{II

w
\

  

WIN}

1

$9335: IIHHIIWWIWWINMNHHHMM

THESQS

Illlllllll llllllll'llllll'll ll'llllllllll

3 1293 01712 9168

This is to certify that the

dissertation entitled

Liana distribution and host relationships in
some temperate versus tropical forest sites

presented by
Christine M. Jarzomski

has been accepted towards fulfillment
of the requirements for

Masters Botany

degree in

M 5/

Major professor

 

 

 

Date l4/30/633
’ I

MSU is an Affirmatiw Action/Equal Opportunity Institun'on 0.12771

 

F iLIBRARv

I Michigan State
l university

 

 

 

PLACE lN RETURN BOX
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

DATE DUE DATE DUE DATE DUE

 

WV 1 9,7003

 

 

{110203

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1198 chlRC/DuoDquS—p.“

LIANA DISTRIBUTION AND HOST RELATIONSHIPS IN SOME TEMPERATE
VERSUS TROPICAL FOREST SITES

By
Christine M. J arzomski

A THESIS
Submitted to
Michigan State University

in partial fulfillment of the requirement
for the degree of

MASTER OF SCIENCE

Department of Botany and Plant Pathology

1998

ABSTRACT

LIANA DISTRIBUTION AND HOST RELATIONSHIPS IN SOME TEMPERATE
VERSUS TROPICAL FOREST SITES

By

Christine M. Jarzomski

Lianas (woody vines) are an important but often neglected component of many types of
forest. This study was designed to document the species of lianas in Carolina mixed
mesophytic forest (Rich Gap and Conley Creek), Carolina xeric white-oak dominated
forest (Cliffside), and Costa Rican seasonally dry tropical forest (Palo Verde). Another
aim of this study was to investigate the liana-host relationship in each of these forests. I
identified and measured the diameter at breast height (dbh) of every tree within 20 x 20 m
plots. I identified and measured the diameter of lianas at 10 cm above the ground, and I
recorded the primary host tree that each liana was utilizing. No lianas were observed
within plots at Clifftop Vista. Palo Verde had the highest species richness in terms of
lianas (15 species) and trees (33 species), and the highest percentage of trees hosting lianas
(55%), compared to 0% at Clifftop Vista, 37% at Rich Gap and 29% at Conley Creek. The
number of lianas per tree was greatest at Palo Verde (mean of 1.68 lianas per tree) and
lower at the two temperate mesic forest sites (0.49 lianas per tree at Rich Gap and 0.48
lianas per tree at Conley Creek). Host tree species was important in explaining liana
distribution at Rich Gap and Conley Creek. Host-tree size was influential at Rich Gap,
Conley Creek and Palo Verde. At Rich Gap, lianas were found in proportion to the host
surface area available to climb (e. g., Vitis), whereas at Conley Creek and Palo Verde,
lianas were restricted to particular sizes of hosts (e.g., Aristolochia at Conley Creek). This
study suggests that lianas are restricted by host tree species and host size. Thus, it seems

that each species of liana may utilize a particular suite of tree species and sizes as hosts.

ACKNOWLEDGMENTS

The William Chambers Coker Botanical Fellowship via the Highlands Biological
Foundation, Inc., and grant-in-aid of research from Highlands Biological Station from
made the North Carolina work possible. This was greatly facilitated by the Executive
Director of the station, Dr. Richard Bruce, who provided incalculable support, and Michael
Baranski, the instructor for the 1996 course “Southern Appalachian Flora,” who provided
valuable guidance for the choice of field sites. The United States Forest Service provided
permission to work in N antahala National Forest, a magical part of the Southern
Appalachians. The Michigan State University Ecology and Evolutionary Biology Program
provided funding to support my travel to and within North Carolina.

The Organization for Tropical Studies was a valuable resource for partial funding
for my participation in OTS 97-3. Michigan State University provided the remainder of the
OTS 97-3 tuition. The Michigan State Department of Botany and Plant Pathology’s Paul
Taylor Fund assisted with travel expenses to Costa Rica.

The Palo Verde project would have been impossible without the dedicated
assistance of many people, including Lucia Lohmann who was essential for project
development, and Victor Carmona who provided invaluable help in the field. Eugenio
Gonzales helped with plant identification and follow-up information. The Palo Verde staff
authorized the use of the old growth plot. Deedra McCleam and Theresa Singer assisted
with project development and reviews of the OTS 97-3 manuscript, which contributed to
this work.

Oliver Schabenberger and Chris Paciorek helped with data analysis and greatly
improved the final product. Several good friends assisted with interpretation, including
Piera Giroux, Heather Govenor, and Ian Ramjohn. The second chapter was improved by

suggestions from Sheila Ebert, Ed Grafius, Piera Giroux, and Nate Siegert.

iii

Finally, my Masters Thesis Guidance Committee, including Frank Ewers, Peter
Murphy and L. Alan Prather, was crucial in my project development, interpretation, draft
reviewal, and valuable general advice and guidance. Pete Murphy especially helped with
coordinating my trip to Costa Rica and L. Alan Prather gave me the opportunity to enhance
my knowledge of plant families through a specially-designed course and an herbarium
assistantship. Special thanks to my major faculty advisor, Frank Ewers, who encouraged

me to explore my own area of interest although it meant never even using the microtome!

iv

TABLE OF CONTENTS

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

LIST OF FIGURES ............................................................................... vii
CHAPTER 1

LITERATURE REVIEW OF LIANA (W OODY VINE) DISTRIBUTION AND HOST

RELATIONSHIPS .................................................................................. 1

The climbing habit .......................................................................... 1

Liana geographic distribution .............................................................. 2

Structural parasites and host preferences ................................................. 3

Effects of lianas on their hosts ............................................................. 7

Host mechanisms to avoid initial liana infestation ....................................... 9

Host mechanisms to remove lianas ...................................................... 10

Suggested future research ................................................................ 12

References .................................................................................. l4
CHAPTER2

LIANA DISTRIBUTION AND HOST RELATIONSHIPS IN CAROLINA TEMPERATE

AND COSTA RICAN SEASONALLY DRY FOREST ....................................... 17

Introduction ................................................................................ 17

Methods ..................................................................................... 19

North Carolina sampling technique ............................................. 19

North Carolina site descriptions ................................................ 21

Rich Gap .................................................................. 23

Conley Creek ............................................................. 25

Clifftop Vista ............................................................. 25

Costa Rica site description and sampling technique .......................... 27

Results ...................................................................................... 31

Liana species diversity and density ............................................. 31

Tree species diversity and density .............................................. 31

Liana host-species preference ................................................... 46

Liana host-size preference ....................................................... 48

Spatial distribution of lianas ..................................................... 53

Relationship between host bark and liana climbing mechanism ............. 57

Discussion .................................................................................. 6O

Liana species diversity and density ............................................. 6O

Liana host-species preference ................................................... 64

Liana host-size preference ....................................................... 65

Spatial distribution of lianas ..................................................... 66

What influences liana distribution? ............................................. 67

References .................................................................................. 7O

Appendices ................................................................................. 73

LIST OF TABLES

Table 1. Summary statistics for three sites in North Carolina and one site in Costa Rica. 32

Table 2. Species of lianas present in 0.6 ha at Rich Gap, NC ............................... 33
Table 3. Species of lianas present in 0.28 ha at Conley Creek, NC ......................... 33
Table 4. Species of lianas present in 0.4 ha at Palo Verde, Costa Rica ..................... 34
Table 5. Tree composition in 0.6 ha at Rich Gap, NC ......................................... 35

Table 6. Tree composition in 0.28 ha at Conley Creek, NC. *=grouped in analysis as
hard-barked oaks. 1;! =grouped in analysis as “other trees.” .................................. 40

Table 7. Tree composition in 0.2 ha at Clifftop Vista, NC. No lianas were present inside
the plots .............................................................................................. 42

Table 8. Tree composition in 0.4 ha at Palo Verde, Costa Rica. (ba=basal area) ......... 43

Table 9. Influence of tree species on liana distribution at Rich Gap (df=11) and Conley
Creek (df=7) when test is based on number of trees available or tree basal circumference
available to climb. Chi-square and P-values are shown ....................................... 46

Table 10. Size classes of trees of all species pooled in 0.6 ha at Rich Gap, NC.
Parentheses indicate exclusion of the end point within the size class, whereas brackets
indicate inclusion ................................................................................... 49

Table 11. Influence of host tree size on liana distribution at Rich Gap, Conley Creek and
Palo Verde when test is based on number of trees available or tree basal circumference
available to climb. Chi-square and P-values are shown. Each case had 5 degrees of
freedom. NS=Not significant .................................................................... 49

Table 12. Size classes of trees of all species pooled in 0.28 ha at Conley Creek, NC.
Brackets indicate inclusion of the end point within the size class, whereas parentheses
indicate exclusion ................................................................................... 51

Table 13. Size classes of trees of all species pooled in 0.2 ha at Clifftop Vista, NC ...... 52

Table 14. Size classes of trees of all species pooled in 0.4 ha at Palo Verde. Parentheses
indicate exclusion of the end point within the size class, whereas brackets indicate
inclusion ............................................................................................. 52

Table 15. Comparison of present results to previous studies: density of climbers 2 2.5 cm
diameter in 0.1 ha, number of climbing species, and number of and families in the
Neotropics and North America. Data from Gentry (1991) indicated by asterisks are
regional averages, including at least two sites from the region ................................ 62

vi

LIST OF FIGURES

Figure 1. General location of field sites in North Carolina and Costa Rica. ................ 20

Figure 2. Map of North Carolina showing relative locations of the three temperate field
sites (Rich Gap, Conley Creek, and Clifftop Vista) ............................................ 22

Figure 3. Rich Gap field site in the Highlands Quadrangle, Macon County, NC. (Source:
United States Geological Service.) ............................................................... 24

Figure 4. Conley Creek field site in the Greens Creek Quadrangle, Swain County, NC.
(Source: United States Geological Service.) .................................................... 26

Figure 5. Clifftop Vista field site in the Highlands Quadrangle, Macon County, NC
(Source: United States Geological Service.) .................................................... 28

Figure 6. Map of Costa Rica showing location of Palo Verde National Wildlife Refuge,
Costa Rica ........................................................................................... 29

Figure 7. Cumulative number of species for cumulative area sampled at Rich Gap for (a)
trees and (b) lianas .................................................................................. 37

Figure 8. Cumulative number of species for cumulative area sampled at Conley Creek for
(a) trees and (b) lianas .............................................................................. 41

Figure 9. Cumulative number of tree species for cumulative area sampled at Clifftop
Vista .................................................................................................. 42

Figure 10. Cumulative number of species for cumulative area sampled at Palo Verde for
(a) trees and (b) lianas .............................................................................. 45

Figure 11. Average number of lianas per tree for tree size classes for three sites. Brackets
indicate inclusion of end points in the size class, whereas parentheses indicate exclusion.
........................................................................................................ 5 5

Figure 12. Percentage of trees with a specified number of lianas per tree predicted by the
negative binomial (clumped) and Poisson distributions (random) given parameters
observed in actual liana distribution (black) for (a) Rich Gap, (b) Conley Creek, and (c)
Palo Verde ........................................................................................... 56

Figure 13. Palo Verde, Costa Rica. (a) Proportion of twining, tendril-climbing or
adhesive-disc climbers that were observed climbing trees with chipping, rough or smooth
(includes peeling and spiny) bark. Also shown is proportion of trees with rough, smooth,
and chipping bark (b) Proportion of trees that hosted lianas that are tendril climbers,
twiners or adhesive disc climbers. Some trees hosted no lianas, others hosted lianas in
more than one category (>1cat). Also shown is proportion of lianas that had various
climbing mechanisms .............................................................................. 59

vii

Chapter 1

Literature review of liana (woody vine) distribution and host relationships

The climbing habit

Plants exhibit various growth forms, one of which is the climbing habit. Vines, climbing
plants that are rooted in the soil, are “structural parasites” that occupy almost every habitat
where there are trees available to climb (Stevens 1987, Dillenburg et al. 1993). The
climbing habit is widespread phylogenetically, and about 133 of the approximately 400
vascular plant families include at least a few climbing species (Gentry 1991). The climbing
habit has evolved independently in many plant lineages including groups as diverse as
gymnosperrns (Gnetaceae), dicotyledons (e. g., Vitaceae), and monocotyledons (e. g.,
Araceae).

Plants use many tactics to climb, and thus climbing plants as a whole form a diverse
group. While discussing this diversity, Gentry (1991) states, “there must be strong
selective pressure favoring evolution of a scandent habit.” Darwin (1867) discusses the
origin of the climbing habit by asserting, “plants become climbers, in order, as it may be
presumed, to reach the light...” The climbing habit allows a plant to reach the canopy with
little expenditure in structural tissue (Darwin 1867 and others). This characteristic is very
important for plants that live in habitats with thick canopies.

The definition of the term “vine” is not always clear. Some species will grow as
shrubs until a trellis becomes available (e.g., Toxicodendron radicans as shown in Gartner
1991). However, many vine seedlings will die if they are unable to access a structural

support before they deplete their seed reserves (examples in Janzen 1983). The definition

of the term “liana,” or “woody vine,” is also sometimes problematic. Some non—
dicotyledonous plants climb into the canopy but do not produce secondary xylem (e. g.,
climbing ferns, climbing monocots, etc.). It is debatable whether to include these species
in the category of liana because they do not produce anatomically true wood. This thesis
will reserve the word “liana” for woody vines that add layers of secondary xylem or wood
in a somewhat predictable fashion.

Lianas are important components of both temperate and tropical forest ecosystems
(Darwin 1867, Putz 1984b, Gentry 1991, Teramura et al. 1991) and can cause silvicultural
problems (e.g., Featherly 1941, Lutz 1943, Trimble and Tryon 1974, Siccama et al. 1976,
Putz 1991). Host-dependent species are crucial for many indigenous Central and South
American peoples (Bennett 1995). Recent interest in tropical biology and development of
rope-assisted tree ascension has promoted the study of lianas by some canopy biologists,
but fundamental questions about vine ecology remain unanswered. The objective of this
literature survey will be to address some of these unanswered questions. I will first
examine the geographic distribution of lianas and address host preferences of structural
parasites. An overview of the effects of lianas on their hosts will be followed by a
discussion of mechanisms that host trees utilize to avoid and then shed lianas. Finally, I

will give suggestions for future liana research.

Liana geographic distribution

Liana species diversity increases as latitude decreases. The proportion of plant species that
climb relative to total number of plant species is higher in tropical compared to temperate
forests. For example, climbing plants contribute only 1.3% to the flora of the Carolinas
and the Southern Appalachian Mountains relative to about 10% of the Neotropical flora

(Gentry 1991).

Tropical forests provide habitat for several types of structural parasites in addition
to lianas, such as epiphytes and hemiepiphytes. This may be explained by the structural
complexity of the tropical canopy. There is more complexity and space for climbing in the
canopy of a tr0pical forest relative to that of a temperate forest. This has most likely
encouraged the growth and diversification of structural parasites in tropical habitats.

There are other advantages to climbing in addition to reaching the well-lit forest
canopy. The elevated position in the canopy improves the possibility of effective wind
pollination and seed dispersal. A sought-after space in the canopy also allows a plant to
escape terrestrial herbivores. Arboreal pollinators and seed dispersers will also be more
readily available in the forest canopy.

It has been suggested that the hydraulic architecture of lianas restricts their
distribution mainly to the tropics (Ewers et al. 1997). The wide vessels that characterize
secondary xylem of lianas allow extremely efficient water conduction with limited stem
width, but these vessels are also very susceptible to freeze-induced embolism (Hargrave et
al. 1994), and thus lianas are less speciose and less densely distributed in frost-prone areas
than in non-frost areas. It seems that lianas can only survive in the temperate zone if they
produce a fresh layer of secondary xylem every spring or produce positive root pressure in

the spring to refill the embolized vessels.

Structural parasites and host preferences

Lianas are structural parasites, plants that rely on other plants for structural support. These
plants do not seem to be distributed randomly amongst available host trees within a forest.
Epiphytes (plants that grow on other plants without attachment to the ground),
hemiepiphytes (plants that have an epiphytic and terrestrial phase), and lianas all show

some host preferences. Epiphytes are not distributed in proportion to host species

abundance (Schlesinger and Marks 1977, Bennett 1986, Benzing 1995). Instead,
microhabitat availability and seed dispersal mechanisms seem to predict where epiphytes
will be successful (Bennett 1986).

The most convincing evidence that structural parasites show some host preferences
comes from studies on hemiepiphytes. Hemiepiphytes are plants that germinate in the
canopy and later establish a connection with the ground (primary hemiepiphytes), or plants
that germinate on the forest floor, climb into the canopy, lose their connection with the
ground, but may later reestablish this connection by way of adventitious roots (secondary
hemiepiphytes). Characteristics of hemiepiphytes have been discussed by Putz and
Holbrook (1986), who mention that hemiepiphytes are among the least understood types of
plants. Guy (1977) surveyed the host usage of plants within the fig genus Ficus
(Moraceae) at the Mana Pools Game Reserve in Rhodesia. He observed that Ficus was
distributed independent of host species abundance, and was found more often on some tree
species than expected based on the relative abundance of these trees in the forest.
Individuals within the host tree species Diospyros mespiliformis (Ebenaceae), were more
likely to serve as hosts to Ficus than expected by their abundance in the forest. Guy (1977)
notes that this tree has rough bark and suggests that rain cannot dislodge seeds as easily
from trees with rough bark compared to trees with smooth bark. He also mentions that D.
mespiliformis produces fruit around the same time as Ficus, so that animals are likely to
deposit the fig seeds on D. mespiliformis after feeding on figs.

Additional evidence that hemiepiphytes are distributed independent of host tree
species abundance came from a study of several species on Barro Colorado Island (BCI) in
Lake Gatun, Panama (Todzia 1986). Only 1% of the individuals within the smooth-barked
tree species, Quararibea asterolepis (Bombacaceae), serve as hosts to hemiepiphytes.
However, 58% of the surveyed trees of the spiny-barked species, Hura crepitans

(Euphorbiaceae), contain hemiepiphytes. Todzia (1986) suggested that spines may

facilitate hemiepiphyte seedling establishment on the host tree. She also showed that
different species of hemiepiphytes on BCI occupy distinct canopy positions.

Additional evidence that hemiepiphytes are not distributed randomly amongst
potential host trees came from a study of Ficus crassiuscula in the Monteverde Cloud
Forest in the Cordillera de Tileran in Costa Rica (Daniels and Lawton 1991). The authors
show that the hemiepiphytes are distributed randomly on potential host species during the
viny juvenile stage of development. In contrast, adult F. crassiuscula are more likely to be
found on certain tree species. This evidence supports the idea that trees are not avoiding
initial establishment of structural parasites, but rather have some mechanism to shed these
plants after they have colonized the host’s canopy. Daniels and Lawton (1991) also
showed that the size of the host tree does not seem to influence its acceptability to
hemiepiphytes.

A study in the Bomean rain forest found some species of Ficus more often on
dipterocarp tree species, but others more often on non-dipterocarp host species (Larnan
1996). Additionally, canopy rnicrohabitats seem important in determining acceptability of
hosts, as different species of figs occupied distinct canopy positions. Based on the above
evidence, it seems that one class of structural parasites, hemiepiphytes, utilize distinct tree
species and perhaps even prefer particular positions within the tree’s canopy. Finally,
Williams-Lindera and Lawton (1995) reassert the existence of herniepiphytic host
preferences in a review paper.

Like the epiphytes and hemiepiphytes discussed above, lianas tend to grow more
successfully on some host tree species than others. Jack Putz has examined some of these
issues on BCI. He examined trellis requirements for lianas with various climbing
mechanisms (Putz 1984b). His work shows that trellises must be strong enough to
support the weight of a climber but he observed no minimum trellis diameter requirement
for any of the lianas he studied. However, although no species of liana shows a minimum

trellis size requirement, some species of lianas do not seem able to climb trellises that

exceed a particular size. Most tendril—climbing lianas utilize hosts less than 10 cm diameter
because they must wrap around their support structure. Twiners can climb larger supports
than tendril climbers. Species that climb by way of adventitious roots or adhesive discs can
climb structures of any size. Putz (1984b) concluded from these observations that support
availability is a major factor limiting lianas. He also mentions that canopy gaps provide
ideal habitat for lianas with host size restrictions. Additionally, lianas can climb other
lianas to access the canopy (Putz 1984a, 1984b, 1995), and lianas are spatially aggregated
at BCI.

Putz (1984b) focused mainly on adult lianas, whereas Collins and Wein (1993)
examined the spatial distribution of juvenile understory vines in a South Carolina hardwood
forest. Collins and Wein (1993) found that spatial distribution differs among juvenile vine
species but that the identity of the nearest support species does not influence juvenile vine
location. This supports the idea that vines are initially equally likely to climb any tree, so
tree mechanisms for shedding lianas become more important than mechanisms to avoid
initial liana growth. The spatial location of juvenile vines are influenced by microhabitat,
especially soil moisture and elevation (Collins and Wein 1993).

A study by Talley, Lawton and Setzer (1996) in an Alabama hardwood forest deals
most directly with the subject of liana host specificity. They showed that poison ivy, Rhus
radicans (Anacardiaceae), was not distributed randomly among potential host species.
Instead, Carya ovata (Juglandaceae) and the hard-barked oaks (Quercus spp.: Fagaceae)
hosted more lianas than expected, whereas Acer rubrum (Aceraceae) hosted fewer than
expected based on relative tree species stem densities. They also showed that host size
influences poison ivy distribution. They observed more lianas than expected on hosts
larger than 60 cm in diameter, but fewer than expected on hosts less than 60 cm diameter.
Additionally, small, erect poison ivy stems are not distributed randomly on the forest floor,
as was also seen by Collins and Wein for juvenile vines (1993, see above). Talley,

Lawton and Setzer suggest that host bark allelopathic secretions may influence host

preferences of poison ivy. They also acknowledge that some tree bark types may offer
better liana adhesive-root attachment success rates. Finally, trees of different species have
different morphological architecture, and thus may provide differing light levels once the
liana has reached the canopy. Theoretically, it would be more advantageous for lianas to
climb species with thinner crowns that allow more light penetration to the liana leaves.
Based on these studies of epiphytes, hemiepiphytes and lianas, it becomes clear that
structural parasites display host preferences. The remainder of this chapter will look more
closely into the relationship between one class of structural parasites, lianas, and their host
trees. In order to understand why lianas are found on some tree species more often than
others, we must now consider why lianas would be selected to exhibit host preferences.

We must also consider the selection pressure on trees to avoid and shed lianas.

Effect of lianas on their hosts
Lianas are intimately associated with their hosts because they are physically in contact with
each other, unlike two mutually competing trees in a given forest. Similar to two
competing trees, lianas and their hosts compete for water and nutrients at the forest floor.
Lianas and other structural parasites can be very detrimental to their host trees. First, liana
infestation can eventually kill host trees. After only two years of study after wild grape
(Vitis spp.: Vitaceae) was released in a West Virginia temperate hardwood forest, Trimble
and Tryon (1974) reported 2% death of trees caused by lianas.

Lianas can have many negative effects on their hosts that are less severe than death.
The shear weight of supporting another plant can cause mechanical damage. Lianas can
structurally weaken their host trees by physically suppressing branches causing them to
break. This additional weight also increases host tree susceptibility to ice damage (Siccama
et al. 1976). Lianas can also cause mechanical abrasion to their hosts, sometimes resulting

in passive strangulation which injures host-tree stems. Through girdling, lianas can

interfere with their hosts’ internal transport (Lutz 1943), restricting internal translocation of
nutrients and water. Lianas can also decrease the stem incremental diameter growth rate of
the host tree (Putz 1984b). The cumulative effects of lianas may result in an overall
deformed tree crown and a weakened host tree.

In addition to structurally weakening or even killing their host trees, lianas also
influence their hosts by competing with them for light, as liana leaves may shade out the
leaves of the host trees, resulting in less photosynthetically active radiation to reach the
leaves of the host tree, reducing its potential for growth and reproduction. Below-ground
competition with lianas for limited nutrients and water can also reduce the host’s growth
(Dillenburg et al. 1993). Lianas may decrease the reproductive output of certain tree
species, as in Bursera simaruba (Bursuraceae; Stevens 1987). Sometimes lianas connect
nearby trees, constructing a canopy trellis providing access to mammalian herbivores.

In addition to the effects that lianas can have on individual trees, they can affect
overall community structure. Lianas or scramblers may compete with tree or shrub
saplings for limited resources. This may result in a “vine-dominated disclimax” (Whigham
1984), where succession is said to be “arrested,” and will temporarily progress no further.

In spite of all of the possible negative effects of lianas on their hosts listed above,
usually only a subset of these factors will actually influence a particular host tree. The rank
of importance of the above negative factors of hosting lianas will probably differ from
species to species and often from tree to tree.

Serving as hosts to lianas may benefit host trees in a few ways. By intertwining
through several trees and connecting trees together, lianas may mechanically stabilize the
canopy within a forest (Putz 1995). This effect of liana presence seems important in thin
tropical soils where trees are often shallowly rooted. However, this may be detrimental if
connected trees fall for any reason. Another important result of lianas in tropical forests is
that they transform a disjunct, multilayered two-dimensional canopy to an intertwined,

three-dimensional connected system which provides coherent routes for seed dispersers to

access hosts. One mammologist who studies arboreal mammals commented that lianas
seem absolutely essential to the mammals that inhabit tropical forest canopies (D.

McCleam, personal communication).

Host mechanisms to avoid initial liana infestation

In light of all the negative effects of lianas on their hosts listed above, it seems that host
trees would be under selection pressure to avoid lianas altogether. Some tree characteristics
have been purported to assist in avoiding or preventing liana infestation. The first idea
proposed in the literature was that buttresses “protect” trees from lianas because twining
vines would have to spend more energy maneuvering around large buttresses to reach the
climbable host stem and would not be likely to succeed in accessing the host tree (Black
and Harper 1979). This idea suggests that perhaps buttresses prevent initial ascent of
twining lianas. This is essentially an extension of the idea that twiners will not be able to
climb trees whose diameters exceed an acceptable maximum size (Putz 1984b). Buttresses
would not protect a potential host tree from lianas that climb by way of adventitious roots
or adhesive discs, which can utilize most tree sizes (Putz 1984b). Additionally, buttresses
would not protect trees from lianas that enter the canopy of one tree from that of other
nearby trees. However, Boom and Mori (1982) found that buttresses do not reduce liana
load in tropical wet forest of Brazil.

Another suggestion is that trees with smooth bark will be less likely to host lianas
than trees with rough bark (Putz 1980, 1984a). The idea is that rough bark provides more
surface area and crevices for a liana to securely attach to the tree stem. Boom and Mori
(1982) rejected this hypothesis since smooth-barked tree species were just as likely to host
lianas as rough-barked trees in Brazil. However, this idea needs further testing.

Another possible mechanism to prevent liana infestation is by utilizing rapid

diameter growth (Putz 1980). Fast-growing trees will theoretically more quickly escape the

window of time when their diameter is climbable by twining lianas, but survey data does
not support this claim (Putz 1984a).

Another recently suggested mechanism possibly utilized by trees to prevent liana
infestation is the secretion of allelopathic toxins from tree bark (Talley et al. 1996). This

idea needs more testing before acceptance.

Host mechanisms to remove lianas

None of the mechanisms mentioned above seem satisfactory in preventing lianas from
climbing an available host tree. Perhaps trees have not been selected to avoid the initial
ascent of lianas (maybe with the exception of trees that produce allelopathic secretions), but
rather to shed lianas (and possibly other types of structural parasites as well) after they have
infested the host tree. What then becomes important is how long the liana can remain
attached to its host, and whether a host tree has mechanisms to shed climbers or other
structural parasites.

One suggestion in the liana-avoidance literature is that trees with peeling bark will
be less likely to host lianas than trees with rough bark (Putz 1980, 1984a). Trees that shed
their peeling bark periodically could dislodge their lianas when they shed their bark. The
bark of the gumbo-limbo tree (Bursera simaruba: Burseraceae) and sycamore (Platanus
spp.: Platanaceae) represent possible examples of this “strategy.” Stevens (1987) mentions
peeling Bursera bark as a mechanism to reduce host liana load. This “exfoliating bark”
hypothesis only holds for lianas that climb by utilizing adhesive discs or adventitious roots,
both of which depend on the suitability of their climbing surface to maintain their grip on
their host. Utilizing exfoliating bark to shed lianas is effective both against lianas climbing
the tree from the ground and against those entering from adjacent canopy.

Another mechanism proposed to help trees rid themselves of lianas is by producing

self-pruning compound leaves (Putz 1980, 1984a). The advantage of large, self-pruning

10

compound leaves, with short retention times, is that large leaves will initially support liana
attachment, but when the tree drops the leaves, the tree also sheds its lianas. Leaf length
seems weakly negatively correlated with liana presence (Putz 1984a). Stevens (1987)
mentions the importance of Bursera compound-leaf shedding as a mechanism to reduce
host liana load.

Spiny bark has been suggested as a mechanism to saw off lianas, but no evidence
supports this claim. Putz (1984a) mechanically agitated saplings of two spiny-barked trees
to see whether they would saw off their connecting lianas, but the results were
inconclusive.

A final related suggested mechanism that would dislodge lianas from two connected
trees would be for the trees to have high mechanical flexibility, as measured by Young’s
modulus (Putz 1980). Theoretically, higher flexibility would allow two interconnected
trees to sway in opposite directions and dislodge or break their connectors. However, this
idea seems improbable because seldom would trees be swaying in opposite directions.

One of the most effective mechanisms that some trees utilize to dislodge lianas from
their crowns is by employing protective symbionts. Some tree species house ants in
hollow bark (e.g., Cecropia spp.: Moraceae) or swollen thorns (e.g., Acacia spp.:
Fabaceae). Azteca ants that colonize Cecropia trees rush out and attack all intruders that
contact the tree, including lianas that begin to climb the trunk or enter the canopy (Janzen
1969). However, this mutualistic relationship gives several advantages to the tree and the
protective symbiont in addition to protecting the tree from lianas, and may be costly for the
host tree involved. It does not seem likely that this relationship arose only in response to
the pressure of avoiding liana colonization or shedding lianas.

None of the above mechanisms for avoiding lianas prior to their colonization, or
shedding lianas after infestation, seems entirely satisfactory to explain what we observe
today. Perhaps different species of host trees utilize some unique combination of the above

mechanisms to avoid or shed lianas. Because trees that can escape lianas will have higher

11

annual basal growth (Putz 1984b) and greater reproductive output (Stevens 1987), it seems

likely that lianas are a selective force influencing trees.

Suggested future research

In order to understand liana distribution and the liana-host relationship, there are many
issues that should be addressed. Perhaps the most effective tree “adaptations” to reduce
liana load differ in different habitats and forest types. Maybe different tree species utilize
different strategies to avoid or shed lianas, and perhaps the arsenal of mechanisms to avoid
or shed lianas is still evolving. It would be interesting to more accurately assess which of
the negative effects of lianas are more detrimental in different habitats or forest types. For
example, vulnerability to seasonal weight of snow and ice, heightened by hosting lianas, is
obviously more influential in temperate areas, whereas light competition with lianas may be
more important in the tropics.

It should also be assessed whether trees preferentially shed liana-laden limbs or
shaded limbs to reduce their liana load. Putz (1995) suggested the construction of model
trees to investigate the influence of surfaces and sizes of available hosts in order to further
our understanding of the relationship between host tree bark and liana host preference.
Artificial support structures with smooth, rough or exfoliating surfaces could be compared
for their acceptability as liana substrate. Similarly, structures with or without wide bases
could be compared for their acceptability to lianas to model the effect of buttresses on host
acceptance. Support structures of different diameters could be offered to juvenile lianas to
experimentally test the influence of host size on initial host colonization.

Finally, more habitats and geographic locations should be examined for liana host
species preferences. Perhaps with additional comparable data sets we will be better able to

assess the relationship between structural parasites, particularly lianas, and their hosts.

12

REFERENCES

13

References

Bennett BC (1986) Patchiness, diversity and abundance relationships of vascular
epiphytes. Selbyana 9:70-75

Bennett BC (1995) Ethnobotany and economic botany of epiphytes, lianas, and other host-
dependent plants: an overview. Pages 547-586 in Lohman MD, Nadkarni NM
(eds) Forest Canopies. Academic Press, San Diego.

Benzing CH (1995) Vascular Epiphytes. Pages 225-254 in Lohman MD, Nadkarni NM
(eds.) Forest Canopies. Academic Press, San Diego.

Black I-IL, Harper KT (1979) The adaptive value of buttresses to tropical trees: additional
hypotheses. Biotropica 11:240

Boom BM, Mori SA (1982) Falsification of two hypotheses on liana exclusion from
tropical trees possessing buttresses and smooth bark. Bulletin of the Torrey
Botanical Club 109:447-450

Collins BS, Wein GR (1993) Understory vines: distribution and relation to environment on
a southern mixed hardwood site. Bulletin of the Torrey Botanical Club 120:38-44

Daniels JD, Lawton R0 (1991) Habitat and host preferences of Ficus crassiuscula, a
Neotropical strangling fig of the lower-montane rain forest. Journal of Ecology
79: 129-141

Darwin C (1867) On the movements and habits of climbing plants. Journal of the Linnean
Society (Botany) 9: 1—1 18

Dillenburg LR, Whigham DF, Teramura AH, Forseth IN (1993) Effects of below- and
aboveground competition from the vines Lonicera japonica and Parthenocissus
quinquefolia on the growth of the tree host Liquidambar styraciflua. Oecologia
93:48-54

Ewers FW, Cochard H, Tyree MT (1997) A survey of root pressure in vines of a tropical
lowland forest. Oecologia l 10: 191-196

Featherly HI ( 1941) The effect of grapevines on trees. Proceedings of the Oklahoma
Academy of Science. 21 :61-62

Gartner BL (1991) Structural stability and architecture of vines vs. shrubs of poison oak,
Toxicodendron diversilobium. Ecology 72: 2005-2015

Gentry AH (1991) The distribution and evolution of climbing plants. Pages 3-52 in Putz
FE and Mooney HA, editors. The biology of vines. Cambridge University Press.
Cambridge

Guy PR (1977) Notes on the host species of epiphytic figs (Ficus spp.) on the flood-plain
of the Mana Pools Game Reserve, Rhodesia. Kirkia 10:559-562

14

Hargrave KR, Kolb KJ, Ewers FW, Davis SD (1994) Conduit diameter and drought-

induoed embolism in Salvia mellifera Greene (Labiatae). New Phytologist 126:
695-705

Janzen DH (1969) Allelopathy by Myrrnecophytes: the ant Azteca as an allelopathic agent
of Cecropia. Ecology 50: 147-153

Janzen DH, ed. (1983) Costa Rican Natural History. University of Chicago Press.
Chicago, IL

Lutz HJ (1943) Injuries to trees caused by Celastrus and Vitis. Bulletin of the Torrey
Botanical Club 70:436-439

McCleam D. Coordinator for Organization for Tropical Studies: Tropical Biology an
Ecological Approach

Muller RA, Oberlander TM (1984) Physical geography today. Random House. New York
Putz FE (1980) Lianas vs. trees. Biotropica 12: 224-225
Putz FE (1984a) How trees avoid and shed lianas. Biotropica 16: 19-23

Putz FE (1984b) The natural history of lianas on Barro Colorado Island, Panama. Ecology
65:1713-1724

Putz FE (1991) Silvicultural effects of lianas. Pages 493-501 in Putz FE, Mooney HA,
editors. The biology of vines. Cambridge University Press. Cambridge, England

Putz FE (1995) Vines in treetops: consequences of mechanical dependence. Pages 311-
324 in Lohman MD, Nadkarni NM (eds.) Forest Canopies. Academic Press, San
Diego.

Putz FE, Holbrook NM (1986) Notes on the natural history of hemiepiphytes. Selbyana
9:61-69

Radford AE, Ahles HE, Bell CR (1968) Manual of the vascular flora of the Carolinas.
University of North Carolina Press, Chapel Hill, USA

Schlesinger WJ, Marks PL (1977) Mineral cycling and the niche of Spanish moss,
Tillandsia usenoides L. American Journal of Botany 64: 1254-1262

Siccama TG, Weir G, Wallace K (1976) Ice damage in a mixed hardwood forest in
Connecticut in relation to Vitis infestation. Bulletin of the Torrey Botanical Club.
103: 180-183

Stevens GC (1987) Lianas as structural parasites: the Bursera simarouba example.
Ecology 68:77-81

Talley SM, Lawton RO, Setzer WN (1996) Host preferences of Rhus radicans
(Anacardiaceae) in a southern deciduous hardwood forest. Ecology 77: 1271-1276

Teramura AH, Gold WG, Forseth m (1991) Physiological ecology of mesic, temperate

woody vines. Pages 245-285 in Putz FE, Mooney HA, editors. The biology of
vines. Cambridge University Press. Cambridge, England

15

Todzia C (1986) Growth habits, host tree species, and density of hemiepiphytes on Barro
Colorado Island, Panama. Biotropica 18:22-27

Trimble GR, Tryon EH (1974) Grapevines a serious obstacle to timber production on good
hardwood sites in Appalachia. Northern Logger and Timber Processor. 23:22-23,
44

Whigham D (1984) The influence of vines on the growth of Liquidambar styraciflua L.
(sweetgum). Canadian Journal of Forest Research 14:37-39

Williams-Linera G, Lawton R0 (1995) The ecology of hemiepiphytes in forest canopies.

Pages 255-323 in Lohman MD, Nadkarni NM (eds.) Forest Canopies. Academic
Press, San Diego.

16

Chapter 2

Liana distribution and host relationships in Carolina temperate and

Costa Rican seasonally dry forest

Introduction

Vines are structural parasites that occupy almost every habitat where trees are
available to climb (Stevens 1987, Dillenburg et al. 1993). Climbing allows a plant to reach
the forest canopy with little expenditure in structural tissue (Darwin 1867 and others). The
climbing habit has evolved independently in several plant lineages including groups as
diverse as gymnosperrns (Gnetaceae), dicotyledons (e.g., Vitaceae), and monocotyledons
(e.g., Araceae).

Liana (woody vine) species diversity increases as latitude decreases. The 48
woody climbers native to the southeastern United States belong to 25 genera in 19 families
(Gentry 1991). Climbing plants contribute 1.3% of the flora of the Carolinas and the
Southern Appalachian Mountains (Radford et al. 1968, in part compiled by D. Boufford as
noted in Gentry 1991). In contrast, lianas contribute about 10% of the Neotropical flora
(Gentry 1991).

Lianas are important components of both temperate and tropical forest ecosystems
(Darwin 1867, Putz 1984b, Gentry 1991, Teramura et al. 1991) and can cause silvicultural
problems (e.g., Featherly 1941, Lutz 1943, Trimble and Tryon 1974, Siccama et al. 1976,
Putz 1991). Recent interest in tropical biology and development of rope-assisted tree
ascension have promoted the study of lianas, but fundamental questions about vine ecology

remain unanswered. Although lianas are common in some habitats, they are often absent in

17

others, and factors detennining local success are unclear. In addition to a vague
understanding of what habitats best support lianas and what factors favor liana growth, it is
difficult to predict how lianas will be distributed in a particular forest. Previous work
suggests that lianas grow more successfully on certain host tree species (Talley et al. 1996,
Putz 1984a). The size of the host trees available may also influence the success of
particular liana species. Although largely untested, host bark characteristics may influence
liana host preference (Putz 1980, 1984a, Stevens 1987). Some work indicates that rough-
barked trees provide a better substrate for hemiepiphytes than do smooth-barked trees (Guy
1977, T odzia 1986), but little data exists to evaluate whether the same is true for lianas.
There is little data to allow predictions of the percentage of trees that will host lianas in a
particular forest. Finally, most data on lianas have come from wet tropical forests, and
very few studies have examined lianas in warm temperate forests or in seasonally dry
tropical forests.

The present study was designed to document the species and densities of lianas
growing in three differing habitats: 1) mixed mesic cove hardwood forest, 2) xeric white-
oak dominated forest, and 3) seasonally dry tropical forest. Another goal was to
investigate the liana-host relationship, including the influence of 1) host species and 2) host
size on lianas, 3) the spatial distribution of lianas, and 4) the relationship between host bark

morphology and liana climbing mechanism.

18

Methods

Three sites were sampled in North Carolina in the United States, and one site in the
Guanacaste Province of Costa Rica (Figure 1). The methods used in North Carolina
differed slightly from those used in Costa Rica, and are described below along with

descriptions of each of the four sites.

North Carolina sampling technique
At each site I arbitrarily designated a starting point along a road or trail (usually an
intersection of one trail or road with another). At least 10 m into the forest from the road or
trail, I established 20 x 20 m plots (each plot = 0.04 ha). I allowed at least 20 m between
each plot, and I established plots in a transect-fashion along the road or trail. I was careful
not to include edges or recently-disturbed areas (e.g., gaps) in the study sites. This
sampling method gave me the freedom to skip areas that contained no or few trees.

In each plot I identified and measured the diameter at breast height (dbh) of each
tree larger than 10 cm dbh for comparability with earlier work (Talley et al. 1996). I
identified each liana stem that was climbing a tree (the tree was considered the “primary”
host for each liana), recorded which tree each liana stern was climbing, measured each liana
stem 10 cm above ground level (as in Talley et al. 1996), and identified the lianas. Rarely,
tree trunks were fused together in the field, so I summed the basal circumferences of each
trunk to calculate one size value for the fused trunks because this was the most reliable
technique for determining how much tree trunk surface area was available for the lianas to
climb. Occasionally it was difficult to tell if each liana was a distinct individual, a problem
other workers have noted in the past (Putz 1984b, Talley et al. 1996). I considered each

independent stem one individual, unless it could be determined otherwise.

19

 

   

”43%:- \
- ‘ g has-

 

 

“.
v‘ - \ \\o
‘ I ‘

 

 

 

 

 

 

 

 

 

 

r
COSTA RICA f

 

 

 

 

 

 

 

 

Figure 1. General location of field sites in North Carolina and Costa Rica.

20

 

North Carolina site descriptions

All three temperate sites were in North Carolina within the Blue Ridge Mountains, which
extend from northern Virginia through eastern Tennessee and western North Carolina to the
extreme west of South Carolina and northeastern Georgia (Martin et al. 1993). These
mountains contain the highest peaks in eastern North America (> 1500 m), and once rivaled
the Rocky Mountains in relief and size prior to erosion (Martin et al. 1993). The Blue
Ridge Mountains form part of the Southern Appalachians which is composed of the
Appalachian (Cumberland) Plateau (extending from West Virginia to Alabama), the Ridge
and Valley Unit (West Virginia to Alabama), and the Blue Ridge region (northern Virginia
through northern Georgia; Martin et al. 1993). Soils of slopes and ridgetops are generally
shallow, with deep alluvial soils in the valleys (Pitillo 1976). The soils are primarily
ultisols; entisols, histosols, and inceptisols are also present (Pitillo 1976).

Nestled within this mountainous terrain is the Highlands quadrangle, in Macon
County, NC, near the Georgia and South Carolina boundaries. The town of Highlands
(elevation 1255 m), is located on the southern rim of the southernmost high plateau of the
Blue Ridge Mountains (Ogbum 1975). The Great Smoky Mountains and the Blue Ridge
Parkway lie to the northwest, the Nantahala Mountains to the west, and the Cowee
Mountains to the north. The Highlands region is known for its unusually high rainfall even
for the Southern Appalachians, and is considered a “frontier” between northern and
southern forms of life (Ogbum 1975). This high rainfall and the many habitats sheltered
by the ridges have provided a refuge to many forms of life and have resulted in a rich and
diverse flora and fauna. The Blue Ridge Province, and thus, the Highlands region, is
considered one of the most floristically diverse regions of the eastern United States
(W offord 1989). I chose the Highlands region for this study because of its floristic
diversity, and because its moderate climate seemed favorable to liana growth. Two of the
sites were in the Highlands quadrangle (Rich Gap and Clifftop Vista), and one was in the
Greens Creek quadrangle (Conley Creek; Figure 2).

21

.335
mecca—O 23 £080 >250 .920 new: 8% 20m 33095» 025 05 ago 80:82 2628 wage? £5030 5.52 no “FE .m 05me

Sec—emu

z<muo <zficm<o
0:595. :58 35.5
- neon—z

 

3‘.
<z3c¢<o 825 3.55 .595 .78 B

39.332 :20 .3390
@550 35.5.

i Hammmzzmh
<~Z~G¢ Cr

22

Rich Gap. The first site was a mesic cove forest. Cove forests, in general, are the most
species-rich forests in eastern North America with no consistently dominant tree species.
Cove hardwood forests are found in sheltered mountain valleys, generally on north- and
east-facing slopes (Clay et al. 1975). The cove forests of the Smoky Mountains contain
more tree species than are found in all of temperate Europe (Clay et al. 1975). Usually the
most dominant trees are: Eastern hemlock (Tsuga canadensis L. Carriere: Pinaceae),
basswood (Tilia heterophylla Ventenat: Tiliaceae), sugar maple (Acer saccharum Marshall:
Aceraceae), silverbell (Halesia carolina L.: Styracaceae), buckeye (Aesculus octandra
Marshall: Hippocastanaceae) and yellow birch (Betula lutea Michaux: Betulaceae)(Clebsch
1989). Beech (Fagus grandifolia Ehrhart: Fagaceae) and tulip tree (Liriodendron tulipifera
L.: Magnoliaceae) are also abundant.

Rich Gap is a mesic cove in the Highlands Quadrangle in Macon County, North

Carolina (approximately 35°2’30” North and 83°10’11” West; southwest of benchmark SN

391 3003), southeast of the town of Highlands. This site is adjacent to Rich Gap Road
(State Road 1710), a gravel road off of Horse Cove Road (Figure 3). One Liriodendron
tulipifera tree at this site is the second largest tree in North Carolina and the third largest tree
in the eastern United States (Highlands Chamber of Commerce 1994). Rich Gap Road
runs north-south along the east-facing slope of Little Fodderstack and Fodderstack
Mountains, part of the Cowee Mountain Range. The elevation ranges from 910 to 1040 m
at this site, and several hillside seepages dissect the area. Macon County receives between
1270 and 2030 mm of precipitation per year (Clay et al. 1975). I selected Rich Gap
because it represented cove hardwood forest which I wanted to sample and also because I

had seen wild grape (Vitis spp.) in the area.

23

 

‘I/IBM SN 3
'2920

 

Figure 3. Rich Gap field site in the Highlands Quadrangle, Macon County, NC. (Source:
United States Geological Service.)

24

At Rich Gap, the canopy includes abundant Liriodendron tulipifera, umbrella tree
(Magnolia fraseri Walter: Magnoliaceae), Halesia carolina, and numerous less-abundant
species. The diverse spring understory in this rich mesic cove hardwood forest includes
several Trillium L. (Liliaceae) species, downy rattlesnake plantain (Goodyera pubescens
[Willdenow] R. Brown: Orchidaceae), New York Fern (Thelypteris noveboracensis [L.]
Nieuwland: Aspidiaceae), bloodroot (Sanguinaria canadensis L.: Papaveraceae), and many
others. I established and surveyed 15 plots at this site (2 0.6 ha) because this sampling
area seemed to include an adequate number of trees and lianas. The first three plots were
west of the road and the remaining 12 were cast of the road. I was unable to identify lianas
in the genus Vitis L. (Vitaceae) to species because there were no leaves, flowers or fruits

when the data were collected (7 through 14 May 1997).

Conley Creek. The second sampling site was another mesic cove forest (see general
description above), but was in Swain County, North Carolina (Figure 4). This site is

adjacent to Conley Creek Road (State Road 1177) near Pigpen Flats by a series of three

switchbacks in the unpaved road, in the Green Creek Quadrangle (35°21 ’North and

83°22’West). In this area the elevation ranges from about 1160 to 1280 m. This site is

located on the east-facing slope of the Alarka Mountains. The canopy includes
Liriodendron tulipifera, Acer rubrum, Magnolia fraseri, and oak species (Quercus: L.
Fagaceae). Swain County receives between 1100 and 1320 mm of precipiation per year
(Clay et al. 1975). I selected Conley Creek because the dutchman’s pipe, Aristolochia
macrophylla is present in the area. Due to time constraints, I established and surveyed only

seven plots at this site (0.28 ha; 21 May through 2 June 1997).

Clifftop Vista. The third site was in a white oak-dominated forest community. White oak
forests usually occur at high elevations and have white oak as the dominant species (Roe

and Mansberg 1984).

25

 

Figure 4. Conley Creek field site in the Greens Creek Quadrangle, Swain County, NC.
(Modified from United States Geological Service.)

26

This particular site was located in Cliffside Recreation Area in the Cowee Mountain Range

(Figure 5), and was in the Highlands quadrangle of Macon County, North Carolina (about
35°5’ North and 83°l4’ West). This xeric site is part of the Van Hook White Oak Stand

located at the summit along the east-facing slope of Clifftop Vista Trail that faces the scenic
Cullasaja River Gorge. This site is located at a moderately high elevation (about 1150 to
1180 m). The canopy is dominated by white oak (Quercus alba L.), but also includes
pignut hickory (Carya glabra (Miller) Sweet: Juglandaceae), scarlet oak (Quercus coccinea
Muenchhausen), white pine (Pinus strobus L.: Pinaceae), and Acer rubrum. The
understory is dense in Vaccinium L. spp. (Ericaceae) thickets, with few herbaceous plants.
Biotite schists and gneisses underlie the area. The strongly acidic soil has a clay-
loam texture, and is in the dry-mesic to dry-xeric moisture class. The dark brown humus
loam topsoil is about 15 cm deep with a pH ranging from 4.3 to 5.3. The yellowish-brown
clay loam subsoil is more than 33 cm deep and has a pH of 5.4 (Roe and Mansburg 1984).
I chose this site to investigate whether lianas would be present in a white-oak dominated
forest. I established and surveyed only five plots at this site because no lianas were found

within the plots (0.2 ha; 9,15 May and 3 June 1997).

Costa Rica site description and sampling technique

A similar sampling technique was repeated in the seasonally dry tropical forest of Palo
Verde in Costa Rica (Figure 6). Palo Verde National Wildlife Refuge is located in the
south-central Guanacaste Province of lowland Costa Rica, and was leased by the
Organization for Tropical Studies in 1968 (Hartshom 1983). The park is considered a

tropical dry forest in the Holdridge life zone system, receiving between 1000 and 1500 mm
of rainfall per year with an average biotemperature above 24°C (Hartshom 1983). The

southward shift in the intertropical convergence zone results in a dry season from

December to March when northeast tradewinds dry the land (Coen 1983).

27

its .......~.H.......3.._

n.\.///\ .... .“U..

1
r l“

\
fl

 

 

NC

28

Figure 5. Clifftop Vista field site in the Highlands Quadrangle, Macon County,

(Source: United States Geological Service.)

NICARAGUA

Palo Verde
./ CARIBBEAN
SEA
COSTA RICA
PACIFIC PANAMA
OCEAN

Figure 6. Map of Costa Rica showing location of Palo Verde National Wildlife Refuge,
Costa Rica.

29

Over time, the limestone cliffs have eroded and produced soil for the lower slopes,
whose colluvial deposits consist of limestone, clay and silt. South of the limestone cliffs,
well-drained colluvial soils support a mature forest (Hartshom 1983). This area is one of
the most mature and least-disturbed patches of tropical dry forest in the Guanacaste
Province. The elevation is approximately sea level.

In January 1970, Gary Hartshom and a group of scientists from the University of
Washington at Seattle established a 4 ha plot in this forest along the apiary road, 5 km west
of the main station. This forest represented primary tropical dry forest. They divided this

plot into 100 subplots of 20 x 20 m each (about 10°21’ North and 85 °23’ West; E.

Gonzales, personal communication). I used the ten northernmost original 20 x 20 m plots
(Plots 000.000 to 000.200) as the study site (Table 1). This sampling technique did not
allow the freedom to skip areas with canopy gaps, which differed from that of the North
Carolina sampling technique.

On 30 June 1997, I identified all trees whose diameter exceeded 10 cm at breast
height and recorded their dbh. For all trees _>_ 10 cm dbh I classified the bark type
according to the following categories: smooth, peeling, chipping (distinct plates of bark
clearly sloughing from tree), rough (sinewy but not ridged, not smooth, bark fragments
seldom chip off), or spiny (=omamented). On 1 and 2 July 1997, I identified all lianas that
were climbing the trees 2 10 cm dbh, and I categorized the climbing mechanism of each
liana as tendril climber, twiner, or adhesive disc climber (similar to classification in Putz
1984b). 1 saw no lianas utilizing adventitious roots for climbing. I included all lianas that
were rooted within 2 m of the base of the tree, and each liana stern was considered a
different individual unless it could be otherwise determined. I measured all lianas at 10 cm

above the ground.

30

Results

Liana species diversity and density

At Rich Gap, I measured 154 lianas of three species in two taxonomic families (Tables 1
and 2); 148 (=96%) of these were grapevines in the genus Vitis, which could not be
identified to species because the vines were just beginning to leaf out during sampling.

At Conley Creek, NC, I measured 63 lianas of three species in two taxonomic
families (Tables 1 and 3). Of these lianas, 45 (=71%) were Aristolochia macrophylla, but
68% of the total liana basal area was contributed by eight Vitis spp. stems (Table 3).

At Clifftop Vista, NC, there were no lianas within the established sampling area.
However, there were a few lianas outside of the plots that I noticed on the hike up to the
field site, including one wild grape (Vitis spp.) stem and several juvenile poison ivy (Rhus
radicans) stems.

At Palo Verde 157 lianas were measured belonging to at least 14 species in nine
taxonomic families (Tables 1 and 4); two species remained unidentified. Bignoniaceae,
represented by 66 lianas within five genera, was the most species-rich family. The other
nine families present were each represented by only one genus. F orsteronia spicata was the
most abundant species in the area.

Overall, at all sites, I measured 374 lianas in 37 plots (=0.04 ha each; total of 1.48

ha) at four sites (Table 1). Palo Verde was the most species-rich site (Table 1).

Tree species diversity and density
At Rich Gap, I measured 311 trees in 21 species within 13 taxonomic families (Tables 1

and 5). One-hundred-fifteen of these trees served as host to one or more lianas (=37%;
Table 5). The estimated tree basal area per hectare was 36.3 m2 (Table 1). Species-area

curves revealed that this sampling area was sufficient to include most of the species present

in this forest for both trees (Figure 7a) and lianas (Figure 7b).

31

 

 

 

 

2: a 9m wdm a: mo 5a mesa: Co C383: :38 :6 Wm N
me o 2 m2 923: mo .38? $55
wfim o QNN Ema a; mo Com mace: mo 528:: :38
awfi H and _ o ANN H 3 .m mm; H 3.3 mm H SE Com was: he hopes: SEE
me; o wed med cob com was“: he Hogan: :88
mw~ c an of $83 “ea com «Ba Ewan new: 588
53 o ohm 3mm $88 «one Ewan new: BEBE
>2 o m@ v2 was“: Co 328::
m Hade— 3 o m N 858$ me 898::
m _ ~32 an o m “mac. 3 m we“: 8 860% new: me .8985: 3.33
gm 5cm Qmm mdm $25.5 a: Ba «Ba EBB cub 83:53
33 e8: 80: New: pee ea 8 «on EB 8: 52:
mmw mag Sn L who“ A83 83 Com unaccomfisoho Ewan cob :38
oemm exec $3 “Sum 35: 89: Ho 28 5:» much we amoeba
v6 @6— od~ th SE com moo: mo coca—E :38
wmmvo fimmmh 333 33.3 $88 «pa Ewan cob 3883
Ommw movw 2: 2 of mm 3.3 oucBoEzoho Ewan cob 3883
mm c an m: mean: 9.58 8 25 55 much we “BEE.
em mm mm L M _m 89: me 898::
cm m S 2 8:85 we confine Eu :8 2A
mm 2 2 _m 860% 8.: Co 89:5: mood.
oEo> 2am Sm; 3&er “.320 5:80 new 50E
SE 880 «5880 55 Z

dog 880 5 8mm one use «ECHO 552 5 8% 025 c8 859% Egg ._ 2an

 

32

Table 2. Species of lianas present in 0.6 ha at Rich Gap, NC.

 

 

family specres number of summed liana mean liana
lianas basal area basal area
(cmz) i SE (cmz)
Anacardiaceae Rhus radicans L. 1 1.5 1.5
Vitaceae Parthenocissus 5 4.5 0.9 i 0.3
quinquefolia (L.) Planchon
Vitis L. spp. 148 2387.9 16.2 i 1.2
total 154 2394

Table 3. Species of lianas present in 0.28 ha at Conley Creek, NC.

     

 

 

family species number of summed liana mean liana
lianas basal area basal area

(cmz) _-i_-_ SE (cmz)

Aristolochiaceae Aristolochia macrophylla Lam. 45 76.6 1.7 _-I; 0.2
Vitaceae Parthenocissus 10 12.5 1 3 :1; 0.9

quinquefolia (L.) Planchon
Vitis L. spp. 8 186.8 23.4 i 8.2
total 63 275.9

33

 

34

 

 

 

En: lg 32
E29 3 n no no m N 565:5
Ewes so so A _ cease:
fies no no c as Seances accuracy. soaemeoe
_EEB _.N _N _ .Em .U.< SE3: 53.388.» Ezmwzemscm oaoowufloomma
.8525 We we _ i— 398; 338.80 becomeowbom
.558 mo H md ad N dam J Eekasem omooecocamam
Ewes ed a 3 do N 950 .m .< 3V Basses nuances: sugarcane
Fee 3 H no on 4 950 .m .< 3C .5:chch assesseszi goaeoaam
.553 a: a cam 58 w .52 Set Senses usesfig aoaaocwa
see we a gm :8 a .4 A532 usages geuao
mowG 038:3 m.~ H N6 flaw o_ 5.5M ExmeocEKEEEQEeU 3882ch0
Bees 3 a mo 3: S 23> Seekers: 329 8085
megs mg a o. _ m v.08 : EA .4 esfiem 808,5
E23 3 n 3 no: om SEC .43. Ce grease 8.4%: aoficoawa
E23 3 n 3 as: mm as: 35% engages 535 amnesia
8525 _d H m6 o.o_ em >02 .0 28m: 88.3% QEESECR becomexooa/w
pee mm a pee

Emgnooe «0.3 Human «0.8 Ewan wand:

mango new: 508 new: BEES. me 398:: mass: bag

.8;— 880 .oEo> 23 “a 2 ed 5 :5on mace: mo 883m .v 038.

 

Table 5. Tree composition in 0.6 ha at Rich Gap, NC.

 

of trees trees
trees with lianas with lianas

 

 

Magnoliaceae Liriodendron tulipifera L. 126 62 49%
Aceraceae Acer rubrum L. 31 13 42%
Styracaceae Halesia carolina L. 26 7 27%
Magnoliaceae Magnolia fraseri Walter 23 5 22%
Juglandaceae Carya glabra (Miller) Sweet 15 3 20%
Fagaceae Quercus prinus L.* 14 1 7%
Fagaceae Quercus rubra L.* 14 2 14%
Comaceae Camus florida L. 13 5 38%
Betulaceae Betula lenta L. 12 3 25%
Pinaceae Tsuga canadensis (L.) Carriere 8 1 13%
Rosaceae Prunus pensylvanica L. 6 3 50%
Tiliaceae Tilia heterophylla Vent. 6 2 33%
01630636 F raxinus pennsylvanica Marshall W 4 0 0%
Fabaceae Robinia pseudo-acacia L. I]! 4 2 50%
Pinaceae Pinus strobus L. y! 2 0 0%
Lauraceae Sassafras albidum (Nuttall) Nees w 2 2 100%
Betulaceae Betula lutea Michaux w 1 0 0%
Juglandaceae Carya ovata (Miller) K. Koch l]! 1 1 100%
Fagaceae Castanea pumila (L.) Miller y! l 1 100%
Rosaceae Prunus serotina Ehrart ‘l’ 1 1 100%
Fagaceae Quercus coccinea Muenchhausen * 1 1 100%
total 311 115

*= grouped in analysis as "hardbarked oaks."
up: grouped in analysis as "other trees."

35

Table 5. (Ctn'd)

 

 

family tree species summed tree number summed liana
basal area basal area
(cmz) (cmz)
Magnoliaceae Liriodendron tulipifera 95747 1399.5
Aceraceae Acer rubrum 33458 204.3
Styracaceae Halesia carolina 5730 120.4
Magnoliaceae Magnolia fi'aseri 26142 207.8
Juglandaceae Carya glabra 14393 38 .8
Fagaceae Quercus prinus* 9303 1 1.9
Fagaceae Quercus rubra* 15130 3 1 . 1
Comaceae Camusflorida 1203 56.8
Betulaceae Betula lenta 3528 1 1.5
Pinaceae Tsuga canadensis 1587 4.9
Rosaceae Prunus pensylvanica 2942 50.2
Tiliaceae Tilia heterophylla 1034 21 .9
Oleaceae F raxinus pennsylvanica I}! 958 0.0
Fabaceae Robinia pseudoacacia I]! 1929 13
Pinaceae Pinus strobusl/I 738 0.0
Lauraceae Sassafras albidumu/ 455 8.0
Betulaceae Betula luteavl 107 0.0
Juglandaceae Carya ovatall/ 754 15 1 .8
Fagaceae Castanea pumilaw 375 23.5
Rosaceae Prunus serotina W 979 9- 1
Fagaceae Quercus coccinea* 1036 30.7
total 217529 2394

*= grouped in analysis as "hardbarked oaks."

w: grouped in analysis as "other trees."

36

 

 

 

A. trees ‘1

 

 

 

 

 

.3

gzs

.3 20 —

E 15

E i

a 10 —~

O)

>

a 54

a 0 : : : i - : 1

3 0.00 0.08 0.16 0.24 0.32 0.40 0.48 0.56
cumulative area (ha)

 

 

 

B. lianas

 

t—tI-‘NN
OUIOUI

 

 

l 1 l J 1

1,

.00 0.08 0.16 0.24 0.32 0.40 0.48 0.56

cumulative area (ha)

cumulative number of species
C O LII

 

 

 

Figure 7. Cumulative number of species for cumulative area sampled at Rich Gap for (a)
trees and (b) lianas.

37

At Conley Creek, I measured 133 trees in 19 species within 10 taxonomic families

(Tables 1 and 6). Thirty-nine of these trees served as host to one or more lianas (=29%;
Tables 1 and 6). The estimated tree basal area per hectare was 33.9 m2 (Table 1). Species-

area curves revealed that this sampling area was sufficient to include most of the species
present in this forest for both trees (Figure 8a) and lianas (Figure 8b).
At Clifftop Vista, I measured 98 trees in 11 species within five taxonomic families

(Tables 1 and 7); zero of the 98 trees served as host to a woody vine. The estimated tree
basal area per hectare was 36.7 m2 (Table 1). Species-area curves revealed that this

sampling area was sufficient to include most of the tree species present in this forest
(Figure 9).
At Palo Verde, I measured 94 trees in 33 species within 20 taxonomic families

(Tables 1 and 8); 52 of the trees served as host to one or more lianas (=55%; Tables 1 and
8). The estimated tree basal area per hectare was 23.6 m2 (Table l). Species-area curves

revealed that this sampling area was insufficient for trees (Figure 10a) and lianas (Figure
10b).

Overall, at all sites, I measured 636 trees, 206 of which hosted one or more lianas
(32%). Palo Verde was the most species-rich site in terms of tree species (Table l). The
highest percentage of trees hosting one or more lianas was also at Palo Verde (Table 1).
However, the largest estimated tree basal area per hectare was at Rich Gap Road and

Clifftop Vista, and the lowest was at Palo Verde (Table 1).

38

2.385 85o:
8 £92.25 5 c&=o&us .mxao woe—392w: mm max—«:5 E gasp—my .0 Z £0er 3:80 am a: mad E 536888 8; .o 03$.

39

 

 

QRN 8 8223 mm 22 2.82
22.22 22 N88 so 22 2 .2. 5.22022 832 3.822 882.2232
”N o 882 .282: 2 2 .2. .2 382 55.5 80822022
22.22 o :m .8 o 2 .2. 22802 .02 822222 8.8 2.8.6 82.82.222.282
22.22 2 82 so 22 2 .2. 232a? 2.8882288: 80822882
2.2 2 £8 28222 2 2 .2. 523m 8228.... .3552 88282
no N $222 $82 2 2 .2. .2 3288-883 2.2332 2.085
22.22 o 22m so o N .2. 22802 .02 225% 2.288288 85.2 2.8.2.5222
N. m m SN ..8222 N N .2 3.2288 8.22.: 2.08825
3 2 82 soon 2 N .2. 2238., 328288.22 8: 382.2222.
22.22 22 cmmN so 22 2. .2. .2 98.258 2.228%: 882.228.2222
ed 22 SN so o 2. .2. .2 82.. 38.20 808228
c. m m mam 88% m o .4 Ezowzufibémm 864.. 808.2812.
2 .o N $2N .89. N 8 228222 222226 2.8.2282 8.8 2.082.222.2292

a2 2 m N2.2oN 8N N N .2 8.2%....86 0882228

53 N mm s: N N2 *2 use .3238 8088.2

8.2: m ow 0N. 2 N m 2.2 *2 3.3 .3228 8088.2
9o 2 8 .8 2 2 28% 2202222): 28222 83.6 8082.22.23

3A 2: N2 sNN N NN .2 3.232... 8.238.222 8082288:

32 NN 2m .82. 2 on .2 see: 8.. 808282..

2.2223 22.5.22 2.2223 35.22 22223 85.22 22223 822

«2a :83 mo no.8 Emma 8922 moo: .«o
2228: BEBE 238:: 8.: 3:22:25 Mo 25822 no 23225: 238:: 860% 85 £22.23

 

 

40

 

 

 

 

 

 

 

 

 

 

 

N
O

 

fl
L11
1

LII
1

 

       
 

4

.00 0.04 0.08

cumulative number of species
fl
o

CO

 

 

 

l

0.12 0.16

cumulative area (ha)

 
 
   

0.20

 

0.24 0.28

F“ 'i
m A. trees i
i 20 l
«3 15 --

*2

:3 10 »

I:

E

E

8 0 4 i i i i L

0.00 0.04 0.08 0.12 0.16 0.20 0.24 0.28
cumulative area (ha)
B. lianas

 

Figure 8. Cumulative number of species for cumulative area sampled at Conley Creek for

(a) trees and (b) lianas.

41

 

Table 7. Tree composition in 0.2 ha at Clifftop Vista, NC. No lianas were present inside

 

 

 

 

the plots.
Tami-1y tree species 11.1111er
of basal area
trees (cmz)
Fagaceae Quercus alba L. 37 27902
Fagaceae Quercus coccinea Muenchh. 22 51 18
Juglandaceae Carya glabra (Miller) Sweet 8 1834
Pinaceae Pinus strobus L. 7 13681
Fagaceae Quercus velutina Lam. 6 7023
Aceraceae Acer rubrum L. 6 5157
Fagaceae Quercus prinus L. 5 270
Fagaceae Quercus muehlenbergi i En gelmann 3 107
Fagaceae Quercus rubra L. 2 5402
Ericaceae Kalmia latifolia L. 1 2105
Juglandaceae Carya tomentosa (Poiret)Nutta11 1 4732
total 98 7333 1

 

u—
N

 

j—d
O
1
T

 

 

 

cumulative number of species

ONAONOO

 

0.04 0.08 0.12

cumulative area (ha)

0.16

 

0.20

 

 

Figure 9. Cumulative number of tree species for cumulative area sampled at Clifftop Vista.

42

Table 8. Tree composition in 0.4 ha at Palo Verde, Costa Rica. (ba=basa1 area)

 

family tree species

Tiliaceae Luehea candida (DC.) Mart.

Bignoniaceae Tabebuia ochracea (Cham.) Standl.

Rubiaceae Calycophyllum candidissimum (Vahl.) DC
Fabaceae Caesalpinia eriostachys Benth.

Rubiaceae Chomelia spinosa J acq.

Theophrastaceae Jacquinia spp.

Fabaceae Pithecellobium lanceolatum (Humb. & Bonpl. ex. Willd.) Ben
Fabaceae Lysiloma divaricatum (J acq.) J. F. Macbr.
Meliaceae Trichilia hirta L.

Rubiaceae Guettarda macrospermum Donn. Sm.

Olacaceae Ximenia americana L.

Anacardiaceae Astronium graveolens Jacq.

Bombacaceae Bombacopsis quinata (J acq.) Dugand
Flaticourtiaceae Casearia tremula (Griseb.) Griseb. ex. W. Wright
Fabaceae Lonchocarpus costaricensis (Donn. Sm.) Pittier
Anacardiaceae Spondias mombin L.

Salviniaceae Thouinidium decandrum (Humb. & Bonpl.) Radlk.
Fabaceae Acacia comigera L.

Fabaceae Albizia adinocephala (Donn. Sm.) Britton & Rose
Fabaceae Albizia caribbea (Urban) Britt.

Burseraceae Bursera simaruba (L.) Sarg.

Polygonaceae Coccoloba Browne spp.

Cochlosperrnaceae Cochlospermum vitifolium (Wilde) Spreng.
Boraginaceae Cordia alliodora (Ruiz & Pav.) Oken
Sterculiaceae Guazuma ulmifolia Lam.

Menispermaceae Hyperbaena tonduzii Diels

Theophrastaceae Jacquinia nervosa C. Presl

Fabaceae Machaerium biovulatum Micheli

Sapotaceae Manilkara zapota (L.) Royen

Theophrastaceae Mantingia calabura L.

Rubiaceae Randia thurberi S. Watson

Simaroubaceae Simaruba glauca DC.

Apocynaceae Stemmadenia obovata (Hook. & Am.) K. Schum.

 

43

Table 8. (Con't)

 

 

tree species number number of percent of tree number liana
of trees trees ba of ba
trees with lianas with lianas (cmz) lianas (cm2)
Luehea candida 14 5 36% 9907 15 209.5
Tabebuia ochracea 8 4 50% 1925 8 194.6
Calyc0phyllum candidissimum 7 6 86% 27897 17 320.7
Caesalpinia eriostachys 6 2 33% 7664 5 233.3
Chomelia spinosa 6 6 100% 2323 23 178.2
Jacquinia pungens 6 4 67% 843 12 27.8
Pithecellobium lanceolatum 5 2 40% 2923 3 0.9
Lysiloma divaricatum 4 3 75% 5757 1 1 53.7
Trichilia hirta 4 2 50% 976 6 3 .5
Guettarda macrospermum 3 1 33 % 1621 3 16. 1
Ximenia americana 3 1 33% 1476 1 10.2
Astronium graveolens 2 2 100% 2268 6 105.4
Bombacopsis quinata 2 1 50% 8586 2 1.2
Casearia tremula 2 0 0% 381 0 0.0
Lonchocarpus costaricensis 2 2 100% 637 2 0.6
Spondias mombin 2 1 50% 1608 13 19.0
Thouim'dium decandrum 2 2 100% 298 4 6.6
Acacia comigera 1 0 0% 156 0 0.0
Albizia adinocephala 1 1 100% 1412 l 9.5
Albizia caribbea 1 0 0% 594 0 0.0
Bursera simaruba l 0 0% 1 13 0 0.0
Coccoloba spp. 1 0 0% 99 0 0.0
Cochlospermum vitifolium 1 0 0% 1750 0 0.0
Cordia alliodora 1 l 100% 275 3 4. 1
Guazuma ulmifolia 1 1 100% 384 2 41.9
Hyperbaena tonduzii 1 1 100% 145 2 119.9
Jacquinia nervosa 1 1 100% 625 5 30.8
Macairea biovulatum 1 l 100% 281 3 50.3
Manilkara zapota 1 0 0% 10696 0 0.0
Mantingia calabura 1 1 100% 123 9 189.1
Randia thurberi 1 0 0% 1 1 1 0 0.0
Simaruba glauca 1 0 0% 194 0 0.0
Stemmadenia obovata 1 1 100% 194 1 0.2
total 94 52 94238 157 1827

 

A. trees
35

30 —»
25 4
20 ..
15 —~
10 ~~
5 1+
0 .1 + i
0.00 0.04 0.08 0.12

 

 

 

 

 

l 1 1 1

0.16 0.20 0.24 0.28 0.32 0.36 0.40

cumulative number of species

cumulative area (ha)

 

 

 

 

 

B. lianas

 

I» D.)
0 UI
l

15 -.
10 ~~

 

 

1 1 1 1 l

.00 0.04 0.08 0.12 0.16 0.20 0.24 0.28 0.32 0.36 0.40

cumulative number of species

 

COLA

cumulative area (ha)

 

 

 

Figure 10. Cumulative number of species for cumulative area sampled at Palo Verde for
(a) trees and (b) lianas.

45

Liana host-species preference

There were two ways to calculate expected frequencies of lianas on potential host tree
species present in a forest. I utilized both the number of tree stems available per species
and the surface area available to climb per species, which is best approximated by tree basal
circumference (Daniels and Lawton 1991, Talley et al. 1996). For statistical validity, the
less abundant tree species were grouped in the analysis. Additionally, the hardebarked
oaks were grouped at Rich Gap and Conley Creek because they have similar non-chipping,
rigid bark morphology, and for comparability with earlier work (Table 5; Talley et al.
1996). At Rich Gap, based on the number of stems available to climb, for all lianas

present, the number of lianas observed climbing each tree species was independent of
relative abundance of the tree species (X211=26.76, P<.005; Table 9). This same trend was
seen for the most abundant vine at Rich Gap, Vitis spp., when examined alone

(X2,,=26.61, P<.006; Table 9).

Table 9. Influence of tree species on liana distribution at Rich Gap (df=11) and Conley
Creek (df=7) when test is based on number of trees available or tree basal circumference
available to climb. Chi-square and P-values are shown.

     

 

 

"_M_ _ —'I“meal—1rccemferen
Rich Gap
All lianas (26.76) <0.005 (28.96) <0.002
Vitis spp. only (26.61) <0.006 (27.35) <0.005
Conley Creek
All lianas (24.68) <0.0009 (27.89) <0.0003
Aristolochia only (35.01) <0.00001 (23.39) <0.002

 

Liriodendron tulipifera served as host to 30% more lianas than expected, contributing 21%

of the Chi-square value. The hard-barked oaks hosted only 23% of that expected based on

46

the number of stems available, contributing 32% of the Chi-square value. Most other tree
species hosted slightly fewer lianas than expected based on their relative abundances.

A similar trend was seen at Rich Gap utilizing summed tree basal circumference,
which represents surface area available to climb. Again, for all lianas, climbers were not

distributed among tree species based on their relative available circumferences

(X211=28.96, P<.002; Table 9). The same was true for the most dominant liana at Rich

Gap, Vitis (X211=27.35, P<.005; Table 9). Comusflorida hosted twice that expected

based on basal circumference, contributing 14% of the Chi-square value. The hard-barked

oaks hosted fivefold fewer lianas than expected based on their summed basal circumference

available to climb, contributing 38% of the Chi-square value. Carya glabra hosted about

half that expected, contributing 10% to the Chi-square value. Other species hosted a few

more or less than expected based on species contribution to surface area available to climb.
At Conley Creek, based on the number of trees available to climb, for all lianas,

Chi-square analysis showed that lianas were distributed on host trees independent of the
relative abundance of potential hosts (X27=24.68, P<.0009; Tables 6 and 9). Similarly, the
most abundant species, Aristolochia macrophylla, when considered alone, was distributed
independent of host relative abundance (X27=35 .01, P< .0000]; Table 9). Carya glabra

hosted only 8% of the lianas expected from its relative stem density, contributing 40% of
the Chi-square value. The hard-barked oaks hosted only 42% of the lianas expected,
contributing 39% of the Chi-square value.

A similar pattern occurred at Conley Creek when the analysis was based on relative

summed tree basal circumference. Lianas were distributed independent of the surface area
available to climb per tree species. This trend held true for all lianas (X27=27.89, P<.0003;
Table 9), and also when the most dominant species, Aristolochia macrophylla, was
considered alone (X27=23.39, P<.002; Table 9). Acer rubrum hosted twice that expected

based on its relative surface area available to climb, contributing 50% of the Chi-square

47

value. Carya glabra hosted 14% the lianas expected from its contribution to surface area
available to climb, contributing 19% of the Chi-square value. The hard-barked oaks hosted
one-half that expected, contributing 17% of the Chi-square value.

At Palo Verde there were too few individuals representing each tree species to
enable examination of the relationship between tree species and liana presence. Overall,
host-tree species was important in explaining liana distribution at the two sites where
statistical analysis was possible (Rich Gap and Conley Creek in NC). This was true
regardless of whether the number of tree stems available to climb or the summed-tree-basal-

circumference available per species was utilized to calculate the number of lianas expected

per species.

Liana host-size preference

In addition to species of host, size of the host tree influenced liana distribution among
potential hosts at Rich Gap. All species of trees were pooled and grouped into 10 cm-
diarneter size classes (Table 10 and Appendix A) and Chi-square analysis was performed
based on the number of trees available to climb per size class and tree basal circumference
per size class. First, expected frequencies of lianas per size class were calculated based on
the number of trees available to climb, and when all lianas were included in the calculation,

lianas were not distributed in proportion to the number of trees in each size class

(X25=24.23, P<.0002; Table 11). This same trend was seen when Vitis was considered

alone (X25=26.68, P<.0001; Table 11). Trees in the smallest size class (105de cm dbh)

hosted 68% fewer lianas than expected, contributing 27% of the Chi-square value.
However, trees in the largest size class (260 cm dbh) hosted three times as many lianas as
expected by the proportion of trees in this size category, contributing 51% of the Chi-
square value. Thus, based on host relative abundance there seemed to be fewer lianas than

expected on smaller hosts, but more lianas than expected on the larger hosts.

48

Table 10. Size classes of trees of all species pooled in 0.6 ha at Rich Gap, NC.
Parentheses indicate exclusion of the end point within the size class, whereas brackets
indicate inclusion.

m
srze class number of summed tree number of summed lrana mean number

 

 

(cm dbh) trees basal area lianas basal area of lianas per

(cmz) (cmz) tree i SE
[10-20) 128 20546 43 452 0.34 _-_l; 0.05
[20-30) 91 43097 49 649 0.54 :t 0.08
[30-40) 48 44331 25 507 0.52 i 0.11
[40-50) 21 34635 15 316 0.71 3; 0.18
[50-60) 18 42193 14 265 0.78 i 0.21
260 5 32727 8 205 1.60 i 0.81
total 31 1 217529 154 2394

Table 11. Influence of host tree size on liana distribution at Rich Gap, Conley Creek and
Palo Verde when test is based on number of trees available or tree basal circumference
available to climb. Chi-square and P-values are shown. Each case had 5 degrees of
freedom. NS=Not significant.

     

 

 

a—— _a1 circumfrence

Rich Gap

All lianas (24.23) <0.0002 (6.34) NS

Vitis spp. only (26.68) <0.0001 (5.07) NS
Conley Creek

All lianas (8.52) NS (45.62) <0.0001

Aristolochia only (14.20) <0.02 (45.33) <0.0001
Palo Verde (35.29) <0.0001 (72.74) <0.0001

 

49

This picture was clarified when I evaluated the effect of tree size on lianas based on

summed tree basal circumference available to climb per size class. For all lianas, lianas

were distributed according to host size (X25=6.34, P>.38; Table 11). Similarly, there was

no effect of size class when only Vitis was considered (X25=5.07, P>.41; Table 11).

Thus, the size of host does not seem to influence liana distribution.

In light of the above findings, it appears that liana density per surface area available
to climb was constant at Rich Gap. Larger trees have more surface area available to climb,
whereas smaller trees have less surface area available to climb. The smallest size classes
had fewer lianas per tree but the number of lianas per size class was proportional to the
relative surface area contribution per size class. One tree species that defied this overall
trend was flowering dogwood, Camusflorida, which hosted more lianas than predicted by
its contribution to forest available tree—surface-area, but all of the individuals were in the
smallest size class.

Similar to Rich Gap, host size influenced liana distribution at Conley Creek. Trees
were grouped into 10 cm diameter size classes (Table 12 and Appendix B) and Chi-square
analysis was done based on number of trees available to climb per size class and tree basal

circumference per size class. Based on number of trees available to climb, lianas were
distributed in proportion to relative abundance of size classes (X25=8.52, P>.12; Table 11).
However, the most common vine, Aristolochia macrophylla, dutchman’s pipe, when
considered alone, was distributed independent of relative abundance of size classes
available to climb (X25=14.20, P<.02; Table 11). For A. macrophylla, the smallest size

class hosted more lianas than expected based on relative stem densities, whereas all larger
size classes hosted fewer than expected. This suggests that A. macrophylla was utilizing
smaller hosts, and the addition of Vitis spp. and Parthenocissus quinquefolia to the analysis
removed the effect of host size because these two genera were distributed independent of

host size.

50

Table 12. Size classes of trees of all species pooled in 0.28 ha at Conley Creek, NC.
Brackets indicate inclusion of the end point within the size class, whereas parentheses
indicate exclusion.

=
srze class number of summed tree number of summed lrana mean number

 

 

(cm dbh) trees basal area lianas basal area of lianas per
(cmz) (cmz) tree i SE
[10-20) 58 9052 34 48 0.59 i 0.14
[20-30) 26 13282 16 145 0.62 i 0.19
[30-40) 25 24088 7 9 0.28 _-l_-_ 0.13
[40-50) 14 22222 4 6 0.29 i 0.22
[50-60) 6 13803 0 0 0
_>_60 4 12612 2 69 0.5 i 0.5
total 133 95059 63 276

At Conley Creek, lianas were distributed independent of the surface area available

to climb per species (for all lianas, X25=45.62, P<.0001, Table 11; Aristolochia

macrophylla only, X25=45.33, P<.0001, Table 11). For all lianas, the two smallest tree

size classes hosted more lianas than expected, whereas the four larger size classes all
hosted fewer lianas than expected. This suggests that liana density is not constant over
surface area available to climb. Instead, lianas at Conley Creek, the most abundant being
A. macrophylla, are preferentially utilizing smaller hosts.
Lianas were not seen at Clifftop Vista. Trees size classes (Table 13) and species

representation amongst tree size classes (Appendix C) are provided.

As with the two above sites in NC where lianas were present, host size
influenced lianas at Palo Verde. Tree size classes were constructed as above (Table 14 and

Appendix D).

51

Table 13. Size classes of trees of all species pooled in 0.2 ha at Clifftop Vista, NC.

size class number of summed tree

 

 

(cm dbh) trees basal area
(cmz)
[10-20) 37 6401
[20-30) 20 9734
[30-40) 28 25074
[40-50) 5 8084
[50-60) 4 9273
_>_60 4 14765
total 98 7333 1

Table 14. Size classes of trees of all species pooled in 0.4 ha at Palo Verde. Parentheses
indicate exclusion of the end point within the size class, whereas brackets indicate
inclusion.

 

 

 

size class summed tree - - a ram mean num.-
(cm dbh) trees basal area lianas basal area of lianas per
(cmz) (cmz) tree i SE
[10-20) 48 8308 74 831 1.54 i 0.36
[20-30) 14 6505 37 273 2.64 i 0.84
[30-40) 10 9848 15 336 1.50 i 0.58
[40-50) 9 13249 11 33 1.22 i 0.88
[SO-60) 5 11433 9 236 1.80 i 0.92
260 8 44895 11 119 1.38 i 0.71
total 94 94238 157 1827

52

Based on number of trees available to climb, lianas were not distributed in proportion to
their relative abundances (for all lianas, X25=35.29, P<.0001; Table 11). The largest

contribution to the Chi-square value was from the smallest size class (105x<20 cm dbh;
72%), which hosted fewer lianas than expected based on the relative host stem densities.
When the test was based on tree basal surface area available to climb the two

smallest size classes hosted more lianas than predicted, whereas all larger size classes
hosted fewer lianas than expected (for all lianas, X25=72.74, P<.0001; Table l 1). The

Smallest size class (105x<20 cm dbh) hosted twice that expected by chance, contributing
38% of the Chi-square value; the largest size class (260 cm dbh) hosting one-fourth that
expected by chance, contributing 27% of the Chi-square value.

To summarize, tree size class was important in explaining liana distribution at all
sites where lianas were present, but with different trends at different sites. At Rich Gap,
lianas were found in proportion to the host surface area available to climb, resulting in more
lianas utilizing trees in larger size classes, where more climbing surface area was available.
At Conley Creek, size class did not influence all lianas. However, when the dominant
liana, Aristolochia macrophylla, was considered alone, it showed a preference for smaller
hosts, independent of surface area available to climb per size class. Lianas at Palo Verde
also showed a preference for smaller hosts, and were not found in proportion to host tree

surface area available to climb amongst various size classes.

Spatial distribution of lianas
At Rich Gap, Conley Creek and Palo Verde, there was no clear relationship between the
tree basal circumference and the number of lianas per tree. There was also no linear
relationship between the tree basal circumference and the liana basal area per tree.

There was a trend toward larger trees hosting more lianas per tree at Rich Gap, but
this was not statistically significant (Figure 11). There was an overall mean of 0.49 lianas

per tree at Rich Gap (Table 1). The observed number of lianas per tree did not differ from

53

that expected from a Poisson random distribution (X23=4.80, P>. 18) nor from that

expected from a negative binomial clumped distribution (X22=4.91, P>.08, k=4.0; Figure

12a). This indicates that at Rich Gap, the number of trees hosting a certain number of
lianas was as expected by chance.

At Conley Creek, there was no trend in the number of lianas per tree with increased
size class (Figure 11), but this is difficult to interpret because Vitis spp. and Aristolochia
macrophylla are both included. There was an overall mean of 0.48 lianas per tree at Conley

Creek (Table 1). The observed number of lianas per tree differed from that expected from a

random Poisson distribution (X25=l 3.09, P<.02), but did not differ from that expected

from a negative binomial clumped distribution (X24=1.67, P>.80, k=0.4; Figure 12b).

This indicates that there were more trees hosting more than one lianas than expected.
There is no trend in the number of lianas per tree with increased size class at Palo
Verde Wildlife Refuge (Figure 11). There was an overall mean of 1.68 lianas per tree.

The observed number of lianas per tree differed from that expected from a Poisson

distribution (X212=419500, P<.00001), but did not differ from that expected from a

negative binomial distribution (X211=14.1, P>.23, k=0.65; Figure 12c).

Palo Verde had the highest mean number of lianas per plot, but this was not
statistically higher than the other two sites with lianas (Table 1). Palo Verde also had the
highest mean number of lianas per tree (Table 1). The number of lianas per tree did not

vary systematically with tree size class except for at Rich Gap Road where larger trees

tended to host more lianas per tree (Figure 11).

54

dean—axe 32862: $355.52 326:3 .326
SB 65 E 358 95 mo comma—o2: 3.865 38285 .235 325 .28 momma—o 82m cob 28 3b 2a «was: Mo 23822: ow86>< .2 6222mm

E23 $26 3% 85

ooA 8062 Smtoi ontofl Smtofl 8WD:

 

 

   

 

   
 

 

 

 

 

 

 

, , - o
m

1... — o

u

m

12. N W

85> 2&- .m

.2826 52228- w

new 2262 n. t m a

55

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

    

 

 

 

 

 

 

 

 

80%
70% ii unegative binomial A
60% d 5 lPoisson
g 50% -- 1‘
‘H 40% _. 3; _ I ranas
g 30% ~~
§ 20% _.
a 10% —~ 9?
0% - *
012345678910111213
predicted and observed number of lianas per tree
80%
70% ‘ nnegative binomial B
m 60% 7" lPoisson
g 50% 4“ Ilianas
a5 40% —~
‘5 30% ~~
32’ 20% *
8. 10% —» 5
0% * E: _ _
012345678910111213
predicted and observed number of lianas per tree
80%
70% -. Dnegative binomial c
m 60% —~ lPoisson
g 50% -~ Ilianas
«5 40% j
a 30% «~
§ 20% _.
8 10% .. g .
0% _L 5': . I: I u .. _ _ _ _ i
012345678910111213
predicted and observed number of lianas per tree

 

 

 

Figure 12. Percentage of trees with a specified number of lianas per tree predicted by the
negative binomial (clumped) and Poisson distributions (random) given parameters
observed in actual liana distribution (black) for (a) Rich Gap, (b) Conley Creek, and (c)
Palo Verde.

56

Relationship between host bark type and liana climbing mechanism

At Palo Verde, I examined the relationship between tree bark type vs. liana climbing
mechanism because the liana flora was diverse with three distinct climbing mechanisms
(Table 4). The data were analyzed in two ways. Each liana could be considered one
individual observation, or each tree, serving as host to zero to many lianas, could be
considered one observation. When each liana was treated as one individual observation

(N=157; Figure 13a), there was a non-zero correlation between liana climbing mechanism
and host bark type (contingency table: X24=14.6, P<.006; non-zero correlation value =

4.020, df=1, P<.05). The greatest contribution to the Chi-square value was from the
association between twining lianas climbing trees with chipping bark (48% of Chi-square
value). However, when each tree is considered as one observation (N=94; Figure 13b),

there was no relationship between tree bark and liana climbing mechanism (contingency

table: X28=10.57, P>.22).

57

0252:0508
wing—0 022202; was 205 00:0: .20 22292022822 0.2 223220 8?. .6820 bowgao 0:0 :05 0208 E 00:0: @0200: $050 .9220:

on 0028: 00022 088 8008:”. 006 0200500 .20 80532 22382.0 2.502202 03 205 00:0: 0028; 205 000.22 mo 222282205 3v 02.223 wcanfio
20:0 .5825 .558 532 00002 00 2222022022 02 :39? 8?. 3.2.23 35% can mam—00m 0002025 50080 20 :wse .wcfimEo 223 000.22
92525—0 002030 0203 205 8022250 006-030.0500 2o wcfifizitccfl .wEEE .20 222202205 3 .003 0280 .028, 0:3 .2 020wE

58

Emamnooe

 

 

 

 

 

 

 

 

 

 

 

 

 

0:550 2a: 25 as 80
8:0: @595 swan: £88m
§o
00:0: 0:. $8
wigs»:— gov
30:8-
momfia ‘ 058
000 “An. $3
$03
m hm _ u: C u: En: on":
093 0:3 005 8030508 35:50 «:0:
000: 08% E250 wEEE
,. ‘ $5
. 0&8
58E: ‘ $3
smack I
mcamfion 0&8
‘ 0&8
< 058“

van: 3 H: 5 _ Ha own:

seen .10 seam] JO afimuaomd

seen .10 scum] J0 afieluaolad

59

Discussion

Species diversity and density

The seasonally dry tropical site, Palo Verde, had the highest species richness in both lianas
and trees. The number of lianas per plot and the mean liana basal area per plot was also
greatest at Palo Verde (Table 1). This follows expectations that the forested tropics are
generally more species rich than even the richest temperate areas. However, it is important
to note that total sampling area differed at the sites, which may influence the number of
species observed at each site. Species-area curves for the three temperate sites suggested
that the sampling area was large enough to include most of the species of lianas present
within these forests (Figures 7b and 8b), whereas the sampling size was not large enough
in Costa Rica for the number of liana species to level off with increased sampling area
(Figure 10b).

The two North Carolina mesic cove forests (Rich Gap and Conley Creek)
supported about the same absolute number of lianas per sampling area (Table 1), which
was expected because they are similar forest types. These two sites also supported the
same number of liana species, but they were nonetheless different species (Tables 2 vs. 3).
Aristolochia macrophylla was present at Conley Creek but absent from Rich Gap. This
species is listed in Radford et al. (1968) as “infrequent,” which indicates that although Rich
Gap may be acceptable habitat, it may not have been dispersed to the area. Similarly,
poison ivy (Rhus radicans) was present at Rich Gap but absent at Conley Creek. This was
not an ideal habitat for this species, which prefers disturbed places.

Palo Verde had the highest mean number of lianas per plot (15.7 lianas per plot;
Table 1). This number is higher than all sites in North Carolina, but is one fourth the 64
lianas per 20 x 20 m that Putz (1984b) reported for Barro Colorado Island (BCI), Panama.
Palo Verde is dryer than BCI, which may explain its lower liana densities (Palo Verde

receives 1500 mm of rain per year (Hartshom 1983) whereas BCI receives 2600 mm

60

(Todzia 1986)). Additionally, Putz included every liana present, rather than excluding
lianas that were not clearly climbing a host as in this study.

It is interesting to examine at each site the number of lianas greater than or equal to
2.5 cm diameter per 0.1 hectare because this figure is comparable with previous studies
(Table 15). This number was highest at Rich Gap (20.8 lianas/ 0.1 ha), but I chose this site
because it had many lianas, which I suspected would yield interesting data for liana
distribution. However, this site could bias comparison with previous studies. Palo Verde
contained the second highest number of lianas larger than 2.5 cm diameter per 0.1 hectare
(10.5 lianas/0.1 ha, Tables 1 and 15). This is one-eighth that observed at Charnela dry
forest where 78 climbers 2 2.5 cm diameter per 0.01 ha were seen (Gentry 1991, Table
15). However, it is unclear whether Gentry measured all lianas or just those clearly
ascending a host, as in the present study, nor is it clear at what height Gentry measured the
plants. The two mesic NC cove forests in the present study, Rich Gap and Conley Creek,
both contained more lianas 2 2.5 cm diameter than the reported North American average of
five lianas per 0.1 ha (Table 15, Gentry 1991). However, one of the sites in the present
study contained no lianas (Tables 1 and 15), suggesting that there is high variance in the
densities of lianas in temperate forests.

It is also useful to compare the percentage of trees 2 10 cm diameter that hosted one
or more lianas to that observed in previous studies. Putz reported that 42% of trees in San
Carlos de Rio Negro, Venezuela (1983), and 30 to 50% at BCI served as host to one or
more lianas (1984b). In the present study, at Palo Verde, 55% of trees served as hosts to
one or more lianas (Table l). A smaller proportion of trees at Rich Gap hosted lianas

(37%, Table 1), and even fewer at Conley Creek (29%, Table 1).

61

Table 15. Comparison of present results to previous studies: density of climbers 2 2.5 cm
diameter in 0.1 ha, number of climbing species, and number of and families in the
Neotropics and North America. Data from Gentry (1991) indicated by asterisks are
regional averages, including at least two sites from the region.

 

(or regional average) species families

Neotropics
New World lowland rainforests*¢ 62.1 38.7 18.6
Chamela dry forest* 78 14.6 1 1
Palo Verde dry forest 10.5 15 9
North temperate
North America* 5 1.9 1.4
Rich Gap-mesic cove forest 20.8 3 3
Conley Creek-mesic cove forest 10 3 2
Clifftop Vista- xeric white-oak 0 0 0

dominated forest

*=data from Gentry ( 1991) and includes hemiepiphytes
¢=includes sites with 2 4000 mm of annual precipitation

Tree species richness varied at the four sites, but these data are difficult to interpret
because the sampling size was different for each site. Species-area curves suggested that
the number of tree species seen in the temperate sites reflected the total number of species
present in the forest (Figures 7a, 8a, and 9), but at Palo Verde the number of tree species
observed was less than had a larger area been sampled (Figure 10a). Palo Verde had the
most tree species, Rich Gap and Conley Creek had an intermediate number of species, and
Cliffside had the fewest species. This trend was as expected. Tropical forests are, in
general, more species-rich than those in the temperate zone, and, mesic coves are more
species rich than white oak-dominated forests (Roe and Mansberg 1984; Clebsch 1989).

Extrapolated tree basal area per hectare in this study was lowest at Palo Verde, but
similar at all three sites in North Carolina (Table 1). It is generally accepted that dry forest
tree basal area will be less than that of temperate forests. Murphy and Lugo (1986) provide

a range of basal areas for tropical dry forests between 17 to 40 m2/ ha. The range of basal

62

area per hectare in temperate forests is about 25 to 63 m2/ ha (DeAngelis et al. 1981). The
mean number of trees per plot shows the same trend as that of tree basal area per hectare
(Table 1). Again, the mean number of trees per plot was lowest at Palo Verde, but higher
and uniform at all three sites in North Carolina.

Although the reasons that we see increased species diversity with decreasing
latitude are often debated among ecologists (e. g., Rosenzweig 1995), perhaps a more
general rule exists for lianas. Lianas are poorly represented as a portion of the flora in
temperate areas, perhaps because the wide vessels that characterize lianas are prone to
freeze-induced embolism (Ewers et al. 1997). Most temperate woody vines must construct
new vascular tissue each spring because, in the absence of positive root pressure to refill
embolized vessels, the previous year’s tissues are often rendered functionless by the
winter’s embolisms. Similarly, it has been shown that larger conduits are more susceptible
to drought-induced embolism (Hargrave et al. 1994). Because lianas have such large
vessels, it follows that drought-induced embolism might restrict lianas to wetter areas.
However, Palo Verde does undergo drought-like conditions in the dry season, and the data
from the present study suggest that these lianas can do very well in a drought-stressed
environment. Lianas at Palo Verde are mainly deciduous in the dry season 03. Gonzales,
personal communication), so perhaps this accounts for liana species success at Palo Verde.
The susceptibility of lianas to drought stress might explain why no lianas were present
within the plots at Clifftop Vista (Tables 1 and 15). The high elevation, coupled with low
moisture and brutal winters with severe winds, may preclude liana survival. Although I
saw no woody vines at Clifftop Vista, I noted a few herbaceous scramblers in the genus
Smilax. These scramblers probably survive by dying back in the winter and sprouting new
growth in the spring. It is interesting to note that Clifftop Vista had the largest extrapolated
tree basal area per hectare (Table 1). In spite of this high measure of tree basal area in
terms of trees, the forest supported no lianas. This indicates that the contribution of liana

basal area to the forest is not predictable by the overall tree basal area. In fact, Palo Verde

63

had the smallest total tree basal area per hectare compared to all other sites, yet it supported

the highest liana basal area per plot (Table 1).

Liana host-species preference

Host tree species was important in explaining liana distribution at all sites where it was
possible to investigate. This was true regardless of whether the number of tree stems
available to climb, or the summed tree basal circumference available for climbing, was
utilized to calculate the number of lianas expected per species (Rich Gap and Conley
Creek). Thus, it seems very clear that host species influences liana distribution, as was
seen in a Tropical Moist Forest at BCI (Putz 1984a). Additionally, host preferences have
been shown in hemiepiphites (Guy 1979, Daniels and Lawton 1991, Todzia 1986, Laman
1996). Why are climbers more likely to be found on certain tree species than others? At
both Rich Gap and Conley Creek, there were fewer lianas than expected based on relative
contribution to surface area available to climb on Carya glabra and on the hard-barked oaks.
However, in a previous study on poison ivy (Talley et al. 1996) in old-growth Alabama
mesophytic forest, these tree species supported more vines than expected based on tree
stern densities. Perhaps liana species and climbing mechanisms contribute to this
discrepancy. Each liana species may have its own suite of suitable characteristics required
to colonize a particular host species. The suggestion that host bark type may influence
lianas differently based on their individual climbing mechanisms is a little-tested
hypothesis. Putz (1980, 1984a) suggested that trees with smooth bark are less likely to
host lianas, but Boom and Mori (1982) showed that smooth-barked tree species were no
less likely to host lianas in Brazil. Putz (1984a) shows that spiny-barked trees were not
likely to “saw off” lianas when the trees were mechanically agitated. The data from the

present study suggest that there was a non-linear correlation between tree bark type and

liana climbing mechanism at Palo Verde. However, this was not a powerful enough test to

definitively answer this question.

Liana host-size preference

Tree size was less consistent than tree species in its influence on lianas across sites
and with different species of climbers. At Rich Gap, where Vitis spp. was the most
common liana, I saw more lianas than expected on large trees, but there were fewer lianas
than expected on small trees based on relative stem abundances. It appears that Vitis spp.
will be found at Rich Gap according to tree trunk surface area available to climb. For
approximately every 160 cm of tree basal surface circumference available to climb, one
Vitis spp. stem was encountered. Perhaps competition for some limiting resource results in
a uniform distribution of Vitis spp. on the tree surface area available to climb.

A different pattern in the influence of tree size on lianas arose at Conley Creek
where Aristolochia macrophylla was the most common liana There were more lianas on
smaller trees and fewer on larger trees than expected based on relative stem abundances.
Also, at Conley Creek, surface area available for climbing did not determine where lianas
were found, strengthening the claim that A. macrophylla is preferentially utilizing smaller
hosts. Thus, A. macrophylla seems better able to utilize smaller trees as support because it
must twine around its substrate. I also noticed that it often twined around other vines, a
condition that has been noted in other liana species by other workers (e. g., Putz 1984a,
1984b, 1995).

At Palo Verde where many liana species were present, there were more lianas than
expected on the two smallest tree size classes based on tree surface area available for
climbing. At this site, 94% of the lianas were tendril climbers and twiners that must coil
around their hosts. Thus, the size of the host seems relevant when considering the type of

climbing mechanism of the vines that are utilizing these host trees. Putz (1984b) showed

65

that tendril climbers utilize small hosts (0 to 7 cm diameter), twiners utilize a larger set of
supports (3 to 16 cm), whereas root- and adhesive-tendril climbers have no size
restrictions. These trends in the influence of host size on lianas may be confounded with
the successional age of each of these forests. Although gaps were avoided while sampling
in North Carolina, they could not be avoided in Costa Rica. Perhaps sites earlier in
succession will contain more smaller trees which are then utilized as hosts because they
dominate the area. Additionally, lianas grow as the trees grow, so as their hosts grow
larger, lianas are already committed to inhabitance and will thus be found on smaller trees
earlier in time but larger trees later in time. Thus, the effects of host size will always be
confounded with the influence of host age in a study that is performed at one point in time.
Long-tenn monitoring of locations of lianas within a forest may help determine whether

lianas show preferences for host size or age.

Spatial distribution oflianas

The number of lianas per tree at each site presents an interesting pattern. At both
Conley Creek and Palo Verde, the number of lianas per tree differed from that expected
from a Poisson or random distribution, suggesting that lianas were not randomly
distributed in these forests but were instead more likely to climb a tree that was already
hosting at least one liana. Putz (1984b) saw that there were more trees hosting two or more
lianas than expected by chance, which is similar to that observed in this study at Palo Verde
(Figure 8). However, at Rich Gap, the liana distribution did not differ from that expected
from a Poisson nor from that expected from a negative binomial distribution. I believe this
is a statistical artifact of the test, because as the mean number of lianas per tree approaches
zero, the distribution of the negative binomial approaches that of the Poisson. Or it may be
that wild grape, Vitis spp., is no more likely to climb a tree that is already serving as host to

one or more lianas than it is to climb an unutilized tree. As the present study suggests, the

66

number of Vitis spp. individuals will increase at a constant rate as climbing surface area
availability increases.

The present study suggests that lianas are spatially clumped amongst the trees
within a forest. Plots at Rich Gap contained anywhere from two to 21 lianas, at Conley
Creek, plots contained between one and 15 lianas, and at Palo Verde the range was from
one to 36 lianas per plot. It is quite clear from a walk through the woods that lianas are
usually spatially clumped, which has been seen in other studies (Putz 1984b, Talley et al.
1996). Collins and Wein (1993) showed that understory vines are spatially distributed
based on soil moisture, ground cover or plant size. Also, an established liana can act as a

trellis for new colonizers.

What influences liana distribution?

It is fairly certain that habitat characteristics and environmental factors will influence liana
presence in a given area. The presence of available substrate is necessary for the success of
climbers. However, once an inhabitable area exists and is colonized by a liana species,
what then determines its success and ultimate distribution within this particular area? This
study suggests that host species and size will influence liana distribution, although these
factors may influence different species of lianas differently. Additionally, preliminary data
suggest that host bark type may influence various species of lianas with different climbing
mechanisms differently. Lianas can also be clumped due to dispersal features.

I think that perhaps with the exception of hosts that are unacceptable due to toxic
effects of allelopathic secretions (as suggested by Talley et al. 1996), most trees serve as
acceptable initial hosts for lianas. What becomes important is how long the liana can hold
onto the hosts, and whether the tree has mechanisms to shed climbers. Fast growth and
large, self-pruning compound leaves seem influential in tree liana avoidance (Putz 1984a).

Spiny bark was also suggested as a mechanism to saw off lianas, but no evidence supports

67

this claim (Putz 1984a). Lianas are most likely a selective force in maintaining tree
diversity by influencing host life histories. Trees that can escape lianas will have higher
annual basal growth (Putz 1984b) and greater reproductive output (Stevens 1987).

In each of the temperate forests that contained lianas in the present study, one
species was clearly the dominant climber in the area. Other species are clearly being
dispersed to the area, because they were seen in these sites but in far lower abundance.
What factors determine the success of a particular species over others that are also present?
I don’t believe that direct competition for hosts would prevent another species from
successfully inhabiting a particular forest because there were always many uninhabited
trees available to climb. Do those trees have properties that make them unacceptable hosts?
Epiphytes and hemiepiphytes are often fairly specific as to what hosts provide suitable
germination sites. Lianas may also perform better on hosts with an acceptable suite of
characteristics. From the present study it seems that each species of liana may utilize a

particular suite of tree species and sizes as hosts.

68

REFERENCES

69

References

Bennett BC ( 1995) Ethnobotany and economic botany of epiphytes, lianas, and other host-
dependent plants: an overview. Pages 547-586 in Lohman MD, Nadkarni NM
(eds.) Forest Canopies. Academic Press, San Diego.

Boom BM, Mori SA (1982) Falsification of two hypotheses on liana exclusion from
tropical trees possessing buttresses and smooth bark. Bulletin of the Torrey
Botanical Club 109:447-450

Clay JW, Orr EM, Stuart AW, eds. ( 1975) North Carolina Atlas: Portrait of a changing
southern state. UNC Press, Chapel Hill, USA

Clebsch EC (1989) Vegetation of the Appalachian mountains of Tennessee east of the Great
Valley. Journal of the Tennessee Academy of Science 64:79-83

Coen E (1983) Chapter 3 in: DH Janzen (ed.), Costa Rican Natural History, pp. 35-46.
University of Chicago Press, Chicago, IL

Collins BS, Wein GR (1993) Understory vines: distribution and relation to environment on
a southern mixed hardwood site. Bulletin of the Torrey Botanical Club 120238-44

Daniels JD, Lawton R0 (1991) Habitat and host preferences of Ficus crassiuscula, a
Neotropical strangling fig of the lower-montane rain forest. Journal of Ecology
79:129-141

Darwin C (1867) On the movements and habits of climbing plants. Journal of the Linnean
Society (Botany) 9:1-118

DeAngelis DL, Gardner RH, Shuggart HH (1981) Productivity of forest ecosystems
studied during IPB: the woodlands data set. in Reichle DE (ed.) Dynamic
properties of forest ecosystems. Cambridge University Press. London

Dillenburg LR, Whigham DF, Teramura AH, Forseth IN (1993) Effects of below- and
aboveground competition from the vines Lonicera japonica and Parthenocissus
quinquefolia on the growth of the tree host Liquidambar styraciflua. Oecologia
93:48-54

Ewers FW, Cochard H, Tyree MT (1997) A survey of root pressure in vines of a tropical
lowland forest. Oecologia 110: 191-196

Featherly HI (1941) The effect of grapevines on trees. Proceedings of the Oklahoma
Academy of Science. 21:61-62

Gentry AH (1991) The distribution and evolution of climbing plants. Pages 3-52 in Putz

FE and Mooney HA, editors. The biology of vines. Cambridge University Press.
Cambridge

70

Gonzales, E Station director at Palo Verde Wildlife Refuge, Guanacaste Province, Costa
Rica

Guy PR (1977) Notes on the host species of epiphytic figs (Ficus spp.) on the flood-plain
of the Mana Pools Game Reserve, Rhodesia. Kirkia 10:559-562

Hargrave KR, Kolb KJ, Ewers FW, Davis SD (1994) Conduit diameter and drought-
induced embolism in Salvia mellifera Greene (Labiatae). New Phytologist 126:
695-705

Hartshom GS (1983) Pages 118-157 in DH Janzen (ed.) Costa Rican Natural History.
University of Chicago Press, Chicago, IL

Highlands Chamber of Commerce (1994) Pamphlet entitled “National Forest Trails”

Lutz HJ ( 1943) Injuries to trees caused by Celastrus and Vitis. Bulletin of the Torrey
Botanical Club 70:436-439

Martin WH, Boyce SG, Echtemacht AC (1993) Biodiversity of the Southeastern United
States lowland terrestrial communities. John Wiley & Sons, Inc. New York, NY

Murphy PG, Lugo AE (1986) Ecology of tropical dry forest. Annual Review of Ecology
and Systematics 17:67-88

Ogbum C (1975) The Southern Appalachians. William Morrow & Co., Inc. New York,
NY

Pitillo D (1976) Potential natural landmarks of the southern Blue Ridge portion of the
Appalachian Ranges Natural Region. Report prepared for the National Park
Service and Western Carolina University.

Putz FE (1980) Lianas vs. trees. Biotropica 12: 224—225

Putz FE ( 1983) Liana biomass and leaf area of a “tierra finne” forest in the Rio Negro
Basin, Venezuela. Biotropica 15: 185-189

Putz FE (1984a) How trees avoid and shed lianas. Biotropica 16: 19-23

Putz FE (1984b) The natural history of lianas on Barro Colorado Island, Panama. Ecology
65: 1713-1724

Putz FE (1991) Silvicultural effects of lianas. Pages 493-501 in Putz FE, Mooney HA,
editors. The biology of vines. Cambridge University Press. Cambridge, England

Putz FE (1995) Vines in treetops: consequences of mechanical dependence. Pages 311-
324 in Lohman MD, Nadkarni NM (eds.) Forest Canopies. Academic Press, San
Diego.

Radford AE, Ahles HE, Bell CR (1968) Manual of the vascular flora of the Carolinas.
University of North Carolina Press, Chapel Hill, USA

Roe C, Mansburg L (1984) Cullasaja Gorge including Cullasaja waterfalls (Dry Falls and

Lower Cullasaja Falls) and Clifftop Vista— Van Hook White Oak Stand in Nantahala
National Forest: Natural Areas. Report prepared for NC Natural Heritage Program.

71

Rosenzweig ML (1995) Species diversity in space and time. Cambridge University Press.
Cambridge, England

Siccama TG, Weir G, Wallace K (1976) Ice damage in a mixed hardwood forest in
Connecticut in relation to Vitis infestation. Bulletin of the Torrey Botanical Club.
103: 180-183

Stevens GC (1987) Lianas as structural parasites: the Bursera simarouba example.
Ecology 68:77-81

Talley SM, Lawton RO, Setzer WN (1996) Host preferences of Rhus radicans
(Anacardiaceae) in a southern deciduous hardwood forest. Ecology 77:1271-1276

Teramura AH, Gold WG, Forseth IN (1991) Physiological ecology of mesic, temperate
woody vines. Pages 245-285 in Putz FE, Mooney HA, editors. The biology of
vines. Cambridge University Press. Cambridge, England

Todzia C (1986) Growth habits, host tree species, and density of hemiepiphytes on Barro
Colorado Island, Panama. Biotropica 18:22-27

Trimble GR, Tryon EH (1974) Grapevines a serious obstacle to timber production on good
hardwood sites in Appalachia. Northern Logger and Timber Processor. 23:22-23,
44

Whigham D (1984) The influence of vines on the growth of Liquidambar styraciflua L.
(sweetgum). Canadian Journal of Forest Research 14:37-39

Wofford BE (1989) Guide to the vascular plants of the Blue Ridge. The University of
Georgia Press. Athens, USA

72

APPENDICES

73

APPENDIX A

74

APPENDIX A

Species representation in tree size classes (cm) for 0.6 ha at Rich Gap, NC.

 

 

 

Acer rubrum 8 9 5 2 4 3 31
Betula lenta 6 6 0 0 0 O 12
Betula lutea 1 O 0 0 O O 1
Carya glabra 4 4 3 1 3 0 15
Carya ovata 0 O l 0 0 0 1
Castanea pumila 0 1 0 0 O O 1
C omus florida 13 O O 0 O O 13
F raxinus pennsylvanica 3 1 0 0 O 0 4
Halesia carolina 19 7 0 O O 0 26
Liriodendron tulipifera 33 44 27 13 8 1 126
Magnolia fraseri l4 7 1 O O l 23
Pinus strobus 1 1 O 0 0 0 2
Prunus pensylvanica 4 1 0 1 0 0 6
Prunus serotina 0 0 1 0 O 0 1
Quercus coccinea 0 0 1 O O 0 1
Quercus prinus 4 4 4 2 0 O 14
Quercus rubra 3 3 3 2 3 14
Robinia pseudoacacia l 2 l O O O 4
Sassafras albidum 2 0 0 O O 0 2
Tilia heterophylla 5 l 0 0 0 O 6
Tsuga canadensis 7 0 1 O O O 8
total 128 91 48 21 18 5 31 1

75

APPENDIX B

76

APPENDIX B

Species representation in tree size classes (cm) for 0.28 ha at Conley Creek, NC.

   

species I I seclass (cm dbh)
[IO-20) [20-30) [30-40) [40—50) [50-60) _>_60 total

 

 

Acer pensylvanica 5 O 1 0 O O 6
Acer rubrum l4 9 3 3 0 1 30
Betula allegheniensis 1 0 0 O O 0 l
Betula lento 1 0 O 0 O 0 1
Carya cordiformis 2 0 0 0 0 0 2
Carya glabra 8 4 2 0 2 0 16
Carya ovata 1 0 0 0 O O l
Carya tomentosa 3 2 0 1 O 0 6
Cornusflorida 7 O O 0 0 O 7
Halesia carolina 2 0 0 0 0 0 2
Liriodendron tulipifera 2 5 9 6 0 0 22
Magnolia acuminata 3 0 0 1 0 O 4
Magnolia fraseri o 1 O 0 0 0 1
Prunus serotina 1 O O 0 O O l
Quercus alba 2 0 O l 0 1 4
Quercus prinus 3 3 7 1 O O 14
Quercus rubra l 2 2 l 4 2 12
Robinia pseudoacacia l 0 0 O 0 O 1
Tilia heterophylla 1 0 l 0 0 0 2
total 5 8 26 25 14 6 4 133

77

APPENDIX C

78

APPENDIX C

Species representation among tree size classes (cm) for 0.2 ha at Clifftop Vista, NC.

 

 

spec1es size class (cm Ebb)

[IO-20) [20-30) [30-40) [40-50) [50-60) 260 total
Acer rubrum 3 1 1 O O l 6
Carya glabra l 4 2 l 0 0 8
Carya tomentosa 1 O 0 0 O 0 1
Kalmia lanfolia 1 0 0 O 0 0 1
Pinus strobus 2 2 2 1 0 O 7
Quercus alba 20 5 4 2 4 2 37
Quercus coccinea 6 5 1 1 0 0 0 22
Quercus muehlenbergii 1 0 2 0 0 0 3
Quercus prinus 0 l 3 1 0 O 5
Quercus rubra O O 2 0 O 0 2
Quercus velutina 2 2 1 0 O l 6
total 37 20 28 5 4 4 98

79

APPENDIX D

80

APPENDIX D

Species representation among tree size classes (cm) for 0.4 ha at Palo Verde, Costa Rica.

spec1es size class (cm dbh)
[10-20) [20-30) [30-40) [40-50) [50-60) 260 total

 

Acacia comigera

Albizia adinocephala
Albizia caribbea
Astronium graveolens
Bombacopsis quinata
Bursera simaruba
Caesalpinia eleostachys
Calycophyllum candidissimum
Casearia tremula

Chomelia spinosa
Coccoloba sp.
Cochlospermum vinfolium
Cordia alliodora

Guazuma ulmifolia
Guettarda macrospermum
Hyperbaena tonduzii
Jacquinia nervosa
Jacquinia pungens
Lonchocarpus costaricensis
Luehea candida

Lysiloma divaricatum
Macairea biovulatum
Manilkara zapota
Muntingia calabura
Pithecellobium lanceolatum
Randia thurberi

Simaruba glauca

Spondias mombin
Stemmadenia obovata
Tabebuia ochracea
Thouinidium decandrum
Trichilia hirta

Ximenia americana

wwNaH——u—-w—o—om—mo-—No~o~ANO-O~O o-—
o—omoooooooo—A-d~oo~ooo~ooooooo u—o
~ooooooo~oooowoooo~cooo~o-oooo oo
ocooo—oo~ooo~oooooooo~ooo-ooo~c o
oooooooooooo~Noooooooooooo~oo~o oo
oooooooooo—oooooooooooooomoowoo oo
m-tsNoo—N——m~——A:Nm~—w~——~0~N\Im—NN— _._.

 

total

A
00
p—n
h
u—a
O
\O
LII
00
\O
.b

81

   

"‘1l11111111111111“