rainforest is an ecosystem that occurs within 28 degrees north or south of the
equator. It experiences high average temperatures, usually a monthly mean exceeding 18°C year
round, and a significant amount of rainfall, averaging between 69 and 79 inches per year.
Because of the high levels of precipitation, tropical rainforests often have soils with poor
nutrients. (Denslow1987).
There are high levels of biodiversity in the rainforest, and rainforests are home to half of
all the living animal and plant species on the planet. “A single hectare of rainforest may contain
42,000 different species of insect, up to 807 trees of 313 species and 1,500 species of higher
plants” (Denslow 1987). Plant alpha diversity is greater in equatorial rain forests than in any
other vegetation type.
Tropical rainforests are among the most threatened ecosystems globally due to human
activity. Logging and agricultural clearance have caused habitat destruction, which is a major
cause of species extinction, and the areas covered by rainforest is rapidly shrinking. (Denslow
1987). Thankfully, there has been an increase in protected areas as public awareness has
increased, especially in the past two decades.
Tropical rainforest species are differentiated by their growth responses to shaded
understory and gap conditions. Species can be divided into the following categories: pioneer
species, small gap specialists, and shade tolerant species (Popma 1988). Pioneer species are
shade intolerant and can only grow in large gaps. Small gap specialists are shade intolerant
during some stages of development, but also need gaps to reach maturity. Shade tolerants are
able to grow under deep shade.
Tropical rainforests can be divided into the following layers: emergent layer, canopy,
understory, and forest floor. Canopy can be defined as the above ground portion of a plant
community formed by mature tree crowns (Campbell 1990). Canopy structure can be assessed by
leaf area index (LAI), which is leaf area per unit ground area. Canopy cover can also be assessed
visually, measuring how much the forest canopy obscures the sky, using either a percentage or
ordinal scale (Jennings 1999). This method was used in my study. Another way to measure
canopy cover is to take aerial photographs and make estimates using a crown density scale.
The canopy is important because it protects understory vegetation from strong winds and
storms, although it intercepts sunlight and precipitation. Because it influences local precipitation,
air movements, and the distribution of light, the canopy largely determines the air humidity,
temperature, and moisture conditions of a given area (Jennings 1999). The canopy of the
rainforest is usually 10m thick, intercepting about 95% of sunlight.
The emergent layer consists of a small number of emergent trees that grow above the
general canopy, usually ranging from 30 to 76 meters. These trees have waxy leaves and are able
to withstand the hot temperatures and strong winds that occur well above the canopy.
In forest ecology, the understory is defined as the plant life that grows beneath the canopy
without penetrating it. The forest floor consists of the following vegetation types: grasses (less
6. 6
than 30 cm), herbs (between 30 cm and 1 meter), shrubs (between 1 and 5 meters), and trees
(greater than 5 meters). Young understory trees often persist for decades as suppressed juveniles
until there is opening in the forest canopy. Understory shrubs complete their life cycles in the
shade of the forest canopy. Forest understories receive less light than plants in the canopy, and
therefore understory plants must be shade tolerant (Kramer 2004) Trees in tropical rainforests are
usually both shade tolerant and gap dependent.
Openings in the canopy are important for the establishment and growth of many
rainforest species, especially trees (Denslow 1989). Without these openings, many plants would
never be able to germinate or grow past sapling size. Trees that die standing often produce small
gaps in the canopy over an extended period of time, but when an emergent tree falls, a gap in the
canopy is formed that is sufficiently large enough for light demanding species. Root and shade
competition from trees forming the canopy layer suppresses understory plants, so when an
emergent tree falls, there is rapid growth among understory plants (Jennings 1999).
Studies have shown that canopy and understory species, from shade tolerant to light
demanding physiologies, have greater growth, survival, and reproduction when they occur near
canopy openings. Part of the reason for this is the sudden increase and intensity of direct sunlight
to the lower strata of the forest. There is also a slight increase in soil nutrients and temperature.
Community composition under an opening in the canopy is usually determined both by
chance, and by interactions among saplings competing for resources (Denslow 1989). The
presence of seedlings already in the place where the gap is formed gives them an advantage, and
this advance regeneration is important for allowing the seedlings to survive within the first few
years. Seed mortality is much lower under gaps as opposed to intact canopy. Although some
species will only grow in large canopy openings, most species in tropical rainforests have some
degree of shade tolerance and also respond positively to canopy openings.
With the growth of vegetation under canopy openings, eventually light availability
declines so that it is no longer sufficient for the germination of light demanding seeds, and
sapling stem densities begin to decline. Studies have shown that temperature and humidity
usually return to pre-gap levels within two years of tree fall, and forest turnover begins.
It is important to understand how different amounts of sunlight affect plant growth
because shade tolerance is critical to determining plant survival in tropical rainforests (Popma
1988). My study investigates how plant communities change under different levels of canopy
cover in Mazumbai Forest Reserve. How much light is able to penetrate the canopy, and how
does this affect the different types of species that are able to grow, the distribution of understory
vegetation? Percent vegetation coverage, the number and types of species, and the number of
trees and their height within each plot will be recorded, representing the composition and
structure of the plant community.
I predict that there will be more variation in species, and the highest vegetation coverage,
under the canopies of emergent trees that cover approximately 33-67% of the plot, because this
provides an ideal habitat for plants that need varying degrees of shade and sun. I also predict that
I will find the least variation in species, and the least amount of vegetation coverage, under the
canopies of emergent trees that cover approximately 67-100% of the plot, because it will be
harder for plants to thrive in an area with very little sunlight. My null hypothesis is that there will
7. 7
be no change in the composition and distribution of vegetation under different levels of canopy
cover.
8. 8
Study Site
Mazumbai Forest Reserve is located in the West Usambara Mountains in Northern
Tanzania. The Usambaras are part of the Eastern Arc Mountains, stretching from Kenya to
Tanzania (Appendix A). The West Usambara Mountains are approximately 90 km long and
range from 600 to 2300 meters above sea level, and they were created during the tertiary tectonic
uplift about two billion years ago that formed the East African Rift (Minderhound 2011). The
soil in the Usambaras is largely derived from ancient granite bedrock because the mountains
were created only by tectonic events.
The Usambara Mountains were originally inhabited by the Bantu, Shambaa, and Massai
people, who were a mix of agriculturalists and pastoralists. The West Usambara mountains are
located near the Indian Ocean. During the late 1800s, many people fled coastal regions and
sought refuge in the Usamabara Mountains as the slave trade intensified, and the sudden increase
in population increased pressure on the land (Conte 2004). When the slave trade died down,
German colonization began, altering the way the land was used.
Historically, farming techniques consisted of agroforestry and land rotation. The land was
cultivated for about two years, and then remained unused for a short period of time to allow the
soil to rest. Under the Germans, however, cash crops plantations of coffee and tea were spread
across the area, using most of the available land and creating land scarcity for local farmers
(Vigiak 2005). Historical farming techniques were abandoned, and cultivation was increased on
steep slopes and on forest edges.
In the 1940’s, concerns about social erosion grew, but soil conservation efforts were
abandoned with the resistance of local villagers. In 1970, after Tanzania’s independence, land
degradation became too severe to ignore, and the government began several conservation
programs.
Today, the population of the Usambaras has one of the highest growth rates (about 4%
compared to the national average of 2.1%), The West Usambara Mountains are highly populated,
with 215 villages distributed around government reserves and protected areas. Most of the
inhabitants are subsistence farmers who rely heavily on the forests around them for timber,
medicinal plants, clearing for agriculture, and fuel wood. Sadly, 70% of the original forest cover
of the Usambaras has been lost, and major land and forest degradation remain a pressing issue
(Conte 2004).
I decided to do my study in Mazumbai because this forest reserve is considered to be a
pristine rainforest, and the heart of it has remained unchanged for the last 30 million years. It is a
protected area of land with very little human disturbance, making it an ideal place to study plant
communities and canopy coverage.
9. 9
Methods
This study took place between November 8 and November 23 in Mazumbai Forest
Reserve, located in the West Usambara Mountains of Northeastern Tanzania. My goal was to set
up three plots 10x10 m each day, chosen non-randomly with opportunistic sampling. I focused
my study at the altitudinal band of 1500-1600m. Metadata for each plot was recorded,
specifically taking into account litter depth and slope.
Levels of canopy cover were determined by estimating the overall crown cover of
emergent trees, which were divided into three categories: 0-33%, 33-67%, and 67-100%. To do
this, I stood in the center of each plot and estimated how much sky was obscured by canopy.
Inside each plot, the percent coverage of vegetation was estimated and recorded for the
following: plants under 30 cm (grasses), plants between 30 cm and 1 meter (herbs), plants
between 1 and 5 meters (shrubs), and plants greater than 5 meters (trees). All the species and
their percent coverage within the plot were recorded. Height was also recorded for each tree, and
even emergent trees outside of the plot that contributed to the canopy inside of the plot had their
species and height recorded.
Plants were identified with the help of Babu Frances, who was my forest guide. I
analyzed my data using ANOVA and descriptive statistics.
10. 10
Results
I recorded percent vegetation coverage for grasses, herbs, shrubs, and trees within each
plot, and I averaged these values for each category of canopy coverage. I also averaged the
number of species per plot, and then averaged these values for each category of canopy coverage
(figure 1.1) After running ANOVA, I found that there was only a significant difference among
percent coverage of trees, between 0-33% and 67-100% canopy cover. Therefore, I decided to
focus most of my results on trees.
I also recorded types of species and their percent coverage within each plot. I took only
species of grasses, herbs, and shrubs that covered at least 50% of the plot into account, since I
had a lot of different species per plot, and the data would have otherwise been overwhelming.
According to figures 1.1, 1.2, and 1.3, there is more variation in species in plots covered by 0-
33% canopy, and the least amount of variation in plots covered by 67-100% canopy cover.
Focusing now on trees, 0-33% canopy coverage had on average 10 species of trees (range
0-23). 33-67% canopy coverage had on average 24.5 species of trees (range 3-136). 67-100%
canopy coverage had on average 20.6 species of trees (range 7-48). There was no significant
difference among the different species (p=0.2651; p>0.05)
Tree heights were grouped into four categories: 5-10m, 10-20m, 20-40m, and 40-60m. I
compared tree heights within each category of canopy coverage (table 1.2) After running
ANOVA, all values were significant. I then compared tree heights among different categories of
canopy coverage (table 1.3) I found that none of the values were significant.
Because there was a significant difference among trees of different heights within each
category of canopy coverage, I graphed the species found within each grouping of tree height
(figures 1.4, 1.5, and 1.6) I only graphed species that occurred in at least two plots, because the
graph would have been overwhelming otherwise, and this allows us to see the most common tree
species found among tree heights. All three categories of canopy coverage were dominated by
trees 5-10 m. There weren’t any tree species 40-60 m common in at least two plots for 0-33%
canopy coverage. Trees 5-10m under 67-100% canopy cover were dominated by msikisiki trees.
Kampindi was found in almost every grouping of tree height and in every category of canopy
coverage.
11. 11
Table 1.1 Comparingaverage percent vegetation cover and average number of species usingANOVA. There was no
significantdifferencebetween percent cover of grasses (p=0.8027;p>0.05), herbs (p=0.3272; p>0.05), shrubs
(p=0.09363; p>0.05) or number of species (p=0.6690; p>0.05). There was a significantdifferencebetween percent
cover of trees, but only between 67-100%and 0-33%canopy cover (p=0.0015053;p<0.05).
Figure 1.1 Average percent vegetation cover for plants (grasses,herbs,and shrubs) thatmade up more than 50%
of the plot under 0-33% canopy.
mshia
4%
tungashianga
4%
tikini
14%
jenny
18%
shiu
9%
tuanangay
4%
mtambagoshwa
9%
tuhutu
4%
shikize
5%
kampindi
5%
kokatongo
9%
gugufa
5%
konyosa
5%
mso
5%
0-33% Canopy Cover
Canopy % Grasses % Herbs % Shrubs % Trees # Species
0-33% 43.8 (0-95%) 58.4 (30-85%) 62.6 (5-90%) 5.78 (0-20%) 24.8 (14-35%)
33-67% 51.8 (10-85%) 65.9 (20-90%) 41.8 (10-85%) 15.7 (3-70%) 23.6 (15-32%)
67-100% 48.8 (10-90%) 52.9 (30-85%) 42.1 (10-90%) 29.8 (5-60%) 23 (19-30%)
12. 12
Figure 1.2 Average percent vegetation coverage for plants (grasses,herbs,and shrubs) thatmade up more than
50% of the plot under 33-67% canopy.
Figure 1.3 Average percent vegetation cover for plants (grasses,herbs,and shrubs) thatmade up more than 50%
of the plot under 67-100% canopy.
tikini
50%
wazuzewakoko
12%
vimbamazwe
12%
mbangwe
13%
shikizi
13%
33-67% Canopy Cover
shikize
9%
mso
9%
ozuzewakoko
19%
konyosa
9%
kokatondo
18%
tikini
18%
kimwe
9%
kiandama
9%
67-100% Canopy Cover
13. 13
Trees
0-33% canopy
(m)
33-67% canopy
(m)
67-100%
canopy (m)
Tree heights P value Tree heights P value Tree heights P value
5-10 vs 10-20 0.001 5-10 vs 10-20 0.001 5-10 vs 10-20 0.001
5-10 vs 20-40 0.001 5-10 vs 20-40 0.001 5-10 vs 20-40 0.001
5-10 vs 40-60 0.001 5-10 vs 40-60 0.001 5-10 vs 40-60 0.001
10-20 vs 20-40 0.001 10-20 vs 20-40 0.001 10-20 vs 20-40 0.001
10-20 vs 40-60 0.001 10-20 vs 40-60 0.001 10-20 vs 40-60 0.001
20-40 vs 40-60 0.001 20-40 vs 40-60 0.001 20-40 vs 40-60 0.001
Table 1.2 Comparingtree heights within each category of canopy coverage usingANOVA. All values aresignificant
(p<0.05).
Canopy % 5-10 m 10-20 m 20-40 m 40-60 m
0-33 vs 33-67 0.438 0.4 0.841 0.097
0-33 vs 67-100 0.19 0.143 0.07 0.252
33-67 vs 67-100 0.874 0.884 0.235 0.523
Table 1.3 Comparingtree heights among categories of canopy cover usingANOVA. All values areinsignificant
(p>0.05).
15. 15
Figure 1.6 Comparingtree species among different heights occurringin atleasttwo plots within 67-100% canopy
cover.
0
10
20
30
40
50
60
70
kampindi
kwate
mkokoko
msikisiki
mkuguma
gwau
kihembile
kikwande
mbamgwe
mcande
mchembechembe
mgoymaze
miasa
mpandakiga
msacasua
mshiwe
mso
mtiwamba
mtonhe
mungu
nyasa
pigamcoffee
sangana
#Trees Comparing Trees 67-100% Canopy Cover
5--10
10--20
20-40
40-60
16. 16
Discussion
I hypothesized that I would find the most variation in plant species, and the most
vegetation percent coverage, in plots that were covered by 33-67% canopy cover. My hypothesis
was incorrect, so we cannot reject the null hypothesis; my results showed that plots covered by
0-33% canopy cover had the most variation in plant species (figure 1.1). Usually the plots I
sampled within the 0-33% canopy cover range had dead trees either within the plot or near it-
suggesting that the vegetation under these gaps was relatively new. These decomposing trees
also increase the nutrient content of the soil. The sudden increase in sunlight stimulated new
plants to grow, explaining why many of my plots had thick vegetation of herbs and shrubs. These
plots had the greatest variation of species, made up of mostly recent vegetation, also explaining
how they had the least amount of trees, which need a longer period of time to mature.
I was surprised that plots covered by 33-67% canopy cover had the least variation in
species (figure 1.2). I had hypothesized that I would see this for plots covered by 67-100%
canopy (figure 1.3). “In large gaps, the growth of existing or newly established shrubs and herbs
rapidly reduce light availability at seedling levels. In small gaps, light levels might change just a
little. These light levels are not sufficient to promote high growth rates.” (Denslow 1998) Plots
covered by 33-67% canopy usually consisted of many smaller gaps in the canopy, whereas plots
covered by 0-33% canopy usually consisted of a very large gap.
Perhaps plots covered by 33-67% canopy did not let in enough light for pioneer species
to grow, while also increasing the mortality rate for shade tolerant species. Canopy closure
occurs relatively quickly, especially when small gaps are closed by lateral canopy growth.
Perhaps this sudden and quick competition for light was ruthless enough that many species died
out.
Could it be lack of sunlight that discourages plant growth in plots covered by 33-67%
canopy, or just not enough trees? Weltzin (1990) hypothesizes that areas under canopies of trees
and shrubs sometimes support dense herbaceous vegetation relative to open areas. He claims that
trees facilitate understory plant growth through increased nutrient availability, because trees act
as a nutrient pump by taking up nutrients from deeper soil layers, or from soil outside the
canopy, and depositing them under their canopy through litter fall or leaching.
Perhaps I found the least variation in species in plots covered by 33-67% canopy because
there is not enough sunlight or tree cover to support many different species of plants. Plots
covered by 0-33% canopy have plenty of sunlight to encourage plant growth, whereas plots
covered by 67-100% canopy have plenty of trees to create a more nutritious soil.
It is puzzling how there was no significant difference between the percentage of
vegetation cover for grasses, herbs, and shrubs among the different levels of canopy cover, but
17. 17
there was a significant difference for the percentage coverage of trees (table 1.1). Not much
research has been done on tropical rainforests, so reasons for this finding remain unknown. There
are many factors that could account for the vegetation cover in each plot. Just to name a few,
vegetation cover is affected by the size of the gap in the canopy (and how much light is let
through), soil content, heights of surrounding trees, and topography.
The greatest percentage of trees was found under the greatest amount of canopy (67-
100%), whereas the smallest percentage of trees was found under the least amount of canopy (0-
33%), and a significant difference was found between them (table 1.1). When large gaps are
formed in the canopy, tree growth is suppressed by competing vegetation and vines (Richards
1996). This would explain why I had so few trees in plots covered by 0-33% canopy. The
thickness of 67-100% canopy cover prevents a lot of light from getting through to the forest
floor, suppressing the growth of vines. This would explain why these plots had the most trees; in
order for plants to get a sufficient amount of light, they need to grow upward.
There was a significant difference among trees of different heights within each category
of canopy coverage (table 1.2). Trees that form the canopy layer are always larger than
understory trees (Richards 1996). Vertical stratification of the forest represents an adaptive
strategy to light conditions under the canopy. In tropical rainforests, the systematic variation of
crown shapes is compelling evidence that trees are fundamentally adapted for particular heights
in the forest (Richards 1996).
All categories of canopy cover were dominated by trees 5-10 m (figures 1.4, 1.5, 1.6). It
is interesting how the species Kampindi is found in all three categories of canopy coverage, and
within every single grouping of tree height (the only exception being trees 40-60m in 0-33%
canopy coverage). Kampindi must be adaptive to many different types of environments. I think it
is also notable how plots covered by 67-100% canopy are dominated by msikisiki trees. Msikisiki
is found in all the categories of canopy coverage, but they do not have nearly as many as in 67-
100% canopy coverage.
There was no significant difference between tree heights among categories of canopy
coverage (table 1.3) This suggests that the vertical stratification of trees was similar in all the
plots.
Because there was no significant difference between percent coverage for grasses, herbs,
and shrubs, nor was there a significant difference between tree heights between my plots, this
leads me to conclude that canopy coverage does not cause a significant difference in community
structure. Although as shown in figures 1.1-1.6, canopy coverage does affect species
composition.
18. 18
Conclusion
I found that canopy cover influences species composition. The number and types of
species differed according to canopy coverage. I found that plots covered by 33-67% canopy
cover had the least variation in species, and plots covered by 0-33% canopy cover had the most
variation in species. This could be explained by levels of sunlight and tree cover. Sunlight is
needed for many plant processes, and more sunlight increases plant growth. Trees can also make
the soil under their canopy more nutritious, so a greater number of trees could increase growth of
understory vegetation.
I also found that canopy cover does not influence the structure of understory vegetation
.There was not much of a difference between tree height and vegetation coverage among
different categories of canopy coverage. Understory trees adapt to a certain height based on light
conditions, creating vertical stratification. There was no significant difference among tree
stratification in my plots across different categories of canopy coverage, but there was a
difference within plots of the same canopy coverage.
It is important to understand the composition and distribution of understory vegetation
relative to canopy cover within tropical rainforests because this gives us some insight into the
functionality of plant communities. How different layers of the rainforest impact one another is
an area that needs further study. As the population of the Usambaras continues to increase, more
and more pressure will be put on Mazumbai and other protected forest reserves. Rainforests are
rapidly disappearing on a global scale, and there is so much that has yet to be learned from them.
19. 19
Limitations and Recommendations
Limitations:
Without an altitude meter, it was hard to know if we were within the designated altitude range,
1500-1600 meters. At times we walked so much uphill it seemed that we had passed 1600
meters.
The vegetation could be super thick and made hard to walk, especially when there were big gaps
in the canopy.
Language was a barrier. My guide only spoke KiSwahili and Kisambaa, making communication
difficult.
Recommendations:
I would recommend repeating this study by looking at how plant communities change according
to altitude, or in different micro spatial habitats.
`
20. 20
Sources:
Campbell, G.S., and J.M. Norman. 1990. The description and measurement of plant canopy
structure. pp. 1-19 In: Russell, G., B. Marshall, and P.G. Jarvis (editors). Plant Canopies: Their
Growth, Form and Function. Cambridge University Press.
Conte, C. 2004. Highland Sanctuary: Environmental History in Tanzania’s Usamabara
Mountains. Ohio University Press. Ohio, USA.
Denslow, J S (1987). "Tropical Rainforest Gaps and Tree Species Diversity". Annual Review of
Ecology and Systematics 18: 431.
Denslow, J.S. 1998. Treefall Gap Size Effects on Above- and Below- Ground Processes in a
Tropical Wet Forest. Journal of Ecology. 86:597-609
Jennings, S.B., Brown N.D., and Sheil D. 1999. Assessing Forest Canopies and Understory
Illumination: Canopy Closure, Canopy Cover and Other Measures. Institute of Charted Foresters.
Kramer, D. M., G. Johnson, O. Kiirats, G. E. Edwards. 2004. New fluorescence parameters for
the determination of Q redox state and excitation energy fluxes. Photosynthesis Research
79:209-218
Minderhound, P. 2011. Historical Soil Erosion in the West Usambara Mountains, Tanzania.
Utrecht University. Netherlands.
Popma, J. and Bongers, F. 1988. The Effect of Canopy Gaps on Growth and Morphology of
Seedlings in Rainforest Species. Oecologia 75: 625-632
Richards, P.W. 1996. The Tropical Rainforest. Cambridge University Press, Cambridge UK.
Vigiak, O. 2005. Modeling spatial patterns of erosion in the West Usambara Mountains of
Tanzania. Wageningen University. Netherlands
Weltzin, J.F. and Coughenour M.B. 1990. Savanna Tree Influence on Understory Vegetation and
Soil Nutrients in Northwestern Kenya. Journal of Vegetation Science. 1:325-334