Citation
Edges, fruits, frugivores, and seed dispersal in a neotropical montane forest

Material Information

Title:
Edges, fruits, frugivores, and seed dispersal in a neotropical montane forest
Creator:
Restrepo, Carla
Publication Date:
Language:
English
Physical Description:
xi, 189 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Anthropometric measurements ( jstor )
Birds ( jstor )
Ecology ( jstor )
Edge effects ( jstor )
Forest growth ( jstor )
Forests ( jstor )
Fruiting ( jstor )
Fruits ( jstor )
Species ( jstor )
Understory ( jstor )
Dissertations, Academic -- Zoology -- UF
Forest ecology -- Colombia ( lcsh )
Mountain ecology -- Colombia ( lcsh )
Zoology thesis, Ph. D
City of Gainesville ( local )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1995.
Bibliography:
Includes bibliographical references (leaves 166-187).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Carla Restrepo.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
022370952 ( ALEPH )
34336937 ( OCLC )

Downloads

This item has the following downloads:


Full Text











EDGES, FRUITS, FRUGIVORES, AND SEED DISPERSAL
IN A NEOTROPICAL MONTANE FOREST















By


CARLA RESTREPO


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



UNIVERSITY OF FLORIDA


1995














ACKNOWLEDGMENTS


From the mountains of La Planada to the concrete of Bartram Hall I have had the fortune of interacting with people who have enriched my life in fundamental ways. As I see it now, this dissertation proved to be a point of convergence of people, situations, and ideas that is leading me to new discoveries. I thank the members of my committee, Douglas J. Levey, H. Jane Brockmann, John Ewel, C. S. "Buzz" Holling, and Frank Slansky, for their continuous encouragement and support throughout the two years I spent in Bartram Hall trying to figure out which direction I wanted my dissertation and myself to go.

It is not easy to collect data over 12 ha of steep land, nor to live in isolation. I am particularly grateful to Natalia G6mez, Sylvia Heredia, and Arlex Vargas for their help and support in the field. I am indebted to the neighbors of La Planada, in particular to Adolfo Ortega, Abelardo Nastacuaz, Demetrio Guanga, Pacho Guanga, Amparo Oliva, and the GELISI, for sharing their life with me. At various points during this research I benefited from help provided by Marta Baena, Girleza Ramirez, Ivan Jimenez, Natalia Arango, Luis F. Citelli, Omaira Ospina, Maria de Restrepo, and Paul Marples. By the end of the field season


ii








several people were instrumental in helping to put together little pieces of my puzzle. J. H. Cock and A. P. Hernandez from CENICANA lend me the LAI-Canopy Analizer. J. Luteyn, T. G. Lammers, P. E. Berry, D. Froding, B. Hammel, J. J. Wurdak, J. Kress, C. Taylor, L. E. Skog, L. R. Landrum, A. M. W. Mennega, J. S. Miller, and T. Croat kindly identified the plant material I collected at La Planada. P. Kubilis and C. Steible provided statistical advice when most needed. L. Walz prepared maps. P. Amezquita counted pollen tubes. I am grateful for their valuable help.

This project was crafted some years ago with the input provided by P. Feinsinger, my former advisor. He presented me with alternative routes that certainly proved fruitful. I am particularly grateful for this.

This study was funded by the Fundaci6n para la Promocidn de la Investigaci6n y la Tecnologia, Banco de la Repiblica, Colombia and the Wildlife Conservation Society (WCS).


iii















TABLE OF CONTENTS


ACKNOWLEDGMENTS ............................................ 1

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

LIST OF FIGURES..... ......................................viii

ABSTRACT ...................................................x

CHAPTERS

1 THE ROLE OF EDGES IN NEOTROPICAL MONTANE LANDSCAPES........1

2 DESCRIPTION OF STUDY AREA

Study Area.......... ......................................5

General Sampling Procedure...............................11

3 FROM FLOWERS TO SEEDLINGS: THE EFFECT OF EDGES AND
TREEFALL GAPS ON TWO UNDERSTORY SHRUBS, Palicourea
aibbosa AND Faramea affinis (RUBIACEAE)

Introduction .............................................15

Methods. --..................................................18

Analyses ..................................................24

Results -.................-- -.................................26

Discussion .................................................34

4 UNDERSTORY FRUIT ABUNDANCE IN A NEOTROPICAL MONTANE
FOREST: THE INFLUENCE OF EDGES AND TREEFALL GAPS

Introduction ..................-...........................42

Methods- ---..................................................45

Analyses.................................................49

Results. --..................................................52

Discussion.- -...............................................72


iv









5 EDGES AND UNDERSTORY BIRDS IN A NEOTROPICAL MONTANE
FOREST

Introduction.............................................82

Methods..................................................84

Data Analysis............................................86

Results..................................................90

Discussion..............................................105

6 FRUGIVOROUS BIRDS IN FRAGMENTED NEOTROPICAL MONTANE
FORESTS: LUMP STRUCTURE IN BODY MASS

Introduction............................................112

Methods.................................................114

Results.................................................126

Discussion..............................................138

7 CONCLUSIONS..............................................148

APPENDIX A

Plant species fruiting in the understory at the
Reserva Natural La Planada..............................151

APPENDIX B

Bird species captured in the understory at the Reserva
Natural La Planada......................................161

LITERATURE CITED...........................................166

BIOGRAPHICAL SKETCH........................................188


V


















LIST OF TABLES


Table Dag


2-1 Water balance for the Reserva Natural La Planada
........... ............................ ... .......... .... 9

2-2 Characteristics of the edges included in this
study .................----.................................. 13

3-1 Results of Mixed Factorial ANOVAs for pollen tube
production in Palicourea gibbosa and Faramea affinis
in relation to distance from forest edge................. 27


3-2 Pollen tube production in Palicourea aibbosa and
Faramea affinis in relation to distance from forest
edge ....................-- -................................. 28


3-3 Results of Mixed Factorial ANOVAs for fruit set
and fruit damage by insects in Palicourea aibbosa and
Faramea affinis in relation to distance from forest
edge .................-..................................... 30


3-4 Results of Mixed Factorial ANOVAs for seed
predation and seed germination in Palicourea gibbosa
and Faramea affinis in relation to distance from
forest edge............... .............................. 35


3-5 Results of Mixed Factorial ANOVAs for relative
growth and leaf production rates in PalicoureA aibbosa
and Faramea affinis in relation to distance from
forest edge ................... .......................... 38


4-1 Results of Repeated Measures ANOVA on leaf area
index (LAI) for old edges.................................. 55


vi








4-2 Results of Mixed Factorial ANOVAs on fruit abundance across pasture-forest edge ................... 59

4-3 Results of Goodness of Fit Test on the number of fruiting individuals of abundant species across pasture-forest edge..................................... 68

4-4 Results of Replicated Goodness of Fit Test on the number of fruiting individuals across pasture-forest edge in old and new edges............................... 69

4-5 Distribution of number of fruiting individuals across pasture-forest edge based on species abundance... 70

4-6 Summary of results of changes in fruit abundance across the pasture-forest edge.......................... 74

5-1 Results of ANOVAs for Mixed Factorial Design on capture rates of understory birds. .................... 92

5-2 Results of Goodness of Fit Test on the number of bird captures of abundant species across pastureforest edge............................................. 101

5-3 Results of the Replicated Goodness of Fit Test on the number of bird captures across pasture-forest edge .................................................... 103

5-4 Distribution of bird captures across pastureforest edge based on species abundance.................. 104

6-1 Description of sites included in lump analyses of body mass of frugivorous birds.......................... 119


vii


















LIST OF FIGURES


2-1 Location of study area and edges included in the study ................................................... 6

2-2 Distribution of mean monthly rainfall and temperature at the Reserva Natural La Planada ............ 8

2-3 Edge indicating general sampling design ............ 14

3-1 Fruit set in Palicourea gibbosa as influenced by distance from forest edge............................... 31

3-2 Seed germination, seed predation, seedling growth and, leaf production in Palicourea aibbosa and Faramea affinis as influenced by distance from forest edge...... 36

3-3 Seed germination, seed predation, seedling growth and, leaf production in Palicourea aibbosa and Faramea affinis as influenced by habitat........................ 39

4-1 Leaf area index (LAI) across the pasture-forest edge at old edges....................................... 54

4-2 Leaf area index (LAI) and fruit abundance .......... 57

4-3 Variation in fruit abundance at the Reserva Natural La Planada in relation to distance from forest edge .................................................... 60

4-4 Variation in fruit abundance at the Reserva Natural La Planada in relation to edge age and distance from forest edge............................... 62

4-5 Variation in fruit abundance at the Reserva Natural La Planada in relation to habitat and distance from forest edge........................................ 63




viii








4-6 Variation in fruit abundance at the Reserva Natural La Planada in relation to edge age and month ... 65

4-7 Variation in fruit abundance at the Reserva Natural La Planada in relation to distance from the forest edge and month................................... 66

5-1 Variation in the distribution of understory birds at La Planada in relation to distance from the forest edge .................................................... 93

5-2 Variation in the distribution of understory birds at La Planada in relation to distance from the forest edge and edge age....................................... 95

5-3 Variation in the distribution of understory birds at La Planada in relation to edge age and month of the year .................................................... 96

5-4 Variation in the distribution of understory birds at La Planada in relation to distance from forest edge and month of the year................................... 98

6-1 Montane habitats of Colombia and sites included in lump analysis of body mass of frugivorous birds...... 116

6-2 Lump analysis for body mass of frugivorous birds of Colombian upper lowland tropical forests showing
(a) body mass distribution vs. rank order and (b) rank size-ordered body mass distribution versus gap rarity indexes................................................. 122

6-3 Lump structure of Colombian montane frugivorous birds according to elevational zone..................... 127

6-4 Lump structure of Colombian frugivorous birds from sites cover mostly by forest to sites highly transformed by human activities within the upper lowland zone............................................ 130

6-5 Lump structure of Colombian frugivorous birds from sites cover mostly by forest to sites highly transformed by human activities within the lower montane zone............................................ 133

6-6 Lump structure of Colombian montane frugivorous birds from sites cover mostly by forest to sites highly transformed by human activities within the upper montane zone...................................... 136

6-7 Relationship between species richness and lump structure in landscapes of variable complexity.......... 145


ix














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


EDGES, FRUITS, FRUGIVORES, AND SEED DISPERSAL
IN A NEOTROPICAL MONTANE FOREST By

Carla Restrepo

December, 1995




Chairman: Douglas J. Levey
Major Department: Zoology


Edges resulting from natural or human disturbances

influence the distribution of organisms as well as ecological processes. One such process is seed dispersal, which in turn may influence may the location of edges through time and even the entire structure of landscapes. In the Reserva Natural La Planada, Colombia, I investigated how edges resulting from human activities influenced seed dispersal. In particular, I examined how distance from forest edge, in combination with edge age and treefall gaps, could affect recruitment rates, fruit abundance, seed movement, and the distribution of birds in the understory of this neotropical montane forest. Sampling took place at three old (>40 yr) and three new (<15 yr) edges and within each edge at four distances (0-10, 3040, 60-70, 190-200 m) from the pasture-forest edge.


x








Edges influenced Palicoura gibbosa and Faramea affinis, the two most common understory plants, at various stages of their life cycle. Seed predation and seed germination but not relative growth or leaf production rates changed across the pasture-forest edge. The latter, however, were influenced by treefall gaps. At the community level, fruit abundance and the distribution of understory birds changed across the pasture-forest edge in complex ways that not always reflected changes due to the presence of edges. This was demonstrated by the fact that (1) two-way interactions between distance, edge age, treefall gaps, and month were significant, and (2) response variables describing fruit and bird abundance at the community level did not show the same trends.

Edges influenced fruit-frugivore interactions at the level of forest stands but also at the level of entire landscapes as demonstrated by an analysis of body mass distribution of frugivorous birds as a function of ecosystem fragmentation. With more edges, entire groups of birds with similar body mass (termed "lumps") disappeared. Nevertheless, the distribution of body mass, i.e., lump structure, remained almost intact under certain land use types. This work suggests that at broad scales edges influence frugivorous birds and, as a result, seed dispersal.


xi














CHAPTER 1
THE ROLE OF EDGES IN NEOTROPICAL MONTANE LANDSCAPES



Edges constitute a common feature of neotropical montane landscapes. Complex topography and climate in conjunction with natural disturbances, ranging from treefall gaps (Murray 1988; Lawton and Putz 1988; Samper 1992), and landslides (Garwood et al. 1979; Lawton and Dryer 1980; Gentry 1992]), to mudflows, have given rise to a heterogeneous landscape in which edges bound the disturbed areas. Over evolutionary time, the dynamic and heterogeneous character of these landscapes may have resulted in the unusually high levels of biodiversity that characterize neotropical montane ecosystems (Terborgh and Winter 1983, Terborgh 1985, Gentry 1986, 19924).

Superimposed on this natural heterogeneity is that

resulting from human activities. In areas where favorable conditions prevail patches of forest are immersed in an agricultural and urban matrix. Conversely, in areas where unfavorable conditions limit the development of economic activities, fields, second growth, and urban areas are immersed in a forest matrix. In both situations the areas modified by human activities are bounded by edges. Over shorter time scales than those defined by large-scale natural disturbances, changes in land use have led to extinctions and


1





2


invasions by an unknown number of species (Henderson et al. 1991, Kattan 1992) .

Even though edges are a prominent feature of neotropical mountains, little is known about how they influence landscape pattern. My dissertation focuses on edges and how they influence seed dispersal through fruit-frugivore interactions. Changes in the nature of fruit-frugivore interactions can provide information on the persistence of edges through time, depending on the ability of plants to produce and disperse their seeds. On the other hand, changes in the composition and number of individuals across edges can provide information on edge structure and productivity.

A common thread among studies focusing on edges is the lack of a common pattern, without which it is difficult to propose underlying causes or even consequences. Each new study has added details at the cost of finding generalities from which testable predictions can be made. In my dissertation I followed two different approaches to the study of edges. Both had as a central theme that of fruitfrugivore interactions. The first generated detailed information on fruit-frugivore interactions at the level of forest stands. Nevertheless, this information precludes generalizations about the role of edges in neotropical montane forests due to the intrinsic characteristics of my study site. The second generated consistent patterns at the level of entire landscapes that differed in degree of transformation by human activities. This portion of the





3


study, therefore, might be useful in generating testable hypotheses that could guide future studies.

My study area and general sampling procedure are

described in detail in chapter 2. I often refer to this chapter because the studies I describe in chapters 3 to 5 were based on the same sampling procedure.

In chapter 3 I focus on two understory plants, Palicoura gibbosa and Faramea affinis (Rubiaceae) and ask how edges may influence different components of the life cycle of these two species and how this may influence recruitment. These two species are the most common understory species of my study site and are bird-dispersed. In particular, I looked at how the combined effect of distance from forest edge and presence of treefall gaps affect pollination, fruit set, seed predation, seed germination, and seedling growth.

In chapter 4 I take a broader approach to the study of

edges and their influence on fruit-frugivore interactions, by focusing on the assemblage of understory plants. I ask whether fruit abundance changes across pasture-forest edges and how treefall gaps and season modify such effects. I look at changes in fruit abundance for the entire assemblage of understory plants and for individual species. Changes in fruit abundance can affect the distribution of birds feeding on them and thus seed dispersal across edges.

In chapter 5 I keep the previous approach to the study of edges and their influence on fruit-frugivore interactions but ask how edges influence the distribution of understory





4


birds, particularly frugivores. I compare the distribution of nectarivorous, insectivorous, and frugivorous birds across the pasture-forest edge to elucidate possible mechanisms underlying observed patterns.

I take a completely different approach in chapter 6 to the study of edges and their influence on fruit-frugivore interactions. Instead of asking how the distribution of frugivorous birds is affected by the creation of edges within forest stands, I ask how the distribution of frugivorous birds is affected by edges within whole landscapes. I compare changes in the distribution of frugivorous birds across sites that have been modified in various ways by human activities. This approach provides the basis for some generalizations and the formulation of testable hypotheses for future work.

Chapter 7 is the place for synthesis and speculation. I emphasize that edges are part of landscapes that are in continuous change and thus have to be seen as dynamic, not fixed entities of landscapes. Seed dispersers and seeds move between forest interior and edge, and sparse species of plants and frugivorous birds are found more often at edges than at forest interior. These results suggest a critical role of edges in landscapes subject to change.














CHAPTER 2
DESCRIPTION OF STUDY AREA


Study Area


I conducted this study at Reserva Natural La Planada and Finca El Bosque, located in the municipality of Ricaurte,

department of Nariflo, SW Colombia (78*00'W and 1*10'N) (Fig. 2-1). Both localities lie on the western slope of the Andes at 1,800 m. The biota of La Planada and its surroundings is one of the most diverse of the northern Andes (Terborgh and Winter 1983; Orejuela 1987) and twenty percent of the plant and animal species reported for the area are endemic (geographical range <50,000 km2, Terborgh and Winter 1983).

Observations were concentrated in the NW and W portions of La Planada and El Bosque, respectively. These two areas lie on the watershed division of the Miraflores and Pialapi rivers. Colonization of this area started in the early 1940s and proceeded from the bottom of the valleys (1,200 m) to the top of the mountains (1,800 m), which are still mostly covered by forest. Small sugarcane plantations, transient corn fields, pastures, fallows, and second-growth vegetation are embedded in the forest matrix. These disturbed areas are concentrated at the bottom of the valleys and range from 1-9


5











Tumac~
**



Chucunes-x COLOMBIA San Isidro




- -El Rollo
JIY











SH ogenes Marcos 0 eCelimd ILa Planadat PialapiEl Bosque Acantayac

Figure 2-1. Location of study area and edges that were sampled. Forest is represented by shaded area (based on a 1981 aerial photograph, approximate scale 1:26,000)





7


ha (mode 3 ha) (G6mez and Palau 1994). The patches of forest are connected by strips of forest that have been left along streams, steep sloptes, and mountain ridges.

The natural disturbance regime is varied. At small

scales, treefall gaps, which occur mostly during the rainy season, are common (Samper 1992). At larger scales landslides, ash rain (e.g., January 1993 volcanic eruption), and strong winds, the latter resulting in the defoliation of large areas (e.g., August 1993), affect whole landscapes.

Unpublished climatological records of La Planada (19851994) show a mean annual rainfall and temperature of 4,437 mm

and 19.2*C, respectively (Fig. 2-2). Rainfall is distributed in two wet seasons, interrupted by a mild dry (FebruaryMarch) and a strong dry (June-August) season (Fig. 2-2). Based on these data, La Planada can be classified as a transitional life zone between tropical premontane rain and wet forest (Holdridge 1967). An important climatological feature of La Planada and its surroundings is the presence of afternoon mist during most of the year.

The water balance for La Planada shows that on average every month has a surplus of water which is lost as soil runoff (Table 2-1). Nevertheless, in some years there might be months in which there is a water deficit. This can explain why during my study period, in particular during the months of July-August of 1993, many plants lost their leaves.





8


o c C 0 3


I I I I I I I I I


J F M A M J


J A


S O N D


Month


Figure 2-2. Distribution of mean monthly rainfall (bars) and temperature (open circles) (1985-1994) at the Reserva Natural La Planada (unpublished data Reserva Natural La Planada). Filled circles represent average values of rainfall for 19921993.


900 800 700 600 500 400 300200 -


E
E


0
C


20


-1 6

CD
-12

CO
8


-4


I A I.,


1000-


--- -- -- -- --


w














Table 2-1. Water balance for Reserva Natural La Planada, transitional life zone between
premontane rain to wet forest (assuming available moisture = 443.7 mm). For details on how to calculate different variables see Ewel and Madriz (1968). Elevation 1,800 m. Based on climatological records from 1985-1995.


J


F M A M


J


J


Biotemperature Potential evapotranspiration (P.ET)
Precipitation Actual ET


*C
mm mm mm


Water surplus mm
Soil moisture change mm
Moisture available in soil end of month mm All runoff mm
Soil moisture deficit mm
Precipitation deficit mm
Total moisture deficit mm


18.9 19.3 95 88
460.1 360.7 95 88
365.1 272.7
0 0
443.7 443.7 365 273
0 0
0 0
0 0


19.5 98
406.5 98
308.5
0
443.7 309
0
0
0


19.6
94.8 451.1 94.8 356.3
0
443.7 356
0
0


19.6 98
405.3 98
307.3
0
443.7 307
0
0


19.3
93.4 292.2
93.4 198.8
0
443.7 199
0
0
0


18.5 92.5
148.5 92.5 56.01
0
443.7 56
0
0
0


0 0
















Table 2-1. (continued)

A S 0 N D TOTAL



Biotemperature 0C 18.8 18.9 19.2 19.2 19.1 19.2
Potential evapotranspiration (P.ET) mm 94 91.5 96 92.9 95.5 1129.6
Precipitation mm 163.2 287.8 527.5 482 491.6 4437.4
Actual ET mm 94 91.5 96 92.9 95.5 1129.6
Water surplus mm 69.17 196.3 431.5 389.1 396.1 3347
Soil moisture change mm 0 0 0 0 0
Moisture available in soil end of month mm 443.7 443.7 443.7 443.7 443.7
All runoff mm 69.2 196 431 389 396.1 3347
Soil moisture deficit mm 0 0 0 0
Precipitation deficit mm 0 0 0 0 0 0
Total moisture deficit mm 0 0 0 0 0





11


The La Planada forest develops on well drained soils (Dystrandept) derived partially from volcanic material, that are moderately acid, with a sandy to clay loam texture (De Las Salas and Ballesteros 1986). The canopy height (average 22 m) and the basal area (dbh > 4 cm; 33.4 m2/ha) of the forest are low and epiphytic and hemiepiphytic plants are very abundant (De Las Salas and Ballesteros 1986; Gentry 1988). Plants (dbh > 2.5 cm) in a 0.1 ha plot were represented by 112 species (Gentry 1992i). The most important trees on this plot were Quararibea sp., Elaegia sp., Hieronyma sp., Alchornea sp., Billia colombiana, Ina sp., Otoba sp., and Ocotea sp.; the most important treelets and shrubs were Faramea eleaans, Prestoea cf. rurpurea, Airhanes sp., Geonoma weberbaueri, Paligourea gibbosa and Miconia sp.; and the most common epiphytes were Philodendron cf. scandens, Srheraedenia stevermarkii and Psammisia sp. (A. Gentry, unpublished data).


General Samolina Procedure


I chose six sites to evaluate how edge age and distance from the edge towards the forest interior influence various components of fruit production and seed dispersal. These sites, hereafter referred to as edges, were active or recently abandoned pastures contiguous with forest. Thus, at most edges there was a sharp delineation between forest and the adjacent pasture (Table 2-2). Four edges lay at La Planada boundaries (Marcos, C6limo I, C61imo II, and





12


Herm6genes), a fifth edge was located within the reserve (Pialapi), and the sixth edge was located at El Bosque (Acantayac) (Fig. 2-2). C6limo I and C6limo II were 400 m apart on the same edge, but because of differences in the weeding regime of the pasture and use of the forest I reasoned that they could represent two independent sampling units. Independence of these two sampling points was particularly important for the part of the work evaluating the influence of edges on the distribution of understory birds (Chapter 5). Recapture frequency between these two sites was <4%, supporting the assumption that these two points represented two independent sampling units.

Three edges, C61imo I, C6limo II, and Pialapi, were

created around 1950 (old edges), when colonists first arrived in the area and cleared the forest to establish pastures. The other three edges, Marcos, Herm6genges, and Acantayac, were created around 1982 (young edges), the year La Planada was established as a private reserve (Table 2-2). At the beginning of the study, I placed barbed wire fences along the edges to keep cattle from penetrating into the forest. I sampled these edges between March 1992 and March 1994.

At each edge I worked in an area of 100 X 200 m (2 ha) and established four strips (100 X 10 m) running parallel to the edge. These strips were located at four different distances from the forest edge towards the forest interior: 0-10 m (D1), 30-40 m (D2), 60-70 m (D3), and 190-200 m (D4).













Table 2-2. Characteristics of the edges included in this study.


Edge T1 02 A3 C4 PS5 G6 CU7 FU8

C6limo I S 40* NE 1953 H 6.0 4.2 Cattle ranching; Extraction of palm hearts and
pasture poles; cattle grazing

C41imo II S 190 NE 1953 M 7.6 7.6 Cattle ranching; Extraction of palm hearts and
pasture poles; cattle grazing

Pialapf F 80 NE 1950 L 9.9 Trail to Pialapf; Selective logging 40 yr ago
second growth

Acantayac F 240 NE 1981 M 3.0 11.0 Cattle ranching; Extraction of palm hearts and
pasture poles; cattle grazing; ancient
graves

Herm6genes M 680 NE 1982 H 3.0 10.0 Cattle ranching; Extraction of palm hearts
pasture

Marcos S 59* NE 1982 H 1.0 7.5 Cattle ranching; Extraction of palm hearts and
pasture poles; cattle grazing; old path
to Pialapf
1 T = Topography; S = steep, M = moderate, F = flat


0 = orientation (position of A = Age (year the forest was C = Edge contrast; H = high,


edges regarding clear cut)


the cardinal points)


M = moderate, L = low


PS = Size of disturbed area (ha) G = Percent of sampling area covered by gaps CU = Use of clear cut area FU = Use of forested area


2
3
4
5
6
7
8


w





14


D1
(0-10 m)




D2
(30-40 m)






D3
(60-70 m)


D4
(190-200


m)


PASTURE


Figure 2-3. Edge indicating general sampling design. Shaded strips represent four distances where sampling took place: Dl (0-10 m), D2 (30-40 m), D3 (60-70 m), and D4 (190-200 m). In strip D1 I show the distribution of subquadrats (1-4) where fruit abundance was evaluated. In strip D2 I show the distribution of each of three pirs of mistnets (perpendicular dark lines). On the left side of the figure I illustrate the orientation of one transect along which LAI was measured.


4 3








I i FOREST




I I
- I


I














CHAPTER 3
FROM FLOWERS TO SEEDLINGS: THE EFFECT OF EDGES AND TREEFALL GAPS ON TWO TROPICAL UNDERSTORY PLANTS, Palicourea gibbosa
AND Faramea affinis (RUBIACEAE)


Introduction


Recruitment rates in plant populations are influenced by success at all stages of the life cycle (Harper 1994). Which stages limit recruitment depends on the requirements of individuals at each stage and the spatial distribution of resources (e.g., Sork 1983, Martinez-Ramos and Soto-Castro 1993, Osunkoya et al. 1994). In tropical areas, for example, treefall gaps influence the distribution of resources at small scales (Denslow and Hartshorn 1994), whereas landslides and forest clearings do so at large scales (Guariguata 1990, Dalling and Tanner 1995). Also, seed predation (e.g., Schupp 1988, Schupp and Frost 1989, Samper 1992), seedling establishment (e.g., King 1990), plant growth (Sizer 1992, Dalling and Tanner 1995), fruit production (Levey 1990) and seed dispersal (Murray 1988) change as a result of such disturbances. Little is known, however, about the combined effect of small and large-scale disturbances on the various stages of the life cycle of plants or how such effects may determine which stage limits recruitment.


15





16


For the most part, studies evaluating the effect of

human disturbances on recruitment rates in tropical plants have focused on single stages of a plant's life cycle (e.g., MacDougall and Kellman 1992, Seizer 1992, Burkey 1993). By looking at several species it has been possible to establish patterns and understand the factors underlying the responses of particular stages (e.g., Sizer 1992). This approach should be complemented with studies focusing on single species to establish the relative contribution of a given stage to the life cycle of a plant (Ellison et al. 1993). A more complete understanding of factors that limit recruitment either in forest fragments or nearby disturbed areas must consider what happens to plants in all stages of their life cycle.

At a neotropical montane site fruiting individuals of Palicourea aibbogs and Faramea affinis were not distributed uniformly across pasture-forest edges (Chapter 4). Here I report results of a study that examined how several stages of the life cycle of these two understory plants were influenced by distance from forest edge, edge age, and treefall gaps. In particular, I wanted to determine how pollination, fruit set, seed dispersal, seed predation, germination, and seedling growth could result in the observed distribution of P. gibbosa and F. affinis across the pasture-forest edge.





17


The Species


Palicourea aibbosa Dwyer and Faramea affinis belong to the Rubiaceae, one of the most speciose and common families of neotropical montane forests (Taylor 1989, Gentry 1992.). Palicourea aibbosa shrubs reach 4 m and are found at middle elevations from Panama to Ecuador (Dwyer 1980; C. Taylor, personal communication), growing in second growth and mature forest (Arias 1993). Faramea affinis treelets reach 9 m and grow in old second growth and mature forests. In a 0.1 ha plot at my study site, 2. aibbosa and E. affinis were the most common species (dbh > 2.5 cm) in the understory (A. Gentry, unpublished data).

Palicourea gibbosa exhibits three flowering periods per year. Its yellow flowers are visited mostly by hummingbirds, including Ocreatus underwoodii, Aalaiocercus coelestis, and Haplophaedia lugens (Arias 1993). Fruits of E. gibbons are dark blue to purple, 7 mm long, and are presented in terminal yellow, erect infructescences containing up to 50 fruits. They contain 1-2 seeds, 5.0 x 4.9 mm. Palicourea gibbosa seeds are dispersed by birds, including Mvadestes ralloides, Pioreola riefferi, Atlapetes brunneinucha, Masius chrysopterus, and Tangara arthus (C. Restrepo and N. Gomez, unpublished data).

Faramea affinis exhibits two flowering periods per year. Its tubular, purple flowers are visited by hummingbirds, including Coeligena wilsoni (Samper 1992). Fruits are blue





18


and are presented in terminal, pendant, green infructescences, containing a maximum of 3 fruits. They measure 20 x 18.2 mm and contain a single seed, 10.4 x 7.6 mm (Samper 1992). Seeds of F. affinis are dispersed by a different set of birds, Andigena laminirostris, Linaugus cryptolophus, Pipreola riefferi, Semnornis ramphastinus, and Trogon personatus (Restrepo 1990, Beltrdn 1991, Samper 1992).


Methods


I evaluated the combined influence of edges and treefall gaps on several stages of the life cycle of Palicourea gibbsag and Faramea affinis by sampling individuals and conducting experiments at six edges (three old and three new) and at four distances from forest edge towards forest interior (0-10 m, 30-40 m, 60-70 m, and 190-200 m) (Chapter 2). Depending on the stage of the life cycle I was examining, I modified the basic sampling design described in Chapter 2. This was due to logistic constraints, including accessibility of the edges.

Beginning in March 1992 I tagged all individuals < 2 m tall in flower and/or in fruit, and classified them as being in gap or interior. An individual was classified as in gap if it was within a gap (sensu Brokaw 1982) or located < 2 m from a gap edge, and as interior if it was located > 2 m from the edge of the nearest treefall gap at the time the study began. I continued tagging individuals throughout the study period as new individuals flowered. For each individual I





19


marked all inflorescences and infructescences and followed them over the entire study period. I monitored individuals for the presence of inflorescences and infructescences on a biweekly basis during the first 6 months (March 1992-August 1992) and on a monthly basis the following 11 months (September 1992-July 1993).


Pollination


To determine the influence of edges on pollination

success I looked at pollen tube production. From June 1993 until November 1993 I checked flowering individuals for four consecutive days to collect an average of 10 flowers per individual. These individuals represent a subset of those that were monitored over the 16-month period. I dissected the flowers and fixed the styles in formalin-acetic acid (FAA) to examine pollen tubes. Pollen tubes were stained (Martin 1959, Feinsinger et al. 1992) and counted under an epifluorescent microscope. Styles were processed by P. Amezquita at the Universidad de Santiago de Compostela, Spain. Pollen tube production per individual was expressed as the percentage of flowers with pollen tubes (F) and as the average number of pollen tubes per flower (P). Fruit Set


For each new inflorescence I counted the number of flower buds and followed them until fruits developed and ripened. I expressed fruit set as the percentage of unripe





20


fruits in relation to the number of flower buds and as the percentage of ripe fruits in relation to the number of unripe fruits counted over the entire study period for each individual. I present results for P. cibbosa only.


Fruit Damaae by Insects


At the same time I monitored infructescences for unripe and ripe fruits, I recorded two types of fruit damage by insects: damage to seeds by wasps (Hymenoptera: Chalcidoidea) and removal of pulp by ants (Hymenoptera: Formicidae: Ponerinae). The former could be recognized by exit holes left by newly emerged adults and the latter by bites taken from fruits. These two types of fruit damage were the most common ones for these two understory plants. I expressed seed and fruit damage as the proportion of unripe fruits exhibiting one of the two types of damage in relation to the total number of unripe fruits produced by an individual over the entire study period. I present results for P. gibbosa only.


Seedlina Growth and Leaf Production


I monitored seedlings of Palicourea gibbosa and Faramea affinis at each of three distances (0-10, 30-40, 60-70 m) at three old edges (C6limo I, C6limo II, and Pialapi) to establish the combined effect of distance from forest edge and treefall gaps on seedlings growth and leaf production (Fig. 2-1, Table 2-2, Chapter 2). In May 1992 I located





21


seedlings of E. aibbosa and F. affinis. I placed by their side a stick with a piece of flagging tape with a distinctive number for each seedling. I recorded whether seedlings were growing in treefall gaps (n = 201 and n = 213, P. gibbose and E. affinis, respectively) or intact forest (n = 210 and n = 221). To standardize measurements I marked the stems of each seedling with yellow vinyl paint (ca. 1.5 cm above soil surface) and the youngest pair of leaves with threads of flagging tape tied around the petioles. With calipers I took a first measurement of the seedling's height from the yellow mark to the base of the meristem and I repeated this procedure five times between May 1992 and October 1993. I also recorded and marked new pairs of leaves.

Seedling growth rate (GR) is expressed as the increment in height between the first (hn) and the last measurement (hn+l) [GR = (hn+l hn/tn+l tn)*(30 days/month)] (Seizer 1992). Leaf production rate (LPR) is expressed as the number of new leaves produced between the first (ln) and last (ln+l) period [LPR = (ln+l ln/tn+l tn)*(30 days/month)].


Field ExDeriments


Experiments on seed predation and seed germination of aibbosA and E. affinis were performed at three edges (Herm6genes, C6limo I, and C6limo II) and that on fruit removal (E. aibbosa) at two edges (Herm6genes and C6limo I) (Fig. 2-1, Chapter 2). Even though these edges represent two





22


different ages (Table 2-1) and edge age is known to influence the effect of distance on vegetation (e.g., Williams-Linera 1990), I chose them to conduct this work because they were close enough to allow frequent monitoring of seeds and fruits. At each of four distances from forest edge (0-10, 30-40, 60-70, and 190-200 m) (Fig. 2-3, Chapter 2) I mapped the treefall gaps and randomly chose 4 of them. At each distance I paired each treefall gap location with an intact forest location.

Seed predation and seed termination

In the seed predation and seed germination experiments I placed an aluminum tray (15 x 7 cm) in each gap and interior site. I punctured the trays to prevent water from accumulating, filled them with soil, and positioned them flush with ground level. I placed 10 seeds of E. gibbosa and

5 seeds of E. affinis in different trays. Seeds were obtained from ripe fruits, and those showing damage by insects were discarded. In total I used 1,920 seeds of E. aibbosa and 960 of E. affinis. I placed 92 trays containing seeds for each edge/species/experiment (32 trays) and simultaneously ran the germination and predation experiments for each species within each edge. The trays containing the seeds sown to evaluate changes in germination rates were covered with galvanized mesh (5 x 5 mm) to protect seeds from vertebrates.

I checked trays on a weekly basis and counted the number of seeds remaining and the number of seeds germinated, i.e.,





23


seeds in which the hypocotyyl was visible (ca. 3 mm long). The seed predation experiment for E. gibbo lasted for 5 days (July 1993-August 1993) and the seed germination experiment for 105 days (July 1993-November 1993). The seed predation experiment for E. affinis lasted for 105 days (August 1993-December 1993) and the seed germination experiment for 252 days (August 1993-April 1994). I concluded the seed germination experiments when 90% of the seeds had germinated and the seed predation experiments when no more seeds were being removed. I assumed that seeds removed from the trays were taken by vertebrates and that this constituted predation.

Fruit removal

In this experiment I placed eight artificial shrubs per distance, 4 at each gap and interior site, for a total of 32 artificial shrubs per edge. Each shrub consisted of a 1.5-mtall bamboo stick to which I attached an artificial infructescence resembling that of E. gibbosa. The artificial infructescences consisted of a 15-cm-long wooden rod from which four pairs of tooth picks extended. The rods and tooth picks were dyed bright yellow, and at the end of each tooth pick I inserted a recently collected ripe fruit of .E. gibbosa. I ran the fruit removal experiment at each edge for four consecutive days. On the morning of the first day (0700) I inserted fresh fruits of P. gibbosa and 24 hours later recorded the number of fruits missing and bitten by ants. All fruits were changed every 24 hours to start a new





24


run of the experiment. I ran this experiment from June 26 1993 to July 3 1993.


Analygs


I analyzed data with ANOVAs for Mixed Factorial designs (Girden 1992). The full design (edge age, distance from forest edge, and treefall gaps) was set up as a split-splitplot design (Winer et al. 1991). The factors of interest were edge age, distance from the edge, and habitat. The edges that I sampled for each level of edge age (old and new) were chosen at random and represented the plot unit. In turn, each edge was divided into four strips (distances from the forest edge towards the forest interior) representing the subplot units. Randomization of the levels of the distance factor was restricted but because the strips were separated in space and I analyzed responses from nonmobile organisms I assumed they represented independent subsampling units. Finally, individuals were classified according to habitat as gap or intact forest, the latter representing the subsubplots.

The design for the seed predation, seed germination, and fruit removal experiments was set up as a split-plot design with one repeated measure (Winer et al. 1991). The factors of interest were distance from the edge, habitat, and time, with time being the repeated measure. In the seed predationseed germination and fruit removal experiments time was represented by weeks and days, respectively. In my design





25


distance (D1 to D4) represents the plot unit, habitat (gap and intact forest) the subplot unit, and edges replicates.

The number of individuals within the gap and interior categories differed at each distance/edge, producing an unbalanced design. I used Type III SS, since it takes into account differences in cell frequencies between treatment combinations (Gagnon et al. 1989; Potvin 1993). To determine whether data satisfied assumptions of an ANOVA, I plotted residuals as a function of fitted Y values. When residuals where not normally distributed, I transformed the data (see type of transformation for each data set).

In all cases, I used an alpha of 10% to increase power of the tests (Zolman 1993). I did so for several reasons. First, the scale at which I worked precluded inclusion of more replicates, which is often the case when dealing with large-scale ecological phenomena (Scheiner 1993). The area encompassed by the six edges was equivalent to 12 ha and access to them was difficult due to steep terrain. Second, in a mixed factorial design the number of degrees of freedom is reduced compared to a factorial design because of multiple nesting (Zolman 1993). In the field, I was limited by the number of edges I could reach within walking distance from the field station, thus I had to set up the design as described. Lastly, the use of Type III SS to analyze unbalanced data sets may lead to Type II errors (Potvin 1993). By increasing the probability of alpha, I compensate





26


for this bias, although it consequently increases Type I errors. In all cases I present P-values.


Results



Pollen Tubes


Distance from forest edge did not influence the

production of pollen tubes in Palicourea gibbosa and Faramea affinis (Table 3-1 and Table 3-2). The percentage of flowers with pollen tubes and the average number of pollen tubes per flower in E. gibbosa, however, was influenced by edge age (ANOVA, F1,4 = 9.2, P = 0.04 and F1,4 = 7.7, P = 0.05, respectively, Table 3-1). At old edges individuals had a higher percentage of flowers with pollen tubes and more pollen tubes per flower (50% 3.1% and 2.7 0.3, n = 105, mean SE, respectively) than those at new edges (36% 4.8% and 2.3 0.5, n = 52, respectively).

Habitat influenced the percentage of E. affinis flowers with pollen tubes and the number of pollen tubes per flower (ANOVA, F1,4 = 8.0, P = 0.02 and F1,4 = 11.8, P = 0.01, respectively, Table 3-1). In intact forest individuals had a higher percentage of flowers with pollen tubes and more pollen tubes per flower (32.5% 3.3% and 0.7 0.06, mean SE, n = 81, respectively) than those in gaps (26.0% 3.1% and 0.5 0.07, n = 48, respectively). The effect of habitat, however, was modified by edge age and distance from forest edge as shown by the significant interaction of












Table 3-1. Results of ANOVAs for Mixed Factorial Designs (Split-split-plot) on percentage of flowers with pollen tubes in relation to the total number of flowers (F) and average number of pollen tubes per flower (P) for Paliourea gibbosa and Faramea affinis.
Significance at 10% (*),5% (**), and 1% (***).


Variable


Age A
df F


Edge (A) Error df SS


Distance
D
df F


D x A df F


DxE (A) Error df SS


df ss


E. gibbosa
F
P

E. affinis
F P


1
1



1
1


9.2** 7.7**


4
4


3.9 1.9


8011.5 3
35.6 3


0.5 0.3


3
3


4 2408.8 3 0.2


4 1.9


3 0.4 3 0.5 12 3.6


1.3 2.5


12 12


3 0.5 3 0.5


11229.4 49.9


12 5054.8 12 3.6


Habitat
H
df F


. aibbosa
F P

E. affinis
F P


1
1


1.0
4.1


H x D df F


3
3


1 8.0**


H x A df F


df F df F df ss


0.5
1.4


1
1


3 1.2


H x D x A df F


0.4 0.3


3
3


1 0.09


1 11.8*** 3 0.9 1 2.4


0.17
0.3


Hx [E (A) xD] Error df SS


16
16


3 2.6
3 4.3**


8968.1
65.04


16 1239.6 16 0.9


3 0.4





28


Table 3-2. Proportion of flowers with pollen tubes in relation to the total number of flowers (F) and average number of pollen tubes per flower (P) in Palicourea gibbosa and Faramea affinis in relation to distance from forest edge. Numbers are the mean i standard error and number of individuals sampled ().


D1 D2 D3 D4

Palicourea gibbosa
F 43.3 4.0 35.7 7.6 52.5 5.0 43.0 6.4
(59) (19) (52) (27)
P 2.7 3.5 2.1 3.7 2.8 2.9 2.4 2.9


Faramea affinis
F 28.7 3.3 25.0 3.7 35.2 5.1 33.3 3.6
(n=39) (n=29) (n=24) (n=37)
P 0.6 0.08 0.4 0.08 0.7 0.1 0.6 0.09





29


habitat, distance, and edge age (ANOVA, F3,7 = 4.3, P = 0.05, Table 3-1).


Fruit Set


In Palicourea gibbosa fruit set was influenced by

distance from forest edge, but the effect depended on edge age, as shown by the significant interaction between distance and edge age (ANOVA, F2,2 = 5.8, P = 0.02, Table 3-3). At D1 and D2 the percentage of developing fruits was greater at new than at old edges but this trend was reversed at D3 where 28%

0.02% (mean SE) of the flower buds resulted in fruits in old edges as compared to 20.0% 0.02% in new edges (Fig. 31). The percentage of ripe fruits was also influenced by distance and similarly depended on edge age (ANOVA F2,2 = 3.9, P = 0.06, Table 3-3). At D1 and D3 the percentage of ripe fruits was greater at new edges but the trend was reversed at D2. At D2 in old edges, 51.0% 0.05% of fruits ripened compared to 42.0 0.04% at new edges (Fig. 3-1). Seed and Fruit Damaae


The percentage of Palicourea gibbosa fruits damaged by

ants and wasps did not differ among the four distances (Table 3-3). Edge age, however, influenced the percentage of fruits eaten by ants but depended on habitat as shown by the significant interaction between edge age and habitat (ANOVA, F1,1 = 4.0, P = 0.07, Table 3-3). At new edges, individuals growing in gaps had a higher percentage of fruits damaged












Table 3-3. Results of ANOVAs for Mixed Factorial Designs (Split-split-plot) on percentage of fruit set (UF/FB), percentage of fruits ripening (RF/UF), percentage of seeds damaged by wasps (UFW/FV), and percentage of fruits damaged by ants (UFA/FV) in Palicoure aibboka. *
Log-transformed data. Significance at 10% (*),5% (**), and 1% (***).


Age
A


Edge (A) Error


Distance
D


D x A


DxE (A) Error


df F df SS df F df F df SS

1 0.0 4 0.1 3 2.8 3 5.8** 12 0.001
1 1.0 4 0.5 3 0.0 3 3.9** 12 0.9
1 0.0 4 0.001 3 1.4 3 0.5 12 0.002
1 0.0 4 0.2 3 0.5 3 1.6 12 0.6





Habitat H x D H x A H x D x A Hx[E(A)xD]
H Error
df F df F df F df F df SS


(Uj


UF/FB
RF/RF UFW/FV


1
1
1


4.7* 1.7
0.001


3 0.9 3 0.4 3 0.7


1
1
1


1.8 0.1 0.3


3
3
3


1.1
0.4 0.3


3 1.4 1 3.9* 3 1.7


16 0.05 16 1.7 16 0.003 16 4.2


Variable




UF/FB*
RF/UF UFW/FV UFA/FV


UFA/FV 11.2





31


0.35 0.3


0.25 02

LU
: 0.15
z

0.1 0.6


EE)




we
Fr


0.55 0.5


0.45 0.4


0.35 -


D1 D2 D3


D1


D2


D3


DISTANCE FROM FOREST EDGE



Figure 3-1. Fruit set in Palicourea gibboa as influenced by distance from forest edge. Points are means and bars standard errors.


.











-


1: New SOld






I





32


by ants (7% 0.01%, mean SE) compared to intact forest (3%

0.01%). This trend was reversed at old edges.


Fruit Removal


Even though on average more fruits of Palicourea gibbosa fruits were removed from the artificial infructescences at D2 (0.4 0.2, mean SE) than at the other distances (D1, 0.3 0.09, D3, 0.3 0.9, and D4, 0.1 0.08), this difference was

not significant (ANOVA, F3,3 = 0.4, P = 0.7). The same was true for habitat where on average more fruits were removed from intact forest (0.3 0.1) than from gaps (0.2 0.06) (ANOVA, F1,52 = 0-6, P = 0.4).


Seed Predation


The number of seeds remaining in the trays averaged over time did not differ among the four distances in Palicourea aibbosa but they did differ in Faramea affinis (ANOVA, F3,6 =

3.5, P = 0.09; Table 3-4, Fig. 3-2a). In E. affinis the number of seeds remaining in the trays decreased from D1 to D4, indicating higher removal rates at the interior. Habitat alone did not have an effect on the number of seeds remaining in the trays for either species (Table 3-4, Fig. 3-3a).

In E. affinis the distance effect was modified by

habitat and by week as shown by the significant distance x

habitat and distance x week interactions (ANOVA, F3,80 = 2.3, P = 0.08 and F51,102 = 2.0, P = 0.002, respectively, Table 34). The number of E. affinis seeds remaining in the trays





33


was similar in gap and intact forest from D1 to D3, but lower in gap (3.5 0.1, mean SE) than in intact forest at D4 (4.4 0.08). Over the 18-week period the mean number of F. affinis seeds remaining in the trays decreased at all four distances but the rate of decline was steeper at D4 and at D2 than at D1 and D3.


Seed Germination


Palicourea gibbosa seeds germinated sooner (week 6) than Faramea affinis seeds (week 18). Averaging over time, germination rates of E. gibbosa seeds were significantly affected by distance from forest edge. The same was not true for E. affinis. Germination rates of 2. aibbosa seeds were greater at D2 than at D1 and D4 (ANOVA, F3,6 = 5.0, P = 0.04; Table 3-4, Fig. 3-2b). Habitat had a significant effect on seed germination rates in 2. gibbosa but not in E. affinis (ANOVA, F1,80 = 7.3, P = 0.008 and F1,78 = 2.3, P = 0.13, respectively). Averaged over time, more seeds of P. aibbosa germinated in gaps than in intact forest (Fig. 3-2b).

The effects of distance and habitat were modified by

time as shown by the significant distance x week (E. affinis) and habitat x time (k. gibbosa) interactions (ANOVA, F42,84 =

2.0, P = 0.003 and F3,1120 = 20.3, P = 0.006, respectively). In the former, germination rates over time were steeper at D2 and D3 than at D1 and D4. In the latter, germination rates were higher at gaps than in intact forest.





34


Seedling Growth Rate


Distance did not influence relative growth rates in

Palicourea gibbosa and Faramea affinis (Table 3-5, Fig. 32c). Habitat, however, had a major effect on both species (Table 3-5). Seedlings showed greater growth rates (ANOVA, E. gibbosa, F1,399 = 38.7, P = 0.0001 and E. affinis, F1,422 =

8.1, P = 0.005; Fig. 3-3c) in gaps than in intact forest. Overall seedlings of E. gibbosa grew faster (5.0 0.17 mm/month, mean SE, n = 411) than those of E. affinis (2.4 +

0.08 mm/month, mean SE, n = 434). Leaf Production


As with growth rate, habitat and not distance from forest edge had a significant effect on leaf production (ANOVA, f. gibbosa, F1,408 = 17.1, P = 0.0001 and E. affinis, F1,424 = 3.4, P = 0.06; Table 3-5, Figs. 3-2d, 3-3d). Seedlings of both species produced more leaves per month in gaps than in intact forest (Fig. 3-3d). Overall, leaf production was greater in seedlings of Palicourea aibbosa (0.8 0.01 pairs of leaves/month, mean SE, n = 420) than in seedlings of Faramea affinis (0.38 0.008 pairs of leaves/month).


Discussion


Of the stages of a plant's life cycle, I examined

pollination, fruit set, seed dispersal, seed predation, seed germination, and seedling growth. An important stage












Table 3-4. Results of Mixed Factorial ANOVAs (Split-splot, 1 repeated measure) on mean numbers of seeds remaining in trays (predation experiment) and mean number of seeds
germinated in trays (seed germination experiment) for Palicourea gibbosa (PG) and Faramea


Significant at 10% (*),5% (**), 1% (***), 0.1% (****)


Distance


FA (Predation) PG (Predation) FA (Germination) PG (Germination)







FA (Predation) PG (Predation) FA (Germination) PG (Germination)


df
3
3
3
3


F
3.5 0.3 0.1
5.0**


D x E


df
6
6
6
6


SS
75.4 182.7
19.9
24.2


Habitat


df
1
1
1
1


F
1
1.5 2.3
7.3***


H x D


df
3
3
3
3


F
2.3* 1.9 0.5 1.6


TxHx(DxE)]
Error df SS 80 1068.9 80 2721.1 78 366.8 80 910.6


CO)
uL


Week x D W x D x E Week W x H W xH x D W x T x[Hx(DxE)]

df F df SS df F df F df F df SS


51 1.9*** 15 0.3 70 2.7***
42 2.0**


102 30
210 84


50.9 60.6
71.4 87.8


17
5
35
14


44.2**** 17 163.7**** 5 304.4**** 35 835.0**** 14


1.7
1.4 1.2
4.3***


51 1.3 15 1.6 105 0.6 42 0.9


1360
400 2730
1120


769.2
845.4 907.2
1300.5


affinis (FA)










8 4


D1 D2 D3 D4
DISTANCE FROM FOREST EDGE (m)


(c)


D2


D3


D1


DISTANCE FROM FOREST EDGE (m)


U


SU ca



.0
E


3



2-


0


D 1



0
z
0
D
0
q: 0.


0.9 S0.8
0


(0.6

0.5

0.4-


0.3


D4


D1


I I
D2 D3
DISTANCE FROM FOREST EDGE (M)


D2


D3


DISTANCE FROM FOREST EDGE (m)


Figure 3-2. Seed germination (a), seed predation (b), seedling growth (c), and leaf production in Palicourea gibbosa and Faramea affinis as influenced by distance from forest edge. Points represent means and bars standard errors.


(a) FA

PG


(b)


a) .0

E


-C
C
0
E

E


7


6


5


4-


3



6


5


4


3


2


1


L

LU LU


z
0



U)


(d)


D4


('3 0~~


D4


4


8


|





37


missing from this analysis is seedling establishment which links seed germination to seedling growth. In discussing my results I assume that all seeds that germinated survived into the seedling stage. I also restrict this discussion to those stages for which I present results for both Palicourea aibbosa and Faramea affinis.

Fruiting individuals of J. gibbosa and E. affinis were not distributed uniformly from pasture to forest interior (Chapter 4). Palicourea gibbosa was more abundant closer to the forest edge (D1-D3) than farther inside the forest (D4), and E. affinis was more abundant at D2 and D3 than at Dl and D4 (Chapter 4). Such distributions suggest that distance from forest edge influences one or more stages in the life cycle of these plants. My results show that not all stages in the life cycle of P. gibbosa and F. affinis are influenced equally by the creation of edges and treefall gaps. In addition, species differed in their response to these two types of disturbance.

Pollination was influenced by habitat and edge age but not by distance from forest edge. The percentage of flowers with pollen tubes and the average number of pollen tubes per flower decreased from intact forest to gaps (F. affinis) and from old edges to new edges (2. gibbosa). Although edges and treefall gaps represent two different scales of disturbance, results for these two species suggest that recently disturbed areas affect pollination levels. My results regarding the effect of distance on pollination levels are similar to those












Table 3-5. Results of Mixed Factorial ANOVAs (Split-splot, 1 repeated measure) on relative growth (mm/month) and relative leaf production (pairs of leaves/month) rates in Palicourea
gibbosa (PG) and Faramea affinis (FA). Significant at 10% (*),5% (**), 1% (***), 0.1%


Distance D x E Edge Habitat H x D Residual

df F df SS df F df F df F df SS


FA (Growth) 2 3.3 4 19.4 2 0.9 1 8.1*** 2 2 422 1053.7
PG (Growth) 2 0.2 4 421 2 1.7 1 38.7**** 2 1.1 399 4101.1
FA (Leaves) 2 1.1 4 0.2 2 5.2*** 1 3.4* 2 0.7 424 10.3
PG (Leaves) 2 0.5 4 1.4 2 2.1 1 17.1**** 2 1.2 408 16


00








7 4


Gap Interior
Habitat




(c)













Gav Interior


Habitat


(a)


0
(D

.

E
:3


m Cn
> FL m


3



2


1-_


0





0.9 0.8 0.7 0. E 0.4 0.


Gap Interior
Habitat



(d)













Gmo Interior


Habitat


Figure 3-3. Seed germination (a), seed predation (b), seedling growth (c), and leaf production in Palicourea gibbosa and Faramea affinis as influenced by habitat. Points represent means and bars standard errors.


Lz
w


CO


6.5


0
-6

Co5.5

5
E
4.5 -


I


(b)



0 FA 0 F


4





7


6


5
0 E4


3


C!,


2


7


4





40


reported by Murcia (1993) for a Colombian site north of La Planada. She found that pollination levels in 11 out of 13 species were not affected by distance from forest edge and explained these results in terms of hummingbirds not being influenced by edges. At La Planada I found that this was the case (Chapter 5). Mean capture rates of nectarivorous birds were not influenced by distance from forest edge. It seems then that pollination can not account for differences in the distribution of fruiting individuals of E. gibbosa and F. affinis across pasture-forest edges.

Seed predation increased from edge towards forest

interior in E. affinis but not in E. aibbosa. The effect of distance from edge on E. affinis, however, was influenced by time. Not only were more seeds removed at D4 but they were removed faster than at the other distances. In both species, seed germination was affected by distance from forest edge but depended on time. Seeds closer to the edge germinated sooner than those in forest interior. Assuming an equal probability of seeds arriving at any of the four distances, it is likely that seed predation and seed germination may limit recruitment rates in these two species across pastureforest edge. Differences in seed predation (E. affinis and E. )ibbosa) and in seed germination (E. affinis) over time may result in fewer individuals of these two species establishing in the forest interior (D4).

In another study of F. affinis conducted at La Planada Samper (1992) found that (1) seed removal rates were not





41


affected by habitat (gaps, edge of gaps, and intact forest),

(2) seed germination rates were faster in gaps (mean = 174 days, n = 148) than along edges of treefall gaps (177, n = 132) and intact forest (187, n = 135), and (3) seedling establishment (i.e., the stage at which seedlings become independent from food reserves contained in the seeds) was not affected by habitat. Samper's work and mine show that the seed and seedling stages in E. affinis are affected differently by treefall gaps and edges resulting from human activities.

Once seeds of E. gibbosa and E. affinis arrive and germinate at any distance from forest edge, treefall gaps seem to have a major influence on these two species by increasing growth and leaf production rates in seedlings. This is not in accordance with results obtained in the Amazon, where relative growth rates of seedlings was greater up to 10 m from forest edge towards forest interior (Seizer 1992).

My study shows that edges can influence recruitment

rates of E. gibbosa and E. affinis through their effect on seed predation and seed germination but not on pollination and the growth of seedlings. On the other hand, treefall gaps influence recruitment rates through their effect on seedling growth.














CHAPTER 4
UNDERSTORY FRUIT ABUNDANCE IN A NEOTROPICAL MONTANE FOREST: THE INFLUENCE OF EDGES AND TREEFALL GAPS Introduction


Fruit abundance can be influenced by disturbances

occurring at various scales. In general, fruit abundance increases in small, natural disturbances, such as treefall gaps (Blake and Hoppes 1986; Levey 1988A,1), in large, natural disturbances, such as patches affected by hurricanes (Walker and Neris 1993) or fire (Fleming 1988), and in large, human-disturbed areas, such as abandoned fields and pastures (Martin 1985; Levey 1988A,h; Blake and Loiselle 1991; Lugo and Frangi 1993; but see Wong 1986). In treefall gaps high fruit production is the result of an increase in the number of fruits produced by individuals growing in the disturbed area compared to conspecifics growing in intact forest and to an increase in the number of fruiting individuals (Pifnero and Sarukhan 1982; Clark and Clark 1987; Levey 1990). In large disturbed areas, high fruit production has been related to the same two factors (Auclair and Cottam 1971; Halls 1973; McDiarmid et al. 1977; Fleming 1988), and to the appearance of pioneer species that typically produce more fruits than late successional or mature-forest species (Martin 1985).


42





43


It has been suggested that the spatial heterogeneity of the fruit resource base, regardless of the scale at which disturbances occur, influences the distribution of organisms feeding on fruits (e.g., Martin 1985; Wong 1986; Levey 1988A,h; Heideman 1989; Blake and Loiselle 1991; Loiselle and Blake 1991) and the resulting dispersal of seeds (Murray 1988). Nevertheless, depending on the scale of disturbance changes in seed dispersal might have different consequences for the plants. For example, small-scale disturbances may influence recruitment within populations whereas large-scale disturbances may influence colonization of new areas (Harper 1994).

One immediate consequence of disturbance is the creation of edges or boundaries. In general, boundaries mediate fluxes between adjacent ecological systems (Margalef 1968; Wiens et al. 1985; Gosz 1991). Moreover, because of differences in their permeability to fluxes of material and energy, boundaries may influence the dynamics of neighboring systems (Correll 1991; Ryszkowski 1992). In this context, edges that bound patches resulting from disturbance might influence the dynamics and structure of the neighboring patches by influencing the distribution of resources and/or movement of organisms (Crist et al. 1992; Johnson et al. 1992; Wiens 1992). In particular, edges resulting from large-scale disturbances may influence the movement of seeds and thus the structure of whole landscapes.





44


Seed movement and fruit abundance are related in at least two ways. First, fruit numbers determine the availability of seeds. Second, fruit and seed availability affect the behavior of the dispersers (Murray 1987; Loiselle and Blake 1993). In spite of these well known relationships between fruit abundance and seed movement, few studies have addressed how edges influence fruit abundance (Blanchard 1992). I explored this question in the understory of a neotropical montane forest using edges resulting from forest clear-cutting. In particular, I documented how fruit abundance and seed movement were affected by distance from the forest edge towards the forest interior for the assemblage of understory shrubs and for individual species. Since edge age (Williams-Linera 1990; Blanchard 1992) and treefall gaps (Janzen 1983; Lovejoy et al. 1986; Noss 1991) can modify the steepness of such a response, I examined how they interacted with distance. By looking at four different scales, edges of different age within the forest matrix, distance from forest edge within edges, treefall gaps and foliage density (LAI) within each distance, I could examine changes in the fruit resource base along edges resulting from large-scale disturbances.





45


CanoDv Structure


I estimated leaf area index (LAI) (m2 foliage area/m2

ground area) to (1) characterize the structure of the forest canopy across the pasture-forest edge and to (2) relate LAI to fruit abundance. Fruit production is strongly influenced by light environment (e.g., May and Antcliff 1963; Jackson and Palmer 1977), which in the subcanopy and on the forest floor is influenced by canopy structure (Norman and Campbell 1991). Thus LAI estimates provide a fine-scale description of the light environment of each point where I sampled fruits.

I used a LAI-2000 Plant Canopy Analyzer (Li-Cor, Inc.) in October 1993 to estimate LAI at the three old edges. Estimates of LAI obtained with this instrument are based on the transmitted fraction of incident radiation on the canopy. At each of the old edges (C41imo I, C6limo II, and Pialapi) I established three transects running perpendicular to the forest edge and extending 10 m into the pastures and 210 m into the forest (Chapter 2, Fig. 2-3). I made readings at intervals of 5 m in the first 50 m of the transect, 10 m in the next 130 m, and 20 m in the next 40 m. For each point along a transect I made four consecutive below-canopy readings that together were paired with a single reading taken in an area devoid of trees and shrubs in the nearby





46


pasture. I always kept the lens and LAI-2050 optical sensor pointing in the same direction and 1.5 m above the ground. I covered the optical sensor with a 450 view cap to block my image and direct beam radiation during clear days (Li-Cor 1992).

Estimates obtained with this instrument often

underestimate true LAI (Chason et al. 1991; Hannan and B6gu6 1995). Nevertheless, in the context of this study these values are useful to describe relative changes in canopy structure across the pasture-forest edge. Fruit Abundance


To establish the influence of edges and treefall gaps on fruit production by understory plants I subdivided each of the four 100 x 10 m strips within each edge into five 20 x 10 m quadrats (Chapter 2, Fig. 2-3). In turn each quadrat was subdivided into four 10 x 5 m subquadrats and for each quadrat I chose at random two subquadrats in which to monitor fruit production (Fig. 2-3).

I used Brokaw's (1982) definition of treefall gap to classify each subquadrat as gap or interior habitat. A subquadrat was classified as gap if it was within a gap or located <5 m from a gap edge, and as interior if it was located >5 m from the edge of the nearest treefall gap at the time the study began. These two categories do not reflect the environmental continuum from the center of the treefall to the intact forest nor do they take into account





47


differences in gap size and shape (Brown 1993; Denslow and Hartshorn 1994). Nevertheless, I was more interested in the possible interaction between distance from the edge and treefall gaps than in the treefall gaps themselves.

I monitored changes in fruit production at each pair of the 10 x 5 m subquadrats over a 12 mo period (September 1992August 1993, excluding December). In each subquadrat I identified and counted individual plants <7 m tall bearing unripe and/or ripe fleshy fruits (Levey 1988A,h; Blake and Loiselle 1991). I also included broken limbs bearing fruits. Most species I recorded complete their life cycle within this arbitrarily set understory stratum. A few species, mostly in the Arecaceae (palms), Rubiaceae, and Melastomataceae, also fruit in higher strata. For each individual, except species in the Araceae, I counted the total number of unripe and ripe fruits every month on a biweekly basis. I averaged these biweekly counts to obtain a single value on fruit abundance for any given month.

I expressed fruit production in four different ways: (1) total number of individuals bearing unripe and/or ripe fruits

(TI); (2) total number of fruits (unripe + ripe fruits) (TF);

(3) total number of ripe fruits (RF), and (4) total number of fruits (unripe + ripe fruits), excluding the Arecaceae (TFA). In all cases, fruit abundance is expressed as the mean number of counts per 50 m2, the area of each 10 x 5 m subquadrat.





48


I included unripe fruits because they constitute a food resource for frugivorous insects. I excluded the Araceae from variables 2-3 because it was difficult to estimate fruit numbers for each infructescence. I excluded the Arecaceae from variable 4 because their high productivity and prolonged fruiting season could mask patterns of fruit production among shrubs producing fewer fruits and fruiting over shorter periods of time.

To explore fruit abundance responses to edges at the

species level I looked at the number of individuals bearing unripe and/or ripe fruits. For each species I pooled this information for all subquadrats and months to obtain a single value for each edge age and distance.

I collected most plant species and deposited voucher specimens at Botany Department Herbarium, Arizona State University (ASU), Herbario Nacional de Colombia (COL), Botany Department Herbarium, Field Museum of Natural History, Chicaco (F), Herbario de la Universidad de Antioquia (HA), Kew Botanical Garden (K), Missouri Botanical Garden (MO), New York Botanical Garden (NY), Herbario Universidad de Narifno (PSO), Utrecht Herbarium (U), Smithsonian Institution (US), and Department of Botany Herbarium, University of Wisconsin (WIS). Family names follow Cronquist (1981). Seed Movement


To evaluate seed movement across edges I counted and

identified seeds contained in bird droppings retrieved from





49


birds captured in mist nets. Birds were sampled at the same edges and distances from the forest edge as were fruiting plants (Chapter 5). After capture, birds were kept in cloth bags lined with filter paper for ca. 20 min. Bird droppings were preserved in alcohol and seeds were compared to a reference collection, compiled during the study period.

This method for evaluating seed movement may have biases in addition to those involved when sampling birds with mist nets (see Chapter 5). In particular, seeds recovered from birds might represent a non-random sample of seeds ingested, since seed handling varies within and among species depending on seed size and other seed characteristics (Levey 1986, 1987). Nevertheless, this method does provide information on seed movement that would be difficult to determine by other means (e.g., seed traps).


Analyses


I used a Repeated Measures ANOVA to analyze LAI. Edge was included as a between factor variable, distance from the edge as a within factor variable, and individual transects as subjects.

I analyzed data on fruit production at the community

level with ANOVAs for mixed factorial designs (Girden 1992). The mixed design was set up as a split-split-plot design with one repeated measure (Winer et al. 1991). The factors of interest were edge age, distance from the edge, habitat, and month, the latter representing the repeated measure. The six





50


edges that I sampled for each level of edge age (old and new) were chosen at random from a population of old and new edges and represented the plot unit. In turn, each edge was divided into four strips, i.e., distances from the forest edge towards the forest interior, representing the subplot units. Randomization of the levels of the distance factor was restricted, but because the strips were separated in space and I analyzed responses from non-mobile organisms I assumed they represented independent subsampling units. This was supported by results of an ANOVA in which distance was included as a repeated measure and the epsilon factor equaled one, indicating no correlation between the levels of the distance factor (Girden 1992). Finally, fruit production was monitored in subquadrats that were chosen at random and classified according to habitat as gap or interior, the latter representing the sub-subplots.

The number of subquadrats falling within the gap and interior categories differed at each distance among the edges, producing an unbalanced design. I used Type III SS since it takes into account differences in cell frequencies between treatment combinations (Gagnon et al. 1989; Potvin 1993).

To determine whether the data satisfied assumptions of an ANOVA, I plotted residuals as a function of fitted Y values. When residuals where not normally distributed, I log-transformed the data. In addition, I verified the assumption of compound symmetry for the repeated measure





51


factor and used a corrected F-ratio (H-F) to interpret the analyses (Girden 1992; von Ende 1993).

I used an alpha of 10%. I set alpha at this level

because my design could lead to increases in Type II errors (reduced power of my tests) (Zolman 1993). Concomitantly I increased the probability of committing Type I errors. First, the scale at which I worked precluded inclusion of more replicates, which is often the case when dealing with large-scale ecological phenomena (Scheiner 1993). The area encompassed by the 6 edges was equivalent to 12 ha and the access to them was difficult due to steep terrain. Second, in a mixed factorial design the number of degrees of freedom is reduced compared to a factorial design because of multiple nesting (Zolman 1993). In the field, I was limited by the number of edges I could reach within walking distance from the field station and thus I had to set up the design as was described above. Lastly, the use of Type III SS to analyze unbalanced data sets may lead to Type II errors (Potvin 1993).

I analyzed fruit abundance data for understory species using a Replicated Goodness of Fit Test (G-statistic) (Sokal and Rohlf 1981) to establish whether distance and edge age affected the number of fruiting individuals. First, I pooled the data for old and new edges and calculated Gp (G-Pooled) to determine if the number of individuals across the four distances departed significantly from a uniform distribution. Second, I compared old and new edges and calculated GH (G-





52


Heterogeneity) to test for homogeneity between the two edge ages. For these two analyses I used only species in which at least 80 percent of the expected cell frequencies were greater than 5, since the G statistic departs from the X2 distribution if this is violated (Siegel and Castellan 1988).


Results



LAI and Distance from the Edge


I measured LAI only at old edges and found that major

changes in LAI were observed at the interface between pasture and forest (Fig. 4-1). Once inside the forest, LAI values were highly variable not only from the edge towards forest interior but also among edges at the same distance (Fig. 41). Averaging over the three edges, variability in LAI measurements for the 0-40 m interval (Coefficient of variation, CV = 0.3) was identical to that for the 50-210 m interval (CV = 0.3).

I did not include pasture LAI values in the ANOVA to

establish the effect of distance on LAI. LAI did not differ significantly either among edges (ANOVA, F3,5 = 0.4, P = 0.75) or in relation to distance from the forest edge towards the forest interior (F23, 115 = 0.47, P = 0.98) (Table 4-1). This indicates that at least along old edges, canopy structure does not vary in a predictable way from the edge towards the forest interior.





53


LAI and Fruit Abundance


To establish whether LAI influences fruit abundance, I averaged values of fruit abundance for each pair of 50 m2 subquadrats where I took LAI measurements. LAI was not significantly correlated with total number of fruits (TF) (Coefficient of determination, r2 = 0.032, n =36), total number of fruits excluding the Arecaceae (TF-A) (r2 = 0.077, n = 36), total number of ripe fruits (RF) (r2 = 0.015, n = 36), or total number of fruiting individuals (TI) (r2 = 0.02, n = 36) (Fig. 4-2).


Fruit Abundance


Plant Assemblages

In the ANOVAs none of the three-way interactions was

significant. Several two-way interactions were significant but not consistently so for the four measurements of fruit abundance. In describing the results I look first at the single effect of distance on fruit abundance and then at the interactions involving this term.

The total number of fruits (TF) and total number of ripe fruits (RF) differed significantly among the four distances on a yearly basis (ANOVA, F3,12 = 4.3, P = 0.03 and F3,12 =

5.4, P = 0.01, respectively) (Table 4-2). For TF and RF the mean number of fruits as well as the variance decreased from forest edge towards forest interior (Fig. 4-3). There are three not mutually exclusive explanations for these results.





54


8

7

6

5 C\1
E
c4 4
E
- 3
..J
2

1

0
-2


DISTANCE (m)


t


Figure 4-1. Leaf area index across the pasture-forest edge, indicated by the arrow (0 m). Points represent the average of three measurements per edge and bars the standard errors. Open squares = Pialapi, open triangles = C6limo II, and filled circles = C61imo I.


iI I I I I I I I
0 0 20 40 60 80 100 120 140 160 180 200 220











Table 4-1. Results of a Mixed Factorial ANOVA on leaf area index for old edges at the Reserva Natural La Planada

Source df SS MS F-Value P-Value H-F

Edge 3 6.768 2.256 0.401 0.7585
Transect (Edge) 5 28.095 5.619
Distance 23 11.64 0.506 0.468 0.981 0.981
Distance Edge 69 76.934 1.115 1.031 0.4367 0.437
Distance Transect (Edge) 115 124.376 1.082

H-F Epsilon
Distance 1.714


LJ L,





56


First, individuals growing at D1 (0-10 m) produced more fruits than those growing at other distances. Support for this comes from the fact that distance did not have an effect on the total number of fruiting individuals (TI) (Table 4-2) and the observation that some species that were heavily represented at D1 (e.g., Clidemia sp. 1 and Palicourea aibbosa) produced larger crops here than at the other distances. Recall that TI eliminates the variability associated with crop size because it considers only the number of fruiting individuals. Second, palm fruits made a disproportionate contribution to overall fruit production at D1. Support for this comes from the fact that distance did not have an effect on total number of fruits excluding the Arecaeae (TF-A) (Table 4-2). Third, some species found only at D1 (0-10 m from the forest edge), including Marcaravia eichleriana, Marcaraviastrum subssesilis, Miconia pseudoradula, M. theaezans, Psammisia ferruginea, Schefflera lasiogyne, Phytolacca rivinoides, produced large fruit crops. Even though the effect of distance alone was not significant for total number of fruits excluding palms (TF-A), the trend was similar to that for the previous two variables, i.e., sharp decrease from forest edge towards forest interior.

The effect of distance on fruit abundance was modified by edge age and habitat, but neither factor alone had an effect on any of the variables describing fruit abundance (Table 4-2). The interaction between edge age and distance from the forest edge was significant for total number of











5000


C'4 4000E
0
LO 3000

2000
0
1000
C
CD
E 0-


-1000 1000


Toa Fut


A L 0o


800


z D=


2 3 4 5 6
LAI


Total Fruits
(excluding palms)




A

O

2 A 4

1 2 3 4 5 6


LAI


Figure 4-2. Leaf area index (LAI) Symbols represent old edges. Open II, and filled circles = C61imo I.


w


z


D


C14
E
600


400
0

C200
E

0


12.5

E 10
0
7.5



0 c,5
Q
C cc cD 2.5

0


w


z



LL


1 2 3 4 5


6


LAI

and mean counts expressing fruit abundance. squares = Pialapi, open tirangles = C61imo


Ripe Fruits



-A







1 2 3 4 5 6
LAI



Total Individuals
0
0%
Ae








3 4 5

-c
Tota Iniviual


w


z


D
EE


C'.
E
0


0 CD

E


750 500 250


0





58


fruits (TF) (ANOVA, F3,12 = 2.56, P = 0.1; Table 4-2) Total number of fruits (TF) showed a sharp decline from the forest edge towards the forest interior at new edges and remained almost unchanged at old edges (Fig. 4-4). The interaction between habitat and distance from the forest edge was significant for total number of fruits (TF) (ANOVA, F3,16 =

2.95, P = 0.06) and for total number of ripe fruits (RF) (ANOVA, F3,16 = 3.5, P = 0.04; Table 4-2). For both variables, fruit abundance at Dl (0-10 m from forest edge) was higher in gaps than in forest interior (Fig. 4-5). These differences disappeared at the other distances. The fact that none of these interactions was significant for total number of fruits excluding palms (TF-A), suggests that palms made an important contribution to these results.

The number of ripe fruits (RF) differed significantly between old and new edges depending on month (ANOVA, F10,40 2.8, P = 0.009; Table 4-2, Fig. 4-6). The total number of fruiting individuals (TI) and the total number of fruits excluding Arecaceae (TF-A) differed across the four distances in some months but not in others as shown by the significant interaction between distance from edge and month (ANOVA, F30,120 = 1.48, P = 0.07 and F30,120 = 1.57, P = 0.04, respectively; Table 4-2, Fig. 4-7). These results contrast with those for TF and RF, in which this interaction was not significant but in which fruit abundance averaged over time was affected by distance (Table 4-2). Recall that the total number of fruiting individuals (TI) and the total number of











Table 4-2. Results of ANOVAs for Mixed Factorial Designs (Split-split-plot, 1 repeated measure) on mean counts/50 m2 of number of fruiting individuals (TI), number of fruits (TF), number of ripe fruits (RF), and number of fruits excluding Arecaceae (TF-A). 0 log-transformed data. Significance at 10% (*),5% (**), and 1% (***).

Variable Age Edge(Age) Distance Distance x Age D x E(A)
(A) Error (D) Error
df F df SS df F df F df SS
TI 1 0.99 4 1256.13 3 1.77 3 0.65 12 1745.05
TF* 1 0.28 4 71.37 3 4.34 ** 3 2.56 12 67.2
TF-A* 1 1.26 4 39.18 3 1.43 3 0.4 12 73.79
RF0 1 0.47 4 40.58 3 5.38 ** 3 0.96 12 31.67
Habitat H x D H x A H x D x A Hx[E(A)xD]
(H) Error
df F df F df F df F df SS
TI 1 0.74 3 2.24 1 0.522 3 0.016 16 1522.51
TF0 1 0.058 3 2.95 1 0.047 3 0.41 16 65.02
TF-A* 1 1.26 3 0.77 1 0.244 3 0.165 16 42.28
RF* 1 0.26 3 3.47 ** 1 0.01 3 1.02 16 29.38
Month x H M x D x H M x A x H M x A x D x H MxHx[DxE(A)]
Error
df F H-F df F df F df F df SS
TI 10 0.86 30 1.27 10 0.31 30 1.28 160 389.1
TF* 10 2.03 ** 30 0.55 10 0.37 30 0.6 160 32.45
TF-A0 10 1.42 30 0.58 10 0.69 30 1.28 160 24.58
RF* 10 2.01 ** 30 0.41 10 0.94 30 0.62 160 54.78
M x A M x E(A) M x D M x D x A M x D x E(A)
Error Error
df F H-F df SS df F H-F df F df SS
TI 10 1.47 40 144.98 30 1.48 30 1.09 120 350.03
TF* 10 0.49 40 21.43 30 1.29 30 0.99 120 25.43
TF-A" 10 0.65 40 17.25 30 1.57 ** 30 0.7 120 22.13
RF* 10 2.83 *** 40 1.27 30 0.76 30 1.17 120 43.2


(-n





60




1100
1000- Total Fruits

900800700600500

Total Fruits 350- (excluding palms)

300250 Q C200z

Q0 150Z LO Ripe Fruits
-=3 200 LL 15010050


Total Individuals


6



5



4 I I I
D1 D2 D3 D4
DISTANCE FROM FOREST EDGE (M)



Figure 4-3. Variation in fruit abundance at the Reserva Natural La Planada in relation to distance from forest edge. Points represent means and bars standard errors.





61


fruits excluding Arecaceae (TF-A) eliminates part of the variability associated with crop size because (1) TI considers only the number of fruiting individuals and (2) TFA includes plant species producing smaller and less persistent fruit crops, i.e., number of fruits per reproductive season, than those of palms and plant species producing similar number of fruits.

Lastly, fruit abundance expressed as the total number of fruits (TF) and total number of ripe fruits (RF) differed significantly between gaps and interior, but depended on month, as shown by the significant interaction between habitat and month (ANOVA, Flo,160 = 2.0, P = 0.03 and Flo,160 = 2.0, P = 0.03, respectively; Table 4-2). The Arecaceae again seemed to be mostly responsible for this result as indicated by the fact that this interaction was not significant for the total number of fruits when the Arecaceae were excluded (TFA) (Table 4-2).

Species Level Responses

I recorded 149 plant species fruiting in the understory of the edges included in this study and classified them in five categories: extremely sparse (1 individual), very sparse (2-5 individuals), sparse (6-20 individuals), abundant (21-50 individuals), and very abundant (>51 individuals). Of 149 species, 26 (17%) were abundant to very abundant (>21 individuals) and 125 (83%) were sparse to extremely sparse (<21 individuals) (Appendix A). The most abundant species





62


1500
Total Fruits
1250 1000750

500

250 3 NE
OLD

Total Fruits 500- _(excluding palms)

400

N 300W E Z C) 200
O
Za, SED CRipe Fruits
300








Total Individuals


6
5 -








4

3
D1 D2 D3 D4
DISTANCE FROM FOREST EDGE (m)



Figure 4-4. Variation in fruit abundance at the Reserva Natural La Planada in relation to edge age and distance from forest edge. Points represent means and bars standard errors.





63


2000
Total Fruits 1500- 0 GP


1000 500



Total Fruits (excluding palms)



300


CJ 200wE


Ripe Fruits
Q C400 <0
300

E 200100


Total Individuals
T
6


5 4


3
D1 D2 D3 D4
DISTANCE FROM FOREST EDGE


Figure 4-5. Variation in fruit abundance at the
Reserva Natural La Planada in relation to habitat
and distance from forest edge. Points represent
means and bars standard errors.





64


were two Rubiaceae, Faramea affinis and Palicourea aibbosa. One hundred species were found both at new and old edges and 49 were exclusive to new (35 species) and old edges (14 species) (Appendix A). In this respect, new edges presented proportionally more species than old edges (Goodness of Fit test, G = 9.3, P < 0.01).

Fruit abundance on a species by species level, expressed as the number of fruiting individuals, varied depending on distance from the edge and edge age. The distribution of fruiting individuals for 16 of the 26 abundant species departed significantly from a uniform distribution across the four distances (Table 4-3). Given a 10% probability of obtaining a species that shows a non-uniform distribution it is very unlikely that 16 or more species out of 26 would have shown a non-uniform distribution by chance alone (Binomial test, P = 2.0 x 10-10). Clearly, the distribution of fruiting individuals of some species is affected by the creation of edges.

For the 16 species showing a non-uniform distribution I used residuals to further determine if they were more abundant at any particular distance from the forest edge (i.e., if at any given distance the observed frequency was greater than the expected frequency). Four species were more abundant at D1 (0-10 m), three species were more abundant at D4 (190-200 m), four species were more abundant at D2 (30-40 m) or D3 (60-70 m), and five species were more abundant at two different distances (e.g., Alloplectus tetragonus and





65


1250

1000 Total Fruits

750 500

o New
250 -Nw
Old 600-
Total Fruits
500- (excluding palms)

400
C~4

E 300zo
0 -. 200600-
< 0
500- Ripe Fruits LL 400E
300200100

8

7 Total Individuals

6

5- T





S 0 N J F M A M J J A MONTH



Figure 4-6. Variation in fruit abundance at the Reserva
Natural La Planada in relation to edge age and month. Points
represent means and bars standard errors.





66


2000


1500


1000


500





750
C\1
o E 500z o
< LO 0 Z c 250LL. 0
D 5
LL() 500 -


300


100




7

6

5

4

3

2-


Total Fruits











Total Fruits
(excluding palms)











Ripe Fruits












Total IndividualsT




w D1 o D3 A D4 S 0 N J F M A M J J A MONTH


Figure 4-7. Variation in fruit abundance at the Reserva Natural La Planada in relation to distance from forest edge and month. Points represent means and bars standard errors.


-





67


Palicourea gibbosa), the latter suggesting a bimodal distribution (Table 4-3). Species showing bimodal distributions, in particular, suggest that distance alone can not explain their distribution across the pasture-forest edge.

I evaluated the combined effect of distance and edge age on the number of fruiting individuals for 16 species (Table 4-4). For nine species there was a significant interaction between distance and edge age, indicating that the distribution of fruiting individuals across the four

distances differed between old and new edges (GHeterogeneity, P < 0.1, Table 4-4). A significant interaction between distance and edge age was irrespective of whether the species showed a non-uniform distribution across the four distances by combining the two types of edges (Gpooled, P < 0.1, Table 4-3) or by looking at old and new edges separately (Gold, P

0.1 and GNew, P < 0.1, Table 4-4).

Most of the fruiting individuals found at my study edges were represented by few individuals, precluding the use of Goodness of Fit Test to establish how distance from forest edge affected their distribution. Instead, I used information on their abundance (Table 4-5) to establish the distance at which sparse species were found more often. I excluded abundant and very abundant species from the analysis and found that for the remaining three groups of plants abundance and distance from the edge were not independent (Test for Independence, X2 = 18.86, P = 0.004). Examination





68


Table 4-3. Distribution of fruiting individuals of abundant ( 21 individuals) plant species in the understory of La Planada in relation to distance from the edge. Numbers represent number of individuals. Distances are 0-10 m (D1), 30-40 m (D2), 60-70 m (D3), and 190-200 m (D4) from forest edge. P < 0.1 (*), P < 0.05 (**), P < 0.01 (***), P < 0.001


Species D1 D2 D3 D4 G-stat
Uniform Distribution


Burmeistera carnosa 10 11 13 14 0.83 ns
Solanum sp.7 14 15 11 11 0.99 ns
Anthurium membranaceum 32 30 27 24 1.31 ns
Burmeistera sp. nov. 7 4 5 9 2.35 ns
Anthurium umbraculum 17 17 13 23 2.87 ns
Columnea cinerea 10 5 4 5 3.34 ns
Soheraedenia stevermarkii 7 2 6 6 3.37 ns
Anthurium cf. oulverulentum 8 6 5 2 3.99 ns
Anthurim ersicolor 18 8 16 14 4.36 ns
Geonoma weberbaueri 19 20 9 14 5.29 ns


Non-Uniform Distribution


Psammisia aff. debilis 20 9 4 8 12.9 *
Besleria solanoides 39 16 14 4 35.5 *
Psvcothria aubletiana 45 14 12 5 40.3 ****
Clidemia sp.1 54 59 27 10 48.7 ****
Solanum sp.5 3 14 14 17 12.2
Chamaedorea nolvchlada 5 7 15 18 10.8 **
Allonlectus tennis 2 2 7 9 8.0 **
Asolundia sp.1 2 10 5 8 6.6 *
Anthuriu umbricol 9 7 22 7 12.2 *
Anthuri cf. marmoratum 10 9 17 5 7.2 *
Faramea affinis 104 146 127 117 7.6 *
Alloolectus teuscheri 22 41 18 30 10.9 *
Anthuriu carchiense 3 15 5 12 11.7 *
Anthuriu cf. melamnvi 20 10 19 32 12.3 *
Palicourea aibbosa 91 65 91 43 23.6 ****
Alloolectus tetraaonus 10 2 7 1 11.7 ***










Table 4-4. Results of Replicated Goodness of Fit Test on the number of individuals in for understory plants at the Reserva Natural La Planada. GHeterogeneity (GH) GTotal Gold edges (GO), and GNew edges (GN) P < 0.1 (*), P < 0.05 (**), P < 0.01 (***), P <


fruit
(GT) ,
0.001


Species df GH df GT df Go df GN


Anthurium versicolor 3 6.64 6 11 3 1.48 ns 3 9.49 **
Anthurium cf. marmoratum 3 7.88 ** 6 15.08 ** 3 10.72 ** 3 43 ns
AnthUr~im umbricolum 3 28.28 6 40.48 **** 3 17.96 **** 3 3.22 ns
Allolectus teuscheri 3 7.72 6 18.6 *** 3 5.06 ns 3 13.52 *
Geonoma weberbauieri 3 17.56 6 22.86 **** 3 3.93 ns 3 18.92 *
Solanum sp.7 3 6.84 6 7.83 ** 3 2.65 ns 3 5.18 ns
Anthurium cf. melamyi 3 21.76 6 34.02 **** 3 17.38 **** 3 16.6 *
Palicmurea aibbosa 3 6.32 6 29.93 3 16.77 **** 3 13.14 *
Solanim sp.5 3 6.24 18.38 **** 7.16 11.19 **

Anthuriuim membranaceunM 3 0.94 ns 6 0.94 ns 3 1.74 ns 3 0.49 ns
Burmesitera carnosa 3 3.47 ns 6 4.32 ns 3 2.39 ns 3 1.89 ns
Clidemia sp.1 3 2.42 ns 6 51.11 3 32.79 **** 3 18.12 *
Psycothria aubletiana 3 3.84 ns 6 44.15 3 24.29 3 19.83 *
Faramea elegans 3 2.24 ns 6 2.25 ns 3 3.35 ns 3 6.48 *
Besleria solanoides 3 4.04 ns 6 39.5 **** 3 35.13 **** 3 4.35 ns
Chamaedorea Dolychlada 3 3.32 ns 6 14.14 ** 3 12.45 *** 3 1.69 ns





70


Table 4-5. Distribution of fruiting individuals in the understory of the Reserva Natural La Planada across pastureforest edge based on species abundance. Numbers represent the number of fruiting individuals.


Number of


Distance


Extremely sparse (1 individual)

Very sparse
(2-5 individuals)

Sparse
(6-20 individuals)

Abundant
(21-50 individuals)

Very abundant (>51 individuals)


Species D1 D2 D3 D4


33 13 4 7 9



40 64 21 28 25



50 167 144 138 120



13 94 99 115 137



12 475 441 384 327





71


of the residuals showed that more individuals of very sparse species were found at Dl and of sparse species at D2. Although individuals of both groups of plants might be found across the four distances, these results show that sparse species are found more often close to the edges. Seed movement

I recovered 393 bird droppings from 19 species of mistnetted frugivorous birds. Seeds of 93 species, of which I was able to identify 65, were represented. Of the species I identified, 52 (80%) were found in the subquadrats where I counted fruits. This figure, compared to the total number of plant species I recorded fruiting in the understory (149), shows that bird droppings represent a subsample of the plant species that I found.

The number of plant species in individual bird droppings ranged from 1 to 7 (mean SD, 1.7 1.0, N = 393). There was no significant difference in the mean number of plant species in droppings recovered at the four distances (ANOVA, F3,393 = 0.8, P = 0.5). In addition, total numbers of seeds in bird droppings was independent of distance (X2 Test for Independence, df = 15, X2 = 12.8, P = 0.6) .

Fourteen of the 26 abundant species (>21 individuals) and 38 of the 124 sparse species (<21 individuals) fruiting in the understory were represented in the bird droppings (Appendix A). Since sparse species were found more often at the forest edge, I compared their distribution against that of droppings containing their seeds to determine whether they




72


were independent of each other. Plants and droppings were classified as (1) "edge" if the number of observations at D1 (0-10 m) and D2 (30-40 m) combined together was >0, (2) "interior" if the number of observations at D3 (60-70 m) and D4 (190-200 m) was >0, and (3) "edge=interior" if the number of observations at D1 and D2 = D3 and D4. I found that the proportion of droppings containing seeds of sparse species was independent of their abundance in edge and interior (Test for Independence, X2 = 2.4, df= 4, P = 0.6). Even though sparse species were found more often at forest edge, their seeds are potentially reaching forest interior.


Discussion


Distance from the edge towards the forest interior had a major effect on fruit abundance but only close to the forest edge. Nevertheless, the facts that (1) the effect of distance was modified by edge age, habitat, and month of the year, and (2) results for the four variables describing fruit abundance differed, indicate that there are complex interactions between edges and fruiting plants (summarized in Table 4-6). I interpret these complex interactions in terms of scales at which fruit abundance changes in relation to the creation of edges.

At the scale defined by the pastures and the forest matrix and by the length of the study (1 yr), new edges generated spatial heterogeneity in fruit abundance within the study area at D1 (0-10 m from forest edge). Fruit abundance,





73


expressed as total number of fruits (TF) and as total number of ripe fruits (RF), was higher in new than in old edges at Dl but these differences were not evident at the other distances. At the scale defined by the edges and by the length of this study, treefall gaps at Dl generated spatial heterogeneity in fruit abundance within edges. Fruit abundance, expressed as total number of fruits (TF) and as total number of ripe fruits (RF) was higher in treefall gaps than in forest interior at Dl but, again, these differences were not found at the other distances.

When I looked at total number of ripe fruits (RF), and

in addition examined the total number of fruits excluding the Arecaceae (TF-A) and the number of fruiting individuals (TI), a different picture emerged. At the scale of the study area and month, RF differed between old and new edges but depended on month. At the scale of edges and month, TF-A and TI differed among the four distances but also depended on month.

These results suggest that changes in fruit abundance

and the magnitude of these changes across pasture-forest edge are related to the size of fruit "patches". In my study area, large patches of fruit were generated by understory palms, which produced large fruit crops that persisted for a long time. These large fruit patches seemed to generate a steep gradient in fruit abundance from the edge towards the forest interior on a yearly basis. Conversely, small patches of fruit seemed to generate gradients that varied in their magnitude depending on month. I will discuss these results





74


Table 4-6. Summary of results of ANOVAs on fruit abundance across the pasture-forest edge for the different response variables. Fruit abundance expressed as total number of fruits (TF), total number of ripe fruits (RF), total number of fruits excluding the Arecaceae (TF-A), and total number of fruiting individuals (TI). Significance at 10% (*), 5% (**), 1% (***)



TF RF TF-A TI
Large Fruit Small Fruit
Crops Crops
ANOVA terms


Distance ** **
Distance x Age *
Distance x Habitat **
Distance x Month ** *
Age x Month


** **


Habitat x Month





75


in terms of (1) the factors that influence fruit abundance and (2) the consequences that these observed changes might have on plants and frugivores.


Factors Influencing Fruit Abundance


Fruit abundance is regulated by abiotic and biotic

factors interacting in complex ways with the flowering and/or fruiting stages of a plant's life cycle (e.g., Marshall and Grace 1992). Abiotic factors that have a direct effect on fruit numbers and fruit size are photoperiod and irradiance (e.g., Auchter et al. 1926; May and Antcliff 1963; Jackson and Palmer 1977; Mathai and Sastry 1988; Tombesi et al. 1994), temperature (e.g., Chaikiattiyos et al. 1994), water availability (e.g., George et al. 1990), and nutrients (e.g., Stephenson 1992). Biotic factors include pollination, predation of flowers, seeds, and fruits, and damage by pathogens (Stephenson 1981). The importance of these factors is likely to differ within and among species, depending on habitat.

Irradiance is an important factor influencing fruit

abundance. Most work that supports this contention is based on the observation that when irradiance increases in a forest as a result of disturbance, so does fruit abundance (Halls 1973; Pinero and Sarukhan 1982; Clark and Clark 1987; Agren 1988; Levey 1990). Although I did not measure irradiance directly, my estimates of leaf area index (LAI) describe indirectly the light environment at my edges. I found that





76


(1) LAI did not correlate with fruit abundance at old edges,

(2) LAI did not change significantly with distance from forest edge, (3) treefall gaps alone did not influence fruit abundance (but see the significant interaction between habitat and distance, and between habitat and month, on TF and RF), and (4) distance from edge influenced fruit abundance in complex ways. Thus, irradiance alone cannot explain my findings. Other abiotic factors, such as temperature, water availability, and nutrients, likely influence fruit abundance across the pasture-forest edge.

In tropical areas flower and fruit abundance are

influenced by low temperatures (Tutin and Fernandez 1993), soil fertility (Gentry and Emmons 1987), water availability (Heideman 1989; Seghieri et al. 1995), and pollination (Compton et al. 1994). During the dry season (June-August), La Planada experiences clear skies, low rainfall, strong winds, and extreme maximum and minimum temperatures. The effect of these factors on understory vegetation may be exacerbated at the pasture-forest edge, thereby influencing changes in fruit abundance. For instance, during this period, leaves of understory plants and vines at the forest edge but not in the interior wilted and abscised. At a montane locality north of La Planada, soil moisture across the pasture-forest edge changed progressively over the dry season and reached its lowest value 10 m from the forest edge towards forest interior (Murcia 1993). At a lowland tropical site, vapor pressure deficit (VPD) in the understory changed





77


across the pasture-forest edge, decreasing towards the forest interior (Kapos 1989; Seizer 1992). More important, however, were the differences found between the wet and dry season, higher VPD values being recorded farther inside the forest during the dry than during the wet season (Seizer 1992). At the same lowland site, Kapos et al. (1993) compared the carbon isotopic composition (a'3C) of leaves of two canopy and two understory species and found that the D13C concentration decreased from the edge towards the forest interior for the understory but not for the canopy species. These changes were more pronounced for Duquetia aff. flagellaris (Annonaceae) than for Astrocarvium social (Arecaceae). The results described by Kapos et al. (1993) indicate that the understory environment might be more sensitive to edge creation than the canopy and that understory species vary in their sensitivity to the factors that influence their distribution.

At La Planada, increases in fruit abundance at the forest edge could also be related to changes in soil fertility as a result of increased litterfall during the dry season, and deposition of volcanic ash at edges. La Planada is influenced by several active volcanoes that release andesitic ash (Mizota and van Reenwyk 1989, cited in van Wambeke 1992) rich in nutrients (Shoji et al. 1993). These air-borne particles might be deposited disproportionately along pasture-forest edges. Some studies have shown that edges alter the deposition of dry airborne material such that





78


deposition rates decrease from the forest edge towards the forest interior (Geiger 1965; Draiijers et al. 1988). In an area like La Planada, which is characterized by nutrient-poor soils, the addition of nutrients could affect the production of fruits.

A combination of several factors then may account for the changes I observed in fruit abundance for individual species and for the entire understory plant assemblage. I propose that changes in TF-A and TI across the four distances depending on month, are the result of within-year variability in environmental conditions. On the other hand, changes in TF and RF over the year may be the result of increased irradiance at the forest edge due to direct exposure of the pasture-forest interface to sunlight. Recall that TF and RF, which included counts of palm fruits, were most influenced by distance from forest edge. This is consistent with Blanchard's (1992) results showing that increased abundance of palm fruits at the forest edge is correlated with increased light levels.


Consequences for Plants and Fruaivores


Studies looking at the effect of edges on plants have focused more on the vegetative (e.g., Ranney et al. 1981; Chen et al. 1992; Seizer 1992; Young 1993; Matlack 1993) than on the reproductive stage (Romano 1990; Willimas-Linera 1990; Blanchard 1992; Murcia 1993) of their life cycle. The former provides information on the structure and productivity of the





79


edge. The latter provides information on the persistence of edges through time, depending on the ability of plants to complete their life cycle, including the production and dispersal of seeds, in a given environment.

Increases in fruit abundance, and thus of seed outputs, can have different consequences for individuals, populations, and assemblages. A numerical increase in fruit production by an individual can affect traits, such as seed size (Agren 1988, 1989), and thus seedling performance (Westoby et al. 1992). Changes in fruit numbers at the population level can affect recruitment rates (e.g., Kellman and Kading 1992, Guimar&es et al. 1994). For plant assemblages, an increase in fruit numbers can affect colonization rates of disturbed areas and thus alter species composition. In all cases, changes in fruit numbers can affect the behavior of dispersers (Murray 1987; Loiselle and Blake 1993).

At La Planada, the distribution of fruiting individuals was influenced by the presence of edges. Changes in the number of fruiting individuals at any given distance from the forest edge may indicate differences in recruitment rates at different distances. Plant establishment, growth, reproduction, and seed dispersal are very likely to be influenced by distance from forest edge in various ways resulting in the observed distributions of fruiting individuals (Chapter 3). It is possible that if edge conditions remain unaltered over time, i.e., pastures and





80


fields are maintained as such, the observed distributions will persist and even become more pronounced.

Sparse species constituted an important component of the assemblage of understory plants of La Planada. Little is known about habitat preferences of these species, but clearly they represent species typical of large disturbed areas, forest gaps, and forest (Appendix A). Sparse species were most abundant close to the pasture-forest interface (Dl and D2) within edges and at new edges within the study area (34 species were found exclusively at new edges compared with 14 at old edges) (Appendix A). In addition, for the few sparse species from which I recovered seeds in bird droppings, I found that the proportion of fruiting individuals at the "edge" and "interior" was independent of the proportion of bird droppings that contained their seeds and were recovered in these two zones. Thus, forest edges are being colonized by sparse species and factors other than seed dispersal might be influencing recruitment rates across the pasture-forest edge.

Changes in fruit abundance across the pasture-forest

edge partially paralleled that of bird captures (Chapter 5). Increased fruit abundance, expressed as total fruits (TF) and total ripe fruits (RF), was mirrored by an increase in frugivore capture rates only at Dl (Chapter 5). The opposite was true at D4 where fruit abundance reached the lowest values but frugivore capture rates were the highest (Chapter 5). The high fruit production at D1 was mainly due to palms.





81


Bird species known to feed on palm fruits at La Planada, including Myadestes ralloides, Lipaugus crvptoloihus, Semnornis ramphastinus, Pipreola riefferi, Entamodestes coracinnus (C. Restrepo personal observation, Restrepo 1990) and Andigena laminirostris (Beltran 1991) did not account for the high capture rates observed at D1. Thus it is unclear whether changes in TF and RF across the pasture-forest edge influenced the behavior of frugivorous birds. The contrary might be true, however, for TF-A and TI. Those species known to feed on fruits other than palms, which were the majority of frugivorous species captured with mist nets, made an important contribution to the high capture rates observed at D1. Moreover, as occurred with TF-A and TI, there was a significant interaction between distance from edge and month on capture rates of frugivores (Chapter 5).

This study showed that edges influenced fruit abundance in different ways. First, fruit production by the assemblage of understory plants changed abruptly from forest edge towards forest interior but depended on edge age, the presence of treefall gaps, the length of the observations, and whether or not palms were taken into accout. Second, fruit production on a species by species level, and expressed as the total number of fruiting individuals, changed across the pasture-forest edge. This was shown both among abundant and sparse species. The latter provided evidence to support the contention that edges represent zones of opportunities for the establishment of a wide range of species.














CHAPTER 5
EDGES AND UNDERSTORY BIRDS IN A NEOTROPICAL MONTANE FOREST


Introduction


Natural disturbances play a major role in maintaining high levels of diversity in tropical ecosystems (e.g., Connell 1978, Salo et al. 1986, Bush 1994, Gentry 1986; but see Haffer 1969, Hubbell and Foster 1987). At regional scales, one result of disturbance is the creation of ecotones, which have been postulated to favor speciation processes (Bush 1994) and high species richness (Terborgh 1977, Bush 1994). Human disturbances, on the other hand, have resulted in a variety of land uses. Focus has now shifted towards understanding how human disturbances affect distributions of species (e.g., Kattan 1992) and how this might impact ecosystem processes (Vitousek 1990, Kruess and Tschnarntke 1994, Tilman and Downing 1994).

One consequence of human activities on landscapes is the creation of sharp edges bounding disturbed areas, such as pastures, logged forest stands, and agricultural fields. These edges, whether found in little or highly modified landscapes, may influence the movement of organisms between the undisturbed and disturbed areas (Wiens et al. 1985, Wiens 1992). It is very likely that edges, by influencing the


82





83


movement of animals, might indirectly affect ecological processes mediated through plant-animal interactions such as pollination and seed dispersal. In tropical systems many organisms are involved in plant-animal interactions. For the most part, work done in the tropics has focused on how edges affect animal distributions (Quintela 1986, Laurance 1990, Malcolm 1994). Less emphasis has been given to how edges influence the distribution of animals mediating ecological processes.

The extent to which edges can affect the distribution of organisms varies with edge age (e.g., Williams-Linera 1990), and land use (e.g., DeGraaf 1992) may determine the degree to which edges can affect organisms. Equally important is the variation among organisms in their response to edges (e.g., Kroodsma 1984, Noss 1991). Such variation can be used to tease apart the mechanisms underlying such responses. Possible mechanisms include changes in the resource base (Malcolm 1991), parasites and predators (e.g., Gates and Gysel 1978, Brittingham and Temple 1983, Loye and Carroll 1995), physiological condition of organisms (Wiens et al. 1985), dispersal, and home range size (Kuitunen and M&kim 1993).

Here I report on how edges influence the distribution of understory birds in a neotropical montane forest. I looked at the effects of distance from the edge towards the forest interior and time since edge creation on birds classified by feeding guilds. I concentrated on frugivores and





84


nectarivores, since a high proportion of understory plants in neotropical cloud forests rely on these two groups of organisms for seed dispersal and pollination (Terborgh 1977, Gentry 1983, Stiles 1985). Thus, changes in their distribution may help explain how edges influence seed dispersal and pollination in highly fragmented habitats.


Methods



Understory Birds


I mist netted birds at six edges, three old and three new, each netting site encompassing an area 100 x 200 m (2 ha) (Chapter 2). Strips at four distances from the forest edge towards the forest interior (D1: 0-10, D2: 30-40, D3: 60-70, and D4: 190-200 m) were divided into 5 plots (20 x 10 m) (Chapter 2) (Fig. 2-3). Three of these plots were chosen at random and one pair of mist nets was placed in each, with one net set perpendicular to the other. Nets were 9 X 2.5 m with a 32 mesh. In each strip, nets were separated by a mean distance of 40 m and positioned 0.5 m above the ground.

I operated 12 pairs of mist nets simultaneously from

0530-1300 for two consecutive days per month per distance per edge, trying to complete when possible 14 hours of mist netting per pair of mist nets. Mist netting started in June 1992 at the old edges. In September 1992 I included the new edges. Thus the six edges were sampled simultaneously from September 1992-August 1993, excluding December 1992 when I





85


did not sample birds. Because the sampling unit was a pair of mist nets, instead of the traditional single net, I define mist net hours as the hours that a pair of nets was opened. In total the mist netting effort was equivalent to 11,892 net-hours. Mist nets were checked every 1-1.5 hours, and for each captured bird I recorded species, mist net number, molt, presence of cloacal protuberance or brood patch, fat, mass, culmen length (total and exposed), culmen height and width, tarsus and tail length, and wing chord. All birds, except hummingbirds, were individually marked with color bands. Hummingbirds were marked temporarily by clipping the tip of their tail and wing feathers to recognize recaptures within a mist netting session. Bird abundance is expressed throughout this paper as capture rates, i.e., number of captures per pair of mist nets per 100 mist net hours (mnh). Recaptures on the same day were excluded from the analyses.

Birds were classified into four feeding guilds:

frugivores, insectivores, nectarivores, and carnivores. Frugivores were defined as species that commonly consumed fruit and/or seeds; most of them also consumed insects to some degree. Insectivores ate primarily insects. Nectarivores relied heavily on nectar and included nectar "thieves." Carnivores primarily preyed on vertebrates. The placement of any given species in one of these categories was based on the analysis of fecal samples, my own observations, and published reports (Miller 1963, Stiles and Skutch 1989, Andrade et al. 1993, Arango 1993).





86


The use of mist nets versus acoustic and/or visual censuses to carry out studies on bird assemblages in the tropics has been widely discussed because of the biases inherent in any sampling method (e.g., Terborgh and Weske 1969, Terborgh 1971, Karr 1971, Remsen and Parker 1983, Karr 1981Ak, Lynch 1989, Remsen 1994). It is accepted that mist nets only sample a proportion of bird species found in an area (Terborgh and Weske 1969, Karr 1981k); that if used over a prolonged period birds learn the position of nets (Terborgh 1977, Bierregaard 1990); and that figures on bird abundance might overestimate the abundance of many species (Karr 1981k, Remsen and Parker 1983, Lynch 1989). In addition, when used to compare habitats that differ markedly in structure, capture rates can be misleading in regard to the presence and abundance of many species (Terborgh 1971, Lynch 1989, Blake et al. 1990). My study, then, only reflects what happens to those birds that are effectively sampled by mist nets in the understory (see also Wong 1986, Levey 1988Ak, Blake and Loiselle 1991, Loiselle and Blake 1991, Poulin et al. 1992). I stress that mist netting took place only inside the forest, and the aim of this study was to compare changes in bird abundance from the edge towards the forest interior. Thus, problems associated with habitat biases are either minimized or held constant.

One possible problem for interpreting the results,

however, relates to the timing of mist netting in old and new edges. In old edges mist netting began in June 1992 and in





87


new edges in September 1992. Higher capture rates in new than old edges in September and October could be attributed to birds having learned the position of nets in the old edges. If this increase in capture rates in new edges was a consequence of a learning process then I would expect (1) a decrease in the proportion of recaptures over time for old and new edges, and (2) a higher proportion of recaptures in new than old edges during these months. The data for all species (excluding hummingbirds), frugivores, and insectivores, do not support these predictions. I conclude that any observable difference can therefore be attributed to differences between the two types of edges. Data Analysis


To establish changes in the abundance of the understory avifauna of La Planada as a function of distance from the edge and edge age I analyzed capture rates for all bird species combined and for three feeding guilds. Carnivores were excluded from the analyses because of small sample size. I used ANOVAs for Mixed Factorial Designs (Girden 1992). Edge age (old and new) was included as a between-subject factor. Month (September 1992-August 1993) and distance (D1D4) were included as within-subject factors or repeated measures. Month and distance were included as within-factors because of restrictions in the randomization procedure when "assigning" month and distance levels to each edge, which can lead to correlations between the observations (Girden 1992,





88


Manly 1992). Edges were treated as subjects because each edge was measured repeatedly for each of the different treatment combinations. Because mist nets are nested within distance and capture rates were zero for many pairs of mist nets at a given month/distance/edge, I averaged capture rates for each three pairs of mist nets/month/distance/edge. This procedure reduced the dimensionality of the data and also made the data more normally distributed by eliminating many zero values. I plotted the residuals as a function of fitted Y values to detect any violation of assumptions (Manly 1992). The data for all four ANOVAs presented in this paper were square-root-transformed. In addition, I verified the assumption of compound symmetry (i.e., the covariation between each pair of treatments is equal for all subjects) for ANOVAs that included within-factors (Girden 1992, Manly 1992). When compound symmetry is violated, the probability of committing a Type I error increases. To account for this, the degrees of freedom have to be corrected by a factor, epsilon, which ranges from 1/(J-1) to 1.0, where J is the number of levels in a treatment. The closer epsilon is to 1.0, the lower is the probability that compound symmetry is being violated (Girden 1992). Epsilon is estimated based on the conservative Geisser-Greenhouse method and the more liberal Huynh-Feldt method (Girden 1992). In this paper I report the corrected F values based on the liberal HuynhFeldt method (H-F).

In addition to the above omnibus ANOVA tests, I





89


specified contrasts of mean differences to test specific hypotheses. These hypotheses included effects involving single factors and interactions (Gagnon et al. 1989, Girden 1992). For the distance effect I specified two contrasts, by comparing mean capture rates at Dl and D4 separately with those of D2 and D3 together. I assumed that changes in bird distribution, if any, would be more marked at the extremes. For the distance x age interaction I specified a single contrast, by comparing mean capture rates at D4 between old and new edges. For the month x age and the month x distance interactions I specified two contrasts for each, comparing dry with wet months. I reasoned that because of marked differences in the rainfall regime at La Planada, changes in bird abundance between habitats (distance or type of edge) were more likely to occur between dry and wet months. Dry months were those exhibiting the lowest rainfall records (February and July) and the previous month when rainfall started to decrease (January and June). Wet months were those that received the highest rainfall (April and October) and the previous month when rainfall started to increase (March and September) (Fig. 2-2). For the interaction between month and age I compared the mean number of captures between old and new edges during the dry and wet months. For the interaction month x distance I compared the mean number of captures between D1 and D4 during the dry and wet months. I report the corrected F-values and associated probability in the results section. All analyses were performed using




Full Text
87
new edges in September 1992. Higher capture rates in new
than old edges in September and October could be attributed
to birds having learned the position of nets in the old
edges. If this increase in capture rates in new edges was a
consequence of a learning process then I would expect (1) a
decrease in the proportion of recaptures over time for old
and new edges, and (2) a higher proportion of recaptures in
new than old edges during these months. The data for all
species (excluding hummingbirds), frugivores, and
insectivores, do not support these predictions. I conclude
that any observable difference can therefore be attributed to
differences between the two types of edges.
Data Analysis
To establish changes in the abundance of the understory
avifauna of La Planada as a function of distance from the
edge and edge age I analyzed capture rates for all bird
species combined and for three feeding guilds. Carnivores
were excluded from the analyses because of small sample size.
I used ANOVAs for Mixed Factorial Designs (Girden 1992).
Edge age (old and new) was included as a between-subject
factor. Month (September 1992-August 1993) and distance (Dl-
D4) were included as within-subject factors or repeated
measures. Month and distance were included as within-factors
because of restrictions in the randomization procedure when
"assigning" month and distance levels to each edge, which can
lead to correlations between the observations (Girden 1992,


Loranthaceae
Alloolectus schultzei Mansf.
Moraceae
Struthanthus aecruatoris Kuiit
Myrsinaceae
Ficus cf. aDollinaris
Ficus aarcia-barriaae
Cvbianthus sorucei (Hook.f.)Aaos
Rubiaceae
Loranthaceae
Haffinania sp.
Lauraceae
Aetanthus sd.
Unknown
F ES
LD-F ES
G ES
G ES
F ES
LD ES
F ES
F ES
BD
160


Solanaceae
Unknown
Unknown
Old Edges
Actinidaceae
Amaralidaceae
Clusiaceae
Ericaceae
Euphorbiaceae
Gesneriaceae
CR 749, CR 637, CR 544
Psvcothria dukei Dwyer
Solanum sp.3
CR 754
Solanum sp.6
CR 536, CR 658, CR 484
Unknown
CR 640, CR 661
CR 665
CR 769
Saurauia parviflora Tr. & Pi.
Bomarea pardina Herbert
Clusia venusta Little
Macleania bullata
Sphvrospermum cordifoliium Bentham
Hveronvma sp.
CR 706
F S
ES
F VS
LD ES
F ES
G ES
LD ES BD
LD ES
F ES
LD S BD
F ES BD
LD-F ES BD
159


52
Heterogeneity) to test for homogeneity between the two edge
ages. For these two analyses I used only species in which at
least 80 percent of the expected cell frequencies were
greater than 5, since the G statistic departs from the X?
distribution if this is violated (Siegel and Castellan 1988).
Resultg
LAI and Distance from the Edge
I measured LAI only at old edges and found that major
changes in LAI were observed at the interface between pasture
and forest (Fig. 4-1). Once inside the forest, LAI values
were highly variable not only from the edge towards forest
interior but also among edges at the same distance (Fig. 4-
1). Averaging over the three edges, variability in LAI
measurements for the 0-40 m interval (Coefficient of
variation, CV = 0.3) was identical to that for the 50-210 m
interval (CV = 0.3).
I did not include pasture LAI values in the ANOVA to
establish the effect of distance on LAI. LAI did not differ
significantly either among edges (ANOVA, F3,5 = 0.4, P = 0.75)
or in relation to distance from the forest edge towards the
forest interior (F23, 115 = 0.47, P = 0.98) (Table 4-1). This
indicates that at least along old edges, canopy structure
does not vary in a predictable way from the edge towards the
forest interior.


81
Bird species known to feed on palm fruits at La Planada,
including Mvadestes ralloides, Lipauous ttyptolQphys,
Semnornis ramohastinus, Piprgpla rieffgri, gnt^mpdeste?
coracinnus (C. Restrepo personal observation, Restrepo 1990)
and Andioena laminirostris (Beltran 1991) did not account for
the high capture rates observed at D1. Thus it is unclear
whether changes in TF and RF across the pasture-forest edge
influenced the behavior of frugivorous birds. The contrary-
might be true, however, for TF-A and TI. Those species known
to feed on fruits other than palms, which were the majority
of frugivorous species captured with mist nets, made an
important contribution to the high capture rates observed at
D1. Moreover, as occurred with TF-A and TI, there was a
significant interaction between distance from edge and month
on capture rates of frugivores (Chapter 5).
This study showed that edges influenced fruit abundance
in different ways. First, fruit production by the assemblage
of understory plants changed abruptly from forest edge
towards forest interior but depended on edge age, the
presence of treefall gaps, the length of the observations,
and whether or not palms were taken into accout. Second,
fruit production on a species by species level, and expressed
as the total number of fruiting individuals, changed across
the pasture-forest edge. This was shown both among abundant
and sparse species. The latter provided evidence to support
the contention that edges represent zones of opportunities
for the establishment of a wide range of species.


AC
70
CA
I
1 0
i i r
J- I i i i i i i i i it
100 316
63
31.6
Body Mass (g)
794
137


50
edges that I sampled for each level of edge age (old and new)
were chosen at random from a population of old and new edges
and represented the plot unit. In turn, each edge was
divided into four strips, i.e., distances from the forest
edge towards the forest interior, representing the subplot
units. Randomization of the levels of the distance factor
was restricted, but because the strips were separated in
space and I analyzed responses from non-mobile organisms I
assumed they represented independent subsampling units. This
was supported by results of an ANOVA in which distance was
included as a repeated measure and the epsilon factor equaled
one, indicating no correlation between the levels of the
distance factor (Girden 1992). Finally, fruit production was
monitored in subquadrats that were chosen at random and
classified according to habitat as gap or interior, the
latter representing the sub-subplots.
The number of subquadrats falling within the gap and
interior categories differed at each distance among the
edges, producing an unbalanced design. I used Type III SS
since it takes into account differences in cell frequencies
between treatment combinations (Gagnon et al. 1989; Potvin
1993) .
To determine whether the data satisfied assumptions of
an ANOVA, I plotted residuals as a function of fitted Y
values. When residuals where not normally distributed, I
log-transformed the data. In addition, I verified the
assumption of compound symmetry for the repeated measure


4-6 Variation in fruit abundance at the Reserva
Natural La Planada in relation to edge age and month ... 65
4-7 Variation in fruit abundance at the Reserva
Natural La Planada in relation to distance from the
forest edge and month 66
5-1 Variation in the distribution of understory birds
at La Planada in relation to distance from the forest
edge 93
5-2 Variation in the distribution of understory birds
at La Planada in relation to distance from the forest
edge and edge age 95
5-3 Variation in the distribution of understory birds
at La Planada in relation to edge age and month of the
year 96
5-4 Variation in the distribution of understory birds
at La Planada in relation to distance from forest edge
and month of the year 98
6-1 Montane habitats of Colombia and sites included
in lump analysis of body mass of frugivorous birds 116
6-2 Lump analysis for body mass of frugivorous birds
of Colombian upper lowland tropical forests showing
(a) body mass distribution vs. rank order and (b) rank
size-ordered body mass distribution versus gap rarity
indexes 122
6-3 Lump structure of Colombian montane frugivorous
birds according to elevational zone 127
6-4 Lump structure of Colombian frugivorous birds
from sites cover mostly by forest to sites highly
transformed by human activities within the upper
lowland zone 13 0
6-5 Lump structure of Colombian frugivorous birds
from sites cover mostly by forest to sites highly
transformed by human activities within the lower
montane zone 1 133
6-6 Lump structure of Colombian montane frugivorous
birds from sites cover mostly by forest to sites
highly transformed by human activities within the
upper montane zone 136
6-7 Relationship between species richness and lump
structure in landscapes of variable complexity 145
IX


FRUIT ABUNDANCE
(counts / 50 m 2)
60
D1 D2 D3 D4
DISTANCE FROM FOREST EDGE (m)
Figure 4-3. Variation in fruit abundance at the Reserva
Natural La Planada in relation to distance from forest
edge. Points represent means and bars standard errors.


100
I I I
63
141
n
631
1585
3981


40
reported by Murcia (1993) for a Colombian site north of La
Planada. She found that pollination levels in 11 out of 13
species were not affected by distance from forest edge and
explained these results in terms of hummingbirds not being
influenced by edges. At La Planada I found that this was the
case (Chapter 5). Mean capture rates of nectarivorous birds
were not influenced by distance from forest edge. It seems
then that pollination can not account for differences in the
distribution of fruiting individuals of P. qjbbosa and F.
affinis across pasture-forest edges.
Seed predation increased from edge towards forest
interior in £. affinis but not in £. qibbosa. The effect of
distance from edge on £. affinis, however, was influenced by
time. Not only were more seeds removed at D4 but they were
removed faster than at the other distances. In both species,
seed germination was affected by distance from forest edge
but depended on time. Seeds closer to the edge germinated
sooner than those in forest interior. Assuming an equal
probability of seeds arriving at any of the four distances,
it is likely that seed predation and seed germination may
limit recruitment rates in these two species across pasture-
forest edge. Differences in seed predation (F. affinis and
£. qibbosa) and in seed germination (F. affinis) over time
may result in fewer individuals of these two species
establishing in the forest interior (D4).
In another study of £. affinis conducted at La Planada
Samper (1992) found that (1) seed removal rates were not


14
Figure 2-3. Edge indicating general sampling design. Shaded
strips represent four distances where sampling took place: D1
(0-10 m), D2 (30-40 m), D3 (60-70 m), and D4 (190-200 m). In
strip DI I show the distribution of subquadrats (1-4) where
fruit abundance was evaluated. In strip D2 I show the
distribution of each of three pirs of mistnets (perpendicular
dark lines). On the left side of the figure I illustrate the
orientation of one transect along which LAI was measured.


LIST OF TABLES
Table page
2-1 Water balance for the Reserva Natural La Planada
9
2-2 Characteristics of the edges included in this
study 13
3-1 Results of Mixed Factorial ANOVAs for pollen tube
production in Palicourea gibbosa and Faramea affinis
in relation to distance from forest edge 27
3-2 Pollen tube production in Palicourea gibbosa and
Faramea affinis in relation to distance from forest
edge 28
3-3 Results of Mixed Factorial ANOVAs for fruit set
and fruit damage by insects in Palicourea gibbosa and
Faramea affinis in relation to distance from forest
edge 30
3-4 Results of Mixed Factorial ANOVAs for seed
predation and seed germination in Palicourea gibbosa
and Faramea affinis in relation to distance from
forest edge 35
3-5 Results of Mixed Factorial ANOVAs for relative
growth and leaf production rates in Palicourea gibbosa
and Faramea affinis in relation to distance from
forest edge 38
4-1 Results of Repeated Measures ANOVA on leaf area
index (LAI) for old edges 55
vi


30
6-3 15.8 39.8
100 251
631 1585
Body Mass (g)
134


97
respectively). Capture rates per 100 mist-net hours for
insectivores were significantly higher at D4 (mean = 8.0)
compared to D2 and D3 combined together (mean = 5.0)
(Contrast of Mean Differences, Fifi2 = 8.3, P = 0.01) but did
not differ between D1 and D2-D3 combined. The omnibus test
showed that distance from the forest edge did not affect
nectarivore capture rates (Table 5-1, Fig. 5-1).
Nevertheless, the specific hypothesis tested by contrast of
mean differences showed that nectarivores were more abundant
at D1 (mean = 5.4) than at D2 and D3 combined (mean = 3.5)
(Contrast of Mean Differences, Fifi2 = 4.9, P = 0.05).
Among frugivores, the distance effect was modified by
edge age, as shown by the significant interaction between
these two factors (ANOVA, F3fi2 = 2.7, P = 0.09; Table 5-1,
Fig. 5-2). Capture rates at DI, D2, and D3 were similar
between old and new edges. However, capture rates at D4 were
significantly higher at new (mean = 9.4) than at old (mean =
4.5) edges (Contrasts of Mean Differences, Fifi2 =7.2, P =
0.02). The interaction between distance from the edge and
edge age in the omnibus test was not significant for
insectivores and nectarivores (Table 5-1, Fig. 5-2).
Nevertheless, when testing specific hypotheses I found that
capture rates for nectarivores at D4 were significantly
higher at new (mean = 5.5) than at old (mean = 3.3) edges
(Contrast of Mean Differences, = 4.0, P = 0.07).


FRUn ABUNDANCE FRUT ABUNDANCE
(mean counts / 50 m2) (mean counts / 50 m2)
5000
4000
3000
2000
1000
0
1000
1 2 3 4 5 6
LAI
Total Fruits
*
& Eif a a
CM
E
o
in
c
3
o
o
c
CD
E
800
600
400
200
0
1 2 3 4 5 6
Ripe Fruits
A.
- -to# tEMfloClA
LAI
1000
750 -
500
250
Total Fruits
(excluding palms)
12.5 -
c\T ^
LU E 10 -
y o
6 in
Total Individuals %
A

d 7.5 -
Z
o
A T?


>
o 5 -
t
tJ c
. V A
o A
A D
C3
a*..,
£ 25-
E.
o J
A J3 A
__ A A
CQ
1 1 1 1
3 4
LAI
LAI
Figure 4-2. Leaf area index (LAI) and mean counts expressing fruit abundance.
Symbols represent old edges. Open squares = Pialapi, open tirangles = Climo
II, and filled circles = Climo I.
~u


89
specified contrasts of mean differences to test specific
hypotheses. These hypotheses included effects involving
single factors and interactions (Gagnon et al. 1989, Girden
1992). For the distance effect I specified two contrasts, by
comparing mean capture rates at D1 and D4 separately with
those of D2 and D3 together. I assumed that changes in bird
distribution, if any, would be more marked at the extremes.
For the distance x age interaction I specified a single
contrast, by comparing mean capture rates at D4 between old
and new edges. For the month x age and the month x distance
interactions I specified two contrasts for each, comparing
dry with wet months. I reasoned that because of marked
differences in the rainfall regime at La Planada, changes in
bird abundance between habitats (distance or type of edge)
were more likely to occur between dry and wet months. Dry
months were those exhibiting the lowest rainfall records
(February and July) and the previous month when rainfall
started to decrease (January and June). Wet months were
those that received the highest rainfall (April and October)
and the previous month when rainfall started to increase
(March and September) (Fig. 2-2). For the interaction
between month and age I compared the mean number of captures
between old and new edges during the dry and wet months. For
the interaction month x distance I compared the mean number
of captures between D1 and D4 during the dry and wet months.
I report the corrected F-values and associated probability in
the results section. All analyses were performed using


167
Andrade, editor. Carpanta Selva Nublada y Pramo.
Fundacin Natura, Santa F de Bogot, Colombia.
Arango, S. 1994. El papel de las aves dispersoras de semillas
en la regeneracin de pastizales en el Alto Quindo,
Andes Centrales, Colombia. Fundacin Herencia Verde,
Cali, Colombia, unpublished report.
Arias, J. C. 1993. Polinizacin y biologa reproductiva de
Palicourea aibbosa Dwyer (Rubiaceae) en bosques de
estados sucesionales diferentes. Tesis Biologa,
Universidad de Antioquia, Colombia.
Auchter, E. C., A. Lee Schrader, F. S. Lagasse, and W. W.
Aldrich. 1926. The effect of shade on the growth, fruit
bud formation and chemical composition of apple trees.
Proceedings of the American Society for Horticultural
Science 23:368-382.
Auclair, A. N. and G. Cottam. 1971. Dynamics of black
cherry (Prunus sertina Ehrh.) in southern Wisconsin oak
forests. Ecological Monographs 41:153-177.
Banguero, H. 1993. La poblacin de Colombia 1938-2025. Una
visin retrospectiva y prospectiva para el pas, los
departamentos y sus municipios. Coleccin de Edicin
Previa, Editorial Universidad del Valle, Cali, Colombia.
Beltrn, W. 1991. Historia natural de Andiaena laminirostris
(Aves: Ramphastidae). Informe Final, Fundacin para la
Promocin de la Investigacin y la Tecnologa, Banco de
la Repblica, Bogot, Colombia.
Bierregaard, R. O. 1990. Avian communities in the understory
of Amazonia forest fragments. Pages 333-343, in A.
Keast, editor. Biogeography and Ecology of Forest Bird
Communities. SPB Academic Publishing bv, Hague, The
Netherlands.
Bierregaard, R. 0., Jr. and T. E. Lovejoy. 1989. Effects of
forest fragmentation on amazonian understory bid
communities. Acta Amaznica 19:215-241.
Blake, J. G. and W. G. Hoppes. 1986. Influence of resource
abundance on use of tree-fall gaps by birds in an
isolated woodlot. Auk 103:328-340.
Blake, J. G. and B. A. Loiselle. 1991. Variation in resource
abundance affects capture rates of birds in three
lowland habitats in Costa Rica. Auk 108:114-130.
Blake, J. G., B. A. Loiselle, T. C. Moermond, D. J. Levey,
and J.S. Denslow. 1990. Quantifying abundance of


24
run of the experiment. I ran this experiment from June 26
1993 to July 3 1993.
Analyses
I analyzed data with ANOVAs for Mixed Factorial designs
(Girden 1992). The full design (edge age, distance from
forest edge, and treefall gaps) was set up as a split-split-
plot design (Winer et al. 1991). The factors of interest
were edge age, distance from the edge, and habitat. The
edges that I sampled for each level of edge age (old and new)
were chosen at random and represented the plot unit. In
turn, each edge was divided into four strips (distances from
the forest edge towards the forest interior) representing the
subplot units. Randomization of the levels of the distance
factor was restricted but because the strips were separated
in space and I analyzed responses from nonmobile organisms I
assumed they represented independent subsampling units.
Finally, individuals were classified according to habitat as
gap or intact forest, the latter representing the sub
subplots .
The design for the seed predation, seed germination, and
fruit removal experiments was set up as a split-plot design
with one repeated measure (Winer et al. 1991). The factors
of interest were distance from the edge, habitat, and time,
with time being the repeated measure. In the seed predation-
seed germination and fruit removal experiments time was
represented by weeks and days, respectively. In my design


CHAPTER 6
FRUGIVOROUS BIRDS IN FRAGMENTED NEOTROPICAL MONTANE FORESTS
LUMP STRUCTURE IN BODY MASS
Introduction
Transformation of tropical landscapes by humans has
influenced plant and animal assemblages in many ways. Most
studies have emphasized how species abundance and richness
change with increasing forest fragmentation (Quintela 1986,
Bierregaard and Lovejoy 1989, Klein 1989, Newmark 1991,
Estrada et al. 1993, Kattan et al. 1994, Malcolm 1994,
Didham, in press, Lynam, in press, Warburton, in press,
Chapter 5) and with transformation of native forests into
second growth and managed ecosystems (Holloway et al. 1992,
Johns 1992, Lambert 1992, Thiollay 1992, Andrade and Rubio
1994, Escobar 1994). Results of these studies vary
considerably, reflecting the complexity of relating habitat
modification to biodiversity loss, but also inherent
differences among study sites, and/or a mismatch between the
scale of the problems being addressed and the methods used.
As a consequence it has been difficult to establish patterns
regarding how habitat modification and biodiversity interact
and moreover, how they relate to ecosystem processes
112


CHAPTER 3
FROM FLOWERS TO SEEDLINGS: THE EFFECT OF EDGES AND TREEFALL
GAPS ON TWO TROPICAL UNDERSTORY PLANTS, Palicourea aibbosa
AND Faramea affinis (RUBIACEAE)
Int reaction
Recruitment rates in plant populations are influenced by
success at all stages of the life cycle (Harper 1994). Which
stages limit recruitment depends on the requirements of
individuals at each stage and the spatial distribution of
resources (e.g., Sork 1983, Martinez-Ramos and Soto-Castro
1993, Osunkoya et al. 1994). In tropical areas, for example,
treefall gaps influence the distribution of resources at
small scales (Denslow and Hartshorn 1994), whereas landslides
and forest clearings do so at large scales (Guariguata 1990,
Dalling and Tanner 1995). Also, seed predation (e.g., Schupp
1988, Schupp and Frost 1989, Samper 1992), seedling
establishment (e.g., King 1990), plant growth (Sizer 1992,
Dalling and Tanner 1995), fruit production (Levey 1990) and
seed dispersal (Murray 1988) change as a result of such
disturbances. Little is known, however, about the combined
effect of small and large-scale disturbances on the various
stages of the life cycle of plants or how such effects may
determine which stage limits recruitment.
15


132
(alpha = 0.1) (Fig. 6-5).
I grouped the lower montane sites according to major
types of land use to describe patterns in lump structure.
Three sites are covered extensively by forest in which second
growth (LP), second growth and forestry plantations (UB), and
second growth, pastures, and weekend cottages (SA) cover less
than 50% of land (Table 6-1). The lump structure of LP and
UB, two sites differing in the number of species, is similar
almost over the entire range of body mass. The lump
structure of SA showed similarities with UB in the body mass
range of 63-313 g, but also differences below 63 g. This
occurred, even though both sites had a similar numbers of
species (Table 6-1).
Three sites are covered half by native forests and half
by pastures (Merenberg, ME), second growth and selectively
logged forests (Miraflores, MI), and tree plantations
established for watershed restoration (Rio Blanco, RB). The
lump structure of RB and ME showed similarities below 39 g
and above 301 g and they differed greatly from that of MI
almost over the entire range of body mass. In the last three
sites the native forest has been replaced almost entirely by
coffee plantations and orchards (Rio Grande, RG) and tree
plantations of exotic species for wood production (Munchique,
MU and Piedras Blancas, PB). RG and MU showed a different
lump structure even though they have the same number of
species. The lump stucture of MU and PB was very similar,
but the lump representing birds >398 g in PB was much smaller


LIST OF FIGURES
Fiq.urg oms.
2-1 Location of study area and edges included in the
study 6
2-2 Distribution of mean monthly rainfall and
temperature at the Reserva Natural La Planada 8
2-3 Edge indicating general sampling design 14
3-1 Fruit set in Palicourea aibbosa as influenced by
distance from forest edge 31
3-2 Seed germination, seed predation, seedling growth
and, leaf production in Palicourea gibbosa and Faramea
affinis as influenced by distance from forest edge 36
3-3 Seed germination, seed predation, seedling growth
and, leaf production in Palicourea aibbosa and Faramea
af finis as influenced by habitat 39
4-1 Leaf area index (LAI) across the pasture-forest
edge at old edges 54
4-2 Leaf area index (LAI) and fruit abundance 57
4-3 Variation in fruit abundance at the Reserva
Natural La Planada in relation to distance from forest
edge 60
4-4 Variation in fruit abundance at the Reserva
Natural La Planada in relation to edge age and
distance from forest edge 62
4-5 Variation in fruit abundance at the Reserva
Natural La Planada in relation to habitat and distance
from forest edge 63
viii


149
The different responses shown by the four variables
describing fruit abundance in the understory and by the
different plant species suggest differences in edge
"penetrability" to the factors regulating fruit production
and seed movement. These differences were manifested across
the four distances within the edges but also among new and
old edges. Steep and permanent gradients from edge towards
forest interior may suggest low penetrability. Conversely,
shallow and more variable gradients may suggest high
penetrability.
My results showed that fruit abundance is influenced by
edges but that the exclusion of certain groups of plants
produces different results. This provided insight regarding
the scales at which edges influence fruit abundance. Future
studies aimed at understanding the role of edges in
landscapes should take the "scale" issue into consideration.
One important finding is that related to the
distribution of sparse species of plants and birds in the
immediate vicinity of pasture-forest edges. These species
may persist and take advantage of changes taking place at
either side of the edge depending on habitat preferences. As
more forest is felled, species characteristic of large
disturbed areas may establish in the recently disturbed area.
On the other hand, as pastures and fields are left abandoned
species characteristic of forest may establish there. Thus,
edges might function as "stepping stones" to recolonization
at both sides and as elements that connect, rather than


122
Figure 6-2. Lump Analysis of body mass for Colombian upper
lowland tropical frugivorous birds (a) body mass distribution
vs. rank order and (b) rank size-ordered body mass
distribution vs. gap rarity indexes (GRI-values). In(b)
potential gaps between the lumps are represented by GRI-
values that exceed the criterion lines (alpha values).


53
LAI and Fruit Abundance
To establish whether LAI influences fruit abundance, I
averaged values of fruit abundance for each pair of 50 m2
subquadrats where I took LAI measurements. LAI was not
significantly correlated with total number of fruits (TF)
(Coefficient of determination, r2 = 0.032, n =36), total
number of fruits excluding the Arecaceae (TF-A) (r2 = 0.077,
n = 36), total number of ripe fruits (RF) (r2 = 0.015, n =
36), or total number of fruiting individuals (TI) (r2 = 0.02,
n = 36) (Fig. 4-2).
Fruit Abundance
Plant Assemblages
In the ANOVAs none of the three-way interactions was
significant. Several two-way interactions were significant
but not consistently so for the four measurements of fruit
abundance. In describing the results I look first at the
single effect of distance on fruit abundance and then at the
interactions involving this term.
The total number of fruits (TF) and total number of ripe
fruits (RF) differed significantly among the four distances
on a yearly basis (ANOVA, ^2,12 4.3, P = 0.03 and F3fi2 =
5.4, P = 0.01, respectively) (Table 4-2). For TF and RF the
mean number of fruits as well as the variance decreased from
forest edge towards forest interior (Fig. 4-3). There are
three not mutually exclusive explanations for these results.


I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor~of Philosophy.
Douglas*'#. Levey J Chair
Associate Profesor of
Zoology
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
H^/Jane Brockmann
Professor of Zoology
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor jaf Philosophy.
C.S. Hollirrg
Arthur R. Marshall, Jr.,
Professor of Ecological
Sciences
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
)hn Ewel
¡sor of Botany
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Frrk Slansky
Professor of Ento:
Nematology
ogy and


123
1.8 2.3 2.8 3.3
0.8
1.3
Log Body Mass
3.8


APPENDIX B
BIRDS CAPTURED IN THE UNDERSTORY
OF THE RESERVA NATURAL LA PLANADA


115
(Irving 1975). I defined montane habitats as those above the
800 m topographic contour line (Fig. 6-1) In the absence of
human disturbance this region would be covered by forest
except for the pramo and small areas affected by rain
shadows (Cuatrecasas 1958). Forest composition, structure,
and physiognomy change along the mountains from the complex
lowland tropical forest to the simpler pramos (Cuatrecasas
1958, Espinal et al. 1977).
The area encompassed by this study represents less than
35% of the total area of Colombia (1,380,000 km2) yet harbors
one of the richest biotas not only of this country but of the
neotropics (Duellman 1979, Henderson et al. 1991, Gentry
1992i,b, Renjifo et al. in press). It has been postulated
that the elevated levels of diversity and endemism of this
area are the result of an intense disturbance regime (Gentry
1992a, J. Luteyn pers. comm.). A complex topography and
geology, combined with high precipitation, generates
landslides, mud flows, avalanches, and volcanic eruptions,
which continuously transform these mountains (Mejia et al.
1994, Velsquez et al. 1994).
Superimposed on the natural disturbance regime is one
generated by human activities. At least 50 per cent of the
total population of Colombia (37 million people) has settled
in montane areas (Banguero 1993) Presently, less than 30
percent of this area is covered by forest, most of which is
found either at elevations >2,500 m or on wetter slopes of
the cordilleras (Cavelier and Etter, in press). The remaining


Table 3-3. Results of ANOVAs for Mixed Factorial Designs (Split-split-plot) on percentage
of fruit set (UF/FB), percentage of fruits ripening (RF/UF), percentage of seeds damaged by
wasps (UFW/FV), and percentage of fruits damaged by ants (UFA/FV) in Palicourea aibbosa.
Log-transformed data. Significance at 10% (*),5% (**) and 1% (***).
Variable
Age
A
df F
Edge(A)
Error
df SS
Distance
D
df F
D x A
df F
DxE(A)
Error
df SS
UF/FB
1
0.0
4
0.1
3
2.8
3
5.8**
12
0.001
RF/UF
1
1.0
4
0.5
3
0.0
3
3.9**
12
0.9
UFW/FV
1
o
o
4
0.001
3
1.4
3
0.5
12
0.002
UFA/FV
1
0.0
4
0.2
3
0.5
3
1.6
12
0.6
OJ
o
Habitat
H x D
H x A
H x D x A
Hx[E(A)xD]
H
Error
df F
df F
df F
df F
df SS
UF/FB
1
4.7*
3
0.9
1
1.8
3
1.1
16
0.05
RF/RF
1
1.7
3
0.4
1
0.1
3
0.4
16
1.7
UFW/FV
1
0.001
3
0.7
1
0.3
3
0.3
16
0.003
UFA/FV
1
1.2
3
1.4
1
3.9*
3
1.7
16
4.2


84
nectarivores, since a high proportion of understory plants in
neotropical cloud forests rely on these two groups of
organisms for seed dispersal and pollination (Terborgh 1977,
Gentry 1983, Stiles 1985). Thus, changes in their
distribution may help explain how edges influence seed
dispersal and pollination in highly fragmented habitats.
Methods
Understorv Birds
I mist netted birds at six edges, three old and three
new, each netting site encompassing an area 100 x 200 m (2
ha) (Chapter 2). Strips at four distances from the forest
edge towards the forest interior (Dl: 0-10, D2: 30-40, D3:
60-70, and D4: 190-200 m) were divided into 5 plots (20 x 10
m) (Chapter 2) (Fig. 2-3). Three of these plots were chosen
at random and one pair of mist nets was placed in each, with
one net set perpendicular to the other. Nets were 9 X 2.5 m
with a 32 mesh. In each strip, nets were separated by a mean
distance of 40 m and positioned 0.5 m above the ground.
I operated 12 pairs of mist nets simultaneously from
0530-1300 for two consecutive days per month per distance per
edge, trying to complete when possible 14 hours of mist
netting per pair of mist nets. Mist netting started in June
1992 at the old edges. In September 1992 I included the new
edges. Thus the six edges were sampled simultaneously from
September 1992-August 1993, excluding December 1992 when I


12
Hermgenes), a fifth edge was located within the reserve
(Pialapi), and the sixth edge was located at El Bosque
(Acantayac) (Fig. 2-2). Climo I and Climo II were 400 m
apart on the same edge, but because of differences in the
weeding regime of the pasture and use of the forest I
reasoned that they could represent two independent sampling
units. Independence of these two sampling points was
particularly important for the part of the work evaluating
the influence of edges on the distribution of understory
birds (Chapter 5). Recapture frequency between these two
sites was <4%, supporting the assumption that these two
points represented two independent sampling units.
Three edges, Climo I, Climo II, and Pialapi, were
created around 1950 (old edges), when colonists first arrived
in the area and cleared the forest to establish pastures.
The other three edges, Marcos, Hermgenges, and Acantayac,
were created around 1982 (young edges), the year La Planada
was established as a private reserve (Table 2-2). At the
beginning of the study, I placed barbed wire fences along the
edges to keep cattle from penetrating into the forest. I
sampled these edges between March 1992 and March 1994.
At each edge I worked in an area of 100 X 200 m (2 ha)
and established four strips (100 X 10 m) running parallel to
the edge. These strips were located at four different
distances from the forest edge towards the forest interior:
0-10 m (Dl), 30-40 m (D2), 60-70 m (D3), and 190-200 m (D4).


76
(1) LAI did not correlate with fruit abundance at old edges,
(2) LAI did not change significantly with distance from
forest edge, (3) treefall gaps alone did not influence fruit
abundance (but see the significant interaction between
habitat and distance, and between habitat and month, on TF
and RF), and (4) distance from edge influenced fruit
abundance in complex ways. Thus, irradiance alone cannot
explain my findings. Other abiotic factors, such as
temperature, water availability, and nutrients, likely
influence fruit abundance across the pasture-forest edge.
In tropical areas flower and fruit abundance are
influenced by low temperatures (Tutin and Fernandez 1993),
soil fertility (Gentry and Emmons 1987), water availability
(Heideman 1989; Seghieri et al. 1995), and pollination
(Compton et al. 1994). During the dry season (June-August),
La Planada experiences clear skies, low rainfall, strong
winds, and extreme maximum and minimum temperatures. The
effect of these factors on understory vegetation may be
exacerbated at the pasture-forest edge, thereby influencing
changes in fruit abundance. For instance, during this
period, leaves of understory plants and vines at the forest
edge but not in the interior wilted and abscised. At a
montane locality north of La Planada, soil moisture across
the pasture-forest edge changed progressively over the dry
season and reached its lowest value 10 m from the forest edge
towards forest interior (Murcia 1993). At a lowland tropical
site, vapor pressure deficit (VPD) in the understory changed


ACKNOWLEDGMENTS
From the mountains of La Planada to the concrete of
Bartram Hall I have had the fortune of interacting with
people who have enriched my life in fundamental ways. As I
see it now, this dissertation proved to be a point of
convergence of people, situations, and ideas that is leading
me to new discoveries. I thank the members of my committee,
Douglas J. Levey, H. Jane Brockmann, John Ewel, C. S. "Buzz"
Holling, and Frank Slansky, for their continuous
encouragement and support throughout the two years I spent in
Bartram Hall trying to figure out which direction I wanted my
dissertation and myself to go.
It is not easy to collect data over 12 ha of steep land,
nor to live in isolation. I am particularly grateful to
Natalia Gmez, Sylvia Heredia, and Arlex Vargas for their
help and support in the field. I am indebted to the
neighbors of La Planada, in particular to Adolfo Ortega,
Abelardo Nastacuaz, Demetrio Guanga, Pacho Guanga, Amparo
Oliva, and the GELISI, for sharing their life with me. At
various points during this research I benefited from help
provided by Marta Baena, Girleza Ramirez, Ivan Jimenez,
Natalia Arango, Luis F. Citelli, Omaira Ospina, Mara de
Restrepo, and Paul Marples. By the end of the field season
11


FRUIT ABUNDANCE
(mean counts / 50 m2
63
D1 D2 D3 D4
DISTANCE FROM FOREST EDGE
Figure 4-5. Variation in fruit abundance at the
Reserva Natural La Planada in relation to habitat
and distance from forest edge. Points represent
means and bars standard errors.


37
missing from this analysis is seedling establishment which
links seed germination to seedling growth. In discussing my
results I assume that all seeds that germinated survived into
the seedling stage. I also restrict this discussion to those
stages for which I present results for both Palicourea
gibbosa and Faramea affinis.
Fruiting individuals of P. aibbosa and £. affinis were
not distributed uniformly from pasture to forest interior
(Chapter 4). Palicourea aibbosa was more abundant closer to
the forest edge (D1-D3) than farther inside the forest (D4),
and £. affinis was more abundant at D2 and D3 than at D1 and
D4 (Chapter 4). Such distributions suggest that distance
from forest edge influences one or more stages in the life
cycle of these plants. My results show that not all stages
in the life cycle of P. aibbosa and £. affinis are influenced
equally by the creation of edges and treefall gaps. In
addition, species differed in their response to these two
types of disturbance.
Pollination was influenced by habitat and edge age but
not by distance from forest edge. The percentage of flowers
with pollen tubes and the average number of pollen tubes per
flower decreased from intact forest to gaps (F. affinis) and
from old edges to new edges (P. aibbosa). Although edges and
treefall gaps represent two different scales of disturbance,
results for these two species suggest that recently disturbed
areas affect pollination levels. My results regarding the
effect of distance on pollination levels are similar to those


CHAPTER 1
THE ROLE OF EDGES IN NEOTROPICAL MONTANE LANDSCAPES
Edges constitute a common feature of neotropical montane
landscapes. Complex topography and climate in conjunction
with natural disturbances, ranging from treefall gaps (Murray
1988; Lawton and Putz 1988; Samper 1992), and landslides
(Garwood et al. 1979; Lawton and Dryer 1980; Gentry 1992&),
to mudflows, have given rise to a heterogeneous landscape in
which edges bound the disturbed areas. Over evolutionary
time, the dynamic and heterogeneous character of these
landscapes may have resulted in the unusually high levels of
biodiversity that characterize neotropical montane ecosystems
(Terborgh and Winter 1983, Terborgh 1985, Gentry 1986,
1992a)
Superimposed on this natural heterogeneity is that
resulting from human activities. In areas where favorable
conditions prevail patches of forest are immersed in an
agricultural and urban matrix. Conversely, in areas where
unfavorable conditions limit the development of economic
activities, fields, second growth, and urban areas are
immersed in a forest matrix. In both situations the areas
modified by human activities are bounded by edges. Over
shorter time scales than those defined by large-scale natural
disturbances, changes in land use have led to extinctions and
1


49
birds captured in mist nets. Birds were sampled at the same
edges and distances from the forest edge as were fruiting
plants (Chapter 5). After capture, birds were kept in cloth
bags lined with filter paper for ca. 20 min. Bird droppings
were preserved in alcohol and seeds were compared to a
reference collection, compiled during the study period.
This method for evaluating seed movement may have biases
in addition to those involved when sampling birds with mist
nets (see Chapter 5). In particular, seeds recovered from
birds might represent a non-random sample of seeds ingested,
since seed handling varies within and among species depending
on seed size and other seed characteristics (Levey 1986,
1987). Nevertheless, this method does provide information on
seed movement that would be difficult to determine by other
means (e.g., seed traps).
Analyses
I used a Repeated Measures ANOVA to analyze LAI. Edge
was included as a between factor variable, distance from the
edge as a within factor variable, and individual transects as
subjects.
I analyzed data on fruit production at the community
level with ANOVAs for mixed factorial designs (Girden 1992).
The mixed design was set up as a split-split-plot design with
one repeated measure (Winer et al. 1991) The factors of
interest were edge age, distance from the edge, habitat, and
month, the latter representing the repeated measure. The six


8
JFMAMJJASO ND
Month
Figure 2-2. Distribution of mean monthly rainfall (bars) and
temperature (open circles) (1985-1994) at the Reserva Natural
La Planada (unpublished data Reserva Natural La Planada).
Filled circles represent average values of rainfall for 1992-
1993 .
Temperature (C)


186
Velsquez, A., H. Meyer, W. Marin, F. Ramrez, A. David, A.
Campos, M. Hermeln, S. 0. Bender, M. Arango and J.
Serje. 1994. Planificacin regional del occidente
colombiano bajo consideracin de las restricciones por
amenazas. Memorias Conferencia Interamericana sobre
Reduccin de los Desastres Naturales, Cartagena de
Indias, Colombia, Marzo 1994. Casa Impresora Pacfico,
Santa F de Bogot, Colombia.
Velsquez, M. P. 1992. Aves frugvoras y su relacin con la
flora en un bosque hmedo, en el municipio de San
Carlos, Antioquia, Colombia. Trabajo de Investigacin,
Departamento de Biologa, Universidad de Antioquia,
Medellin, Colombia.
Velez, B. E. 1987. Contribucin al estudio avifaunstico del
Santuario de Fauna y Flora de Iguaque, Boyac.Tesis,
Departamento de Biologa, Pontificia Universidad
Javeriana, Santa F de Bogot, Colombia.
Vitousek, P. M. 1990. Biological invasions and ecosystem
processes: towards an integration of population biology
and ecosystem studies. Oikos 57:7-13.
von Ende, C. N. 1993. Repeated-measures analysis: Growth and
other time-dependent measures. Pages 113-137, in S.M.
Scheiner and J. Gurevitch, editors. Design and Analysis
of Ecological Experiments, Chapman Hall, New York, NY.
Walker, L. R. and L. E. Neris. 1993. Posthurricane seed rain
dynamics in Puerto Rico. Biotropica 25:408-418.
Warburton, N. Structure and conservation of forest avifauna
in isolated rainforest remnants in tropical Australia.
in W. F. Laurance and R. O. Bierregaard, Jr., editors.
Tropical Forest Remnants: Ecology, Management and
Conservation of Fragmented Communities, in press.
Westoby, M., E. Jurado, and M. Leishman. 1992. Comparative
evolutionary ecology of seed size. TREE 7:368-372.
Wiens, J. A. 1992. Ecological flows across landscape
boundaries: a conceptual overview. Pages 217-235, in
A. J. Hansen and F. di Castri, editors. Landscape
Boundaries: Consequences for Biotic Diversity and
Ecological Flows. Springer-Verlag, New York, NW.
Wiens, J. A., C. S. Crawford and J. R. Gosz. 1985. Boundary
dynamics: a conceptual framework for studying landscape
ecosystems. Oikos 45:421-427.
Wilcove, D. S., C. H. McLellan, and A. P. Dobson. 1986.
Habitat fragmentation in the temperate zone. Pages 237-
256, in M.E. Soul, editor. Conservation Biology: The


146
The above model generates a set of testable hypotheses
that could contribute to our understanding of how landscape
pattern, biodiversity, and ecosystem processes interact at
large scales. First, there is a threshold in species numbers
below which lump structure changes dramatically, as indicated
by a decrease in the number of lumps. Second, lump structure
is maintained by the persistance of some species that might
function as attractors. Third, lump structure of plant and
animal assemblages reflects the resilience of a given
ecosystem. The removal of species in ecosystems depicted by
the top triangles may have a lesser impact on lump structure
than the removal of species in ecosystems depicted by the the
bottom ones.
Natural and human disturbances, either alone or in
concert, can affect landscapes from hundreds of meters to
hundreds of kilometers. The inherent complexity of
ecological systems defined by this spatial domain has called
for new approaches and methods. Rather than concentrating on
individual parts, these new lines of inquiry concentrate on
aggregates of parts and key processes that structure
ecosystems (Turner et al. 1995, Holling et al. 1995). The
"lump" approach represents one of these new lines of inquiry.
General Implications
In the mountains of Colombia, changes in landscape
structure have dramatic consequences on assemblages of
frugivorous birds. Big changes in land use result in the


189
newsletters, participating in human right groups, and
discovering the pleasure of observing birds. For this I have
to give credit to my former advisor, Humberto Alvarez-L.
Birds and a two-month field season at the Sierra Nevada de
Santa Marta were my gate into Ecology, which I have not
abandoned since then.
The snowball has grown bigger, with lumps here and
there, reminding me that my existence has not been plain.
The time I spent at the University of Florida pursuing my
Ph.D degree certainly shaped this snowball. The core remains
pretty much unaltered, perhaps strengthened, and that is why
some influential people in my life roll their eyes when I
tell them my whereabouts. I do not know if this behavior is
a warning signal "Carla you are getting into trouble" or a
trusting signal "Carla you are going to make it". Perhaps it
is a combination of both.
It seems that the snowball will continue to grow, as my
eyes have widened so has the scale I want to look at things.
I will spend the next two years doing post-doc work at
Stanford University and at the University of Florida, adding
further lumps to my life.


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EAB1SUU7A_AVEP40 INGEST_TIME 2015-04-14T18:12:41Z PACKAGE AA00029844_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


5 EDGES AND UNDERSTORY BIRDS IN A NEOTROPICAL MONTANE
FOREST
Introduction 82
Methods 84
Data Analysis 86
Results 90
Discussion 105
6 FRUGIVOROUS BIRDS IN FRAGMENTED NEOTROPICAL MONTANE
FORESTS: LUMP STRUCTURE IN BODY MASS
Introduction 112
Methods 114
Results 126
Discussion 138
7 CONCLUSIONS 148
APPENDIX A
Plant species fruiting in the understory at the
Reserva Natural La Planada 151
APPENDIX B
Bird species captured in the understory at the Reserva
Natural La Planada 161
LITERATURE CITED 166
BIOGRAPHICAL SKETCH 188
v


25
distance (D1 to D4) represents the plot unit, habitat (gap
and intact forest) the subplot unit, and edges replicates.
The number of individuals within the gap and interior
categories differed at each distance/edge, producing an
unbalanced design. I used Type III SS, since it takes into
account differences in cell frequencies between treatment
combinations (Gagnon et al. 1989; Potvin 1993). To
determine whether data satisfied assumptions of an ANOVA, I
plotted residuals as a function of fitted Y values. When
residuals where not normally distributed, I transformed the
data (see type of transformation for each data set).
In all cases, I used an alpha of 10% to increase power
of the tests (Zolman 1993). I did so for several reasons.
First, the scale at which I worked precluded inclusion of
more replicates, which is often the case when dealing with
large-scale ecological phenomena (Scheiner 1993). The area
encompassed by the six edges was equivalent to 12 ha and
access to them was difficult due to steep terrain. Second,
in a mixed factorial design the number of degrees of freedom
is reduced compared to a factorial design because of multiple
nesting (Zolman 1993). In the field, I was limited by the
number of edges I could reach within walking distance from
the field station, thus I had to set up the design as
described. Lastly, the use of Type III SS to analyze
unbalanced data sets may lead to Type II errors (Potvin
1993) By increasing the probability of alpha, I compensate


147
disappearence of lumps or complete suites of species with
similar mass. The disappearence of particular lumps in body-
size of frugivorous birds in neotropical montane ecosystems
may reflect important changes in seed dispersal and thus
regeneration trajectories of vegetation after disturbance.
There is some indication that assemblages of neotropical
montane frugivorous birds, depending on degree of habitat
modification, are robust to human disturbance. This is based
on the fact lump structure varied little between similar
sites that differ in the number of species. The
fragmentation and transformation of neotropical montane
ecosystems does not seem to generate the same patterns in
assemblages of frugivorous birds in low to middle high
altitudes and high altitudes. This may have important
consequences for the conservation and management of
ecosystems along the altitudinal gradient.


176
Johnson, A. R., B. T. Milne, and J. A. Wiens. 1992.
Diffusion in fractal landscapes: simulations and
experimental studies of Tenebronid beetle movements.
Ecology 73:1968-1983.
Rapos, V. 1989. Effects of isolation on the water status of
forest patches in the Brazilian Amazon. Journal of
Tropical Ecology 5:173-185.
Rapos, V., E. Wandelli, J. L. Camargo, and G. Ganade. Edge-
related changes in environment and plant responses due
to forest fragmentation in Central Amazonia. in W. F.
Laurance and R. 0. Bierregaard, Jr., editors. Tropical
Forest Remnants: Ecology, Management and Conservation of
Fragmented Communities, in press.
Rapos, V. G. Ganade, E. Matsui, and R. L. Victoria. 1993.
313C as an indicator of edge effects in tropical
rainforest reserves. Journal of Ecology 81:425-432.
Karr, J. R. 1971. Structure of avian communities in selected
Panama and Illinois habitats. Ecological Monographs
41:207-233 .
Karr, J. R. 1981a. Surveying birds in the tropics. Studies
in Avian Biology 6:548-553.
Karr, J. R. 1981b. Surveying birds with mist nets. Studies
in Avian Biology 6:62-67.
Karr, J. R. and R. R. Roth. 1971. Vegetation structure and
avian diversity in several new world areas. American
Naturalist 105:423-435.
Kattan, G. 1992. Rarity and vulnerability: The Birds of the
Cordillera Central of Colombia. Conservation Biology
6:64-70.
Kattan, G., C. Restrepo, and M. Giraldo. 1984. Estructura de
un bosque de niebla en la Cordillera Occidental, Valle
del Cauca, Colombia. Cespedesia 13:23-43.
Kattan, G. H., H. Alvarez-L. and M. Giraldo. 1994. Forest
fragmentation and bird extinctions: San Antonio eighty
years later. Conservation Biology 8:138-146.
Kellman, M. and M. Kading. 1992. Faciliation of tree
establishment in a sand dune succession. Journal of
Vegetation Science 3:679-688.
King, D. A. 1991. Allometry of saplings and understory trees
of a Panamanian forest. Oecologia 51:351-356.


58
fruits (TF) (ANOVA, F3/i2 = 2.56, P = 0.1; Table 4-2). Total
number of fruits (TF) showed a sharp decline from the forest
edge towards the forest interior at new edges and remained
almost unchanged at old edges (Fig. 4-4). The interaction
between habitat and distance from the forest edge was
significant for total number of fruits (TF) (ANOVA, F3(i6 =
2.95, P = 0.06) and for total number of ripe fruits (RF)
(ANOVA, F3(i6 = 3.5, P = 0.04; Table 4-2). For both
variables, fruit abundance at D1 (0-10 m from forest edge)
was higher in gaps than in forest interior (Fig. 4-5) These
differences disappeared at the other distances. The fact
that none of these interactions was significant for total
number of fruits excluding palms (TF-A), suggests that palms
made an important contribution to these results.
The number of ripe fruits (RF) differed significantly
between old and new edges depending on month (ANOVA, Fio,40 =
2.8, P = 0.009; Table 4-2, Fig. 4-6). The total number of
fruiting individuals (TI) and the total number of fruits
excluding Arecaceae (TF-A) differed across the four distances
in some months but not in others as shown by the significant
interaction between distance from edge and month (ANOVA,
f30,120 = 1-48, P = 0.07 and F3o,i20 = 1.57< P = 0.04,
respectively; Table 4-2, Fig. 4-7). These results contrast
with those for TF and RF, in which this interaction was not
significant but in which fruit abundance averaged over time
was affected by distance (Table 4-2). Recall that the total
number of fruiting individuals (TI) and the total number of


145
Number of Species
Figure 6-7. Relationship between species richness and lump
structure in landscapes of variable complexity. Triangles
represent ecosystems from the most complex (El) to the
simplest (E4). Points represent changes in the number of
species within and between ecosystems.


26
for this bias, although it consequently increases Type I
errors. In all cases I present P-values.
Results
Pollen Tubes
Distance from forest edge did not influence the
production of pollen tubes in Palicourea aibbosa and Farrea
affinis (Table 3-1 and Table 3-2). The percentage of flowers
with pollen tubes and the average number of pollen tubes per
flower in £. aibbosa. however, was influenced by edge age
(ANOVA, Fi(4 = 9.2, P = 0.04 and Fi,4 = 7.7, P = 0.05,
respectively, Table 3-1). At old edges individuals had a
higher percentage of flowers with pollen tubes and more
pollen tubes per flower (50% 3.1% and 2.7 0.3, n = 105,
mean SE, respectively) than those at new edges (36% 4.8%
and 2.3 0.5, n = 52, respectively).
Habitat influenced the percentage of E. affinis flowers
with pollen tubes and the number of pollen tubes per flower
(ANOVA, Fi,4 = 8.0, P = 0.02 and Fif4 = 11.8, P = 0.01,
respectively, Table 3-1) In intact forest individuals had a
higher percentage of flowers with pollen tubes and more
pollen tubes per flower (32.5% +3.3% and 0.7 0.06, mean
SE, n = 81, respectively) than those in gaps (26.0% 3.1%
and 0.5 0.07, n = 48, respectively). The effect of
habitat, however, was modified by edge age and distance from
forest edge as shown by the significant interaction of


SEEDS LEFT
(number of seeds/tray)
Figure 3-3. Seed germination (a), seed predation (b), seedling growth (c), and leaf
production in Palicourea aibbosa and Faramea affinis as influenced by habitat. Points
represent means and bars standard errors.
u>
vo


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor in Philosophy
EDGES, FRUITS, FRUGIVORES, AND SEED DISPERSAL
IN A NEOTROPICAL MONTANE FOREST
By
Carla Restrepo
December, 1995
Chairman: Douglas J. Levey
Major Department: Zoology
Edges resulting from natural or human disturbances
influence the distribution of organisms as well as ecological
processes. One such process is seed dispersal, which in turn
may influence may the location of edges through time and even
the entire structure of landscapes. In the Reserva Natural
La Planada, Colombia, I investigated how edges resulting from
human activities influenced seed dispersal. In particular, I
examined how distance from forest edge, in combination with
edge age and treefall gaps, could affect recruitment rates,
fruit abundance, seed movement, and the distribution of birds
in the understory of this neotropical montane forest.
Sampling took place at three old (>40 yr) and three new (<15
yr) edges and within each edge at four distances (0-10, 30-
40, 60-70, 190-200 m) from the pasture-forest edge.
x


102
than expected, i.e., decreasing from edge to interior (n =
4), and (3) species for which the observed number of captures
at D1 and at D4 was higher than the expected, i.e., increase
both at the edge and forest interior, (n = 5) (Table 5-2).
These patterns suggest that edges might influence the
distribution of birds in complex ways, such that some species
avoid edges, others are attracted to edges, and still others
are influenced by factors other than distance from forest
edge. This latter situation is suggested by those five
species showing an increase both at forest edge and forest
interior.
Fourteen species were evaluated to determine the
combined effect of distance from the edge and edge age on
their abundance (Table 5-3). I recognized two main groups:
(1) species showing a significant interaction between
distance from the edge and edge age (Gneterogeneity/ P < 0.1)
and (2) species showing a non-significant interaction between
distance from the edge and edge age (Gneterogeneity> P > 0.1;
Table 5-3). Within these two groups species showed different
responses across the four distances depending on edge age
(Gold and Gftew, P < 0.1; Table 5-3). Thus, the distribution
of individuals within a species not only changes from edge
towards forest interior but also varies with edge age.
For all species combined, I found a significant
association between species abundance and distance from the
edge (%2 = 24.4, df = 12, P = 0.02; Table 5-4). Moreover,
upon examination of the residuals I found that the


4
birds, particularly frugivores. I compare the distribution
of nectarivorous, insectivorous, and frugivorous birds across
the pasture-forest edge to elucidate possible mechanisms
underlying observed patterns.
I take a completely different approach in chapter 6 to
the study of edges and their influence on fruit-frugivore
interactions. Instead of asking how the distribution of
frugivorous birds is affected by the creation of edges within
forest stands, I ask how the distribution of frugivorous
birds is affected by edges within whole landscapes. I
compare changes in the distribution of frugivorous birds
across sites that have been modified in various ways by human
activities. This approach provides the basis for some
generalizations and the formulation of testable hypotheses
for future work.
Chapter 7 is the place for synthesis and speculation. I
emphasize that edges are part of landscapes that are in
continuous change and thus have to be seen as dynamic, not
fixed entities of landscapes. Seed dispersers and seeds move
between forest interior and edge, and sparse species of
plants and frugivorous birds are found more often at edges
than at forest interior. These results suggest a critical
role of edges in landscapes subject to change.


62
DISTANCE FROM FOREST EDGE (m)
Figure 4-4. Variation in fruit abundance at the Reserva
Natural La Planada in relation to edge age and distance
from forest edge. Points represent means and bars
standard errors.


173
Girden, E. R. 1992. ANOVA: Repeated Measures. Sage
University Paper Series on Quantitative Applications in
the Social Sciences, 07-84, Newbury Park, CA.
Gmez, M. E. and C. Palau. 1994. El campesino de San
Isidro, Nario: Condiciones de vida y participacin en
programas de desarrollo en la zona de influencia de la
Reserva Natural La Planada. Tesis Pregrado, Universidad
del Valle, Cali, Colombia.
Goodwin, D. 1976. Crows of the World. Comstock Publishers
Associate, Ithaca, New York, USA.
Gosz, J. R. 1991. Fundamental ecological characteristics of
landscape boundaries. Pages 8-30 in M. J. Holland, P.
G. Risser, and R. J. Naiman, editors. Ecotones: The
Role of Landscape Boundaries in the Management and
Restoration of Changing Environments. Chapman Hall, New
York, USA.
Grubb, P. J. 1977. Control of forest growth and
distribution on wet tropical mountains: with special
reference to mineral nutrition. Annual Review of
Ecology and Systematics 8:83-107.
Guariguata, M. R. Landslide disturbance and forest
regeneration in the upper Luquillo mountains of Puerto
Rico. Journal of Ecology 78:814-832.
Guimares, I. C., C. Uhl, and D. Nepstad. 1994. The role of
the shrub Cordia multispicata Cham as a succcession
facilitator in an abandoned pasture, Paragominas,
Amazonia. Vegetatio 115:91-99.
Gunderson, L. H. 1992. Spatial and temporal dynamics in the
Everglades ecosystem with implications for water
deliveries to Everglades National Park. Ph.D.
Dissertation, University of Florida, Gainesville, FL.
Haffer, J. 1969. Speciation in Amazonian birds. Science
165:131-137 .
Halls, L. K. 1973. Flowering and fruiting of Southern browse
species. U.S. Forest Service Research Paper S0-90:l-10.
Hannan, N. P. and A. Bgu. 1995. A method to estimate
instantaneous and daily intercepted photosynthetically
active radiation using a hemispherical sensor.
Agricultural and forest Meteorology 74:155-168.
Hansen, A. J., D. L. Urban, and B. Marks. 1992. Avian
community dynamics: the interplay of landscape
trajectories and species life histories. Pages 170-195,
in A.J. Hansen and F. di Castri, editors. Landscape


85
did not sample birds. Because the sampling unit was a pair
of mist nets, instead of the traditional single net, I define
mist net hours as the hours that a pair of nets was opened.
In total the mist netting effort was equivalent to 11,892
net-hours. Mist nets were checked every 1-1.5 hours, and for
each captured bird I recorded species, mist net number, molt,
presence of cloacal protuberance or brood patch, fat, mass,
culmen length (total and exposed), culmen height and width,
tarsus and tail length, and wing chord. All birds, except
hummingbirds, were individually marked with color bands.
Hummingbirds were marked temporarily by clipping the tip of
their tail and wing feathers to recognize recaptures within a
mist netting session. Bird abundance is expressed throughout
this paper as capture rates, i.e., number of captures per
pair of mist nets per 100 mist net hours (mnh). Recaptures
on the same day were excluded from the analyses.
Birds were classified into four feeding guilds:
frugivores, insectivores, nectarivores, and carnivores.
Frugivores were defined as species that commonly consumed
fruit and/or seeds; most of them also consumed insects to
some degree. Insectivores ate primarily insects.
Nectarivores relied heavily on nectar and included nectar
"thieves." Carnivores primarily preyed on vertebrates. The
placement of any given species in one of these categories was
based on the analysis of fecal samples, my own observations,
and published reports (Miller 1963, Stiles and Skutch 1989,
Andrade et al. 1993, Arango 1993).


142
In addition, within forest fragments wind shear-forces,
atmospheric humidity, and soil moisture can alter the
vertical structure of vegetation (Esseen 1994, Laurance 1994,
Kapos et al., in press). Horizontal structure refers to the
spatial array of vegetation types that results from changes
in abiotic conditions or disturbance (Wiens et al. 1985). In
Colombia, horizontal structure of the vegetation is simpler
as one moves from the lowlands to the pramo or from little
to highly modified landscapes. At lower and middle
elevations, landslides generate spatial heterogeneity locally
(Gardwood et al. 1979, Mejia et al. 1994, Velsquez et al.
1994) At higher altitudes, however, changes in climate and
soil conditions determine the presence of forest and pramo
(Cuatrecasas 1958, Espinal et al. 1977). Along the gradient
of land-use, abiotic and socioeconomic factors determine not
only rates of deforestation but also the matrix in which
forest fragments are embedded.
Examination of the results for two sites in the lower
montane zone, Rio Grande and Munchique, may help understand
how vertical and horizontal structure interplay and how they
relate to lump structure. Rio Grande and Munchique represent
highly modified landscapes where fragments of degraded native
forest are interspersed with orchards and pastures (RG) and
pine plantations (MU) (Munves 1975, Mondragn 1989). In
these two sites horizontal structure of the vegetation is
similar. The same is not true, however, for vertical
structure which if sampled over several points along


PrymQrtya warszewicziana Hanst.
G
S
Gasteranthus aff. wendladianus (Hanst.)
LD-F
S
Wiehler
Unknown
F
S
CR 582, CR 647, CR 541, CR 499
Marcgraviaceae
Melastornataceae
Meliaceae
Marcaravia eichleriana Wittmaok
F
VS
BD
Marcgraviastrum subssesilis (Benth)
F
VS
Bedell
Blakea cf. stiDulacea Wurdack
G
S
BD
Blakea punctulata (Triana) Wurdack
G
VS
Clidemia sp.l
LD-F
VA
BD
Clidemia sp.2
F
S
BD
Miconia aff. neurotricha
F
s
Miconia loreyoides Triana
LD
vs
BD
Miconia pseudoradula Coan. & Gleason ex
LD
vs
BD
Gleason
Miconia smaraqdina Naudin
LD
vs
BD
Miconia sp.5
LD
vs
BD
CR 533, CR 745, CR 602
Miconia theaezans (Bonpl.) Coan.
LD
vs
BD
Ossaea micrantha (Sw.) Macf. ex Coan.
F
vs
BD
Tooobea oittieri Coan.
F
s
BD
Tooobea sp.
F
vs
CR 263
Unknown
F
vs
CR 676, CR 415, CR 433
Ruaaea qlabra
F
vs
155


Table 2-1. Water balance for Reserva Natural La Planada, transitional life zone between
premontane rain to wet forest (assuming available moisture = 443.7 mm). For details on how
to calculate different variables see Ewel and Madriz (1968). Elevation 1,800 m. Based on
climatological records from 1985-1995.
j
F
M
A
M
J
J
Biotemperature
C
18.9
19.3
19.5
19.6
19.6
19.3
18.5
Potential evapotranspiration (P.ET)
mm
95
88
98
94.8
98
93.4
92.5
Precipitation
mm
460.1
360.7
406.5
451.1
405.3
292.2
148.5
Actual ET
mm
95
88
98
94.8
98
93.4
92.5
Water surplus
mm
365.1
272.7
308.5
356.3
307.3
198.8
56.01
Soil moisture change
mm
0
0
0
0
0
0
0
Moisture available in soil end of month
mm
443.7
443.7
443.7
443.7
443.7
443.7
443.7
All runoff
mm
365
273
309
356
307
199
56
Soil moisture deficit
mm
0
0
0
0
0
0
0
Precipitation deficit
mm
0
0
0
0
0
0
0
Total moisture deficit
mm
0
0
0
0
0
0
0


95
Figure 5-2. Variation in the distribution of understory birds
at La Planada in relation to edge age and distance from the
forest edge. Points represent means and bars standard errors.


88
Manly 1992). Edges were treated as subjects because each
edge was measured repeatedly for each of the different
treatment combinations. Because mist nets are nested within
distance and capture rates were zero for many pairs of mist
nets at a given month/distance/edge, I averaged capture rates
for each three pairs of mist nets/month/distance/edge. This
procedure reduced the dimensionality of the data and also
made the data more normally distributed by eliminating many
zero values. I plotted the residuals as a function of fitted
Y values to detect any violation of assumptions (Manly 1992).
The data for all four ANOVAs presented in this paper were
square-root-transformed. In addition, I verified the
assumption of compound symmetry (i.e., the covariation
between each pair of treatments is equal for all subjects)
for ANOVAs that included within-factors (Girden 1992, Manly
1992). When compound symmetry is violated, the probability
of committing a Type I error increases. To account for this,
the degrees of freedom have to be corrected by a factor,
epsilon, which ranges from 1/(J-1) to 1.0, where J is the
number of levels in a treatment. The closer epsilon is to
1.0, the lower is the probability that compound symmetry is
being violated (Girden 1992). Epsilon is estimated based on
the conservative Geisser-Greenhouse method and the more
liberal Huynh-Feldt method (Girden 1992). In this paper I
report the corrected F values based on the liberal Huynh-
Feldt method (H-F).
In addition to the above omnibus ANOVA tests, I


172
Garwood, N., D. P. Janos, and N. Brokaw. 1979. Earthquake-
caused landslides: a major disturbance to tropical
forests. Science 205:997-999.
Gaston, K. J. and T. M. Blackburn. 1995. Birds, body size
and the threat of extinction. Philosophical
Transactions of the Royal Society of London B 347:205-
212.
Gates, J. E. and L. W. Gysel. 1978. Avian nest dipsersion
and fledgling success in field-forest ecotones. Ecology
59:871-883.
Geiger, R. 1965. The Climate Near the Ground. Harvard
University Press, Cambridge, MA.
Gentry, A. H. 1983. Dispersal ecology and diversity in
neotropical forest communities. Sonderband
naturwisseschaften Ver. Hamburg 7:303-314.
Gentry, A. H. 1986. Endemism in tropical versus temperate
plant communities. Pages 153-181, in M. Soul (editor).
Conservation Biology: The Science of Scaracity and
Diversity. Sunderland, MA.
Gentry, A. H. 1988. Changes in plant communty diversity and
floristic composition on environmental and geographical
gradients. Annals of the Missouri Botanical Garden
75:1-34.
Gentry, A. H. 1992. Diversity and composition of Andean
forests of Peru and adjacent countries: implications for
their conservation. Memorias del Museo de Historia
Natural U.N.M.S.M. (Lima) 21:11-29.
Gentry, A. H. 1992b. Tropical forest biodiversity:
distributional patterns and their conservational
significances. Oikos 63:19-28.
Gentry, A. H. and L. H. Emmons. 1987. Geographical variation
in fertility, phenology, and composition of understory
neotropical forests. Biotropica 19:216-227.
George, A. P., R. J. Nissen, and C. Howitt. 1990. Effects
of environmental variables and cropping on leaf
conductance of custard apple (Annona cherimola x Annona
squamosa) 'African Pride'. Scientia Horticulturae
45:137-147.
Gibbs, J. P. 1991. Avian est predation in tropical wet
forest: an experimental study. Oikos 60:155-161.


RIPE FRUITS/UNRIPE FRUITS UNRIPE FRUITS/FLOWER BUDS
31
0.35
0.3
0.25
0.2
0.15
0.1
0.6
0.55
0.5
0.45
0.4
0.35
DISTANCE FROM FOREST EDGE
D1
D2
D3
n 1 1
D1 D2 D3
Figure 3-1. Fruit set in Palicourea qibbosa as influenced
by distance from forest edge. Points are means and bars
standard errors.


182
Piero, D. and J. Sarukhan. 1982. Reproductive behaviour and
its individual variability in a tropical palm,
Astrocarvum mexicanum. Journal of Ecology 70:461-472.
Potvin, C. 1993. Experiments in controlled environments.
Pages 46-68, in S.M. Scheiner and J. Gurevitch, editors.
Design and Analysis of Ecological Experiments, Chapman
Hall, New York, NY.
Poulin, B., G. Lefebvre, and. R. McNeil. 1992. Tropical
avian phenology in relation to abundance and
exploitation of food resources. Ecology 73:2295-2309.
Quintela, C. 1986. Forest fragmentation and differential use
of natural and man-made edges by understory birds in
Central Amazonia. M.Sc. Thesis, University of Chicago,
Chicago, IL.
Rangel, J. 0., editor. 1994. Ucumar: Un caso tpico de la
diversidad bitica Andina, Corporacin Autnoma Regional
de Risaralda, Colombia.
Ranney, J. W., M. C. Bruner, and J. B. Levenson. 1981. The
importance of edge in the structure and dynamics of
forest islands. Pages 67-95, in R. L. Burguess and D. M.
Sharpe, editors. Forest Island Dynamics in Man-
Dominated Landscapes. Springer-Verlag, New York, NY.
Remsen, J.V., Jr. 1994. Use and misuse of bird lists in
community ecology and conservation. Auk 111:225-227.
Remsen, J. V. Jr. and T. A. Parker. 1983. Contribution of
river-created habitats to bird species richness in
Amazonia. Biotropica 15:223-231.
Renjifo, L. M. 1988. Composicin y estructura de la comunidad
aviaria de bosque andino primario y secundario en la
Reserva del Alto Quindo Acaime, Colombia. Tesis,
Departamento de Biologa, Pontificia Universidad
Javeriana, Santa F de Bogot, Colombia.
Renjifo, L. M., G. P. Servat, J. M. Goerck, B. A. Loiselle
and J. G. Blake. Patterns of species composition and
endemism in the northern Neotropics: a case for
conservation of montane avifaunas, in J. V. Remsen,
editor. Natural History and Conservation of Neotropical
Birds, Ted Parker Memorial Volume, Ornithological
Monographs, in press.
Restrepo, C. 1990. Ecology and cooperative breeding in a
frugivorous bird, Semnornis ramohastinus. M.Sc. Thesis,
University of Florida, Gainesville, FL.


Figure 2-1. Location of study area and edges that were sampled. Forest
is represented by shaded area (based on a 1981 aerial photograph,
approximate scale 1:26,000)


Table 6-1. Sites included in this study.
site
Coordinates Elevation Life Zone-*- Land Use2 References
Holdriqe
Upper Lowland (600/800 to 1,400/1,600 m)
Anchicaya-Alto Yunda, PNN
332 1N
7648 W
1,050
TP-rf
F/F, SG
7,
29
Farallones de Cali (AN)
Reserva Forestal Yotoco (YO)
352 1N
7633 'W
1,500
TP-df/mf
F/P, CT
14,
31
Represa San Carlos (SC)
613 N
7451W
750
TP-mf/wf
SGa/F,P
23
Finca La Esmeralda (LE)
1,250
TP-wf
CT/CT, P
4
Finca El Ocaso (EO)
1,000
TP-wf
P/Fb
4
Lower Montane (1,400-1,600 to 2,300/2,600
m)
Reserva Natural La Planada (LP)
110 N
7800'W
1,800
TP-rf
F/SG
6,
15,
19,27
Parque Regional Ucumari,
447 1N
7532 W
1,850
TLM-wf
F/SG, TPEC>
e 6,
13,
16, 26,
28
Ucumar Bajo (UB)
Bosque Protector San Antonio
329 'N
7638 W
2,000
TLM-wf
F/SG, U
6,
9,
10, 26,
27,
(SA)
30
Empresas Publicas de Manizales
528 1N
7532 W
2,400
SG,TPN^, P
22,
25
, 28
(RB)
Finca Merenberg (ME)
214 1N
7608 W
2,300
TLM-wf
P/F
6,
17,
20, 32
Represa Miraflores (MI)
645 1N
7520 W
2,130
TLM-wf
SG/P
5
Finca Rancho Grande (RG)
436 N
7420'W
1,700
P/CT,SG,U
12
Finca Mirador, Munchique (MU)
230'N
76591W
2,300
TLM-wf
TPEe/SG, P
11
Piedras Blancas (PB)
618 N
7530 W
2,350
TLM-wf
TPEe
8
Upper Montane (2,300/2,600 to 3,100/3,400
m)
Reserva Natural Carpanta-
434 'N
7341'W
2,700
F/SG
2
Estacin Sietecuerales (CA)
Reserva Natural Alto Quindio
437 'N
7520 W
2,800
TM-wf
SG/TPNd, F,
3,
18
Acaime (AC)
P
Parque Regional Ucumari,
447 N
7532 W
2,500
SG/TPNd, F,
6,
13,
16, 21,
Ucumar Alto (UA)
P
26,
28
Santuario de Flora y Fauna
540 'N
7330'W
2,600
TLM/TM-mf
SG/P, Fa
1,
21,
24,25
Iguaque-Caon Mamarramos (IG)
Paramo (3,100/3,400 to 4,800 m)
119


100
Species Level Responses
From June 1992 through August 1993 I captured 2,101
birds of 82 species. Ninety percent of the captures
represented 22 species that were classified as abundant (21-
50 captures) and very abundant (>51 captures) (Appendix B).
Insectivores accounted for 39% of all bird captures,
frugivores for 32%, nectarivores for 28%, and carnivores for
1%.
Twenty-four species and the complex Tanaara spp. (which
includes T. arthus, T. labradoridsS> and T. niqroviridis)
were evaluated to establish variation in their abundance with
distance from the edge. The distribution of bird captures
for 17 out of the 24 common species departed significantly
from a uniform distribution across the four distances (Table
5-2). Given a 10% probability of obtaining a species that
shows a non-uniform distribution, it is very unlikely that 17
or more species out of 24 would have shown a non-uniform
distribution just by chance alone (Binomial Test, P = 1.6 x
10-12) # i conclude that the distribution of understory birds
is affected by edges.
Based on the observed and expected cell frequencies I
further divided the species exhibiting a non-uniform
distribution into three groups: (1) species for which the
observed number of captures at D4 was higher than expected,
i.e., increasing from edge to interior, (n = 8), (2) species
for which the observed number of captures at D1 was higher


91
(2-5 captures), sparse (6-20 captures), abundant (21-50
captures), and very abundant (> 51 captures). I assigned
capture numbers for each category to the four distances. I
used a Chi-square test to evaluate the association between
bird abundance and distance and used the residual values to
determine the contribution of each cell to the overall result
(Siegel and Castellan 1988) .
I used an alpha of 10%. The sampling procedure used in
my study resulted in an increase of Type II errors and low
power of my tests (see Chapter 4). Increasing alpha
counterbalances these effects at the cost of increasing Type
I errors (Zolman 1993).
Results
Understorv Birds
From September 1992 through August 1993 I accumulated
1,789 captures of 80 species. Bird captures differed
significantly among the four distances, suggesting that the
abundance of understory birds at La Planada is affected by
the presence of edges (ANOVA, F4#i6 = 7.6, P = 0.004; Table 5-
1, Fig. 5-1) Mean capture rates were significantly higher
at D1 (mean = 16.9) and D4 (mean = 19.3) than at D2 and D3
combined together (mean = 12.1) (Contrast of Mean
Differences, Fi/i2 = 10.7, P = 0.007 and Fi,i2 = 18.4, P =
0.001, respectively). Edge age, however, modified the
distance effect as shown by the significant distance x edge


Edges influenced Palicoura aibbosa and Faramea affinis.
the two most common understory plants, at various stages of
their life cycle. Seed predation and seed germination but
not relative growth or leaf production rates changed across
the pasture-forest edge. The latter, however, were
influenced by treefall gaps. At the community level, fruit
abundance and the distribution of understory birds changed
across the pasture-forest edge in complex ways that not
always reflected changes due to the presence of edges. This
was demonstrated by the fact that (1) two-way interactions
between distance, edge age, treefall gaps, and month were
significant, and (2) response variables describing fruit and
bird abundance at the community level did not show the same
trends.
Edges influenced fruit-frugivore interactions at the
level of forest stands but also at the level of entire
landscapes as demonstrated by an analysis of body mass
distribution of frugivorous birds as a function of ecosystem
fragmentation. With more edges, entire groups of birds with
similar body mass (termed "lumps") disappeared.
Nevertheless, the distribution of body mass, i.e., lump
structure, remained almost intact under certain land use
types. This work suggests that at broad scales edges
influence frugivorous birds and, as a result, seed dispersal.
xi


Paramo
(P)
Upper
Montane
(UM)
Lower
Montane
(LM)
Upper
Lowland
(UL)
6.3 15.8 39.8 100 251
Body Mass (g)
62
192
351
395
i i i i i i i i i i i
631
1585
3981
128


16
For the most part, studies evaluating the effect of
human disturbances on recruitment rates in tropical plants
have focused on single stages of a plant's life cycle (e.g.,
MacDougall and Kellman 1992, Seizer 1992, Burkey 1993). By
looking at several species it has been possible to establish
patterns and understand the factors underlying the responses
of particular stages (e.g., Sizer 1992). This approach
should be complemented with studies focusing on single
species to establish the relative contribution of a given
stage to the life cycle of a plant (Ellison et al. 1993). A
more complete understanding of factors that limit recruitment
either in forest fragments or nearby disturbed areas must
consider what happens to plants in all stages of their life
cycle.
At a neotropical montane site fruiting individuals of
PsiliCQurea gibbosa and Faramea affinis were not distributed
uniformly across pasture-forest edges (Chapter 4). Here I
report results of a study that examined how several stages of
the life cycle of these two understory plants were influenced
by distance from forest edge, edge age, and treefall gaps.
In particular, I wanted to determine how pollination, fruit
set, seed dispersal, seed predation, germination, and
seedling growth could result in the observed distribution of
P. gibbosa and F. affinis across the pasture-forest edge.


Figure 6-6. Lump structure of Colombian montane frugivorous birds from sites covered mostly
by forest (bottom) to sites highly transformed by human activities (top) within the upper
montane zone. Carpanta (CA), Acaime (AC), Ucumar Alto (UA) Iguaque (IG). Each box
represents a lump and the space between the boxes represent gaps in the distribution of body
mass. The different shades indicate the proportion of species falling within lumps:(1) 0-5,
(2) 5-10, (3) 10-20, (4) 20-30, (5) 30-45, (6) 45-60, and (7) 60-100 % of species. Vertical
lines also represent 0-5% of species. Numbers on the right side represent number of species
for the corresponding data set.
1 2 3 4 5 6 7


143
transects would show a greater variability at RG than at MU.
Even though the two sites have a similar number of species,
RG has more lumps, and the species are more evenly
distributed among lumps, than in MU (Fig. 6-5).
I do not know yet, on a quantitative basis, how vertical
and horizontal structure interact to produce changes in lump
structure of animal and plant assemblages or if they entrain
some other feature of landscapes to which frugivorous birds
are responding. Other features include size of fruit patches
and of seeds, the latter representing a measure of both the
dispersal and regeneration mode of plants (Salisbury 1974,
Hughes et al. 1994, Osunkoya et al. 1994). To my knowledge
there is no published account relating changes in seed size
to altitude in tropical ecosystems. However, in a lowland
neotropical site Martin (1975) found that mean size of seeds
was smaller in second growth areas, compared to mature
forest.
A Model Linking Lumps and Species Diversity in Landscapes
My results showed relationships between the number of
lumps and the number of species along a gradient of
structural complexity of landscapes. In addition, they hint
at a relationship between lump structure and the resilience
of ecosystems. Resilience as defined by Holling (1973) is a
measure of the amount of disturbance and/or change that an
ecosystem can absorb before turning into a different one.


184
Schupp, E. W., H. F. Howe, C. K. Augspurger, and D. J. Levey.
1989. Arrival and survival in tropical treefall gaps.
Ecology 30:562-564.
Seghieri, J., Ch. Floret, and R. Pontanier. 1995. Plant
phenology in relation to water availability: herbaceous
and woody species in the savannas of northern Cameroon.
Journal of Tropical Ecology 11:237-254.
Seizer, N. G. 1992. The impact of edge formation on
regeneration and litterfall in a tropical rain forest
fragment in Amazonia. Ph.D. Dissertation, University of
Cambridge, Cambridge, UK.
Shoji, S., M. Nanzyo, and R. A. Dahlgren. 1993. Volcanic ash
soils: Development, properties, and utilization.
Developments in Soil Science 21. Elsevier Science
Publishers B.V., Amsterdam, The Netherlands.
Siegel, S. and N. J. Castellan. 1988. Nonparametric
Statistics for the Behavioral Sciences. McGraw-Hill Book
Company, New York, NY.
Silverman, B. W. 1986. Density Estimation for Statistics and
Data Analysis. Monographs on Statistics and Applied
Probability 26, Chapman Hall, London, UK.
Sokal, R. R. and F. J. Rohlf. 1981. Biometry. Freeman, New
York, NY.
Sork, V. L. 1983. Distribution of pignut hickory (Carva
glabra) along a forest to edge transect, and factors
affecting seedling recruitment. Bulletin Torrey
Botanical Club 110:494-506.
Stephenson, A. G. 1981. Flower and fruit abortion: proximate
causes and ultimate functions. Annual Review of Ecology
and Systematics 12:253-279.
Stephenson, A. G. 1992. The regulation of maternal
investment in plants. Pages 151-172, in C. Marshall and
J. Grace, editors. Fruit and Seed Production: Aspects
of Development, Environmental Physiology and Ecology.
Cambridge University Press, Cambridge, UK.
Stiles, F. G. 1985. On the role of birds in the dynamics of
neotropical forests. ICBP Technical Publication 4:49-
59.
Stiles, F. G. and A. Skutch. 1989. A Guide to the Birds of
Costa Rica. Cornell University Press, Ithaca, New York,
USA.


138
and gaps when compared to the other sites.
Discussion
In part because of the exploratory nature of this work,
and in part because of the early stage of development of the
techniques to test ecosystem "lumpiness", my interpretation
of the results are intended as hypotheses rather than
conclusions. By exploratory I mean that I made use of
information that was already available and thus could not
control for many factors that might confound the results,
including size of the area surveyed, hunting, and differences
in vegetation types. However, the repetition of some
patterns among my four analyses suggests that local
differences in assemblages of frugivorous birds might be
overridden by general processes that impose structure on the
landscape.
Patterns in Lump Structure
In general, the number of lumps, i.e., aggregates of
species having a similar body mass, decreased from areas
covered by continuous native forest to areas where forest has
been replaced by simpler vegetation types. It can be argued
that this trend simply reflects a decrease in the number of
species which in turn may reflect a decrease in habitat
complexity (e.g., Karr and Roth 1971, Terborgh 1977). That
is, lump structure reflects biases resulting from sampling
procedures. Several of my data sets, however, did not


Ericaceae
CR 789
Spheraedenia stevermarkii
Cavendishia enaleriana Hoer.
Cavendishia tarapotana (Benth.) Meisner
CR 666, CR 657, CR 590
Macleania stricta A.A. Smith
Psammisia aff. debilis Sleumer sp. nov.
Psammisia cf. dolichcpoda A.A. Smith
Psammisia cf. ulbrichiana Hoerold.
Psammisia ferruainea A.A. Smith
Psammisia sodiroi Hoerold.
Gesneriaceae
Alloplectus sp.
CR 475, CR 761
Alloplectus sp.1
CR 790, CR 654
Alloplectus tenuis Benth.
Alloplectus tetraaonus (Hanst.) Hanst.
Alloplectus teuscheri (Raymond) Wiehler
B^sleria solanoides
Besleria sp.
CR 759
Columnea bvrnsina (Wiehler) L.P. Kvist &
L.E. Skog
Columnea cinerea
Columnea ebrnea (Wiehler) L.P. Kvist &
L.E. Skog
Columnea eubracteata
Columnea qiaantifolia
Columnea minor
Drvmonia sp.
CR 559, CR 688, CR 540, CR 400
LD-F A BD
F S BD
G VS BD
LD VS
LD A
F S
G S
G VS
F S
G S
F VS
G S
G S
F VA
LD VA BD
LD-G S
LD S BD
LD A BD
F VS
LD S BD
F S BD
LD S
F S BD
154


33
was similar in gap and intact forest from D1 to D3, but lower
in gap (3.5 0.1, mean SE) than in intact forest at D4
(4.4 0.08). Over the 18-week period the mean number of F.
affinis seeds remaining in the trays decreased at all four
distances but the rate of decline was steeper at D4 and at D2
than at D1 and D3.
Seed Germination
Palicourea aibbosa seeds germinated sooner (week 6) than
Faramea affinis seeds (week 18). Averaging over time,
germination rates of £. aibbosa seeds were significantly
affected by distance from forest edge. The same was not true
for £. affinis. Germination rates of £. aibbosa seeds were
greater at D2 than at D1 and D4 (ANOVA, F3,6 = 5.0, P = 0.04;
Table 3-4, Fig. 3-2b). Habitat had a significant effect on
seed germination rates in £. aibbosa but not in £. affinis
(ANOVA, Fi(8o = 7.3, P = 0.008 and Fi/78 = 2.3, P = 0.13,
respectively). Averaged over time, more seeds of £. aibbosa
germinated in gaps than in intact forest (Fig. 3-2b).
The effects of distance and habitat were modified by
time as shown by the significant distance x week (£. affinis)
and habitat x time (£. aibbosa) interactions (ANOVA, F42.84 =
2.0, P = 0.003 and F3(n2o = 20.3, P = 0.006, respectively).
In the former, germination rates over time were steeper at D2
and D3 than at D1 and D4. In the latter, germination rates
were higher at gaps than in intact forest.


HeligdQxa imperatrix
LafrggnaY9 lafresnavi
Ocreatus underwoodii
Phaetornis svrmatophorus
Schistes aeoffrovi
Urosticte beniamini
Trogonidae
Troqon personatus
Capitonidae
Semnornis ramphastinus
Ramphastidae
Andiaena laminirostris
Picidae
Campephilus pollens
Dendrocolaptidae
Dendrocincla tvrannina
Glvphorvnchus spirurus
bepidbCQlapte? affinis
Xiphocolaptes promeropirhvnchus
Furnariidae
Anabacerthia varieqaticeps
Cranioleuca ervthrops?
Marqarornis stellatus
Premnoplex brunnescens
Premnornis quttuliqera
Schizoeaca fuliginosa
Sclerurus mexicanus
Svndactvla subalaris
Thripadectes iqnobilis
Thripadectes virqaticeps
Formicariidae
VS
N
8.2
vs
N
5.0
s
N
2.9
VA
N
6.0
VS
N
4.0
ES
N
4.4
S
F
58.6
VS
F
92.1
VS
F
349.5
ES
I
206.0
VS
I
53.4
VA
I
14.3
VS
I
29.0
ES
I
161.0
ES
I
25.0
ES
I
14.0
S
I
20.7
VA
I
15.9
VA
I
14.8
ES
I
15.5
S
I
22.3
A
I
32.9
S
I
45.5
S
I
58.9
163


54
DISTANCE (m)
Figure 4-1. Leaf area index across the pasture-forest edge,
indicated by the arrow (0 m). Points represent the average
of three measurements per edge and bars the standard errors.
Open squares = Pialapi, open triangles = Climo II, and
filled circles = Climo I.


Table 5-1. Results of ANOVAs for Mixed Factorial Designs (1 between-, 2 within-factors) on
mean capture rates x 100 mist netting hours (mnh) of all birds, frugivores, insectivores,
and nectarivores. The month and month x distance x age effects were excluded from this
table. () based on square root-transformed data. Significance at 10% (*), 5% (**), 1%
(***) .
Age (A)
Error
Edge (Age)
Distance
Distance x Age
Error
D x E (A)
df
F
df
SS
df
F
H-F
df
F
H-F
df
SS
All Birds
1
2.89
ns
4
11.39
3
7.63

3
3.54

12
22.39
Frugivores
1
1.77
ns
4
14.27
3
4.85
*
3
2.75
*
12
25.03
Insectivores
1
0.24
ns
4
18.42
3
2.79

3
1.32
ns
12
27.86
Nectarivores
1
0.05
ns
4
17.78
3
1.78
ns
3
1.83
ns
12
24.67
Month x Age
Error
M x E (A)
Month x
Distance
Error
D X M X E (A)
df
F
H-F
df
SS
df
F
H-F
df
SS
All Birds
10
1.78

40
30.07
30
1.10
ns
120
85.84
Frugivores
10
0.79
ns
40
33.99
30
1.93
*
120
106.78
Insectivores
10
3.23
*
40
39.91
30
0.78
ns
120
114.94
Nectarivores
10
0.52
ns
40
41.85
30
0.82
ns
120
147.25


Table 4-1
Source
Edge
Transect
Distance
Distance
Distance
. Results of a Mixed Factorial ANOVA on leaf area index for old edges at the
Reserva Natural La Planada
df
SS
MS
F-Value
P-Value
H-F
3
6.768
2.256
0.401
0.7585
(Edge)
5
28.095
5.619
23
11.64
0.506
0.468
0.981
0.981
* Edge
69
76.934
1.115
1.031
0.4367
0.437
* Transect (Edge)
115
124.376
1.082
H-F Epsilon
1.714
Distance


170
Dalling, J. W. and E. V. J. Tanner. 1995. An experimental
study of regeneration on landslides in montane rain
forest in Jamaica. Journal of Ecology 83:55-64.
De las Salas, G. and M. Ballesteros. 1986. Tipos de
comunidades forestales de la Reserva Natural La Planada,
Nario, Colombia. Informe Final FES, Cali, Colombia
DeGraaf, R. M. 1992. Effects of even-aged management on
forest birds at northern hardwood stand interfaces.
Forest Ecology and Management 46:95-110.
del Hoyo, J., A. Elliott, and I. Sargatal, editors. 1992.
Handbook of the Birds of the World. Lynx Ediciones,
Barcelona, Espaa.
Denslow, J. S. and G. S. Hartshorn. 1994. Tree-fall gap
environments and forest dynamic processes. Pages 120-
127, in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and
G. S. Hartshorn, editors. La Selva: Ecology and Natural
History of a Neotropical Rain Forest. The University of
Chicago Press, Chicago, IL.
Denslow, J. S. and A. E. Gomez-Diaz. 1990. Seed rain to
treefall gaps in a neotropical forest. Canadian Journal
of Forest Researach 20:642-648.
Didham, R. The influence of edge effects and forest
fragmentation on leaf litter invertebrates in central
Amazonia. in W.F. Laurance and R.O. Bierregaard, Jr.,
editors. Tropical Forest Remnants: Ecology, Management
and Conservation of Fragmented Communities, in press.
Draiijers, G. P. J., W. P. M. F. Ivens, and W. Bleuten. 1988.
Atmospheric deposition in forest edges measured by
monitoring canopy throughfall. Water, Air, and Soil
Pollution 42:129-136.
Duellman, W. E. 1979. The herpetofauna of the Andes:
patterns of distribution, origin, differentiation, and
present communities. Pages 371-45 in W.E. Duellman,
editors. The South American Herpetofauna: Its Origin,
Evolution, and Dispersal. Museum of Natural History,
Monograph 7, University of Kansas, Lawrence, Kansas.
Dunning, J. B., Jr. 1993. Handbook of Avian Body Masses.
CRC Press, Boca Raton, FL.
Dwyer, J. D. 1980. Family 179. Rubiaceae, Part II. Annals
of the Missourit Botanical Garden 67:257-522.


180
Martnez-Ramos, M. and A. Soto-Castro. 1993. Seed rain and
advanced regeneration in a tropical rain forest.
Vegetatio 107/108:299-318.
Mathai, C. K. and K. S. K. Sastry. 1988. Productivity of
black pepper vines (Piper niarum L.) as influences by
the light availability during pre-flowering stage.
Comparative Physiological Ecology 13:97-102.
Matlack, G. R. 1993. Microenvironment variation within and
among forest edge sites in the Eastern United States.
Biological Conservation 66:185-194.
May, P. and A. J. Antcliff. 1963. The effect of shading on
fruitfulness and yield in the sultana. Journal of
Horticultural Science 38:85-94.
McDiarmid, R. W., R. E. Ricklefs, and M. S. Foster. 1977.
Dispersal of Stemmadenia dQnnelljl-smlthii (Apocynaceae)
by birds. Biotropica 9:9-25.
Mejia, J., H. Meyer, A. Campos, A. Gallego, and A. Velsquez.
1994. El terremoto de Paez, anatoma de un sismo, Parte
I: generalidades y movimientos fuertes. Revista de la
Asociacin de Ingenieros del Valle (472):14-15.
Miller, A. H. 1963. Seasonal activity and ecology of the
avifauna of an american equatorial cloud forest.
University of California Publications in Zoology 66:1-
78.
Moermond, T. C. and J. S. Denslow. 1985. Neotropical avian
frugivores: patterns of behavior, morphology, and
nutrition, with consequences for fruit selection.
Neotropical Monographs 36:865-897.
Moller, A. P. 1989. Nest site selection across field-
woodland ecotones: the effect of nest predation. Oikos
56:240-246.
Mondragn, M. L. 1989. Estructura de la comunidad aviaria
en bosques de coniferas y en bosques aledaos de
vegetacin nativa. Trabajo de Grado, Departamento de
Biologa, Universidad del Valle, Cali, Colombia.
Munves, J. 1975. Birds of highland clearing in
Cundinamarca, Colombia. Auk 92:307-321.
Murcia, C. 1993. Edge Effects on the Pollination of Tropical
Cloud Forest Plants. Ph.D. Dissertation, University of
Florida, Gainesville, FL.


Table 4-2. Results of ANOVAs for Mixed Factorial Designs (Split-split-plot, 1 repeated measure) on mean
counts/50 m^ of number of fruiting individuals (TI), number of fruits (TF), number of ripe fruits (RF), and
number of fruits excluding Arecaceae (TF-A). log-transformed data. Significance at 10% (*),5% (**), and
1% (***).
Variable
Age
Edge(Age)
Distance
Distance x
Age
D x E(A)
(A)
Error
(D)
Error
df
F
df
SS
df
F
df
F
df
SS
TI
1
0.99
4
1256.13
3
1.77
3
0.65
12
1745.05
TF
1
0.28
4
71.37
3
4.34 **
3
2.56
*
12
67.2
TF-A
1
1.26
4
39.18
3
1.43
3
0.4
12
73.79
RF
1
0.47
4
40.58
3
5.38 **
3
0.96
12
31.67
Habitat
H x D
H
X A
H
x D x A
Hx[E(A)xD]
(H)
Error
df
F
df
F
df
F
df
F
df
SS
TI
1
0.74
3
2.24
1
0.522
3
0.016
16
1522.51
TF
1
0.058
3
2.95 *
1
0.047
3
0.41
16
65.02
TF-A
1
1.26
3
0.77
1
0.244
3
0.165
16
42.28
RF
1
0.26
3
3.47 **
1
0.01
3
1.02
16
29.38
Month x
H
M
x D x H
M
X
A x H
M X
A x D x
H
MxHx[DxE(A)]
Error
df
F
H-F
df
F
df
F
df
F
df
SS
TI
10
0.86
30
1.27
10
0.31
30
1.28
160
389.1
TF
10
2.03
*
30
0.55
10
0.37
30
0.6
160
32.45
TF-A
10
1.42
30
0.58
10
0.69
30
1.28
160
24.58
RF
10
2.01
* *
30
0.41
10
0.94
30
0.62
160
54.78
M
x A
M x E(A)
M
x D
M
x D x A
M X
D x E (A)
Error
Error
df
F
H-F
df
SS
df
F H-F
df
F
df
SS
TI
10
1.47
40
144.98
30
1.48 *
30
1.09
120
350.03
TF
10
0.49
40
21.43
30
1.29
30
0.99
120
25.43
TF-A
10
0.65
40
17.25
30
1.57 **
30
0.7
120
22.13
RF
10
2.83

40
1.27
30
0.76
30
1.17
120
43.2


47
differences in gap size and shape (Brown 1993; Denslow and
Hartshorn 1994). Nevertheless, I was more interested in the
possible interaction between distance from the edge and
treefall gaps than in the treefall gaps themselves.
I monitored changes in fruit production at each pair of
the 10 x 5 m subquadrats over a 12 mo period (September 1992-
August 1993, excluding December). In each subquadrat I
identified and counted individual plants <7 m tall bearing
unripe and/or ripe fleshy fruits (Levey 1988^.,^; Blake and
Loiselle 1991). I also included broken limbs bearing fruits.
Most species I recorded complete their life cycle within this
arbitrarily set understory stratum. A few species, mostly in
the Arecaceae (palms), Rubiaceae, and Melastomataceae, also
fruit in higher strata. For each individual, except species
in the Araceae, I counted the total number of unripe and ripe
fruits every month on a biweekly basis. I averaged these
biweekly counts to obtain a single value on fruit abundance
for any given month.
I expressed fruit production in four different ways: (1)
total number of individuals bearing unripe and/or ripe fruits
(TI); (2) total number of fruits (unripe + ripe fruits) (TF);
(3) total number of ripe fruits (RF), and (4) total number of
fruits (unripe + ripe fruits), excluding the Arecaceae (TF-
A). In all cases, fruit abundance is expressed as the mean
number of counts per 50 m2, the area of each 10 x 5 m
subquadrat.


19
marked all inflorescences and infructescences and followed
them over the entire study period. I monitored individuals
for the presence of inflorescences and infructescences on a
biweekly basis during the first 6 months (March 1992-August
1992) and on a monthly basis the following 11 months
(September 1992-July 1993).
Pollination
To determine the influence of edges on pollination
success I looked at pollen tube production. From June 1993
until November 1993 I checked flowering individuals for four
consecutive days to collect an average of 10 flowers per
individual. These individuals represent a subset of those
that were monitored over the 16-month period. I dissected
the flowers and fixed the styles in formalin-acetic acid
(FAA) to examine pollen tubes. Pollen tubes were stained
(Martin 1959, Feinsinger et al. 1992) and counted under an
epifluorescent microscope. Styles were processed by P.
Amezquita at the Universidad de Santiago de Compostela,
Spain. Pollen tube production per individual was expressed
as the percentage of flowers with pollen tubes (F) and as the
average number of pollen tubes per flower (P).
Fruit Set
For each new inflorescence I counted the number of
flower buds and followed them until fruits developed and
ripened. I expressed fruit set as the percentage of unripe


71
of the residuals showed that more individuals of very sparse
species were found at Dl and of sparse species at D2.
Although individuals of both groups of plants might be found
across the four distances, these results show that sparse
species are found more often close to the edges.
Seed movement
I recovered 393 bird droppings from 19 species of mist-
netted frugivorous birds. Seeds of 93 species, of which I
was able to identify 65, were represented. Of the species I
identified, 52 (80%) were found in the subquadrats where I
counted fruits. This figure, compared to the total number of
plant species I recorded fruiting in the understory (149),
shows that bird droppings represent a subsample of the plant
species that I found.
The number of plant species in individual bird droppings
ranged from 1 to 7 (mean SD, 1.7 1.0, N = 393). There
was no significant difference in the mean number of plant
species in droppings recovered at the four distances (ANOVA,
f3,393 = 0.8, P = 0.5). In addition, total numbers of seeds
in bird droppings was independent of distance (x2 Test for
Independence, df = 15, x2 = 12.8, P = 0.6) .
Fourteen of the 26 abundant species (>21 individuals)
and 38 of the 124 sparse species (<21 individuals) fruiting
in the understory were represented in the bird droppings
(Appendix A). Since sparse species were found more often at
the forest edge, I compared their distribution against that
of droppings containing their seeds to determine whether they


177
Klein, B. C. 1989. Effects of forest fragmentation on dung
and carrion beetle communities in Central Amazonia.
Ecology 70:1715-1725.
Kroodsma, R. L. 1984. Effect of edge on breeding forest bird
species. Wilson Bulletin 96:426-436.
Kruess, A. and T. Tscharntke. 1994. Habitat fragmentation,
species loss, and biological control. Science 264:1581-
1584.
Kuitunen, M. and M. Mkinen. 1993. An experiment on nest
site choice of the Common Treecreeper in fragmented
boreal forest. Ornis Fennica 70:163-167.
Lambert, F. R. 1992. The consequences of selective logging
for Bornean lowland forest birds. Philosophical
Transactions of the Royal Society of London B 335:443-
457.
Laurance, W. F. 1990. Comparative responses of five arboreal
marsupials to tropical forest fragmentation. Journal of
Mammalogy 71:641-653.
Laurance, W. F. 1991. Ecological correlates of extinction
proneness in Australian tropical rain forest mammals.
Conservation Biology 5:79-89.
Laurance, W. F. 1993. Avian nest predation in modified and
natural habitats in tropical Queensland: an experimental
study. Wildlife Research 20:711-723.
Laurance, W. F. 1994. Rainforest fragmentation and the
structure of small mammal communities in tropical
Queensland. Biological Conservation 69:23-32.
Lawton, J. H., S. Nee, A. J. Letcher, and P. H Harvey. 1994.
Animal distributions: patterns and processes. Pages 41-
58 in P. J. Edwards, R. M. May, and N. R. Webb, editors.
Large-scale Ecology and Conservation Biology. Blackwell
Scientific Publications, Oxford, UK.
Lawton, R. and V. Dryer. 1980. The vegetation of the
Monteverde Cloud Forest Reserve. Brenesia 18:101-116.
Lawton, R. 0. and F. E. Putz. 1988. Natural disturbance and
gap-phase regeneration in a wind-exposed tropical cloud
forest. Ecology 69:764-777.
Leigh, E. G., Jr. 1975. Structure and climate in tropical
rain forest. Annual Review of Ecology and Systematics.
6:67-86.


Araliaceae
Mgngtera sp.
F VS
Scheffiera aff. lasioavne Harms
Scheffiera lasioavne Harms
Arecaceae
VS
LD VS BD
Aiphanes sp.
Chamaedorea polvchlada
Chamaedorea sp.
CR 466, CR 467
Geonoma weberbaueri
Prestoea aff. purpurea
Boraginaceae
Tournefortia aiaantifolia Killip
Bromeliaceae
F S BD
F A
F S BD
F VA BD
F S
F VS
Ronnberaia aff. deleani L.B. Smith
Campanulaceae
Burmeistera aff. lonaifolia Gleason
Burmeistera carnosa Gleason
Burmeistera sp.
Burmeistera sp. nov.
CR 543, CR 716, CR 734, CR 502
Clusiaceae
Clusia sect. Anandrogyne, "A. multiflora
H.B.K.group"CR 796
Cyclanthaceae
Asolundia sp.1
Asplundia stenophvlla
CR 553, CR 680, CR 579, CR 766
Soheraedenia sp.
F VS
F S
LD A
LD-F S BD
LD-F A BD
LD-F VS BD
F A
F S
F
S
153


79
edge. The latter provides information on the persistence of
edges through time, depending on the ability of plants to
complete their life cycle, including the production and
dispersal of seeds, in a given environment.
Increases in fruit abundance, and thus of seed outputs,
can have different consequences for individuals, populations,
and assemblages. A numerical increase in fruit production by
an individual can affect traits, such as seed size (Agren
1988, 1989), and thus seedling performance (Westoby et al.
1992). Changes in fruit numbers at the population level can
affect recruitment rates (e.g., Kellman and Kading 1992,
Guimaraes et al. 1994). For plant assemblages, an increase
in fruit numbers can affect colonization rates of disturbed
areas and thus alter species composition. In all cases,
changes in fruit numbers can affect the behavior of
dispersers (Murray 1987; Loiselle and Blake 1993).
At La Planada, the distribution of fruiting individuals
was influenced by the presence of edges. Changes in the
number of fruiting individuals at any given distance from the
forest edge may indicate differences in recruitment rates at
different distances. Plant establishment, growth,
reproduction, and seed dispersal are very likely to be
influenced by distance from forest edge in various ways
resulting in the observed distributions of fruiting
individuals (Chapter 3). It is possible that if edge
conditions remain unaltered over time, i.e., pastures and


28
Table 3-2. Proportion of flowers with pollen tubes in
relation to the total number of flowers (F) and average
number of pollen tubes per flower (P) in Palicourea qibbosa
and Faramea affinis in relation to distance from forest edge.
Numbers are the mean standard error and number of
individuals sampled ().
D1 D2 D3 D4
Palicourea gibbosa
F
P
43.3 4.0
(59)
2.7 3.5
35.7 7.6
(19)
2.1 3.7
52.5 5.0
(52)
2.8 2.9
43.0 6.4
(27)
2.4 2.9
Faramea affinis
F
p
28.7 3.3
(n=39)
0.6 0.08
25.0 3.7
(n=29)
0.4 0.08
35.2 5.1
(n=24)
0.7 0.1
33.3 3.6
(n=37)
0.6 0.09


Table 2-1. (continued)
A
S
0
N
D
TOTAL
Biotemperature
C
18.8
18.9
19.2
19.2
19.1
19.2
Potential evapotranspiration (P.ET)
mm
94
91.5
96
92.9
95.5
1129.6
Precipitation
mm
163.2
287.8
527.5
482
491.6
4437.4
Actual ET
mm
94
91.5
96
92.9
95.5
1129.6
Water surplus
mm
69.17
196.3
431.5
389.1
396.1
3347
Soil moisture change
mm
0
0
0
0
0
Moisture available in soil end of month
mm
443.7
443.7
443.7
443.7
443.7
All runoff
mm
69.2
196
431
389
396.1
3347
Soil moisture deficit
mm
0
0
0
0
Precipitation deficit
mm
0
0
0
0
0
0
Total moisture deficit
mm
0
0
0
0
0


152
APPENDIX A
PLANT SPECIES FRUITING IN THE UNDERSTORY
OF THE RESERVA NATURAL LA PLANADA (SEPTEMBER 1992-AUGUST 1993)
Habitat (H): forest (F), treefall gaps (G), large-disturbed areas, including second
growth, road sides (LD). Abundance of fruiting individuals at the edges included in this
study (AB): very abundant (VA), abundant (A), sparse (S), very sparse (VS), extremely
sparse (ES). Seeds found in bird droppings (BD). Letters followed by numbers indicate
collection number.
FAMILY SPECIES H AB BD
Old and New Edges
Araceae
Anthurium
andinum Encrl .
F
S
BD
Anthurium
carchiense Croat
LD
A
Anthurium
cf. marmoratum Sodiro
F
A
Anthurium
cf. melampyi Croat
F
VA
BD
Anthurium
cf. oulverulentum Sodiro
F
A
Anthurium
lancea Sodiro
F
VS
Anthurium
loncricaudatum Encrl.
F
S
BD
Anthurium
membranaceum Sodiro
LD
VA
BD
Anthurium
mindense Sodiro
S
Anthurium
ovatifolium Encrl.
F
S
BD
Anthurium
sp.
LD
VS
BD
Anthurium
terracolum Croat
F
S
Anthurium
trinerve Mig.
F
vs
Anthurium
umbraculum Sodiro
F-LD
VA
BD
Anthurium
umbricolum Engl.
F
A
BD
Anthurium
versicolor Sodiro
F
VA
CR 556,
CR 583, CR 670, CR 791


183
Ridgely, R. S. and G. Tudor. 1989. The birds of South
America. VI. The Oscine Passerines. University of
Texas Press, Austin, TX.
Ridgely, R. S. and G. Tudor. 1994. The birds of South
America. V 2. The Suboscine Passerines. University of
Texas Press, Austin, TX.
Ridgely, R. S. and S. J. C. Gaulin. 1980. The birds of finca
Merenberg, Huila department, Colombia. Condor 82:379-
391.
Romano, G. B. 1990. Invasibility of a Mixed Hardwood Forest
by Euoatorium caoillifolium and E. comoositifolium.
M.Sc. Thesis, Universtiy of Florida, Gainesville, FL.
Rosas, M. L. 1986. Estudio de la estructura de la comunidad
de frugvoros en sotobosques del Can de Mamarramos en
el Santuario de Fauna y Flora de Iguaque, Boyac.Tesis,
Departamento de Biologa, Pontificia Universidad
Javeriana, Santa F de Bogot, Colombia.
Ryszkowski, L. 1992. Energy and material flows across
boundaries in agricultural landscapes. Pages 270-284 in
A.J. Hansen and F. di Castri, editors. Landscapes
Boundaries: Consequences for Biotic Diversity and
Ecological Flows. Springer-Verlag, New York, NY.
Salisbury, E. 1974. Seed size and mass in relation to
environment. Proceedings of the Royal Society, London B
186:83-88.
Salo, J., R. Kallioloa, R. Hkkinen, Y. Mkinen, P. Niemela,
M. Puhakka, and P. D. Coley. 1986. River dynamics and
the diversity of Amazon lowland forest. Nature 322:254-
258.
Samper, C. 1992. Natural disturbance and plant establishment
in an Andean cloud forest. Ph.D. Dissertation,
University of Harvard, Cambridge, MA.
Scheiner, S. M. 1993. Introduction: Theories, hypotheses,
and statistics. Pages 1-13, in S. M. Scheiner and J.
Gurevitch, editors. Design and Analysis of Ecological
Experiments, Chapman Hall, New York, NY.
Schupp, E. W. 1988. Seed and early seedling predation in the
forest understory and in treefall gaps. Oikos 51:71-78.
Schupp, E. W. and E. J. Frost. 1989. Differential predation
o Welfia qeorqii seeds in treefall gaps and the forest
understory. Biotropica 21:200-203.


11
The La Planada forest develops on well drained soils
(Dystrandept) derived partially from volcanic material,
that are moderately acid, with a sandy to clay loam texture
(De Las Salas and Ballesteros 1986). The canopy height
(average 22 m) and the basal area (dbh > 4 cm; 33.4 m2/ha) of
the forest are low and epiphytic and hemiepiphytic plants are
very abundant (De Las Salas and Ballesteros 1986; Gentry
1988) Plants (dbh > 2.5 cm) in a 0.1 ha plot were
represented by 112 species (Gentry 1992a). The most
important trees on this plot were Ouararibea sp., Elaeaia
sp., BierQnymft sp., Alghgrngfr sp., Billia colombiana. Inga
sp., Otoba sp., and Ocotea sp.; the most important treelets
and shrubs were Faramea elegans. Prestoea cf. purpurea.
Ajghgngi? sp., Geonoma weberbaueri. Palicourea qibbosa and
Miconia sp.; and the most common epiphytes were Philodendron
cf. sggndgns, Spheraedenia stevermarkii and Psammisia sp. (A.
Gentry, unpublished data).
General Sampling Procedure
I chose six sites to evaluate how edge age and distance
from the edge towards the forest interior influence various
components of fruit production and seed dispersal. These
sites, hereafter referred to as edges, were active or
recently abandoned pastures contiguous with forest. Thus, at
most edges there was a sharp delineation between forest and
the adjacent pasture (Table 2-2). Four edges lay at La
Planada boundaries (Marcos, Climo I, Climo II, and


67
Palicourea aibbosa), the latter suggesting a bimodal
distribution (Table 4-3). Species showing bimodal
distributions, in particular, suggest that distance alone can
not explain their distribution across the pasture-forest
edge.
I evaluated the combined effect of distance and edge age
on the number of fruiting individuals for 16 species (Table
4-4). For nine species there was a significant interaction
between distance and edge age, indicating that the
distribution of fruiting individuals across the four
distances differed between old and new edges (Gneterogeneity p
< 0.1, Table 4-4). A significant interaction between
distance and edge age was irrespective of whether the species
showed a non-uniform distribution across the four distances
by combining the two types of edges (Gp00ied P < 0.1, Table
4-3) or by looking at old and new edges separately (Gold P <
0.1 and GNew p ^ 0.1, Table 4-4) .
Most of the fruiting individuals found at my study edges
were represented by few individuals, precluding the use of
Goodness of Fit Test to establish how distance from forest
edge affected their distribution. Instead, I used
information on their abundance (Table 4-5) to establish the
distance at which sparse species were found more often. I
excluded abundant and very abundant species from the analysis
and found that for the remaining three groups of plants
abundance and distance from the edge were not independent
(Test for Independence, x2 = 18.86, P = 0.004). Examination


BIOGRAPHICAL SKETCH
I was born in the midst of a summer in Los Angeles, CA,
to a Colombian couple. At that time my father was pursuing
his Ph.D. degree in Mathematics and my mother was taking care
of their two daughters. I got to spend a lot of time with my
mother and she would tell me stories about her family,
Colombia, and the activities of my father as a researcher.
She also took the time to answer my questions, many of which
had to do with the whys and hows of animal and plant life.
Later on it was my mother's turn to complete her degree in
the Social Sciences and for my father to take care of us. He
would tell us stories of his own invention. It was from one
of these that I learned that birds migrate and that not all
individuals succeed. My compassion for other organisms was
partially motivated by the little swallow that could never
make it through its long journey.
By the time my parents moved back to Colombia, and after
living in Mexico and Puerto Rico, I was conscious of my
fascination for the living world. I knew that I wanted to
unravel some of its misteries and one way to do so was to
study Biology. At the Universidad del Valle I had plenty of
opportunities to learn about biological principles, natural
history, and politics. I found myself editing environmental
188


77
across the pasture-forest edge, decreasing towards the forest
interior (Kapos 1989; Seizer 1992). More important, however,
were the differences found between the wet and dry season,
higher VPD values being recorded farther inside the forest
during the dry than during the wet season (Seizer 1992). At
the same lowland site, Kapos et al. (1993) compared the
carbon isotopic composition (313C) of leaves of two canopy and
two understory species and found that the 013C concentration
decreased from the edge towards the forest interior for the
understory but not for the canopy species. These changes
were more pronounced for Duauetia aff. flaoellaris
(Annonaceae) than for Astrocarvium sociale (Arecaceae). The
results described by Kapos et al. (1993) indicate that the
understory environment might be more sensitive to edge
creation than the canopy and that understory species vary in
their sensitivity to the factors that influence their
distribution.
At La Planada, increases in fruit abundance at the
forest edge could also be related to changes in soil
fertility as a result of increased litterfall during the dry
season, and deposition of volcanic ash at edges. La Planada
is influenced by several active volcanoes that release
andesitic ash (Mizota and van Reenwyk 1989, cited in van
Wambeke 1992) rich in nutrients (Shoji et al. 1993). These
air-borne particles might be deposited disproportionately
along pasture-forest edges. Some studies have shown that
edges alter the deposition of dry airborne material such that


Figure 6-3. Lump structure of Colombian montane frugivorous birds according to elevational
zone from forest (bottom) to paramo (top). Upper lowland (UL), lower montane (LM), upper
montane (UM), and pramo (P). Each box represents a lump and the space between the boxes
represent gaps in the distribution of body mass. The different shades indicate the
proportion of species falling within lumps: (1) 0-5, (2) 5-10, (3) 10-20, (4) 20-30, (5) SO
IS, (6) 45-60, and (7) 60-100 % of species. Vertical lines represent 0-5% of species.
Numbers on the right side represent number of species for the corresponding data set.
1 2 3 4 5 6 7


110
(Chapter 4). Capture rates of understory frugivorous birds
followed a similar trend up to 70 m from the edge. Fruit
abundance (measured as total number of fruits excluding palms
and as total number of fruiting individuals), differed across
pasture-forest edge depending on month (Chapter 4).
Frugivores, but not insectivores or nectarivores, changed in
the same fashion, as indicated by the significant distance x
month interaction.
Two other factors influenced the distribution of
frugivores across pasture-forest edge. First, at D4 fruit
abundance was the lowest, yet capture rates of frugivores
were the highest. This increase in capture rates was
probably influenced by the presence of two leks (Mionectes
striaticollis and Masius chrvsooterus) in the vicinity of D4
in one of my study edges (Hermgenes). It is impossible to
tell if the establishment of these leks was related to edge
creation 12 years ago. The presence of the leks, however,
certainly acts as an attractor, thus affecting the
distribution of M. striaticollis and M. chrvsooterus in one
of the edges. Second, the sharp increase in capture rates
of insectivores at D4 suggests that some structural feature
of the forest and/or resource covarying with structure
changed at D4 and that both frugivores and insectivores
responded to this. A gentler topography and the presence of
larger gaps at D4 when compared to the other distances may
result in changes in forest structure. Thus, as pointed out
by Wiens (1992), the distribution of organisms across edges


TABLE OF CONTENTS
ACKNOWLEDGMENTS ii
LIST OF TABLES vi
LIST OF FIGURES viii
ABSTRACT x
CHAPTERS
1 THE ROLE OF EDGES IN NEOTROPICAL MONTANE LANDSCAPES 1
2 DESCRIPTION OF STUDY AREA
Study Area 5
General Sampling Procedure 11
3 FROM FLOWERS TO SEEDLINGS: THE EFFECT OF EDGES AND
TREEFALL GAPS ON TWO UNDERSTORY SHRUBS, Palicourea
qikkQSfl AND Faramea affinis (RUBIACEAE)
Introduction 15
Methods 18
Analyses 24
Results 26
Discussion 34
4 UNDERSTORY FRUIT ABUNDANCE IN A NEOTROPICAL MONTANE
FOREST: THE INFLUENCE OF EDGES AND TREEFALL GAPS
Introduction 42
Methods 45
Analyses 49
Results 52
Discussion 72
IV


75
in terms of (1) the factors that influence fruit abundance
and (2) the consequences that these observed changes might
have on plants and frugivores.
Factors Influencing Fruit Abundance
Fruit abundance is regulated by abiotic and biotic
factors interacting in complex ways with the flowering and/or
fruiting stages of a plant's life cycle (e.g., Marshall and
Grace 1992). Abiotic factors that have a direct effect on
fruit numbers and fruit size are photoperiod and irradiance
(e.g., Auchter et al. 1926; May and Antcliff 1963; Jackson
and Palmer 1977; Mathai and Sastry 1988; Tombesi et al.
1994), temperature (e.g., Chaikiattiyos et al. 1994), water
availability (e.g., George et al. 1990), and nutrients (e.g.,
Stephenson 1992). Biotic factors include pollination,
predation of flowers, seeds, and fruits, and damage by
pathogens (Stephenson 1981). The importance of these factors
is likely to differ within and among species, depending on
habitat.
Irradiance is an important factor influencing fruit
abundance. Most work that supports this contention is based
on the observation that when irradiance increases in a forest
as a result of disturbance, so does fruit abundance (Halls
1973; Piero and Sarukhan 1982; Clark and Clark 1987; Agren
1988; Levey 1990). Although I did not measure irradiance
directly, my estimates of leaf area index (LAI) describe
indirectly the light environment at my edges. I found that


124
body masses.
Sample size strongly influences lump structure. In
data sets with a large number of observations a small value
of alpha might reveal a strong pattern of gaps whereas a
large value might reveal a weak pattern of lumps especially
at the lower end of the body mass range (Fig.2b). In this
situation, reducing alpha reduces the probability of
detecting gaps that might not exist (reduction of Type I
error). In small data sets a small value of alpha might not
reveal any pattern whereas a large alpha value might reveal
strong pattern. In this situation, increasing alpha reduces
the probability of not detecting gaps that exist (reduction
of Type II error). Thus, the gap/lump structure of a given
data set is determined by the chosen alpha level (Fig. 6-2b).
This interplay between sample size and the two types of
statistical error should be taken into account when comparing
multiple data sets. Lipsey (1990) gives an excellent
discussion of the importance of using different values of
alpha when detection of pattern is important. In this paper
and for simplicity I kept the alpha level constant within
each comparison. Depending on the data sets I used alpha
levels of 0.05 (within comparison average sample size >81,
range 30-395) and 0.1 (within comparison average sample size
<81, range 30-141).
I set up nested comparisons and derived four continuous
unimodal distributions. The four elevational zones were
compared using a null distribution generated from the


Ill
is influenced not only by distance from forest edge but also
by species characteristics, such as their behavior.
The distribution of nectarivores showed a sharp decrease
from edge to forest interior, even though the omnibus test
was not significant. In Brazil, Quintela's (1986) data show
the same trend but it was significant. In the lowlands of
Costa Rica, Blake and Loiselle (1991) found that capture
rates of nectarivores decreased from young second growth to
old second growth to forest. In Monteverde (Costa Rica)
cloud forest, Feinsinger et al. (1987) found that the
frequency of visits by hummingbirds decreased from large gaps
to small gaps to forest. At La Planada, the distribution of
hummingbird-pollinated plants showed a significant decrease
from edge towards forest interior (Goodness of Fit Test, y2 =
23.01, df = 3, P < 0.05), suggesting that resources might be
responsible for the observed trend in nectarivores. This at
least can explain why very sparse and sparse nectarivores
were captured more often at D1 than at the other distances.


EDGES
FRUITS, FRUGIVORES, AND SEED DISPERSAL
IN A NEOTROPICAL MONTANE FOREST
By
CARLA RESTREPO
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR IN PHILOSOPHY
UNIVERSITY OF FLORIDA
1995

ACKNOWLEDGMENTS
From the mountains of La Planada to the concrete of
Bartram Hall I have had the fortune of interacting with
people who have enriched my life in fundamental ways. As I
see it now, this dissertation proved to be a point of
convergence of people, situations, and ideas that is leading
me to new discoveries. I thank the members of my committee,
Douglas J. Levey, H. Jane Brockmann, John Ewel, C. S. "Buzz"
Holling, and Frank Slansky, for their continuous
encouragement and support throughout the two years I spent in
Bartram Hall trying to figure out which direction I wanted my
dissertation and myself to go.
It is not easy to collect data over 12 ha of steep land,
nor to live in isolation. I am particularly grateful to
Natalia Gmez, Sylvia Heredia, and Arlex Vargas for their
help and support in the field. I am indebted to the
neighbors of La Planada, in particular to Adolfo Ortega,
Abelardo Nastacuaz, Demetrio Guanga, Pacho Guanga, Amparo
Oliva, and the GELISI, for sharing their life with me. At
various points during this research I benefited from help
provided by Marta Baena, Girleza Ramirez, Ivan Jimenez,
Natalia Arango, Luis F. Citelli, Omaira Ospina, Mara de
Restrepo, and Paul Marples. By the end of the field season
11

several people were instrumental in helping to put together
little pieces of my puzzle. J. H. Cock and A. P. Hernandez
from CENICANA lend me the LAI-Canopy Analizer. J. Luteyn, T.
G. Lammers, P. E. Berry, D. Froding, B. Hammel, J. J. Wurdak,
J. Kress, C. Taylor, L. E. Skog, L. R. Landrum, A. M. W.
Mennega, J. S. Miller, and T. Croat kindly identified the
plant material I collected at La Planada. P. Kubilis and C.
Steible provided statistical advice when most needed. L.
Walz prepared maps. P. Amezquita counted pollen tubes. I am
grateful for their valuable help.
This project was crafted some years ago with the input
provided by P. Feinsinger, my former advisor. He presented
me with alternative routes that certainly proved fruitful. I
am particularly grateful for this.
This study was funded by the Fundacin para la Promocin
de la Investigacin y la Tecnologa, Banco de la Repblica,
Colombia and the Wildlife Conservation Society (WCS).
iii

TABLE OF CONTENTS
ACKNOWLEDGMENTS ii
LIST OF TABLES vi
LIST OF FIGURES viii
ABSTRACT x
CHAPTERS
1 THE ROLE OF EDGES IN NEOTROPICAL MONTANE LANDSCAPES 1
2 DESCRIPTION OF STUDY AREA
Study Area 5
General Sampling Procedure 11
3 FROM FLOWERS TO SEEDLINGS: THE EFFECT OF EDGES AND
TREEFALL GAPS ON TWO UNDERSTORY SHRUBS, Palicourea
qikkQSfl AND Faramea affinis (RUBIACEAE)
Introduction 15
Methods 18
Analyses 24
Results 26
Discussion 34
4 UNDERSTORY FRUIT ABUNDANCE IN A NEOTROPICAL MONTANE
FOREST: THE INFLUENCE OF EDGES AND TREEFALL GAPS
Introduction 42
Methods 45
Analyses 49
Results 52
Discussion 72
IV

5 EDGES AND UNDERSTORY BIRDS IN A NEOTROPICAL MONTANE
FOREST
Introduction 82
Methods 84
Data Analysis 86
Results 90
Discussion 105
6 FRUGIVOROUS BIRDS IN FRAGMENTED NEOTROPICAL MONTANE
FORESTS: LUMP STRUCTURE IN BODY MASS
Introduction 112
Methods 114
Results 126
Discussion 138
7 CONCLUSIONS 148
APPENDIX A
Plant species fruiting in the understory at the
Reserva Natural La Planada 151
APPENDIX B
Bird species captured in the understory at the Reserva
Natural La Planada 161
LITERATURE CITED 166
BIOGRAPHICAL SKETCH 188
v

LIST OF TABLES
Table page
2-1 Water balance for the Reserva Natural La Planada
9
2-2 Characteristics of the edges included in this
study 13
3-1 Results of Mixed Factorial ANOVAs for pollen tube
production in Palicourea gibbosa and Faramea affinis
in relation to distance from forest edge 27
3-2 Pollen tube production in Palicourea gibbosa and
Faramea affinis in relation to distance from forest
edge 28
3-3 Results of Mixed Factorial ANOVAs for fruit set
and fruit damage by insects in Palicourea gibbosa and
Faramea affinis in relation to distance from forest
edge 30
3-4 Results of Mixed Factorial ANOVAs for seed
predation and seed germination in Palicourea gibbosa
and Faramea affinis in relation to distance from
forest edge 35
3-5 Results of Mixed Factorial ANOVAs for relative
growth and leaf production rates in Palicourea gibbosa
and Faramea affinis in relation to distance from
forest edge 38
4-1 Results of Repeated Measures ANOVA on leaf area
index (LAI) for old edges 55
vi

4-2 Results of Mixed Factorial ANOVAs on fruit
abundance across pasture-forest edge 59
4-3 Results of Goodness of Fit Test on the number of
fruiting individuals of abundant species across
pasture-forest edge 68
4-4 Results of Replicated Goodness of Fit Test on the
number of fruiting individuals across pasture-forest
edge in old and new edges 69
4-5 Distribution of number of fruiting individuals
across pasture-forest edge based on species abundance... 70
4-6 Summary of results of changes in fruit abundance
across the pasture-forest edge 74
5-1 Results of ANOVAs for Mixed Factorial Design on
capture rates of understory birds 92
5-2 Results of Goodness of Fit Test on the number of
bird captures of abundant species across pasture-
forest edge 101
5-3 Results of the Replicated Goodness of Fit Test on
the number of bird captures across pasture-forest
edge 103
5-4 Distribution of bird captures across pasture-
forest edge based on species abundance 104
6-1 Description of sites included in lump analyses of
body mass of frugivorous birds 119
vi 1

LIST OF FIGURES
Fiq.urg oms.
2-1 Location of study area and edges included in the
study 6
2-2 Distribution of mean monthly rainfall and
temperature at the Reserva Natural La Planada 8
2-3 Edge indicating general sampling design 14
3-1 Fruit set in Palicourea aibbosa as influenced by
distance from forest edge 31
3-2 Seed germination, seed predation, seedling growth
and, leaf production in Palicourea gibbosa and Faramea
affinis as influenced by distance from forest edge 36
3-3 Seed germination, seed predation, seedling growth
and, leaf production in Palicourea aibbosa and Faramea
af finis as influenced by habitat 39
4-1 Leaf area index (LAI) across the pasture-forest
edge at old edges 54
4-2 Leaf area index (LAI) and fruit abundance 57
4-3 Variation in fruit abundance at the Reserva
Natural La Planada in relation to distance from forest
edge 60
4-4 Variation in fruit abundance at the Reserva
Natural La Planada in relation to edge age and
distance from forest edge 62
4-5 Variation in fruit abundance at the Reserva
Natural La Planada in relation to habitat and distance
from forest edge 63
viii

4-6 Variation in fruit abundance at the Reserva
Natural La Planada in relation to edge age and month ... 65
4-7 Variation in fruit abundance at the Reserva
Natural La Planada in relation to distance from the
forest edge and month 66
5-1 Variation in the distribution of understory birds
at La Planada in relation to distance from the forest
edge 93
5-2 Variation in the distribution of understory birds
at La Planada in relation to distance from the forest
edge and edge age 95
5-3 Variation in the distribution of understory birds
at La Planada in relation to edge age and month of the
year 96
5-4 Variation in the distribution of understory birds
at La Planada in relation to distance from forest edge
and month of the year 98
6-1 Montane habitats of Colombia and sites included
in lump analysis of body mass of frugivorous birds 116
6-2 Lump analysis for body mass of frugivorous birds
of Colombian upper lowland tropical forests showing
(a) body mass distribution vs. rank order and (b) rank
size-ordered body mass distribution versus gap rarity
indexes 122
6-3 Lump structure of Colombian montane frugivorous
birds according to elevational zone 127
6-4 Lump structure of Colombian frugivorous birds
from sites cover mostly by forest to sites highly
transformed by human activities within the upper
lowland zone 13 0
6-5 Lump structure of Colombian frugivorous birds
from sites cover mostly by forest to sites highly
transformed by human activities within the lower
montane zone 1 133
6-6 Lump structure of Colombian montane frugivorous
birds from sites cover mostly by forest to sites
highly transformed by human activities within the
upper montane zone 136
6-7 Relationship between species richness and lump
structure in landscapes of variable complexity 145
IX

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor in Philosophy
EDGES, FRUITS, FRUGIVORES, AND SEED DISPERSAL
IN A NEOTROPICAL MONTANE FOREST
By
Carla Restrepo
December, 1995
Chairman: Douglas J. Levey
Major Department: Zoology
Edges resulting from natural or human disturbances
influence the distribution of organisms as well as ecological
processes. One such process is seed dispersal, which in turn
may influence may the location of edges through time and even
the entire structure of landscapes. In the Reserva Natural
La Planada, Colombia, I investigated how edges resulting from
human activities influenced seed dispersal. In particular, I
examined how distance from forest edge, in combination with
edge age and treefall gaps, could affect recruitment rates,
fruit abundance, seed movement, and the distribution of birds
in the understory of this neotropical montane forest.
Sampling took place at three old (>40 yr) and three new (<15
yr) edges and within each edge at four distances (0-10, 30-
40, 60-70, 190-200 m) from the pasture-forest edge.
x

Edges influenced Palicoura aibbosa and Faramea affinis.
the two most common understory plants, at various stages of
their life cycle. Seed predation and seed germination but
not relative growth or leaf production rates changed across
the pasture-forest edge. The latter, however, were
influenced by treefall gaps. At the community level, fruit
abundance and the distribution of understory birds changed
across the pasture-forest edge in complex ways that not
always reflected changes due to the presence of edges. This
was demonstrated by the fact that (1) two-way interactions
between distance, edge age, treefall gaps, and month were
significant, and (2) response variables describing fruit and
bird abundance at the community level did not show the same
trends.
Edges influenced fruit-frugivore interactions at the
level of forest stands but also at the level of entire
landscapes as demonstrated by an analysis of body mass
distribution of frugivorous birds as a function of ecosystem
fragmentation. With more edges, entire groups of birds with
similar body mass (termed "lumps") disappeared.
Nevertheless, the distribution of body mass, i.e., lump
structure, remained almost intact under certain land use
types. This work suggests that at broad scales edges
influence frugivorous birds and, as a result, seed dispersal.
xi

CHAPTER 1
THE ROLE OF EDGES IN NEOTROPICAL MONTANE LANDSCAPES
Edges constitute a common feature of neotropical montane
landscapes. Complex topography and climate in conjunction
with natural disturbances, ranging from treefall gaps (Murray
1988; Lawton and Putz 1988; Samper 1992), and landslides
(Garwood et al. 1979; Lawton and Dryer 1980; Gentry 1992&),
to mudflows, have given rise to a heterogeneous landscape in
which edges bound the disturbed areas. Over evolutionary
time, the dynamic and heterogeneous character of these
landscapes may have resulted in the unusually high levels of
biodiversity that characterize neotropical montane ecosystems
(Terborgh and Winter 1983, Terborgh 1985, Gentry 1986,
1992a)
Superimposed on this natural heterogeneity is that
resulting from human activities. In areas where favorable
conditions prevail patches of forest are immersed in an
agricultural and urban matrix. Conversely, in areas where
unfavorable conditions limit the development of economic
activities, fields, second growth, and urban areas are
immersed in a forest matrix. In both situations the areas
modified by human activities are bounded by edges. Over
shorter time scales than those defined by large-scale natural
disturbances, changes in land use have led to extinctions and
1

2
invasions by an unknown number of species (Henderson et al.
1991, Rattan 1992) .
Even though edges are a prominent feature of neotropical
mountains, little is known about how they influence landscape
pattern. My dissertation focuses on edges and how they
influence seed dispersal through fruit-frugivore
interactions. Changes in the nature of fruit-frugivore
interactions can provide information on the persistence of
edges through time, depending on the ability of plants to
produce and disperse their seeds. On the other hand, changes
in the composition and number of individuals across edges can
provide information on edge structure and productivity.
A common thread among studies focusing on edges is the
lack of a common pattern, without which it is difficult to
propose underlying causes or even consequences. Each new
study has added details at the cost of finding generalities
from which testable predictions can be made. In my
dissertation I followed two different approaches to the study
of edges. Both had as a central theme that of fruit-
frugivore interactions. The first generated detailed
information on fruit-frugivore interactions at the level of
forest stands. Nevertheless, this information precludes
generalizations about the role of edges in neotropical
montane forests due to the intrinsic characteristics of my
study site. The second generated consistent patterns at the
level of entire landscapes that differed in degree of
transformation by human activities. This portion of the

3
study, therefore, might be useful in generating testable
hypotheses that could guide future studies.
My study area and general sampling procedure are
described in detail in chapter 2. I often refer to this
chapter because the studies I describe in chapters 3 to 5
were based on the same sampling procedure.
In chapter 3 I focus on two understory plants, Palicoura
aibbosa and Faramea affinis (Rubiaceae) and ask how edges may
influence different components of the life cycle of these two
species and how this may influence recruitment. These two
species are the most common understory species of my study
site and are bird-dispersed. In particular, I looked at how
the combined effect of distance from forest edge and presence
of treefall gaps affect pollination, fruit set, seed
predation, seed germination, and seedling growth.
In chapter 4 I take a broader approach to the study of
edges and their influence on fruit-frugivore interactions, by
focusing on the assemblage of understory plants. I ask
whether fruit abundance changes across pasture-forest edges
and how treefall gaps and season modify such effects. I look
at changes in fruit abundance for the entire assemblage of
understory plants and for individual species. Changes in
fruit abundance can affect the distribution of birds feeding
on them and thus seed dispersal across edges.
In chapter 5 I keep the previous approach to the study
of edges and their influence on fruit-frugivore interactions
but ask how edges influence the distribution of understory

4
birds, particularly frugivores. I compare the distribution
of nectarivorous, insectivorous, and frugivorous birds across
the pasture-forest edge to elucidate possible mechanisms
underlying observed patterns.
I take a completely different approach in chapter 6 to
the study of edges and their influence on fruit-frugivore
interactions. Instead of asking how the distribution of
frugivorous birds is affected by the creation of edges within
forest stands, I ask how the distribution of frugivorous
birds is affected by edges within whole landscapes. I
compare changes in the distribution of frugivorous birds
across sites that have been modified in various ways by human
activities. This approach provides the basis for some
generalizations and the formulation of testable hypotheses
for future work.
Chapter 7 is the place for synthesis and speculation. I
emphasize that edges are part of landscapes that are in
continuous change and thus have to be seen as dynamic, not
fixed entities of landscapes. Seed dispersers and seeds move
between forest interior and edge, and sparse species of
plants and frugivorous birds are found more often at edges
than at forest interior. These results suggest a critical
role of edges in landscapes subject to change.

CHAPTER 2
DESCRIPTION OF STUDY AREA
Study Areft
I conducted this study at Reserva Natural La Planada and
Finca El Bosque, located in the municipality of Ricaurte,
department of Nario, SW Colombia (7800'W and 110'N) (Fig.
2-1). Both localities lie on the western slope of the Andes
at 1,800 m. The biota of La Planada and its surroundings is
one of the most diverse of the northern Andes (Terborgh and
Winter 1983; Orejuela 1987) and twenty percent of the plant
and animal species reported for the area are endemic
(geographical range <50,000 km^, Terborgh and Winter 1983).
Observations were concentrated in the NW and W portions
of La Planada and El Bosque, respectively. These two areas
lie on the watershed division of the Miraflores and Pialapi
rivers. Colonization of this area started in the early 1940s
and proceeded from the bottom of the valleys (1,200 m) to the
top of the mountains (1,800 m) which are still mostly
covered by forest. Small sugarcane plantations, transient
corn fields, pastures, fallows, and second-growth vegetation
are embedded in the forest matrix. These disturbed areas are
concentrated at the bottom of the valleys and range from 1-9
5

Figure 2-1. Location of study area and edges that were sampled. Forest
is represented by shaded area (based on a 1981 aerial photograph,
approximate scale 1:26,000)

7
ha (mode 3 ha) (Gmez and Palau 1994) The patches of forest
are connected by strips of forest that have been left along
streams, steep sloptes, and mountain ridges.
The natural disturbance regime is varied. At small
scales, treefall gaps, which occur mostly during the rainy
season, are common (Samper 1992). At larger scales
landslides, ash rain (e.g., January 1993 volcanic eruption),
and strong winds, the latter resulting in the defoliation of
large areas (e.g., August 1993), affect whole landscapes.
Unpublished climatological records of La Planada (1985-
1994) show a mean annual rainfall and temperature of 4,437 mm
and 19.2C, respectively (Fig. 2-2). Rainfall is distributed
in two wet seasons, interrupted by a mild dry (February-
March) and a strong dry (June-August) season (Fig. 2-2).
Based on these data, La Planada can be classified as a
transitional life zone between tropical premontane rain and
wet forest (Holdridge 1967). An important climatological
feature of La Planada and its surroundings is the presence of
afternoon mist during most of the year.
The water balance for La Planada shows that on average
every month has a surplus of water which is lost as soil
runoff (Table 2-1). Nevertheless, in some years there might
be months in which there is a water deficit. This can
explain why during my study period, in particular during the
months of July-August of 1993, many plants lost their leaves.

8
JFMAMJJASO ND
Month
Figure 2-2. Distribution of mean monthly rainfall (bars) and
temperature (open circles) (1985-1994) at the Reserva Natural
La Planada (unpublished data Reserva Natural La Planada).
Filled circles represent average values of rainfall for 1992-
1993 .
Temperature (C)

Table 2-1. Water balance for Reserva Natural La Planada, transitional life zone between
premontane rain to wet forest (assuming available moisture = 443.7 mm). For details on how
to calculate different variables see Ewel and Madriz (1968). Elevation 1,800 m. Based on
climatological records from 1985-1995.
j
F
M
A
M
J
J
Biotemperature
C
18.9
19.3
19.5
19.6
19.6
19.3
18.5
Potential evapotranspiration (P.ET)
mm
95
88
98
94.8
98
93.4
92.5
Precipitation
mm
460.1
360.7
406.5
451.1
405.3
292.2
148.5
Actual ET
mm
95
88
98
94.8
98
93.4
92.5
Water surplus
mm
365.1
272.7
308.5
356.3
307.3
198.8
56.01
Soil moisture change
mm
0
0
0
0
0
0
0
Moisture available in soil end of month
mm
443.7
443.7
443.7
443.7
443.7
443.7
443.7
All runoff
mm
365
273
309
356
307
199
56
Soil moisture deficit
mm
0
0
0
0
0
0
0
Precipitation deficit
mm
0
0
0
0
0
0
0
Total moisture deficit
mm
0
0
0
0
0
0
0

Table 2-1. (continued)
A
S
0
N
D
TOTAL
Biotemperature
C
18.8
18.9
19.2
19.2
19.1
19.2
Potential evapotranspiration (P.ET)
mm
94
91.5
96
92.9
95.5
1129.6
Precipitation
mm
163.2
287.8
527.5
482
491.6
4437.4
Actual ET
mm
94
91.5
96
92.9
95.5
1129.6
Water surplus
mm
69.17
196.3
431.5
389.1
396.1
3347
Soil moisture change
mm
0
0
0
0
0
Moisture available in soil end of month
mm
443.7
443.7
443.7
443.7
443.7
All runoff
mm
69.2
196
431
389
396.1
3347
Soil moisture deficit
mm
0
0
0
0
Precipitation deficit
mm
0
0
0
0
0
0
Total moisture deficit
mm
0
0
0
0
0

11
The La Planada forest develops on well drained soils
(Dystrandept) derived partially from volcanic material,
that are moderately acid, with a sandy to clay loam texture
(De Las Salas and Ballesteros 1986). The canopy height
(average 22 m) and the basal area (dbh > 4 cm; 33.4 m2/ha) of
the forest are low and epiphytic and hemiepiphytic plants are
very abundant (De Las Salas and Ballesteros 1986; Gentry
1988) Plants (dbh > 2.5 cm) in a 0.1 ha plot were
represented by 112 species (Gentry 1992a). The most
important trees on this plot were Ouararibea sp., Elaeaia
sp., BierQnymft sp., Alghgrngfr sp., Billia colombiana. Inga
sp., Otoba sp., and Ocotea sp.; the most important treelets
and shrubs were Faramea elegans. Prestoea cf. purpurea.
Ajghgngi? sp., Geonoma weberbaueri. Palicourea qibbosa and
Miconia sp.; and the most common epiphytes were Philodendron
cf. sggndgns, Spheraedenia stevermarkii and Psammisia sp. (A.
Gentry, unpublished data).
General Sampling Procedure
I chose six sites to evaluate how edge age and distance
from the edge towards the forest interior influence various
components of fruit production and seed dispersal. These
sites, hereafter referred to as edges, were active or
recently abandoned pastures contiguous with forest. Thus, at
most edges there was a sharp delineation between forest and
the adjacent pasture (Table 2-2). Four edges lay at La
Planada boundaries (Marcos, Climo I, Climo II, and

12
Hermgenes), a fifth edge was located within the reserve
(Pialapi), and the sixth edge was located at El Bosque
(Acantayac) (Fig. 2-2). Climo I and Climo II were 400 m
apart on the same edge, but because of differences in the
weeding regime of the pasture and use of the forest I
reasoned that they could represent two independent sampling
units. Independence of these two sampling points was
particularly important for the part of the work evaluating
the influence of edges on the distribution of understory
birds (Chapter 5). Recapture frequency between these two
sites was <4%, supporting the assumption that these two
points represented two independent sampling units.
Three edges, Climo I, Climo II, and Pialapi, were
created around 1950 (old edges), when colonists first arrived
in the area and cleared the forest to establish pastures.
The other three edges, Marcos, Hermgenges, and Acantayac,
were created around 1982 (young edges), the year La Planada
was established as a private reserve (Table 2-2). At the
beginning of the study, I placed barbed wire fences along the
edges to keep cattle from penetrating into the forest. I
sampled these edges between March 1992 and March 1994.
At each edge I worked in an area of 100 X 200 m (2 ha)
and established four strips (100 X 10 m) running parallel to
the edge. These strips were located at four different
distances from the forest edge towards the forest interior:
0-10 m (Dl), 30-40 m (D2), 60-70 m (D3), and 190-200 m (D4).

Table 2-2. Characteristics of the edges included in this study.
Edge
T1
O2
A3
C4
PS5
G6
o
G
G
00
Climo I
S
40 NE
1953
H
6.0
4.2
Cattle ranching;
pasture
Extraction of
poles; cattle
palm hearts and
grazing
Climo II
S
19 NE
1953
M
7.6
7.6
Cattle ranching;
pasture
Extraction of
poles; cattle
palm hearts and
grazing
Pialapi
F
8 NE
1950
L
9.9
Trail to Pialapi;
second growth
Selective logging 40 yr ago
Acantayac
F
24 NE
1981
M
3.0
11.0
Cattle ranching;
pasture
Extraction of
poles; cattle
graves
palm hearts and
grazing; ancient
Hermgenes
M
68 NE
1982
H
3.0
10.0
Cattle ranching;
pasture
Extraction of
palm hearts
Marcos
S
59 NE
1982
H
1.0
7.5
Cattle ranching;
pasture
Extraction of
poles; cattle
to Pialapi
palm hearts and
grazing; old path
1 T = Topography; S = steep, M = moderate, F = flat
2 0 = Orientation (position of edges regarding the cardinal points)
3 A = Age (year the forest was clear cut)
* C = Edge contrast; H = high, M = moderate, L = low
5 PS = Size of disturbed area (ha)
6 G = Percent of sampling area covered by gaps
7 CU = Use of clear cut area
FU = Use of forested area

14
Figure 2-3. Edge indicating general sampling design. Shaded
strips represent four distances where sampling took place: D1
(0-10 m), D2 (30-40 m), D3 (60-70 m), and D4 (190-200 m). In
strip DI I show the distribution of subquadrats (1-4) where
fruit abundance was evaluated. In strip D2 I show the
distribution of each of three pirs of mistnets (perpendicular
dark lines). On the left side of the figure I illustrate the
orientation of one transect along which LAI was measured.

CHAPTER 3
FROM FLOWERS TO SEEDLINGS: THE EFFECT OF EDGES AND TREEFALL
GAPS ON TWO TROPICAL UNDERSTORY PLANTS, Palicourea aibbosa
AND Faramea affinis (RUBIACEAE)
Int reaction
Recruitment rates in plant populations are influenced by
success at all stages of the life cycle (Harper 1994). Which
stages limit recruitment depends on the requirements of
individuals at each stage and the spatial distribution of
resources (e.g., Sork 1983, Martinez-Ramos and Soto-Castro
1993, Osunkoya et al. 1994). In tropical areas, for example,
treefall gaps influence the distribution of resources at
small scales (Denslow and Hartshorn 1994), whereas landslides
and forest clearings do so at large scales (Guariguata 1990,
Dalling and Tanner 1995). Also, seed predation (e.g., Schupp
1988, Schupp and Frost 1989, Samper 1992), seedling
establishment (e.g., King 1990), plant growth (Sizer 1992,
Dalling and Tanner 1995), fruit production (Levey 1990) and
seed dispersal (Murray 1988) change as a result of such
disturbances. Little is known, however, about the combined
effect of small and large-scale disturbances on the various
stages of the life cycle of plants or how such effects may
determine which stage limits recruitment.
15

16
For the most part, studies evaluating the effect of
human disturbances on recruitment rates in tropical plants
have focused on single stages of a plant's life cycle (e.g.,
MacDougall and Kellman 1992, Seizer 1992, Burkey 1993). By
looking at several species it has been possible to establish
patterns and understand the factors underlying the responses
of particular stages (e.g., Sizer 1992). This approach
should be complemented with studies focusing on single
species to establish the relative contribution of a given
stage to the life cycle of a plant (Ellison et al. 1993). A
more complete understanding of factors that limit recruitment
either in forest fragments or nearby disturbed areas must
consider what happens to plants in all stages of their life
cycle.
At a neotropical montane site fruiting individuals of
PsiliCQurea gibbosa and Faramea affinis were not distributed
uniformly across pasture-forest edges (Chapter 4). Here I
report results of a study that examined how several stages of
the life cycle of these two understory plants were influenced
by distance from forest edge, edge age, and treefall gaps.
In particular, I wanted to determine how pollination, fruit
set, seed dispersal, seed predation, germination, and
seedling growth could result in the observed distribution of
P. gibbosa and F. affinis across the pasture-forest edge.

17
The Species
Palicourea qibbosa Dwyer and Far ame, a f finis belong to
the Rubiaceae, one of the most speciose and common families
of neotropical montane forests (Taylor 1989, Gentry 1992a.) .
Palicourea aibbosa shrubs reach 4 m and are found at middle
elevations from Panama to Ecuador (Dwyer 1980; C. Taylor,
personal communication), growing in second growth and mature
forest (Arias 1993). Faramea affinis treelets reach 9 m and
grow in old second growth and mature forests. In a 0.1 ha
plot at my study site, £. qibbosa and £. affinis were the
most common species (dbh > 2.5 cm) in the understory (A.
Gentry, unpublished data).
Palicourea qibbosa exhibits three flowering periods per
year. Its yellow flowers are visited mostly by hummingbirds,
including Qgrg^tV? underwoodii. Aglaiocercus coelestis. and
Haploohaedia luqens (Arias 1993). Fruits of £. qibbosa are
dark blue to purple, 7 mm long, and are presented in terminal
yellow, erect infructescences containing up to 50 fruits.
They contain 1-2 seeds, 5.0 x 4.9 mm. Palicourea qibbosa
seeds are dispersed by birds, including Mvadestes ralloides.
PiprgQla riefferj, AUepete? brunneinucha. Masius
chrvsopterus. and Tanqara arthus (C. Restrepo and N. Gomez,
unpublished data).
Faramea affinis exhibits two flowering periods per year.
Its tubular, purple flowers are visited by hummingbirds,
including Coeliqena wilsoni (Samper 1992). Fruits are blue

18
and are presented in terminal, pendant, green
infructescences, containing a maximum of 3 fruits. They
measure 20 x 18.2 mm and contain a single seed, 10.4 x 7.6 mm
(Samper 1992). Seeds of F. affinis are dispersed by a
different set of birds, Andiqena laminirostris, pipauqus
crvotolophus. Pipreola riefferi, Semnornis p^mph^gtings, and
Troaon personatus (Restrepo 1990, Beltrn 1991, Samper 1992).
Methods
I evaluated the combined influence of edges and treefall
gaps on several stages of the life cycle of Palicourea
aibbosa and Faramea affinis by sampling individuals and
conducting experiments at six edges (three old and three new)
and at four distances from forest edge towards forest
interior (0-10 m, 30-40 m, 60-70 m, and 190-200 m) (Chapter
2). Depending on the stage of the life cycle I was
examining, I modified the basic sampling design described in
Chapter 2. This was due to logistic constraints, including
accessibility of the edges.
Beginning in March 1992 I tagged all individuals <. 2 m
tall in flower and/or in fruit, and classified them as being
in gap or interior. An individual was classified as in gap
if it was within a gap (sensu Brokaw 1982) or located < 2 m
from a gap edge, and as interior if it was located > 2 m from
the edge of the nearest treefall gap at the time the study
began. I continued tagging individuals throughout the study
period as new individuals flowered. For each individual I

19
marked all inflorescences and infructescences and followed
them over the entire study period. I monitored individuals
for the presence of inflorescences and infructescences on a
biweekly basis during the first 6 months (March 1992-August
1992) and on a monthly basis the following 11 months
(September 1992-July 1993).
Pollination
To determine the influence of edges on pollination
success I looked at pollen tube production. From June 1993
until November 1993 I checked flowering individuals for four
consecutive days to collect an average of 10 flowers per
individual. These individuals represent a subset of those
that were monitored over the 16-month period. I dissected
the flowers and fixed the styles in formalin-acetic acid
(FAA) to examine pollen tubes. Pollen tubes were stained
(Martin 1959, Feinsinger et al. 1992) and counted under an
epifluorescent microscope. Styles were processed by P.
Amezquita at the Universidad de Santiago de Compostela,
Spain. Pollen tube production per individual was expressed
as the percentage of flowers with pollen tubes (F) and as the
average number of pollen tubes per flower (P).
Fruit Set
For each new inflorescence I counted the number of
flower buds and followed them until fruits developed and
ripened. I expressed fruit set as the percentage of unripe

20
fruits in relation to the number of flower buds and as the
percentage of ripe fruits in relation to the number of unripe
fruits counted over the entire study period for each
individual. I present results for P. cribbosa only.
Fruit Damage by Insects
At the same time I monitored infructescences for unripe
and ripe fruits, I recorded two types of fruit damage by
insects: damage to seeds by wasps (Hymenoptera: Chalcidoidea)
and removal of pulp by ants (Hymenoptera: Formicidae:
Ponerinae). The former could be recognized by exit holes
left by newly emerged adults and the latter by bites taken
from fruits. These two types of fruit damage were the most
common ones for these two understory plants. I expressed
seed and fruit damage as the proportion of unripe fruits
exhibiting one of the two types of damage in relation to the
total number of unripe fruits produced by an individual over
the entire study period. I present results for P. cribbosa
only.
Seedling Growth and Leaf Production
I monitored seedlings of Palicourea aibbosa and Faramea
affinis at each of three distances (0-10, 30-40, 60-70 m) at
three old edges (Climo I, Climo II, and Pialapi) to
establish the combined effect of distance from forest edge
and treefall gaps on seedlings growth and leaf production
(Fig. 2-1, Table 2-2, Chapter 2). In May 1992 I located

21
seedlings of P. aibbosa and F. affinis. I placed by their
side a stick with a piece of flagging tape with a distinctive
number for each seedling. I recorded whether seedlings were
growing in treefall gaps (n = 201 and n = 213, £. aibbosa and
F. affinis. respectively) or intact forest
(n = 210 and n = 221) To standardize measurements I marked
the stems of each seedling with yellow vinyl paint (ca. 1.5
cm above soil surface) and the youngest pair of leaves with
threads of flagging tape tied around the petioles. With
calipers I took a first measurement of the seedling's height
from the yellow mark to the base of the meristem and I
repeated this procedure five times between May 1992 and
October 1993. I also recorded and marked new pairs of
leaves.
Seedling growth rate (GR) is expressed as the increment
in height between the first (hn) and the last measurement
(hn+l) [GR = (hn+i hn/tn+i tn)*(30 days/month)] (Seizer
1992). Leaf production rate (LPR) is expressed as the number
of new leaves produced between the first (ln) and last (ln+l)
period [LPR = (ln+l ln/tn+l tn)*(30 days/month)].
Field Experiments
Experiments on seed predation and seed germination of £.
aibbosa and £. affinis were performed at three edges
(Hermgenes, Climo I, and Climo II) and that on fruit
removal (£. aibbosa) at two edges (Hermgenes and Climo I)
(Fig. 2-1, Chapter 2). Even though these edges represent two

22
different ages (Table 2-1) and edge age is known to influence
the effect of distance on vegetation (e.g., Williams-Linera
1990), I chose them to conduct this work because they were
close enough to allow frequent monitoring of seeds and
fruits. At each of four distances from forest edge (0-10,
30-40, 60-70, and 190-200 m) (Fig. 2-3, Chapter 2) I mapped
the treefall gaps and randomly chose 4 of them. At each
distance I paired each treefall gap location with an intact
forest location.
Sged._pr.edation an In the seed predation and seed germination experiments I
placed an aluminum tray (15 x 7 cm) in each gap and interior
site. I punctured the trays to prevent water from
accumulating, filled them with soil, and positioned them
flush with ground level. I placed 10 seeds of £. cribbosa and
5 seeds of £. affinis in different trays. Seeds were
obtained from ripe fruits, and those showing damage by
insects were discarded. In total I used 1,920 seeds of £.
gibbosa and 960 of F. affinis. I placed 92 trays containing
seeds for each edge/species/experiment (32 trays) and
simultaneously ran the germination and predation experiments
for each species within each edge. The trays containing the
seeds sown to evaluate changes in germination rates were
covered with galvanized mesh (5x5 mm) to protect seeds from
vertebrates.
I checked trays on a weekly basis and counted the number
of seeds remaining and the number of seeds germinated, i.e.,

23
seeds in which the hypocotyyl was visible (ca. 3 mm long).
The seed predation experiment for £. aibbosa lasted for 5
days (July 1993-August 1993) and the seed germination
experiment for 105 days (July 1993-November 1993). The seed
predation experiment for £. affinis lasted for 105 days
(August 1993-December 1993) and the seed germination
experiment for 252 days (August 1993-April 1994) I
concluded the seed germination experiments when 90% of the
seeds had germinated and the seed predation experiments when
no more seeds were being removed. I assumed that seeds
removed from the trays were taken by vertebrates and that
this constituted predation.
Fruit rempv^l
In this experiment I placed eight artificial shrubs per
distance, 4 at each gap and interior site, for a total of 32
artificial shrubs per edge. Each shrub consisted of a 1.5-m-
tall bamboo stick to which I attached an artificial
infructescence resembling that of £. aibbosa. The
artificial infructescences consisted of a 15-cm-long wooden
rod from which four pairs of tooth picks extended. The rods
and tooth picks were dyed bright yellow, and at the end of
each tooth pick I inserted a recently collected ripe fruit of
£ aibbosa. I ran the fruit removal experiment at each edge
for four consecutive days. On the morning of the first day
(0700) I inserted fresh fruits of £. aibbosa and 24 hours
later recorded the number of fruits missing and bitten by
ants. All fruits were changed every 24 hours to start a new

24
run of the experiment. I ran this experiment from June 26
1993 to July 3 1993.
Analyses
I analyzed data with ANOVAs for Mixed Factorial designs
(Girden 1992). The full design (edge age, distance from
forest edge, and treefall gaps) was set up as a split-split-
plot design (Winer et al. 1991). The factors of interest
were edge age, distance from the edge, and habitat. The
edges that I sampled for each level of edge age (old and new)
were chosen at random and represented the plot unit. In
turn, each edge was divided into four strips (distances from
the forest edge towards the forest interior) representing the
subplot units. Randomization of the levels of the distance
factor was restricted but because the strips were separated
in space and I analyzed responses from nonmobile organisms I
assumed they represented independent subsampling units.
Finally, individuals were classified according to habitat as
gap or intact forest, the latter representing the sub
subplots .
The design for the seed predation, seed germination, and
fruit removal experiments was set up as a split-plot design
with one repeated measure (Winer et al. 1991). The factors
of interest were distance from the edge, habitat, and time,
with time being the repeated measure. In the seed predation-
seed germination and fruit removal experiments time was
represented by weeks and days, respectively. In my design

25
distance (D1 to D4) represents the plot unit, habitat (gap
and intact forest) the subplot unit, and edges replicates.
The number of individuals within the gap and interior
categories differed at each distance/edge, producing an
unbalanced design. I used Type III SS, since it takes into
account differences in cell frequencies between treatment
combinations (Gagnon et al. 1989; Potvin 1993). To
determine whether data satisfied assumptions of an ANOVA, I
plotted residuals as a function of fitted Y values. When
residuals where not normally distributed, I transformed the
data (see type of transformation for each data set).
In all cases, I used an alpha of 10% to increase power
of the tests (Zolman 1993). I did so for several reasons.
First, the scale at which I worked precluded inclusion of
more replicates, which is often the case when dealing with
large-scale ecological phenomena (Scheiner 1993). The area
encompassed by the six edges was equivalent to 12 ha and
access to them was difficult due to steep terrain. Second,
in a mixed factorial design the number of degrees of freedom
is reduced compared to a factorial design because of multiple
nesting (Zolman 1993). In the field, I was limited by the
number of edges I could reach within walking distance from
the field station, thus I had to set up the design as
described. Lastly, the use of Type III SS to analyze
unbalanced data sets may lead to Type II errors (Potvin
1993) By increasing the probability of alpha, I compensate

26
for this bias, although it consequently increases Type I
errors. In all cases I present P-values.
Results
Pollen Tubes
Distance from forest edge did not influence the
production of pollen tubes in Palicourea aibbosa and Farrea
affinis (Table 3-1 and Table 3-2). The percentage of flowers
with pollen tubes and the average number of pollen tubes per
flower in £. aibbosa. however, was influenced by edge age
(ANOVA, Fi(4 = 9.2, P = 0.04 and Fi,4 = 7.7, P = 0.05,
respectively, Table 3-1). At old edges individuals had a
higher percentage of flowers with pollen tubes and more
pollen tubes per flower (50% 3.1% and 2.7 0.3, n = 105,
mean SE, respectively) than those at new edges (36% 4.8%
and 2.3 0.5, n = 52, respectively).
Habitat influenced the percentage of E. affinis flowers
with pollen tubes and the number of pollen tubes per flower
(ANOVA, Fi,4 = 8.0, P = 0.02 and Fif4 = 11.8, P = 0.01,
respectively, Table 3-1) In intact forest individuals had a
higher percentage of flowers with pollen tubes and more
pollen tubes per flower (32.5% +3.3% and 0.7 0.06, mean
SE, n = 81, respectively) than those in gaps (26.0% 3.1%
and 0.5 0.07, n = 48, respectively). The effect of
habitat, however, was modified by edge age and distance from
forest edge as shown by the significant interaction of

Table 3-1. Results of ANOVAs for Mixed Factorial Designs (Split-split-plot) on percentage
of flowers with pollen tubes in relation to the total number of flowers (F) and average
number of pollen tubes per flower (P) for Palicourea aibbosa and Faramea affinis.
Significance
at 10%
(*)/5% (**),
and
1% (***).
Variable
Age
A
Edge(A)
Error
Distance
D
D
X A
DxE(A)
Error
df
F
df
ss
df F
df
F
df
SS
£ aibbosa
F
1
9.2**
4
8011.5
3 0.5
3
1.3
12
11229.4
P
1
7.7**
4
35.6
3 0.3
3
2.5
12
49.9
£. affinis
F
1
3.9
4
2408.8
3 0.2
3
0.5
12
5054.8
P
1
1.9
4
1.9
3 0.4
3
0.5
12
3.6
Habitat
H
H x
D
H x A
H :
< D X A
Hx[E(A)xD]
Error
df
F
df
F
df F
df
F
df
SS
£. aibbosa
F
1
1.0
3
0.5
1 0.4
3
0.17
16
8968.1
P
1
4.1
3
1.4
1 0.3
3
0.3
16
65.04
£ affinis
F
1
8.0**
3
1.2
1 0.09
3
2.6
16
1239.6
P
1
11.8***
3
0.9
1 2.4
3
4.3**
16
0.9

28
Table 3-2. Proportion of flowers with pollen tubes in
relation to the total number of flowers (F) and average
number of pollen tubes per flower (P) in Palicourea qibbosa
and Faramea affinis in relation to distance from forest edge.
Numbers are the mean standard error and number of
individuals sampled ().
D1 D2 D3 D4
Palicourea gibbosa
F
P
43.3 4.0
(59)
2.7 3.5
35.7 7.6
(19)
2.1 3.7
52.5 5.0
(52)
2.8 2.9
43.0 6.4
(27)
2.4 2.9
Faramea affinis
F
p
28.7 3.3
(n=39)
0.6 0.08
25.0 3.7
(n=29)
0.4 0.08
35.2 5.1
(n=24)
0.7 0.1
33.3 3.6
(n=37)
0.6 0.09

29
habitat, distance, and edge age (ANOVA, F3(7 = 4.3, P = 0.05,
Table 3-1).
Fruit Set
In Palicourea qibbosa fruit set was influenced by
distance from forest edge, but the effect depended on edge
age, as shown by the significant interaction between distance
and edge age (ANOVA, F2,2 = 5.8, P = 0.02, Table 3-3). At Dl
and D2 the percentage of developing fruits was greater at new
than at old edges but this trend was reversed at D3 where 28%
0.02% (mean SE) of the flower buds resulted in fruits in
old edges as compared to 20.0% 0.02% in new edges (Fig. 3-
1). The percentage of ripe fruits was also influenced by
distance and similarly depended on edge age (ANOVA F2,2 = 3.9,
P = 0.06, Table 3-3). At Dl and D3 the percentage of ripe
fruits was greater at new edges but the trend was reversed at
D2. At D2 in old edges, 51.0% 0.05% of fruits ripened
compared to 42.0 0.04% at new edges (Fig. 3-1).
Seed and Fruit Damage
The percentage of Palicourea aibbosa fruits damaged by
ants and wasps did not differ among the four distances (Table
3-3). Edge age, however, influenced the percentage of fruits
eaten by ants but depended on habitat as shown by the
significant interaction between edge age and habitat (ANOVA,
Fi,i = 4.0, P = 0.07, Table 3-3). At new edges, individuals
growing in gaps had a higher percentage of fruits damaged

Table 3-3. Results of ANOVAs for Mixed Factorial Designs (Split-split-plot) on percentage
of fruit set (UF/FB), percentage of fruits ripening (RF/UF), percentage of seeds damaged by
wasps (UFW/FV), and percentage of fruits damaged by ants (UFA/FV) in Palicourea aibbosa.
Log-transformed data. Significance at 10% (*),5% (**) and 1% (***).
Variable
Age
A
df F
Edge(A)
Error
df SS
Distance
D
df F
D x A
df F
DxE(A)
Error
df SS
UF/FB
1
0.0
4
0.1
3
2.8
3
5.8**
12
0.001
RF/UF
1
1.0
4
0.5
3
0.0
3
3.9**
12
0.9
UFW/FV
1
o
o
4
0.001
3
1.4
3
0.5
12
0.002
UFA/FV
1
0.0
4
0.2
3
0.5
3
1.6
12
0.6
OJ
o
Habitat
H x D
H x A
H x D x A
Hx[E(A)xD]
H
Error
df F
df F
df F
df F
df SS
UF/FB
1
4.7*
3
0.9
1
1.8
3
1.1
16
0.05
RF/RF
1
1.7
3
0.4
1
0.1
3
0.4
16
1.7
UFW/FV
1
0.001
3
0.7
1
0.3
3
0.3
16
0.003
UFA/FV
1
1.2
3
1.4
1
3.9*
3
1.7
16
4.2

RIPE FRUITS/UNRIPE FRUITS UNRIPE FRUITS/FLOWER BUDS
31
0.35
0.3
0.25
0.2
0.15
0.1
0.6
0.55
0.5
0.45
0.4
0.35
DISTANCE FROM FOREST EDGE
D1
D2
D3
n 1 1
D1 D2 D3
Figure 3-1. Fruit set in Palicourea qibbosa as influenced
by distance from forest edge. Points are means and bars
standard errors.

32
by ants (7% 0.01%, mean SE) compared to intact forest (3%
0.01%). This trend was reversed at old edges.
Fruit Removal
Even though on average more fruits of Palicourea aibbosa
fruits were removed from the artificial infructescences at D2
(0.4 0.2, mean SE) than at the other distances (Dl, 0.3
0.09, D3, 0.3 0.9, and D4, 0.1 0.08), this difference was
not significant (ANOVA, F3,3 = 0.4, P = 0.7). The same was
true for habitat where on average more fruits were removed
from intact forest (0.3 0.1) than from gaps (0.2 0.06)
(ANOVA, Fi(52 = 0.6, P = 0.4).
Seed Predation
The number of seeds remaining in the trays averaged over
time did not differ among the four distances in Palicourea
qibbcsa but they did differ in Faramea affinis (ANOVA, F3,6 =
3.5, P = 0.09; Table 3-4, Fig. 3-2a). In E- affinis the
number of seeds remaining in the trays decreased from Dl to
D4, indicating higher removal rates at the interior. Habitat
alone did not have an effect on the number of seeds remaining
in the trays for either species (Table 3-4, Fig. 3-3a).
In £. affinis the distance effect was modified by
habitat and by week as shown by the significant distance x
habitat and distance x week interactions (ANOVA, F3(so = 2.3,
P = 0.08 and Fsi/io2 =2.0, P = 0.002, respectively, Table 3-
4). The number of E. affinis seeds remaining in the trays

33
was similar in gap and intact forest from D1 to D3, but lower
in gap (3.5 0.1, mean SE) than in intact forest at D4
(4.4 0.08). Over the 18-week period the mean number of F.
affinis seeds remaining in the trays decreased at all four
distances but the rate of decline was steeper at D4 and at D2
than at D1 and D3.
Seed Germination
Palicourea aibbosa seeds germinated sooner (week 6) than
Faramea affinis seeds (week 18). Averaging over time,
germination rates of £. aibbosa seeds were significantly
affected by distance from forest edge. The same was not true
for £. affinis. Germination rates of £. aibbosa seeds were
greater at D2 than at D1 and D4 (ANOVA, F3,6 = 5.0, P = 0.04;
Table 3-4, Fig. 3-2b). Habitat had a significant effect on
seed germination rates in £. aibbosa but not in £. affinis
(ANOVA, Fi(8o = 7.3, P = 0.008 and Fi/78 = 2.3, P = 0.13,
respectively). Averaged over time, more seeds of £. aibbosa
germinated in gaps than in intact forest (Fig. 3-2b).
The effects of distance and habitat were modified by
time as shown by the significant distance x week (£. affinis)
and habitat x time (£. aibbosa) interactions (ANOVA, F42.84 =
2.0, P = 0.003 and F3(n2o = 20.3, P = 0.006, respectively).
In the former, germination rates over time were steeper at D2
and D3 than at D1 and D4. In the latter, germination rates
were higher at gaps than in intact forest.

34
Seedling Growth Rate
Distance did not influence relative growth rates in
Palicourea gibbPSfr and Fflrgmea ftffjni? (Table 3-5, Fig. 3-
2c). Habitat, however, had a major effect on both species
(Table 3-5). Seedlings showed greater growth rates (ANOVA,
£. gibbp.Sfl, Fi(399 = 38.7, P = 0.0001 and £. gtffipjg, Fi(422 =
8.1, P = 0.005; Fig. 3-3c) in gaps than in intact forest.
Overall seedlings of P. gibbosa grew faster (5.0 0.17
mm/month, mean SE, n = 411) than those of F. affinis (2.4
0.08 mm/month, mean SE, n = 434).
Leaf Production
As with growth rate, habitat and not distance from forest
edge had a significant effect on leaf production (ANOVA, £.
ai&kosa, Fi,408 = 17.1, P = O.OOOl and £. affinis. Fi>424 =
3.4, P = 0.06; Table 3-5, Figs. 3-2d, 3-3d). Seedlings of
both species produced more leaves per month in gaps than in
intact forest (Fig. 3-3d). Overall, leaf production was
greater in seedlings of Palicourea gibbosa (0.8 0.01 pairs
of leaves/month, mean SE, n = 420) than in seedlings of
Faramea affinis (0.38 0.008 pairs of leaves/month).
Discussion
Of the stages of a plant's life cycle, I examined
pollination, fruit set, seed dispersal, seed predation, seed
germination, and seedling growth. An important stage

Table 3-4. Results of Mixed Factorial ANOVAs (Split-splot, 1 repeated measure) on mean
numbers of seeds remaining in trays (predation experiment) and mean number of seeds
germinated in trays (seed germination experiment) for Palicourea aibbosa (PG) and Faramea
affinis (FA). Significant at 10% (*),5% (**), 1% (***), 0.1% (****)
Distance
df F
D
df
i x E
SS
Habitat
df F
H
df
x D
F
TxHx(DxE)]
Error
df SS
FA
(Predation)
3
3.5
6
75.4
1
1
3
2.3*
80
1068.9
PG
(Predation)
3
0.3
6
182.7
1
1.5
3
1.9
80
2721.1
FA
(Germination)
3
0.1
6
19.9
1
2.3
3
0.5
78
366.8
PG
(Germination)
3
5.0**
6
24.2
1
7.3***
3
1.6
80
910.6
Week x D
W X D X E
Week
W x H
W xH x D
W x T x[Hx(DxE)]
df F
df SS
df F
df F
df F
df SS
FA
(Predation)
51
1.9***
102
50.9
17
44.2****
17
1.7
51
1.3
1360
769.2
PG
(Predation)
15
0.3
30
60.6
5
163.7****
5
1.4
15
1.6
400
845.4
FA
(Germination)
70
2.7***
210
71.4
35
304.4****
35
1.2
105
0.6
2730
907.2
PG
(Germination)
42
2.0**
84
87.8
14
835.0****
14
4.3***
42
0.9
1120
1300.
u>
cn

SEEDLING GROWTH SEEDS LEFT
(mm/month) (number of seeds/tray)
OJ
cn
Figure 3-2. Seed germination (a), seed predation (b), seedling growth (c), and leaf
production in Palicourea aibbosa and Faramea affinis as influenced by distance from forest
edge. Points represent means and bars standard errors.

37
missing from this analysis is seedling establishment which
links seed germination to seedling growth. In discussing my
results I assume that all seeds that germinated survived into
the seedling stage. I also restrict this discussion to those
stages for which I present results for both Palicourea
gibbosa and Faramea affinis.
Fruiting individuals of P. aibbosa and £. affinis were
not distributed uniformly from pasture to forest interior
(Chapter 4). Palicourea aibbosa was more abundant closer to
the forest edge (D1-D3) than farther inside the forest (D4),
and £. affinis was more abundant at D2 and D3 than at D1 and
D4 (Chapter 4). Such distributions suggest that distance
from forest edge influences one or more stages in the life
cycle of these plants. My results show that not all stages
in the life cycle of P. aibbosa and £. affinis are influenced
equally by the creation of edges and treefall gaps. In
addition, species differed in their response to these two
types of disturbance.
Pollination was influenced by habitat and edge age but
not by distance from forest edge. The percentage of flowers
with pollen tubes and the average number of pollen tubes per
flower decreased from intact forest to gaps (F. affinis) and
from old edges to new edges (P. aibbosa). Although edges and
treefall gaps represent two different scales of disturbance,
results for these two species suggest that recently disturbed
areas affect pollination levels. My results regarding the
effect of distance on pollination levels are similar to those

/
Table 3-5. Results of Mixed Factorial ANOVAs (Split-splot, 1 repeated measure) on relative
growth (iran/month) and relative leaf production (pairs of leaves/month) rates in Palicourea
gibbosa (PG) and Faramea affinis (FA). Significant at 10% (*),5% (**), 1% (***), 0.1%
C
) .
Distance
df F
D
df
x E
SS
Edge
df F
Habitat
df F
H x D
df F
Residual
df SS
FA
(Growth)
2
3.3
4
19.4
2
0.9
1
8.1***
2
2
422
1053.7
PG
(Growth)
2
0.2
4
421
2
1.7
1
38.7****
2
1.1
399
4101.1
FA
(Leaves)
2
1.1
4
0.2
2
5.2***
1
3.4*
2
0.7
424
10.3
PG
(Leaves)
2
0.5
4
1.4
2
2.1
1
17.1****
2
1.2
408
16
U>
00

SEEDS LEFT
(number of seeds/tray)
Figure 3-3. Seed germination (a), seed predation (b), seedling growth (c), and leaf
production in Palicourea aibbosa and Faramea affinis as influenced by habitat. Points
represent means and bars standard errors.
u>
vo

40
reported by Murcia (1993) for a Colombian site north of La
Planada. She found that pollination levels in 11 out of 13
species were not affected by distance from forest edge and
explained these results in terms of hummingbirds not being
influenced by edges. At La Planada I found that this was the
case (Chapter 5). Mean capture rates of nectarivorous birds
were not influenced by distance from forest edge. It seems
then that pollination can not account for differences in the
distribution of fruiting individuals of P. qjbbosa and F.
affinis across pasture-forest edges.
Seed predation increased from edge towards forest
interior in £. affinis but not in £. qibbosa. The effect of
distance from edge on £. affinis, however, was influenced by
time. Not only were more seeds removed at D4 but they were
removed faster than at the other distances. In both species,
seed germination was affected by distance from forest edge
but depended on time. Seeds closer to the edge germinated
sooner than those in forest interior. Assuming an equal
probability of seeds arriving at any of the four distances,
it is likely that seed predation and seed germination may
limit recruitment rates in these two species across pasture-
forest edge. Differences in seed predation (F. affinis and
£. qibbosa) and in seed germination (F. affinis) over time
may result in fewer individuals of these two species
establishing in the forest interior (D4).
In another study of £. affinis conducted at La Planada
Samper (1992) found that (1) seed removal rates were not

41
affected by habitat (gaps, edge of gaps, and intact forest),
(2) seed germination rates were faster in gaps (mean = 174
days, n = 148) than along edges of treefall gaps (177, n =
132) and intact forest (187, n = 135), and (3) seedling
establishment (i.e., the stage at which seedlings become
independent from food reserves contained in the seeds) was
not affected by habitat. Samper's work and mine show that
the seed and seedling stages in £. affinis are affected
differently by treefall gaps and edges resulting from human
activities.
Once seeds of £. aibbosa and £. affinis arrive and
germinate at any distance from forest edge, treefall gaps
seem to have a major influence on these two species by
increasing growth and leaf production rates in seedlings.
This is not in accordance with results obtained in the
Amazon, where relative growth rates of seedlings was greater
up to 10 m from forest edge towards forest interior (Seizer
1992).
My study shows that edges can influence recruitment
rates of P. qibbosa and F. affinis through their effect on
seed predation and seed germination but not on pollination
and the growth of seedlings. On the other hand, treefall
gaps influence recruitment rates through their effect on
seedling growth.

CHAPTER 4
UNDERSTORY FRUIT ABUNDANCE IN A NEOTROPICAL MONTANE FOREST:
THE INFLUENCE OF EDGES AND TREEFALL GAPS
Introduction
Fruit abundance can be influenced by disturbances
occurring at various scales. In general, fruit abundance
increases in small, natural disturbances, such as treefall
gaps (Blake and Hoppes 1986; Levey 1988&,k), in large,
natural disturbances, such as patches affected by hurricanes
(Walker and Neris 1993) or fire (Fleming 1988), and in large,
human-disturbed areas, such as abandoned fields and pastures
(Martin 1985; Levey 1988a,Blake and Loiselle 1991; Lugo
and Frangi 1993; but see Wong 1986). In treefall gaps high
fruit production is the result of an increase in the number
of fruits produced by individuals growing in the disturbed
area compared to conspecifics growing in intact forest and to
an increase in the number of fruiting individuals (Piero and
Sarukhan 1982; Clark and Clark 1987; Levey 1990). In large
disturbed areas, high fruit production has been related to
the same two factors (Auclair and Cottam 1971; Halls 1973;
McDiarmid et al. 1977; Fleming 1988), and to the appearance
of pioneer species that typically produce more fruits than
late successional or mature-forest species (Martin 1985).
42

43
It has been suggested that the spatial heterogeneity of
the fruit resource base, regardless of the scale at which
disturbances occur, influences the distribution of organisms
feeding on fruits (e.g., Martin 1985; Wong 1986; Levey
1988,b; Heideman 1989; Blake and Loiselle 1991; Loiselle and
Blake 1991) and the resulting dispersal of seeds (Murray
1988). Nevertheless, depending on the scale of disturbance
changes in seed dispersal might have different consequences
for the plants. For example, small-scale disturbances may
influence recruitment within populations whereas large-scale
disturbances may influence colonization of new areas (Harper
1994).
One immediate consequence of disturbance is the creation
of edges or boundaries. In general, boundaries mediate
fluxes between adjacent ecological systems (Margalef 1968;
Wiens et al. 1985; Gosz 1991). Moreover, because of
differences in their permeability to fluxes of material and
energy, boundaries may influence the dynamics of neighboring
systems (Correll 1991; Ryszkowski 1992). In this context,
edges that bound patches resulting from disturbance might
influence the dynamics and structure of the neighboring
patches by influencing the distribution of resources and/or
movement of organisms (Crist et al. 1992; Johnson et al.
1992; Wiens 1992). In particular, edges resulting from
large-scale disturbances may influence the movement of seeds
and thus the structure of whole landscapes.

44
Seed movement and fruit abundance are related in at
least two ways. First, fruit numbers determine the
availability of seeds. Second, fruit and seed availability
affect the behavior of the dispersers (Murray 1987; Loiselle
and Blake 1993) In spite of these well known relationships
between fruit abundance and seed movement, few studies have
addressed how edges influence fruit abundance (Blanchard
1992). I explored this question in the understory of a
neotropical montane forest using edges resulting from forest
clear-cutting. In particular, I documented how fruit
abundance and seed movement were affected by distance from
the forest edge towards the forest interior for the
assemblage of understory shrubs and for individual species.
Since edge age (Williams-Linera 1990; Blanchard 1992) and
treefall gaps (Janzen 1983; Lovejoy et al. 1986; Noss 1991)
can modify the steepness of such a response, I examined how
they interacted with distance. By looking at four different
scales, edges of different age within the forest matrix,
distance from forest edge within edges, treefall gaps and
foliage density (LAI) within each distance, I could examine
changes in the fruit resource base along edges resulting from
large-scale disturbances.

45
Methpflg
Canopy Structure
I estimated leaf area index (LAI) (m^ foliage area/m^
ground area) to (1) characterize the structure of the forest
canopy across the pasture-forest edge and to (2) relate LAI
to fruit abundance. Fruit production is strongly influenced
by light environment (e.g., May and Antcliff 1963; Jackson
and Palmer 1977), which in the subcanopy and on the forest
floor is influenced by canopy structure (Norman and Campbell
1991) Thus LAI estimates provide a fine-scale description
of the light environment of each point where I sampled
fruits.
I used a LAI-2000 Plant Canopy Analyzer (Li-Cor, Inc.)
in October 1993 to estimate LAI at the three old edges.
Estimates of LAI obtained with this instrument are based on
the transmitted fraction of incident radiation on the canopy.
At each of the old edges (Climo I, Climo II, and Pialap) I
established three transects running perpendicular to the
forest edge and extending 10 m into the pastures and 210 m
into the forest (Chapter 2, Fig. 2-3). I made readings at
intervals of 5 m in the first 50 m of the transect, 10 m in
the next 130 m, and 20 m in the next 40 m. For each point
along a transect I made four consecutive below-canopy
readings that together were paired with a single reading
taken in an area devoid of trees and shrubs in the nearby

46
pasture. I always kept the lens and LAI-2050 optical sensor
pointing in the same direction and 1.5m above the ground. I
covered the optical sensor with a 45 view cap to block my
image and direct beam radiation during clear days (Li-Cor
1992) .
Estimates obtained with this instrument often
underestimate true LAI (Chason et al. 1991; Hannan and Bgu
1995) Nevertheless, in the context of this study these
values are useful to describe relative changes in canopy
structure across the pasture-forest edge.
Fruit Abundance
To establish the influence of edges and treefall gaps on
fruit production by understory plants I subdivided each of
the four 100 x 10 m strips within each edge into five 20 x 10
m quadrats (Chapter 2, Fig. 2-3). In turn each quadrat was
subdivided into four 10 x 5 m subquadrats and for each
quadrat I chose at random two subquadrats in which to monitor
fruit production (Fig. 2-3).
I used Brokaw's (1982) definition of treefall gap to
classify each subquadrat as gap or interior habitat. A
subquadrat was classified as gap if it was within a gap or
located <5 m from a gap edge, and as interior if it
was located >5 m from the edge of the nearest treefall gap at
the time the study began. These two categories do not
reflect the environmental continuum from the center of the
treefall to the intact forest nor do they take into account

47
differences in gap size and shape (Brown 1993; Denslow and
Hartshorn 1994). Nevertheless, I was more interested in the
possible interaction between distance from the edge and
treefall gaps than in the treefall gaps themselves.
I monitored changes in fruit production at each pair of
the 10 x 5 m subquadrats over a 12 mo period (September 1992-
August 1993, excluding December). In each subquadrat I
identified and counted individual plants <7 m tall bearing
unripe and/or ripe fleshy fruits (Levey 1988^.,^; Blake and
Loiselle 1991). I also included broken limbs bearing fruits.
Most species I recorded complete their life cycle within this
arbitrarily set understory stratum. A few species, mostly in
the Arecaceae (palms), Rubiaceae, and Melastomataceae, also
fruit in higher strata. For each individual, except species
in the Araceae, I counted the total number of unripe and ripe
fruits every month on a biweekly basis. I averaged these
biweekly counts to obtain a single value on fruit abundance
for any given month.
I expressed fruit production in four different ways: (1)
total number of individuals bearing unripe and/or ripe fruits
(TI); (2) total number of fruits (unripe + ripe fruits) (TF);
(3) total number of ripe fruits (RF), and (4) total number of
fruits (unripe + ripe fruits), excluding the Arecaceae (TF-
A). In all cases, fruit abundance is expressed as the mean
number of counts per 50 m2, the area of each 10 x 5 m
subquadrat.

48
I included unripe fruits because they constitute a food
resource for frugivorous insects. I excluded the Araceae
from variables 2-3 because it was difficult to estimate fruit
numbers for each infructescence. I excluded the Arecaceae
from variable 4 because their high productivity and prolonged
fruiting season could mask patterns of fruit production among
shrubs producing fewer fruits and fruiting over shorter
periods of time.
To explore fruit abundance responses to edges at the
species level I looked at the number of individuals bearing
unripe and/or ripe fruits. For each species I pooled this
information for all subquadrats and months to obtain a single
value for each edge age and distance.
I collected most plant species and deposited voucher
specimens at Botany Department Herbarium, Arizona State
University (ASU), Herbario Nacional de Colombia (COL), Botany
Department Herbarium, Field Museum of Natural History,
Chicaco (F), Herbario de la Universidad de Antioquia (HA),
Kew Botanical Garden (K), Missouri Botanical Garden (MO), New
York Botanical Garden (NY), Herbario Universidad de Nario
(PSO), Utrecht Herbarium (U), Smithsonian Institution (US),
and Department of Botany Herbarium, University of Wisconsin
(WIS). Family names follow Cronquist (1981).
Seed Movement
To evaluate seed movement across edges I counted and
identified seeds contained in bird droppings retrieved from

49
birds captured in mist nets. Birds were sampled at the same
edges and distances from the forest edge as were fruiting
plants (Chapter 5). After capture, birds were kept in cloth
bags lined with filter paper for ca. 20 min. Bird droppings
were preserved in alcohol and seeds were compared to a
reference collection, compiled during the study period.
This method for evaluating seed movement may have biases
in addition to those involved when sampling birds with mist
nets (see Chapter 5). In particular, seeds recovered from
birds might represent a non-random sample of seeds ingested,
since seed handling varies within and among species depending
on seed size and other seed characteristics (Levey 1986,
1987). Nevertheless, this method does provide information on
seed movement that would be difficult to determine by other
means (e.g., seed traps).
Analyses
I used a Repeated Measures ANOVA to analyze LAI. Edge
was included as a between factor variable, distance from the
edge as a within factor variable, and individual transects as
subjects.
I analyzed data on fruit production at the community
level with ANOVAs for mixed factorial designs (Girden 1992).
The mixed design was set up as a split-split-plot design with
one repeated measure (Winer et al. 1991) The factors of
interest were edge age, distance from the edge, habitat, and
month, the latter representing the repeated measure. The six

50
edges that I sampled for each level of edge age (old and new)
were chosen at random from a population of old and new edges
and represented the plot unit. In turn, each edge was
divided into four strips, i.e., distances from the forest
edge towards the forest interior, representing the subplot
units. Randomization of the levels of the distance factor
was restricted, but because the strips were separated in
space and I analyzed responses from non-mobile organisms I
assumed they represented independent subsampling units. This
was supported by results of an ANOVA in which distance was
included as a repeated measure and the epsilon factor equaled
one, indicating no correlation between the levels of the
distance factor (Girden 1992). Finally, fruit production was
monitored in subquadrats that were chosen at random and
classified according to habitat as gap or interior, the
latter representing the sub-subplots.
The number of subquadrats falling within the gap and
interior categories differed at each distance among the
edges, producing an unbalanced design. I used Type III SS
since it takes into account differences in cell frequencies
between treatment combinations (Gagnon et al. 1989; Potvin
1993) .
To determine whether the data satisfied assumptions of
an ANOVA, I plotted residuals as a function of fitted Y
values. When residuals where not normally distributed, I
log-transformed the data. In addition, I verified the
assumption of compound symmetry for the repeated measure

51
factor and used a corrected F-ratio (H-F) to interpret the
analyses (Girden 1992; von Ende 1993).
I used an alpha of 10%. I set alpha at this level
because my design could lead to increases in Type II errors
(reduced power of my tests) (Zolman 1993). Concomitantly I
increased the probability of committing Type I errors.
First, the scale at which I worked precluded inclusion of
more replicates, which is often the case when dealing with
large-scale ecological phenomena (Scheiner 1993). The area
encompassed by the 6 edges was equivalent to 12 ha and the
access to them was difficult due to steep terrain. Second,
in a mixed factorial design the number of degrees of freedom
is reduced compared to a factorial design because of multiple
nesting (Zolman 1993). In the field, I was limited by the
number of edges I could reach within walking distance from
the field station and thus I had to set up the design as was
described above. Lastly, the use of Type III SS to analyze
unbalanced data sets may lead to Type II errors (Potvin
1993).
I analyzed fruit abundance data for understory species
using a Replicated Goodness of Fit Test (G-statistic) (Sokal
and Rohlf 1981) to establish whether distance and edge age
affected the number of fruiting individuals. First, I pooled
the data for old and new edges and calculated Gp (G-Pooled)
to determine if the number of individuals across the four
distances departed significantly from a uniform distribution.
Second, I compared old and new edges and calculated Gh (G-

52
Heterogeneity) to test for homogeneity between the two edge
ages. For these two analyses I used only species in which at
least 80 percent of the expected cell frequencies were
greater than 5, since the G statistic departs from the X?
distribution if this is violated (Siegel and Castellan 1988).
Resultg
LAI and Distance from the Edge
I measured LAI only at old edges and found that major
changes in LAI were observed at the interface between pasture
and forest (Fig. 4-1). Once inside the forest, LAI values
were highly variable not only from the edge towards forest
interior but also among edges at the same distance (Fig. 4-
1). Averaging over the three edges, variability in LAI
measurements for the 0-40 m interval (Coefficient of
variation, CV = 0.3) was identical to that for the 50-210 m
interval (CV = 0.3).
I did not include pasture LAI values in the ANOVA to
establish the effect of distance on LAI. LAI did not differ
significantly either among edges (ANOVA, F3,5 = 0.4, P = 0.75)
or in relation to distance from the forest edge towards the
forest interior (F23, 115 = 0.47, P = 0.98) (Table 4-1). This
indicates that at least along old edges, canopy structure
does not vary in a predictable way from the edge towards the
forest interior.

53
LAI and Fruit Abundance
To establish whether LAI influences fruit abundance, I
averaged values of fruit abundance for each pair of 50 m2
subquadrats where I took LAI measurements. LAI was not
significantly correlated with total number of fruits (TF)
(Coefficient of determination, r2 = 0.032, n =36), total
number of fruits excluding the Arecaceae (TF-A) (r2 = 0.077,
n = 36), total number of ripe fruits (RF) (r2 = 0.015, n =
36), or total number of fruiting individuals (TI) (r2 = 0.02,
n = 36) (Fig. 4-2).
Fruit Abundance
Plant Assemblages
In the ANOVAs none of the three-way interactions was
significant. Several two-way interactions were significant
but not consistently so for the four measurements of fruit
abundance. In describing the results I look first at the
single effect of distance on fruit abundance and then at the
interactions involving this term.
The total number of fruits (TF) and total number of ripe
fruits (RF) differed significantly among the four distances
on a yearly basis (ANOVA, ^2,12 4.3, P = 0.03 and F3fi2 =
5.4, P = 0.01, respectively) (Table 4-2). For TF and RF the
mean number of fruits as well as the variance decreased from
forest edge towards forest interior (Fig. 4-3). There are
three not mutually exclusive explanations for these results.

54
DISTANCE (m)
Figure 4-1. Leaf area index across the pasture-forest edge,
indicated by the arrow (0 m). Points represent the average
of three measurements per edge and bars the standard errors.
Open squares = Pialapi, open triangles = Climo II, and
filled circles = Climo I.

Table 4-1
Source
Edge
Transect
Distance
Distance
Distance
. Results of a Mixed Factorial ANOVA on leaf area index for old edges at the
Reserva Natural La Planada
df
SS
MS
F-Value
P-Value
H-F
3
6.768
2.256
0.401
0.7585
(Edge)
5
28.095
5.619
23
11.64
0.506
0.468
0.981
0.981
* Edge
69
76.934
1.115
1.031
0.4367
0.437
* Transect (Edge)
115
124.376
1.082
H-F Epsilon
1.714
Distance

56
First, individuals growing at Dl (0-10 m) produced more
fruits than those growing at other distances. Support for
this comes from the fact that distance did not have an effect
on the total number of fruiting individuals (TI) (Table 4-2)
and the observation that some species that were heavily
represented at Dl (e.g., Clidemia sp. 1 and Palicourea
aibbosa) produced larger crops here than at the other
distances. Recall that TI eliminates the variability
associated with crop size because it considers only the
number of fruiting individuals. Second, palm fruits made a
disproportionate contribution to overall fruit production at
Dl. Support for this comes from the fact that distance did
not have an effect on total number of fruits excluding the
Arecaeae (TF-A) (Table 4-2) Third, some species found only
at Dl (0-10 m from the forest edge), including Marcaravia
eichleriana. Marcqraviastrum subssesilis. HidPhia
pseudoradula. M. theaezans. p¡?ammiia ferryiginga, Sghgfflgra
lasiocrvne. Phytolacca rivinoides. produced large fruit crops.
Even though the effect of distance alone was not significant
for total number of fruits excluding palms (TF-A), the trend
was similar to that for the previous two variables, i.e.,
sharp decrease from forest edge towards forest interior.
The effect of distance on fruit abundance was modified
by edge age and habitat, but neither factor alone had an
effect on any of the variables describing fruit abundance
(Table 4-2). The interaction between edge age and distance
from the forest edge was significant for total number of

FRUn ABUNDANCE FRUT ABUNDANCE
(mean counts / 50 m2) (mean counts / 50 m2)
5000
4000
3000
2000
1000
0
1000
1 2 3 4 5 6
LAI
Total Fruits
*
& Eif a a
CM
E
o
in
c
3
o
o
c
CD
E
800
600
400
200
0
1 2 3 4 5 6
Ripe Fruits
A.
- -to# tEMfloClA
LAI
1000
750 -
500
250
Total Fruits
(excluding palms)
12.5 -
c\T ^
LU E 10 -
y o
6 in
Total Individuals %
A

d 7.5 -
Z
o
A T?


>
o 5 -
t
tJ c
. V A
o A
A D
C3
a*..,
£ 25-
E.
o J
A J3 A
__ A A
CQ
1 1 1 1
3 4
LAI
LAI
Figure 4-2. Leaf area index (LAI) and mean counts expressing fruit abundance.
Symbols represent old edges. Open squares = Pialapi, open tirangles = Climo
II, and filled circles = Climo I.
~u

58
fruits (TF) (ANOVA, F3/i2 = 2.56, P = 0.1; Table 4-2). Total
number of fruits (TF) showed a sharp decline from the forest
edge towards the forest interior at new edges and remained
almost unchanged at old edges (Fig. 4-4). The interaction
between habitat and distance from the forest edge was
significant for total number of fruits (TF) (ANOVA, F3(i6 =
2.95, P = 0.06) and for total number of ripe fruits (RF)
(ANOVA, F3(i6 = 3.5, P = 0.04; Table 4-2). For both
variables, fruit abundance at D1 (0-10 m from forest edge)
was higher in gaps than in forest interior (Fig. 4-5) These
differences disappeared at the other distances. The fact
that none of these interactions was significant for total
number of fruits excluding palms (TF-A), suggests that palms
made an important contribution to these results.
The number of ripe fruits (RF) differed significantly
between old and new edges depending on month (ANOVA, Fio,40 =
2.8, P = 0.009; Table 4-2, Fig. 4-6). The total number of
fruiting individuals (TI) and the total number of fruits
excluding Arecaceae (TF-A) differed across the four distances
in some months but not in others as shown by the significant
interaction between distance from edge and month (ANOVA,
f30,120 = 1-48, P = 0.07 and F3o,i20 = 1.57< P = 0.04,
respectively; Table 4-2, Fig. 4-7). These results contrast
with those for TF and RF, in which this interaction was not
significant but in which fruit abundance averaged over time
was affected by distance (Table 4-2). Recall that the total
number of fruiting individuals (TI) and the total number of

Table 4-2. Results of ANOVAs for Mixed Factorial Designs (Split-split-plot, 1 repeated measure) on mean
counts/50 m^ of number of fruiting individuals (TI), number of fruits (TF), number of ripe fruits (RF), and
number of fruits excluding Arecaceae (TF-A). log-transformed data. Significance at 10% (*),5% (**), and
1% (***).
Variable
Age
Edge(Age)
Distance
Distance x
Age
D x E(A)
(A)
Error
(D)
Error
df
F
df
SS
df
F
df
F
df
SS
TI
1
0.99
4
1256.13
3
1.77
3
0.65
12
1745.05
TF
1
0.28
4
71.37
3
4.34 **
3
2.56
*
12
67.2
TF-A
1
1.26
4
39.18
3
1.43
3
0.4
12
73.79
RF
1
0.47
4
40.58
3
5.38 **
3
0.96
12
31.67
Habitat
H x D
H
X A
H
x D x A
Hx[E(A)xD]
(H)
Error
df
F
df
F
df
F
df
F
df
SS
TI
1
0.74
3
2.24
1
0.522
3
0.016
16
1522.51
TF
1
0.058
3
2.95 *
1
0.047
3
0.41
16
65.02
TF-A
1
1.26
3
0.77
1
0.244
3
0.165
16
42.28
RF
1
0.26
3
3.47 **
1
0.01
3
1.02
16
29.38
Month x
H
M
x D x H
M
X
A x H
M X
A x D x
H
MxHx[DxE(A)]
Error
df
F
H-F
df
F
df
F
df
F
df
SS
TI
10
0.86
30
1.27
10
0.31
30
1.28
160
389.1
TF
10
2.03
*
30
0.55
10
0.37
30
0.6
160
32.45
TF-A
10
1.42
30
0.58
10
0.69
30
1.28
160
24.58
RF
10
2.01
* *
30
0.41
10
0.94
30
0.62
160
54.78
M
x A
M x E(A)
M
x D
M
x D x A
M X
D x E (A)
Error
Error
df
F
H-F
df
SS
df
F H-F
df
F
df
SS
TI
10
1.47
40
144.98
30
1.48 *
30
1.09
120
350.03
TF
10
0.49
40
21.43
30
1.29
30
0.99
120
25.43
TF-A
10
0.65
40
17.25
30
1.57 **
30
0.7
120
22.13
RF
10
2.83

40
1.27
30
0.76
30
1.17
120
43.2

FRUIT ABUNDANCE
(counts / 50 m 2)
60
D1 D2 D3 D4
DISTANCE FROM FOREST EDGE (m)
Figure 4-3. Variation in fruit abundance at the Reserva
Natural La Planada in relation to distance from forest
edge. Points represent means and bars standard errors.

61
fruits excluding Arecaceae (TF-A) eliminates part of the
variability associated with crop size because (1) TI
considers only the number of fruiting individuals and (2) TF-
A includes plant species producing smaller and less
persistent fruit crops, i.e., number of fruits per
reproductive season, than those of palms and plant species
producing similar number of fruits.
Lastly, fruit abundance expressed as the total number of
fruits (TF) and total number of ripe fruits (RF) differed
significantly between gaps and interior, but depended on
month, as shown by the significant interaction between
habitat and month (ANOVA, Fio,i60 = 2.0, P = 0.03 and Fio,i60 =
2.0, P = 0.03, respectively; Table 4-2). The Arecaceae again
seemed to be mostly responsible for this result as indicated
by the fact that this interaction was not significant for the
total number of fruits when the Arecaceae were excluded (TF-
A) (Table 4-2).
Species Lgygi Responses
I recorded 149 plant species fruiting in the understory
of the edges included in this study and classified them in
five categories: extremely sparse (1 individual), very sparse
(2-5 individuals), sparse (6-20 individuals), abundant (21-50
individuals), and very abundant (>51 individuals). Of 149
species, 26 (17%) were abundant to very abundant (>21
individuals) and 125 (83%) were sparse to extremely sparse
(<21 individuals) (Appendix A). The most abundant species

62
DISTANCE FROM FOREST EDGE (m)
Figure 4-4. Variation in fruit abundance at the Reserva
Natural La Planada in relation to edge age and distance
from forest edge. Points represent means and bars
standard errors.

FRUIT ABUNDANCE
(mean counts / 50 m2
63
D1 D2 D3 D4
DISTANCE FROM FOREST EDGE
Figure 4-5. Variation in fruit abundance at the
Reserva Natural La Planada in relation to habitat
and distance from forest edge. Points represent
means and bars standard errors.

64
were two Rubiaceae, Faramea affinis and Palicourea aibbosa.
One hundred species were found both at new and old edges and
49 were exclusive to new (35 species) and old edges (14
species) (Appendix A). In this respect, new edges presented
proportionally more species than old edges (Goodness of Fit
test, G = 9.3, P < 0.01).
Fruit abundance on a species by species level, expressed
as the number of fruiting individuals, varied depending on
distance from the edge and edge age. The distribution of
fruiting individuals for 16 of the 26 abundant species
departed significantly from a uniform distribution across the
four distances (Table 4-3). Given a 10% probability of
obtaining a species that shows a non-uniform distribution it
is very unlikely that 16 or more species out of 26 would have
shown a non-uniform distribution by chance alone (Binomial
test, P = 2.0 x 10-10). Clearly, the distribution of fruiting
individuals of some species is affected by the creation of
edges.
For the 16 species showing a non-uniform distribution I
used residuals to further determine if they were more
abundant at any particular distance from the forest edge
(i.e., if at any given distance the observed frequency was
greater than the expected frequency). Four species were more
abundant at D1 (0-10 m) three species were more abundant at
D4 (190-200 m) four species were more abundant at D2 (30-40
m) or D3 (60-70 m), and five species were more abundant at
two different distances (e.g., Alloolectus tetraaonus and

FRUIT ABUNDANCE
(mean counts / 50 m 2)
65
Figure 4-6. Variation in fruit abundance at the Reserva
Natural La Planada in relation to edge age and month. Points
represent means and bars standard errors.

FRUIT ABUNDANCE
(mean counts / 50 m 2
66
MONTH
Figure 4-7. Variation in fruit abundance at the Reserva
Natural La Planada in relation to distance from forest
edge and month. Points represent means and bars standard
errors.

67
Palicourea aibbosa), the latter suggesting a bimodal
distribution (Table 4-3). Species showing bimodal
distributions, in particular, suggest that distance alone can
not explain their distribution across the pasture-forest
edge.
I evaluated the combined effect of distance and edge age
on the number of fruiting individuals for 16 species (Table
4-4). For nine species there was a significant interaction
between distance and edge age, indicating that the
distribution of fruiting individuals across the four
distances differed between old and new edges (Gneterogeneity p
< 0.1, Table 4-4). A significant interaction between
distance and edge age was irrespective of whether the species
showed a non-uniform distribution across the four distances
by combining the two types of edges (Gp00ied P < 0.1, Table
4-3) or by looking at old and new edges separately (Gold P <
0.1 and GNew p ^ 0.1, Table 4-4) .
Most of the fruiting individuals found at my study edges
were represented by few individuals, precluding the use of
Goodness of Fit Test to establish how distance from forest
edge affected their distribution. Instead, I used
information on their abundance (Table 4-5) to establish the
distance at which sparse species were found more often. I
excluded abundant and very abundant species from the analysis
and found that for the remaining three groups of plants
abundance and distance from the edge were not independent
(Test for Independence, x2 = 18.86, P = 0.004). Examination

68
Table 4-3. Distribution of fruiting individuals of
abundant (>21 individuals) plant species in the understory of
La Planada in relation to distance from the edge. Numbers
represent number of individuals. Distances are 0-10 m (Dl) ,
30-40 m (D2), 60-70 m (D3), and 190-200 m (D4) from forest
edge. P < 0.1 (*), P < 0.05
(****)
Species
Uniform Distribution
Burmeistera carnosa
Solanum sp.7
Anthurium membranaceum
Burmeistera sp. nov.
Anthurium umbraculum
columnea cinerea
Spheraedenia stCYermarkii
Anthurium cf. pulverulentum
Anthurium versicolor
Geonoma weberbaueri
Non-Uniform Distribution
Psammisia aff. debilis
Besleria solanoides
Psvcothria aubletiana
Clidemia sp.l
Solanum sp.5
Chamaedorea polvchlada
Alloplectus tenuis
Asnlundia sp.1
Anthurium umbricclum
Anthurium cf. marmoratum
Faramea affinis
AllQBl.ectUS teuscheri
Anthurium carchiense
Anthurium cf. melampvi
Palicourea aibbosa
Allpplechus tetraqcnus
(**) P < 0.01 (***), P < 0.001
Dl
D2
D3
D4
G-stat
10
11
13
14
0.83 ns
14
15
11
11
0.99 ns
32
30
27
24
1.31 ns
7
4
5
9
2.35 ns
17
17
13
23
2.87 ns
10
5
4
5
3.34 ns
7
2
6
6
3.37 ns
8
6
5
2
3.99 ns
18
8
16
14
4.36 ns
19
20
9
14
5.29 ns
20
9
4
8
12.9
***
39
16
14
4
35.5

45
14
12
5
40.3

54
59
27
10
48.7

3
14
14
17
12.2

5
7
15
18
10.8

2
2
7
9
8.0

2
10
5
8
6.6

9
7
22
7
12.2

10
9
17
5
7.2

104
146
127
117
7.6

22
41
18
30
10.9

3
15
5
12
11.7
* *
20
10
19
32
12.3
*
91
65
91
43
23.6

10
2
7
1
11.7


Table 4-4. Results of Replicated Goodness of Fit Test on the number of individuals in fruit
for understory plants at the Reserva Natural La Planada. Gneterogeneity (Gh) / GTotal (Gt)
Gold edges (G0), and GNew edges (GN) P < 0.1 (*), P < 0.05 (**), P < 0.01 (***), P < 0.001
(****)#
Species
df
gh
df
Gt
df
Go
df
Gn
AnLlmrium versicolor
3
6.64

6
11 *
3
1.48 ns
3
9.49 **
Anthurium cf. marmoratum
3
7.88

6
15.08 **
3
10.72 **
3
4>3 ns
Anthurium umbricolum
3
28.28

6
40.48 ****
3
17.96 ****
3
3.22 ns
Alloolectus teuscheri
3
7.72
*
6
18.6 ***
3
5.06 ns
3
13.52 ***
Gegnoma weberbaueri
3
17.56

6
22.86 ****
3
3.93 ns
3
18.92 ****
Solanum sp.7
3
6.84
*
6
7.83 **
3
2.65 ns
3
5.18 ns
Anthurium cf. melampvi
3
21.76
*
6
34.02 ****
3
17.38 ****
3
16.6 ****
Palicourea aibbosa
3
6.32

6
29.93 ****
3
16.77 ****
3
13.14 ***
Solanum sp.5
3
6.24
*
18.38 ****
7.16 *
11.19 **
Anthurium membranaceum
3
0.94
ns
6
0.94 ns
3
1.74 ns
3
0.49 ns
Burmesitera carnosa
3
3.47
ns
6
4.32 ns
3
2.39 ns
3
1.89 ns
Clidemia sn.l
3
2.42
ns
6
51.11 ****
3
32.79 ****
3
18.12 ****
Psvcothria aubletiana
3
3.84
ns
6
44.15 ****
3
24.29 ****
3
19.83 ****
Faramea eleaans
3
2.24
ns
6
2.25 ns
3
3.35 ns
3
6.48 *
Besleria solanoides
3
4.04
ns
6
39.5 ****
3
35.13 ****
3
4.35 ns
Chamaedorea Dolvchlada
3
3.32
ns
6
14.14 **
3
12.45 ***
3
1.69 ns

70
Table 4-5. Distribution of fruiting individuals in the
understory of the Reserva Natural La Planada across pasture-
forest edge based on species abundance. Numbers represent
the number of fruiting individuals.
Number of Distance
Species D1 D2 D3 D4
Extremely sparse
(1 individual)
33 13
Very sparse
(2-5 individuals)
40 64 21 28 25
Sparse
(6-20 individuals)
50 167 144 138 120
Abundant
(21-50 individuals)
13 94 99 115 137
Very abundant
(>51 individuals)
12 475 441 384 327

71
of the residuals showed that more individuals of very sparse
species were found at Dl and of sparse species at D2.
Although individuals of both groups of plants might be found
across the four distances, these results show that sparse
species are found more often close to the edges.
Seed movement
I recovered 393 bird droppings from 19 species of mist-
netted frugivorous birds. Seeds of 93 species, of which I
was able to identify 65, were represented. Of the species I
identified, 52 (80%) were found in the subquadrats where I
counted fruits. This figure, compared to the total number of
plant species I recorded fruiting in the understory (149),
shows that bird droppings represent a subsample of the plant
species that I found.
The number of plant species in individual bird droppings
ranged from 1 to 7 (mean SD, 1.7 1.0, N = 393). There
was no significant difference in the mean number of plant
species in droppings recovered at the four distances (ANOVA,
f3,393 = 0.8, P = 0.5). In addition, total numbers of seeds
in bird droppings was independent of distance (x2 Test for
Independence, df = 15, x2 = 12.8, P = 0.6) .
Fourteen of the 26 abundant species (>21 individuals)
and 38 of the 124 sparse species (<21 individuals) fruiting
in the understory were represented in the bird droppings
(Appendix A). Since sparse species were found more often at
the forest edge, I compared their distribution against that
of droppings containing their seeds to determine whether they

72
were independent of each other. Plants and droppings were
classified as (1) "edge" if the number of observations at D1
(0-10 m) and D2 (30-40 m) combined together was >0, (2)
"interior" if the number of observations at D3 (60-70 m) and
D4 (190-200 m) was >0, and (3) "edge=interior" if the number
of observations at D1 and D2 = D3 and D4. I found that the
proportion of droppings containing seeds of sparse species
was independent of their abundance in edge and interior (Test
for Independence, X2 = 2.4, df= 4, P = 0.6). Even though
sparse species were found more often at forest edge, their
seeds are potentially reaching forest interior.
DiSPUggipp
Distance from the edge towards the forest interior had a
major effect on fruit abundance but only close to the forest
edge. Nevertheless, the facts that (1) the effect of
distance was modified by edge age, habitat, and month of the
year, and (2) results for the four variables describing fruit
abundance differed, indicate that there are complex
interactions between edges and fruiting plants (summarized in
Table 4-6). I interpret these complex interactions in terms
of scales at which fruit abundance changes in relation to the
creation of edges.
At the scale defined by the pastures and the forest
matrix and by the length of the study (1 yr), new edges
generated spatial heterogeneity in fruit abundance within the
study area at D1 (0-10 m from forest edge). Fruit abundance,

73
expressed as total number of fruits (TF) and as total number
of ripe fruits (RF), was higher in new than in old edges at
D1 but these differences were not evident at the other
distances. At the scale defined by the edges and by the
length of this study, treefall gaps at D1 generated spatial
heterogeneity in fruit abundance within edges. Fruit
abundance, expressed as total number of fruits (TF) and as
total number of ripe fruits (RF) was higher in treefall gaps
than in forest interior at D1 but, again, these differences
were not found at the other distances.
When I looked at total number of ripe fruits (RF), and
in addition examined the total number of fruits excluding the
Arecaceae (TF-A) and the number of fruiting individuals (TI),
a different picture emerged. At the scale of the study area
and month, RF differed between old and new edges but depended
on month. At the scale of edges and month, TF-A and TI
differed among the four distances but also depended on month.
These results suggest that changes in fruit abundance
and the magnitude of these changes across pasture-forest edge
are related to the size of fruit "patches". In my study
area, large patches of fruit were generated by understory
palms, which produced large fruit crops that persisted for a
long time. These large fruit patches seemed to generate a
steep gradient in fruit abundance from the edge towards the
forest interior on a yearly basis. Conversely, small patches
of fruit seemed to generate gradients that varied in their
magnitude depending on month. I will discuss these results

74
Table 4-6. Summary of results of ANOVAs on fruit abundance
across the pasture-forest edge for the different response
variables. Fruit abundance expressed as total number of
fruits (TF), total number of ripe fruits (RF), total number
of fruits excluding the Arecaceae (TF-A), and total number of
fruiting individuals (TI). Significance at 10% (*), 5% (**),
1% (***).
TF RF
TF-A TI
Large Fruit
Small Fruit
Crops
Crops
ANOVA terms
Distance
*
Distance x Age

Distance x Habitat
*
Distance x Month

Age x Month

Habitat x Month


75
in terms of (1) the factors that influence fruit abundance
and (2) the consequences that these observed changes might
have on plants and frugivores.
Factors Influencing Fruit Abundance
Fruit abundance is regulated by abiotic and biotic
factors interacting in complex ways with the flowering and/or
fruiting stages of a plant's life cycle (e.g., Marshall and
Grace 1992). Abiotic factors that have a direct effect on
fruit numbers and fruit size are photoperiod and irradiance
(e.g., Auchter et al. 1926; May and Antcliff 1963; Jackson
and Palmer 1977; Mathai and Sastry 1988; Tombesi et al.
1994), temperature (e.g., Chaikiattiyos et al. 1994), water
availability (e.g., George et al. 1990), and nutrients (e.g.,
Stephenson 1992). Biotic factors include pollination,
predation of flowers, seeds, and fruits, and damage by
pathogens (Stephenson 1981). The importance of these factors
is likely to differ within and among species, depending on
habitat.
Irradiance is an important factor influencing fruit
abundance. Most work that supports this contention is based
on the observation that when irradiance increases in a forest
as a result of disturbance, so does fruit abundance (Halls
1973; Piero and Sarukhan 1982; Clark and Clark 1987; Agren
1988; Levey 1990). Although I did not measure irradiance
directly, my estimates of leaf area index (LAI) describe
indirectly the light environment at my edges. I found that

76
(1) LAI did not correlate with fruit abundance at old edges,
(2) LAI did not change significantly with distance from
forest edge, (3) treefall gaps alone did not influence fruit
abundance (but see the significant interaction between
habitat and distance, and between habitat and month, on TF
and RF), and (4) distance from edge influenced fruit
abundance in complex ways. Thus, irradiance alone cannot
explain my findings. Other abiotic factors, such as
temperature, water availability, and nutrients, likely
influence fruit abundance across the pasture-forest edge.
In tropical areas flower and fruit abundance are
influenced by low temperatures (Tutin and Fernandez 1993),
soil fertility (Gentry and Emmons 1987), water availability
(Heideman 1989; Seghieri et al. 1995), and pollination
(Compton et al. 1994). During the dry season (June-August),
La Planada experiences clear skies, low rainfall, strong
winds, and extreme maximum and minimum temperatures. The
effect of these factors on understory vegetation may be
exacerbated at the pasture-forest edge, thereby influencing
changes in fruit abundance. For instance, during this
period, leaves of understory plants and vines at the forest
edge but not in the interior wilted and abscised. At a
montane locality north of La Planada, soil moisture across
the pasture-forest edge changed progressively over the dry
season and reached its lowest value 10 m from the forest edge
towards forest interior (Murcia 1993). At a lowland tropical
site, vapor pressure deficit (VPD) in the understory changed

77
across the pasture-forest edge, decreasing towards the forest
interior (Kapos 1989; Seizer 1992). More important, however,
were the differences found between the wet and dry season,
higher VPD values being recorded farther inside the forest
during the dry than during the wet season (Seizer 1992). At
the same lowland site, Kapos et al. (1993) compared the
carbon isotopic composition (313C) of leaves of two canopy and
two understory species and found that the 013C concentration
decreased from the edge towards the forest interior for the
understory but not for the canopy species. These changes
were more pronounced for Duauetia aff. flaoellaris
(Annonaceae) than for Astrocarvium sociale (Arecaceae). The
results described by Kapos et al. (1993) indicate that the
understory environment might be more sensitive to edge
creation than the canopy and that understory species vary in
their sensitivity to the factors that influence their
distribution.
At La Planada, increases in fruit abundance at the
forest edge could also be related to changes in soil
fertility as a result of increased litterfall during the dry
season, and deposition of volcanic ash at edges. La Planada
is influenced by several active volcanoes that release
andesitic ash (Mizota and van Reenwyk 1989, cited in van
Wambeke 1992) rich in nutrients (Shoji et al. 1993). These
air-borne particles might be deposited disproportionately
along pasture-forest edges. Some studies have shown that
edges alter the deposition of dry airborne material such that

78
deposition rates decrease from the forest edge towards the
forest interior (Geiger 1965; Draiijers et al. 1988). In an
area like La Planada, which is characterized by nutrient-poor
soils, the addition of nutrients could affect the production
of fruits.
A combination of several factors then may account for
the changes I observed in fruit abundance for individual
species and for the entire understory plant assemblage. I
propose that changes in TF-A and TI across the four distances
depending on month, are the result of within-year variability
in environmental conditions. On the other hand, changes in
TF and RF over the year may be the result of increased
irradiance at the forest edge due to direct exposure of the
pasture-forest interface to sunlight. Recall that TF and RF,
which included counts of palm fruits, were most influenced by
distance from forest edge. This is consistent with
Blanchard's (1992) results showing that increased abundance
of palm fruits at the forest edge is correlated with
increased light levels.
Consequences for Plants and Fruaivores
Studies looking at the effect of edges on plants have
focused more on the vegetative (e.g., Ranney et al. 1981;
Chen et al. 1992; Seizer 1992; Young 1993; Matlack 1993) than
on the reproductive stage (Romano 1990; Willimas-Linera 1990;
Blanchard 1992; Murcia 1993) of their life cycle. The former
provides information on the structure and productivity of the

79
edge. The latter provides information on the persistence of
edges through time, depending on the ability of plants to
complete their life cycle, including the production and
dispersal of seeds, in a given environment.
Increases in fruit abundance, and thus of seed outputs,
can have different consequences for individuals, populations,
and assemblages. A numerical increase in fruit production by
an individual can affect traits, such as seed size (Agren
1988, 1989), and thus seedling performance (Westoby et al.
1992). Changes in fruit numbers at the population level can
affect recruitment rates (e.g., Kellman and Kading 1992,
Guimaraes et al. 1994). For plant assemblages, an increase
in fruit numbers can affect colonization rates of disturbed
areas and thus alter species composition. In all cases,
changes in fruit numbers can affect the behavior of
dispersers (Murray 1987; Loiselle and Blake 1993).
At La Planada, the distribution of fruiting individuals
was influenced by the presence of edges. Changes in the
number of fruiting individuals at any given distance from the
forest edge may indicate differences in recruitment rates at
different distances. Plant establishment, growth,
reproduction, and seed dispersal are very likely to be
influenced by distance from forest edge in various ways
resulting in the observed distributions of fruiting
individuals (Chapter 3). It is possible that if edge
conditions remain unaltered over time, i.e., pastures and

80
fields are maintained as such, the observed distributions
will persist and even become more pronounced.
Sparse species constituted an important component of the
assemblage of understory plants of La Planada. Little is
known about habitat preferences of these species, but clearly
they represent species typical of large disturbed areas,
forest gaps, and forest (Appendix A). Sparse species were
most abundant close to the pasture-forest interface (D1 and
D2) within edges and at new edges within the study area (34
species were found exclusively at new edges compared with 14
at old edges) (Appendix A). In addition, for the few sparse
species from which I recovered seeds in bird droppings, I
found that the proportion of fruiting individuals at the
"edge" and "interior" was independent of the proportion of
bird droppings that contained their seeds and were recovered
in these two zones. Thus, forest edges are being colonized
by sparse species and factors other than seed dispersal might
be influencing recruitment rates across the pasture-forest
edge.
Changes in fruit abundance across the pasture-forest
edge partially paralleled that of bird captures (Chapter 5).
Increased fruit abundance, expressed as total fruits (TF) and
total ripe fruits (RF), was mirrored by an increase in
frugivore capture rates only at D1 (Chapter 5). The opposite
was true at D4 where fruit abundance reached the lowest
values but frugivore capture rates were the highest (Chapter
5). The high fruit production at Dl was mainly due to palms.

81
Bird species known to feed on palm fruits at La Planada,
including Mvadestes ralloides, Lipauous ttyptolQphys,
Semnornis ramohastinus, Piprgpla rieffgri, gnt^mpdeste?
coracinnus (C. Restrepo personal observation, Restrepo 1990)
and Andioena laminirostris (Beltran 1991) did not account for
the high capture rates observed at D1. Thus it is unclear
whether changes in TF and RF across the pasture-forest edge
influenced the behavior of frugivorous birds. The contrary-
might be true, however, for TF-A and TI. Those species known
to feed on fruits other than palms, which were the majority
of frugivorous species captured with mist nets, made an
important contribution to the high capture rates observed at
D1. Moreover, as occurred with TF-A and TI, there was a
significant interaction between distance from edge and month
on capture rates of frugivores (Chapter 5).
This study showed that edges influenced fruit abundance
in different ways. First, fruit production by the assemblage
of understory plants changed abruptly from forest edge
towards forest interior but depended on edge age, the
presence of treefall gaps, the length of the observations,
and whether or not palms were taken into accout. Second,
fruit production on a species by species level, and expressed
as the total number of fruiting individuals, changed across
the pasture-forest edge. This was shown both among abundant
and sparse species. The latter provided evidence to support
the contention that edges represent zones of opportunities
for the establishment of a wide range of species.

CHAPTER 5
EDGES AND UNDERSTORY BIRDS IN
A NEOTROPICAL MONTANE FOREST
Introduction
Natural disturbances play a major role in maintaining
high levels of diversity in tropical ecosystems (e.g.f
Connell 1978, Salo et al. 1986, Bush 1994, Gentry 1986; but
see Haffer 1969, Hubbell and Foster 1987). At regional
scales, one result of disturbance is the creation of
ecotones, which have been postulated to favor speciation
processes (Bush 1994) and high species richness (Terborgh
1977, Bush 1994). Human disturbances, on the other hand,
have resulted in a variety of land uses. Focus has now
shifted towards understanding how human disturbances affect
distributions of species (e.g., Kattan 1992) and how this
might impact ecosystem processes (Vitousek 1990, Kruess and
Tschnarntke 1994, Tilman and Downing 1994).
One consequence of human activities on landscapes is the
creation of sharp edges bounding disturbed areas, such as
pastures, logged forest stands, and agricultural fields.
These edges, whether found in little or highly modified
landscapes, may influence the movement of organisms between
the undisturbed and disturbed areas (Wiens et al. 1985, Wiens
1992) It is very likely that edges, by influencing the
82

83
movement of animals, might indirectly affect ecological
processes mediated through plant-animal interactions such as
pollination and seed dispersal. In tropical systems many
organisms are involved in plant-animal interactions. For
the most part, work done in the tropics has focused on how
edges affect animal distributions (Quintela 1986, Laurance
1990, Malcolm 1994). Less emphasis has been given to how
edges influence the distribution of animals mediating
ecological processes.
The extent to which edges can affect the distribution of
organisms varies with edge age (e.g., Williams-Linera 1990),
and land use (e.g., DeGraaf 1992) may determine the degree to
which edges can affect organisms. Equally important is the
variation among organisms in their response to edges (e.g.,
Kroodsma 1984, Noss 1991). Such variation can be used to
tease apart the mechanisms underlying such responses.
Possible mechanisms include changes in the resource base
(Malcolm 1991), parasites and predators (e.g., Gates and
Gysel 1978, Brittingham and Temple 1983, Loye and Carroll
1995), physiological condition of organisms (Wiens et al.
1985), dispersal, and home range size (Kuitunen and Makirn
1993) .
Here I report on how edges influence the distribution of
understory birds in a neotropical montane forest. I looked
at the effects of distance from the edge towards the forest
interior and time since edge creation on birds classified by
feeding guilds. I concentrated on frugivores and

84
nectarivores, since a high proportion of understory plants in
neotropical cloud forests rely on these two groups of
organisms for seed dispersal and pollination (Terborgh 1977,
Gentry 1983, Stiles 1985). Thus, changes in their
distribution may help explain how edges influence seed
dispersal and pollination in highly fragmented habitats.
Methods
Understorv Birds
I mist netted birds at six edges, three old and three
new, each netting site encompassing an area 100 x 200 m (2
ha) (Chapter 2). Strips at four distances from the forest
edge towards the forest interior (Dl: 0-10, D2: 30-40, D3:
60-70, and D4: 190-200 m) were divided into 5 plots (20 x 10
m) (Chapter 2) (Fig. 2-3). Three of these plots were chosen
at random and one pair of mist nets was placed in each, with
one net set perpendicular to the other. Nets were 9 X 2.5 m
with a 32 mesh. In each strip, nets were separated by a mean
distance of 40 m and positioned 0.5 m above the ground.
I operated 12 pairs of mist nets simultaneously from
0530-1300 for two consecutive days per month per distance per
edge, trying to complete when possible 14 hours of mist
netting per pair of mist nets. Mist netting started in June
1992 at the old edges. In September 1992 I included the new
edges. Thus the six edges were sampled simultaneously from
September 1992-August 1993, excluding December 1992 when I

85
did not sample birds. Because the sampling unit was a pair
of mist nets, instead of the traditional single net, I define
mist net hours as the hours that a pair of nets was opened.
In total the mist netting effort was equivalent to 11,892
net-hours. Mist nets were checked every 1-1.5 hours, and for
each captured bird I recorded species, mist net number, molt,
presence of cloacal protuberance or brood patch, fat, mass,
culmen length (total and exposed), culmen height and width,
tarsus and tail length, and wing chord. All birds, except
hummingbirds, were individually marked with color bands.
Hummingbirds were marked temporarily by clipping the tip of
their tail and wing feathers to recognize recaptures within a
mist netting session. Bird abundance is expressed throughout
this paper as capture rates, i.e., number of captures per
pair of mist nets per 100 mist net hours (mnh). Recaptures
on the same day were excluded from the analyses.
Birds were classified into four feeding guilds:
frugivores, insectivores, nectarivores, and carnivores.
Frugivores were defined as species that commonly consumed
fruit and/or seeds; most of them also consumed insects to
some degree. Insectivores ate primarily insects.
Nectarivores relied heavily on nectar and included nectar
"thieves." Carnivores primarily preyed on vertebrates. The
placement of any given species in one of these categories was
based on the analysis of fecal samples, my own observations,
and published reports (Miller 1963, Stiles and Skutch 1989,
Andrade et al. 1993, Arango 1993).

86
The use of mist nets versus acoustic and/or visual
censuses to carry out studies on bird assemblages in the
tropics has been widely discussed because of the biases
inherent in any sampling method (e.g., Terborgh and Weske
1969, Terborgh 1971, Karr 1971, Remsen and Parker 1983, Karr
1981a,t, Lynch 1989, Remsen 1994) It is accepted that mist
nets only sample a proportion of bird species found in an
area (Terborgh and Weske 1969, Karr 1981&); that if used over
a prolonged period birds learn the position of nets (Terborgh
1977, Bierregaard 1990); and that figures on bird abundance
might overestimate the abundance of many species (Karr 1981£,
Remsen and Parker 1983, Lynch 1989). In addition, when used
to compare habitats that differ markedly in structure,
capture rates can be misleading in regard to the presence and
abundance of many species (Terborgh 1971, Lynch 1989, Blake
et al. 1990). My study, then, only reflects what happens to
those birds that are effectively sampled by mist nets in the
understory (see also Wong 1986, Levey 1988,b, Blake and
Loiselle 1991, Loiselle and Blake 1991, Poulin et al. 1992).
I stress that mist netting took place only inside the forest,
and the aim of this study was to compare changes in bird
abundance from the edge towards the forest interior. Thus,
problems associated with habitat biases are either minimized
or held constant.
One possible problem for interpreting the results,
however, relates to the timing of mist netting in old and new
edges. In old edges mist netting began in June 1992 and in

87
new edges in September 1992. Higher capture rates in new
than old edges in September and October could be attributed
to birds having learned the position of nets in the old
edges. If this increase in capture rates in new edges was a
consequence of a learning process then I would expect (1) a
decrease in the proportion of recaptures over time for old
and new edges, and (2) a higher proportion of recaptures in
new than old edges during these months. The data for all
species (excluding hummingbirds), frugivores, and
insectivores, do not support these predictions. I conclude
that any observable difference can therefore be attributed to
differences between the two types of edges.
Data Analysis
To establish changes in the abundance of the understory
avifauna of La Planada as a function of distance from the
edge and edge age I analyzed capture rates for all bird
species combined and for three feeding guilds. Carnivores
were excluded from the analyses because of small sample size.
I used ANOVAs for Mixed Factorial Designs (Girden 1992).
Edge age (old and new) was included as a between-subject
factor. Month (September 1992-August 1993) and distance (Dl-
D4) were included as within-subject factors or repeated
measures. Month and distance were included as within-factors
because of restrictions in the randomization procedure when
"assigning" month and distance levels to each edge, which can
lead to correlations between the observations (Girden 1992,

88
Manly 1992). Edges were treated as subjects because each
edge was measured repeatedly for each of the different
treatment combinations. Because mist nets are nested within
distance and capture rates were zero for many pairs of mist
nets at a given month/distance/edge, I averaged capture rates
for each three pairs of mist nets/month/distance/edge. This
procedure reduced the dimensionality of the data and also
made the data more normally distributed by eliminating many
zero values. I plotted the residuals as a function of fitted
Y values to detect any violation of assumptions (Manly 1992).
The data for all four ANOVAs presented in this paper were
square-root-transformed. In addition, I verified the
assumption of compound symmetry (i.e., the covariation
between each pair of treatments is equal for all subjects)
for ANOVAs that included within-factors (Girden 1992, Manly
1992). When compound symmetry is violated, the probability
of committing a Type I error increases. To account for this,
the degrees of freedom have to be corrected by a factor,
epsilon, which ranges from 1/(J-1) to 1.0, where J is the
number of levels in a treatment. The closer epsilon is to
1.0, the lower is the probability that compound symmetry is
being violated (Girden 1992). Epsilon is estimated based on
the conservative Geisser-Greenhouse method and the more
liberal Huynh-Feldt method (Girden 1992). In this paper I
report the corrected F values based on the liberal Huynh-
Feldt method (H-F).
In addition to the above omnibus ANOVA tests, I

89
specified contrasts of mean differences to test specific
hypotheses. These hypotheses included effects involving
single factors and interactions (Gagnon et al. 1989, Girden
1992). For the distance effect I specified two contrasts, by
comparing mean capture rates at D1 and D4 separately with
those of D2 and D3 together. I assumed that changes in bird
distribution, if any, would be more marked at the extremes.
For the distance x age interaction I specified a single
contrast, by comparing mean capture rates at D4 between old
and new edges. For the month x age and the month x distance
interactions I specified two contrasts for each, comparing
dry with wet months. I reasoned that because of marked
differences in the rainfall regime at La Planada, changes in
bird abundance between habitats (distance or type of edge)
were more likely to occur between dry and wet months. Dry
months were those exhibiting the lowest rainfall records
(February and July) and the previous month when rainfall
started to decrease (January and June). Wet months were
those that received the highest rainfall (April and October)
and the previous month when rainfall started to increase
(March and September) (Fig. 2-2). For the interaction
between month and age I compared the mean number of captures
between old and new edges during the dry and wet months. For
the interaction month x distance I compared the mean number
of captures between D1 and D4 during the dry and wet months.
I report the corrected F-values and associated probability in
the results section. All analyses were performed using

90
SuperANOVA (Gagnon et al. 1989).
To determine if individual species were affected by the
presence of edges, I performed two related tests based on
data collected between June 1992 and August 1993. First, I
used a Goodness of Fit test (G-statistic) to determine if the
number of captures across the four distances departed
significantly from a uniform distribution. Second, I used a
Replicated Test of Goodness of Fit (G-statistic) (Sokal and
Rohlf 1981) to establish, in addition to the distance effect,
an edge age effect on the number of captures. For both
tests, monthly captures for each species were pooled, keeping
separate the information on old and young edges. For these
analyses I chose those species in which at least 80 percent
of the expected cell frequencies were greater than 5, since
the statistic G departs from the X2 distribution if this is
not the case (Siegel and Castellan 1988). Five species
(Lipaugus cryptolophus. Ocreatus underwoodii. Tanqara rthys,
T. labradorides, and T. nigroviridis) did not meet this
criterion but were included in the analyses. The first two
were included because of the clear trends they exhibited.
The last three species were lumped in the Tanaara spp. for
the analyses because of similarities in many features of
their life history (Isler and Isler 1987).
A vast majority of the species did not meet the above
criterion. To evaluate the influence of edges on these
species I classified birds into five categories according to
capture frequency: extremely sparse (1 capture), very sparse

91
(2-5 captures), sparse (6-20 captures), abundant (21-50
captures), and very abundant (> 51 captures). I assigned
capture numbers for each category to the four distances. I
used a Chi-square test to evaluate the association between
bird abundance and distance and used the residual values to
determine the contribution of each cell to the overall result
(Siegel and Castellan 1988) .
I used an alpha of 10%. The sampling procedure used in
my study resulted in an increase of Type II errors and low
power of my tests (see Chapter 4). Increasing alpha
counterbalances these effects at the cost of increasing Type
I errors (Zolman 1993).
Results
Understorv Birds
From September 1992 through August 1993 I accumulated
1,789 captures of 80 species. Bird captures differed
significantly among the four distances, suggesting that the
abundance of understory birds at La Planada is affected by
the presence of edges (ANOVA, F4#i6 = 7.6, P = 0.004; Table 5-
1, Fig. 5-1) Mean capture rates were significantly higher
at D1 (mean = 16.9) and D4 (mean = 19.3) than at D2 and D3
combined together (mean = 12.1) (Contrast of Mean
Differences, Fi/i2 = 10.7, P = 0.007 and Fi,i2 = 18.4, P =
0.001, respectively). Edge age, however, modified the
distance effect as shown by the significant distance x edge

Table 5-1. Results of ANOVAs for Mixed Factorial Designs (1 between-, 2 within-factors) on
mean capture rates x 100 mist netting hours (mnh) of all birds, frugivores, insectivores,
and nectarivores. The month and month x distance x age effects were excluded from this
table. () based on square root-transformed data. Significance at 10% (*), 5% (**), 1%
(***) .
Age (A)
Error
Edge (Age)
Distance
Distance x Age
Error
D x E (A)
df
F
df
SS
df
F
H-F
df
F
H-F
df
SS
All Birds
1
2.89
ns
4
11.39
3
7.63

3
3.54

12
22.39
Frugivores
1
1.77
ns
4
14.27
3
4.85
*
3
2.75
*
12
25.03
Insectivores
1
0.24
ns
4
18.42
3
2.79

3
1.32
ns
12
27.86
Nectarivores
1
0.05
ns
4
17.78
3
1.78
ns
3
1.83
ns
12
24.67
Month x Age
Error
M x E (A)
Month x
Distance
Error
D X M X E (A)
df
F
H-F
df
SS
df
F
H-F
df
SS
All Birds
10
1.78

40
30.07
30
1.10
ns
120
85.84
Frugivores
10
0.79
ns
40
33.99
30
1.93
*
120
106.78
Insectivores
10
3.23
*
40
39.91
30
0.78
ns
120
114.94
Nectarivores
10
0.52
ns
40
41.85
30
0.82
ns
120
147.25

93
Figure 5-1. Variation in the distribution of understory birds
at La Planada in relation to distance from the forest edge.
Points represent means and bars standard errors.

94
interaction (ANOVA, F4,i6 = 3.5, P = 0.05; Table 5-1, Fig. 5-
2). Capture rates remained very similar at DI, D2, and D3
for old and new edges but differed at D4, where they were
higher at new edges (mean = 24.2) than at old edges (mean =
14.1) (Contrast of Mean Differences, Fi,i2 = 12.0, P = 0.005).
The effect of month on capture rates for all birds was
significant (ANOVA, Fi5,5g = 2.9, P = 0.008). However,
changes in bird captures over the year depended on edge age
as shown by the significant edge age x month interaction
(ANOVA, F = 15,59 = 2.1, P = 0.09; Table 5-1, Fig. 5-3). This
suggests that edge features, such as edge age, can influence
the distribution of birds, depending on season. Old and new
edges did not differ significantly in regard to bird captures
during the dry months (Contrast of Mean Differences, Fi,4o =
1.5, P = 0.2) but they did differ during the wet months
(Contrast of Mean Differences, Fi,4o = 13.1, P = 0.0008).
Feeding Guild Responses to Edges
The abundance of frugivores and insectivores differed
significantly with distance (ANOVA, F3,i2 = 4.9, P = 0.02 and
F3,i2 = 2.1, P = 0.08, respectively, Table 5-1, Fig. 5-1).
Capture rates for frugivores were significantly higher at D1
(mean = 5.6) and D4 (mean = 7.0) compared to D2 and D3
combined (mean = 3.5) (Contrast of Mean Differences, Fi,i2 =
7.0, P = 0.02 and Fi,i2 = 12.0, P = 0.005,

95
Figure 5-2. Variation in the distribution of understory birds
at La Planada in relation to edge age and distance from the
forest edge. Points represent means and bars standard errors.

96
MONTH
Figure 5-3. Variation in the distribution of understory birds
at La Planada in relation to edge age and month. Points
represent means and bars standard errors.

97
respectively). Capture rates per 100 mist-net hours for
insectivores were significantly higher at D4 (mean = 8.0)
compared to D2 and D3 combined together (mean = 5.0)
(Contrast of Mean Differences, Fifi2 = 8.3, P = 0.01) but did
not differ between D1 and D2-D3 combined. The omnibus test
showed that distance from the forest edge did not affect
nectarivore capture rates (Table 5-1, Fig. 5-1).
Nevertheless, the specific hypothesis tested by contrast of
mean differences showed that nectarivores were more abundant
at D1 (mean = 5.4) than at D2 and D3 combined (mean = 3.5)
(Contrast of Mean Differences, Fifi2 = 4.9, P = 0.05).
Among frugivores, the distance effect was modified by
edge age, as shown by the significant interaction between
these two factors (ANOVA, F3fi2 = 2.7, P = 0.09; Table 5-1,
Fig. 5-2). Capture rates at DI, D2, and D3 were similar
between old and new edges. However, capture rates at D4 were
significantly higher at new (mean = 9.4) than at old (mean =
4.5) edges (Contrasts of Mean Differences, Fifi2 =7.2, P =
0.02). The interaction between distance from the edge and
edge age in the omnibus test was not significant for
insectivores and nectarivores (Table 5-1, Fig. 5-2).
Nevertheless, when testing specific hypotheses I found that
capture rates for nectarivores at D4 were significantly
higher at new (mean = 5.5) than at old (mean = 3.3) edges
(Contrast of Mean Differences, = 4.0, P = 0.07).

98
MONTH
Figure 5-4. Variation in the distribution of understory birds
at La Planada in relation to distance from forest edge and month.
Points represent means and bars standard errors.

99
The distance effect was also affected by month for
frugivores as shown by the significant interaction between
these two factors (ANOVA, F45fi8i =1.93, P = 0.007; Table 5-
1, Fig. 5-4). I found that capture rate of frugivores was
significantly higher at D4 (mean = 8.4) than at D1 (mean =
4.7) during the dry months (Contrast of Mean
Differences, Fifi20 = 7.3, df = 1, P < 0.008) but not during
the wet months. The distance x month interaction in the
omnibus test was not significant for insectivores and
nectarivores (Table 5-1, Fig. 5-4). However, when I tested
specific hypotheses I found that capture rates of
insectivores were higher at D4 (mean = 8.0 and 8.4 for dry
and wet season, respectively) than at D1 (mean = 4.2 and 6.6
for dry and wet season, respectively) in both seasons
(Contrast of Mean Differences, Fi>i20 = 6.7, P = 0.01). For
nectarivores I found that capture rates were higher at D1
(mean = 5.6) than at D4 (mean = 3.2) only during the dry
season (Contrast of Mean Differences, Fifi20 = 5.7, P = 0.02).
The abundance of insectivores changed in old and new
edges depending on the month of the year, as shown by the
significant interaction between month and edge age (ANOVA,
Fl6, 64 = 3.24, P = 0.004; Table 5-1, Fig. 5-3). Insectivores
were more abundant at new edges (mean = 7.9) than at old
edges (mean = 5.4) only during the wet season (Contrast of
Mean Differences, Fi(4q = 7.0, P = 0.01).

100
Species Level Responses
From June 1992 through August 1993 I captured 2,101
birds of 82 species. Ninety percent of the captures
represented 22 species that were classified as abundant (21-
50 captures) and very abundant (>51 captures) (Appendix B).
Insectivores accounted for 39% of all bird captures,
frugivores for 32%, nectarivores for 28%, and carnivores for
1%.
Twenty-four species and the complex Tanaara spp. (which
includes T. arthus, T. labradoridsS> and T. niqroviridis)
were evaluated to establish variation in their abundance with
distance from the edge. The distribution of bird captures
for 17 out of the 24 common species departed significantly
from a uniform distribution across the four distances (Table
5-2). Given a 10% probability of obtaining a species that
shows a non-uniform distribution, it is very unlikely that 17
or more species out of 24 would have shown a non-uniform
distribution just by chance alone (Binomial Test, P = 1.6 x
10-12) # i conclude that the distribution of understory birds
is affected by edges.
Based on the observed and expected cell frequencies I
further divided the species exhibiting a non-uniform
distribution into three groups: (1) species for which the
observed number of captures at D4 was higher than expected,
i.e., increasing from edge to interior, (n = 8), (2) species
for which the observed number of captures at D1 was higher

101
Table 5-2. Distribution of common (>21 captures) understory
bird species of La Planada in relation to distance from edge.
Numbers are number of captures.
P< 0.01 (***), P < 0.001 (****)
P <
0.1
(*) ,
P < 0.
05 (**),
Distance
G-stat
P
D1
D2
D3
D4
Uniform Distribution
Allocotopterus deliciosus
13
10
11
12
0.42
ns
Eunhonia xanthocraster
32
22
17
24
4.16
ns
Pipreola riefferii
7
5
4
5
0.89
ns
Tanqara spp.
8
4
4
11
5.08
ns
Basileuterus tristriatus
23
14
25
20
3.54
ns
Grallaricula flavirostris
9
13
19
21
6.14
ns
Mviphobus flavicans
9
6
3
5
3.27
ns
Preumgrnis guttulioera
25
33
25
27
1.52
ns
Non-Uniform Distribution
Increase from Edge to Interior
Atlaoetes brunneinucha
3
4
11
10
7.52

Lipauqus crvptolophus
1
2
5
8
7.78

Mionectes striaticollis
30
16
15
61
41.63

Glvohorrvnchus soirurus
10
9
10
30
18.17
****
Mviotriccus ornatus
11
6
9
23
12.51
*
Premnoplex brunnescens
8
20
25
40
23.94

Pseudotriccus oelzelni
13
14
15
40
21.58

Aalaiocercus coelestis
50
53
68
85
11.82
*
Decrease from Edge to Interior
Chlorosoinaus semifuscus
13
2
3
5
13.64

Cgeliqena wilsoni
42
25
18
28
10.49

Haoloohaedia luaens
20
11
13
7
6.87

Ocreatus underwoodii
10
3
4
1
9.6

Increase at Edge and Interior
Mas,US chrvsopterus
37
27
25
58
17.47

Mvadestes ralloides
25
16
9
37
20.58

Henicorhna leucoohrvs
38
19
24
34
8.19
* *
Svndactvla subalaris
14
6
6
14
6.59

Phaetgrnis svrmatophorus
62
21
34
47
23.28


102
than expected, i.e., decreasing from edge to interior (n =
4), and (3) species for which the observed number of captures
at D1 and at D4 was higher than the expected, i.e., increase
both at the edge and forest interior, (n = 5) (Table 5-2).
These patterns suggest that edges might influence the
distribution of birds in complex ways, such that some species
avoid edges, others are attracted to edges, and still others
are influenced by factors other than distance from forest
edge. This latter situation is suggested by those five
species showing an increase both at forest edge and forest
interior.
Fourteen species were evaluated to determine the
combined effect of distance from the edge and edge age on
their abundance (Table 5-3). I recognized two main groups:
(1) species showing a significant interaction between
distance from the edge and edge age (Gneterogeneity/ P < 0.1)
and (2) species showing a non-significant interaction between
distance from the edge and edge age (Gneterogeneity> P > 0.1;
Table 5-3). Within these two groups species showed different
responses across the four distances depending on edge age
(Gold and Gftew, P < 0.1; Table 5-3). Thus, the distribution
of individuals within a species not only changes from edge
towards forest interior but also varies with edge age.
For all species combined, I found a significant
association between species abundance and distance from the
edge (%2 = 24.4, df = 12, P = 0.02; Table 5-4). Moreover,
upon examination of the residuals I found that the

Table 5-3. Results of Replicated Goodness of Fit Test on the number of captures for
understory birds at the Reserva Natural La Planada. Gneterogeneity (Gh) GTotal (Gt) #
Gold edges (G0), GNew edges (GN) P< 0.1 (*), P< 0.05 (**), P< 0.01 (***), and P< 0.001
(****)
df
gh
df
Gt
df
Go
df
Gn
MiQnectes striaticollis
3
19.89****
6
61.53****
3
6.43*
3
55.07****
Mvadestes ralloides
3
8.76**
6
29.31****
3
17.95****
3
13.16***
PremnQPlex brunnescens
3
13.14***
6
37.08****
3
9.15**
3
27.90****
Aglaiocercus coelestis
3
6.56*
6
18.37***
3
9.74**
3
8.6**
Euphonia xanthoaaster
3
13.50***
6
17.74***
3
1.22
3
16.51****
AllocotoDterus deliciosus
3
1.46
6
1.88
3
0.44
3
1.43
Pasiieuteru? tristriatus
3
2.66
6
6.19
3
1.42
3
4.77
Srallaricula flavirostris
3
0.30
6
6.42
3
3.72
3
2.71
Premnprnis auttuliaera
3
1.50
6
3.02
3
1.18
3
1.83
Pseudotriccus pelzelni
3
2.12
6
23.70****
3
11.64***
3
12.05***
Phastornis svrmatODhorus
3
4.24
6
27.5****
3
13.97***
3
13.56***
Masius chrvsoDterus
3
4.22
6
21.73***
3
7.23*
3
14.42***
Henicorhina leucoDhrvs
3
3.72
6
11.92*
3
1.48
3
10.41**
Cp.eliqpna wilsoni
3
0.79
6
11.26*
3
8.26**
3
3.0
103

104
Table 5-4. Distribution of understory birds across the
pasture forest-edge. Birds were classified according to
abundance. Numbers represent number of captures. Carnivores
were excluded because of small number of captures.
Distance
D1 D2 D3 D4
Extremely sparse
(1 capture)
All Birds
12
3
6
4
Frugivores
5
2
0
0
Insectivores
3
1
3
1
Nectarivores
3
0
3
2
Very sparse
(2-5 captures)
All Birds
31
11
12
14
Frugivores
10
3
8
2
Insectivores
9
3
5
8
Nectarivores
11
0
1
4
Sparse
(6-20 captures) All Birds
41
32
24
43
Frugivores
9
13
10
23
Insectivores
17
13
9
19
Nectarivores
15
6
4
1
Abundant
(21-50 captures)
All Birds
70
39
46
75
Frugivores
36
21
28
33
Insectivores
34
18
18
42
Nectarivores
Very abundant
(>.51 captures)
All Birds
423
314
342
557
Frugivores
124
81
66
179
Insectivores
126
122
143
211
Nectarivores
173
111
133
167

105
contribution of extremely sparse and very sparse species at
D1 was disproportionally high. There was also a significant
association between species abundance and distance from the
edge for frugivores and nectarivores (x2 = 32.8, df = 12, P =
0.001 and X2 = 27.0, df = 9, P = 0.01, respectively; Table
5-4). An examination of the residuals for frugivores showed
that (1) extremely sparse species were captured more
frequently than expected at D1 than at the other three
distances, (2) very sparse and abundant species were captured
more frequently than expected at D3 than at the other
distances, and (3) very abundant species were captured more
frequently than expected at D4 than at the other distances.
Among nectarivores the residuals showed that (1) very sparse
and sparse species were captured more frequently than
expected at D1 than to the other distances and (2) very
abundant species were captured more frequently than expected
at D4 than to the other distances.
Discussion
The distribution of understory birds at La Planada
varied across pasture-forest edges in complex ways, not
always reflecting changes due to the presence of edges. This
complexity is demonstrated by (1) differences among the
response variables and (2) significant interactions between
distance from edge, edge age, and month. Capture rates for
all birds, frugivores, and insectivores changed across the
pasture-forest edge, but the same was not true for

106
nectarivores. In my study area, and averaging over time,
capture rates for all birds and frugivores showed bimodal
distributions peaking, at D1 (0-10 m) and D4 (190-200 m).
Capture rates for insectivores showed an abrupt increase at
D4. Time since edge creation, however, had a major effect on
capture rates for all birds and frugivores across the
pasture-forest edge, especially at D3 (60-70 m) and D4 (Fig.
5-2). Within edges, capture rates of frugivores varied among
the four distances on a monthly basis. Between edges,
capture rates of insectivores varied between old and new
edges on a monthly basis. In discussing these findings I
will (1) compare them against results from studies conducted
in other tropical areas and (2) look at possible factors that
might influence the distribution of birds across pasture-
forest edges.
Patterns of Bird Distribution across Edges
My results agree with those of other studies on edges in
that (1) abundance (or density) of organisms changes from
edge towards forest interior (Quintela 1986, Laurance 1990,
Malcolm 1994), (2) functional or taxonomic groups of
organisms, as well as individual species, respond in
different ways to the creation of edges (e.g., Noss 1991),
and (3) the magnitude and direction of the responses to the
creation of edges varies depending on time since edge
creation and season.
A study conducted in the Amazon showed that overall

107
abundance of birds in forest stands increased from the edge
towards forest interior (Quintela 1986; but see Kroodsma 1984
and Lopez de Casenave et al. ms for opposite results). Edge
age and land use can affect the magnitude of such effects
(DeGraaf 1992). At larger scales, landscape configuration
determines the overall effect of edges on the distribution of
organisms (Hansen et al. 1992).
I reanalyzed Quintela's data on bird captures and found
that, in central Amazon, understory insectivores (Goodness of
Fit Test, G = 49.6, P < 0.001) and frugivores increased
(Goodness of Fit Test, G = 5.8, P <0.1) from edge towards the
forest interior (500 m from forest edge), whereas
nectarivores decreased (Goodness of Fit Test, G = 7.1, P
<0.05). In the Chaco Argentino, bird captures of terrestrial
and arboreal granivores, i.e., doves and parrots, was higher
at forest edge than at forest interior (Lopez de Casenave et
al. ms). Mist netting did not reveal differences among the
other guilds. The Amazon site and La Planada differ markedly
in their understory avifauna (Bierregard and Lovejoy 1989).
Nevertheless, results of both studies show that (1) abundance
of all birds, frugivores, and insectivores is highest in the
forest interior and (2) abundance of nectarivores is highest
at the forest edge. In both areas, pastures and second
growth areas are embedded in a forest matrix (Lovejoy et al.
1986, Chapter 2).
In central Sweden, bird density across clear cut-forest
edges changed over a year (Hansson 1983). Bird density was

108
higher at the "edge" (50 m from edge), than at the forest
interior (>50 m from edge) and these differences were more
pronounced in the summer than in the winter (Hansson 1983).
In central Florida, overall bird density was higher near the
forest edge (50 m from edge) than farther away, and this
difference was greater during winter than at other times of
the year at east-facing edges (Noss 1991). Noss also found
variability among years in the response of birds to edges.
At La Planada, capture rates for all birds and for frugivores
were greater at D4 (190-200 m) than at D1 (0-10 m) during dry
months. Capture rates for insectivores were higher at D4
than at D1 and they did not differ between the wet and dry-
season. For nectarivores capture rates were higher at D1
than at D4 only during the dry season. The within-year
variability in bird abundance across forest edges suggests
that edges are dynamic. In addition, this variability is
suggestive of possible mechanisms underlying the observed
patterns.
Factors Influencing the Distribution of Birds across Edges
Environmental factors and intrinsic features of
organisms may influence their distribution across edges
(Wiens 1992). Most of these factors have been examined in
temperate zones, especially in areas where forest fragments
are embedded in an agricultural matrix. They include nest
parasitism and egg predation (Gates and Gysel 1978, Wilcove
et al. 1986, Andrn and Algestam 1988, Moller 1989), and

109
predation and parasitism of adults (Wiens et al. 1985, Loye
and Carroll 1995). In addition, edges can directly influence
the distribution of organisms by affecting their
physiological condition (Wiens et al. 1985), probability of
establishment in a given area depending on home range size
(Kuitunen and Makirn 1993), and dispersal abilities.
In tropical areas there has been little effort to
elucidate factors underlying changes in the distribution of
animals across edges (Malcolm 1991). Some studies using
artificial nests have shown that nest predation increases
towards forest edges where second growth vegetation abuts
undisturbed forest (Gibbs 1991, Burkey 1993). However, nest
predation rates did not change across pasture-forest edges at
three other sites, including La Planada (Arango 1991,
Laurance 1993, C. Restrepo and C. Samper, unpublished data).
In an Amazonian site, Malcolm (1991) found that the
distribution of small mammals changed between "interior" and
"edge" in continuous tracts of forest. These changes
mirrored those in forest structure and abundance of insects,
and he concluded that changes in resource levels could
explain changes in small-mammal assemblages in the Amazon
across pasture-forest edges.
At La Planada the distribution of frugivores might be
partially influenced by the distribution of resources across
the pasture-forest edge. Fruit abundance (measured as total
number of fruits and as number of ripe fruits), decreased
significantly from the edge towards the forest interior

110
(Chapter 4). Capture rates of understory frugivorous birds
followed a similar trend up to 70 m from the edge. Fruit
abundance (measured as total number of fruits excluding palms
and as total number of fruiting individuals), differed across
pasture-forest edge depending on month (Chapter 4).
Frugivores, but not insectivores or nectarivores, changed in
the same fashion, as indicated by the significant distance x
month interaction.
Two other factors influenced the distribution of
frugivores across pasture-forest edge. First, at D4 fruit
abundance was the lowest, yet capture rates of frugivores
were the highest. This increase in capture rates was
probably influenced by the presence of two leks (Mionectes
striaticollis and Masius chrvsooterus) in the vicinity of D4
in one of my study edges (Hermgenes). It is impossible to
tell if the establishment of these leks was related to edge
creation 12 years ago. The presence of the leks, however,
certainly acts as an attractor, thus affecting the
distribution of M. striaticollis and M. chrvsooterus in one
of the edges. Second, the sharp increase in capture rates
of insectivores at D4 suggests that some structural feature
of the forest and/or resource covarying with structure
changed at D4 and that both frugivores and insectivores
responded to this. A gentler topography and the presence of
larger gaps at D4 when compared to the other distances may
result in changes in forest structure. Thus, as pointed out
by Wiens (1992), the distribution of organisms across edges

Ill
is influenced not only by distance from forest edge but also
by species characteristics, such as their behavior.
The distribution of nectarivores showed a sharp decrease
from edge to forest interior, even though the omnibus test
was not significant. In Brazil, Quintela's (1986) data show
the same trend but it was significant. In the lowlands of
Costa Rica, Blake and Loiselle (1991) found that capture
rates of nectarivores decreased from young second growth to
old second growth to forest. In Monteverde (Costa Rica)
cloud forest, Feinsinger et al. (1987) found that the
frequency of visits by hummingbirds decreased from large gaps
to small gaps to forest. At La Planada, the distribution of
hummingbird-pollinated plants showed a significant decrease
from edge towards forest interior (Goodness of Fit Test, y2 =
23.01, df = 3, P < 0.05), suggesting that resources might be
responsible for the observed trend in nectarivores. This at
least can explain why very sparse and sparse nectarivores
were captured more often at D1 than at the other distances.

CHAPTER 6
FRUGIVOROUS BIRDS IN FRAGMENTED NEOTROPICAL MONTANE FORESTS
LUMP STRUCTURE IN BODY MASS
Introduction
Transformation of tropical landscapes by humans has
influenced plant and animal assemblages in many ways. Most
studies have emphasized how species abundance and richness
change with increasing forest fragmentation (Quintela 1986,
Bierregaard and Lovejoy 1989, Klein 1989, Newmark 1991,
Estrada et al. 1993, Kattan et al. 1994, Malcolm 1994,
Didham, in press, Lynam, in press, Warburton, in press,
Chapter 5) and with transformation of native forests into
second growth and managed ecosystems (Holloway et al. 1992,
Johns 1992, Lambert 1992, Thiollay 1992, Andrade and Rubio
1994, Escobar 1994). Results of these studies vary
considerably, reflecting the complexity of relating habitat
modification to biodiversity loss, but also inherent
differences among study sites, and/or a mismatch between the
scale of the problems being addressed and the methods used.
As a consequence it has been difficult to establish patterns
regarding how habitat modification and biodiversity interact
and moreover, how they relate to ecosystem processes
112

113
(Vitousek 1990, Kruess and Tschnarntke 1994, Pimm and Sugden
1994, Tilman and Downing 1994, Turner et al. 1995).
Holling et al. (1995) provided a novel conceptual
framework that may help integrate these issues. First, long
term research has shown that ecosystems are structured by a
few processes operating at various spatial and temporal
scales (e.g., Clark et al. 1979, Harris 1980, Gunderson
1992). Second, the behavior and morphology of organisms
reflect the discontinuous nature of ecosystems. In
particular, body mass of boreal birds and mammals was found
to be discontinuously distributed, such that species of
similar mass tend to aggregate or lump together (Holling
1992) By focusing on lumps, rather than species, it may be
possible to detect patterns at larger spatial scales such as
those defined by the impact of humans on landscapes. Third,
the morphology of organisms reflects differences in landscape
pattern. In particular, the distribution of body mass in
boreal forest and boreal prairie birds differed (Holling
1992). By focusing on the distribution of body mass, i.e.,
lump structure, it could be possible to investigate the
interaction between landscape pattern and the structure of
animal and plant assemblages.
In this paper I ask how changes in landscape pattern
affect assemblages of frugivorous birds in neotropical
mountains. I concentrate on frugivorous birds because seed
dispersal by birds is especially important in neotropical
mountains compared to the lowlands (Terborgh 1977, Stiles

114
1985, Gentry 1988, Renjifo et al., in press), offering an
opportunity to relate the process of seed dispersal to
landscape pattern. I focus on mountains because natural and
human disturbances have generated complex landscapes over
small areas (Haslett 1994), offering an ideal opportunity to
assess how changes in landscape pattern influence animal
assemblages.
I make four comparisons representing two different
scales of inquiry and address how the distribution of body
mass, i.e., lump structure, in frugivorous birds changes from
areas covered mostly by forest to areas covered by open
vegetation. The first scale is defined by elevational zones
within the mountains of Colombia, the second by sites within
elevational zones that have been differently affected by
human activities. I use body mass as an attribute that
reflects information not only on life-history traits, such as
dispersal (Laurance 1991, Lawton et al. 1994, Brown 1995,
Gaston and Blackburn 1995), but also on foraging behavior,
such as size of seeds being dispersed (Moermond and Denslow
1985).
MethQd?
Study Area
My study focused on the Andes of Colombia, South
America. This system consists of three mountain ranges of
different geological origin, each running in a S-N direction

115
(Irving 1975). I defined montane habitats as those above the
800 m topographic contour line (Fig. 6-1) In the absence of
human disturbance this region would be covered by forest
except for the pramo and small areas affected by rain
shadows (Cuatrecasas 1958). Forest composition, structure,
and physiognomy change along the mountains from the complex
lowland tropical forest to the simpler pramos (Cuatrecasas
1958, Espinal et al. 1977).
The area encompassed by this study represents less than
35% of the total area of Colombia (1,380,000 km2) yet harbors
one of the richest biotas not only of this country but of the
neotropics (Duellman 1979, Henderson et al. 1991, Gentry
1992i,b, Renjifo et al. in press). It has been postulated
that the elevated levels of diversity and endemism of this
area are the result of an intense disturbance regime (Gentry
1992a, J. Luteyn pers. comm.). A complex topography and
geology, combined with high precipitation, generates
landslides, mud flows, avalanches, and volcanic eruptions,
which continuously transform these mountains (Mejia et al.
1994, Velsquez et al. 1994).
Superimposed on the natural disturbance regime is one
generated by human activities. At least 50 per cent of the
total population of Colombia (37 million people) has settled
in montane areas (Banguero 1993) Presently, less than 30
percent of this area is covered by forest, most of which is
found either at elevations >2,500 m or on wetter slopes of
the cordilleras (Cavelier and Etter, in press). The remaining

116
78 76 74 72 70 65
Figure 6-1. Montane habitats of Colombia (>800 m) showing
the location of sites included in this study.

117
area has been transformed into pastures, cultivated fields,
coffee and tree plantations, and urban areas. A recent surge
in demand for opium derivatives has prompted forest clear-
cutting at higher altitudes to grow poppy (Papaver
somniferum: Cavelier and Etter, in press).
Fruaivorous Birds
I included all species reported to consume fruits and/or
seeds to any degree (Fitzpatrick 1980, Hilty and Brown 1986,
Isler and Isler 1987, Renjifo 1988, Ridgely and Tudor 1989,
Stiles and Skutch 1989, Fjelds and Krabe 1990, Velsquez
1992, Arango 1993, 1994, Ridgely and Tudor 1994, L.M. Renjifo
and C. Restrepo personal observations) and found at an
elevation >800 m. Thus, my data combine seed dispersers and
seed consumers.
Body mass data were obtained from published records
(Goodwin 1976, Isler and Isler 1987, Stiles and Skutch 1989,
Dunning 1993, del Hoyo et al. 1992, Arango 1993), museum
specimens (Coleccin de Ornitologa, Universidad del Valle,
Cali, Colombia), and my own field observations. For bird
species that I could not obtain mass measurements, I averaged
the available mass for congeners of the same length. I could
not estimate body mass for a small fraction (2%) of the
species and these were not included in the analyses.
To explore the relationship between body mass and
landscape pattern along the elevational gradient, I
classified birds into four groups based on their elevational

118
ranges: upper lowland (from ca. 800 m to 1,500 m) lower
montane (from ca. 1,500 m to 2,400 m) upper montane (from
ca. 2,400 m to 3,400 m) and pramo (from ca. 3,400 to 4,800
m) species (Table 6-1). Note that some species fell in more
than one elevation zone and were entered into the analyses
two or more times. Elevational zones follow Chapman (1917).
Elevational ranges were taken from Hilty and Brown (1986).
To explore the relationship between body mass and
landscape pattern along a gradient of land-use, I obtained
bird inventories for 18 sites (Table 6-1). All inventories
were conducted by experienced ornithologists over periods of
>1 yr and included visual, auditory, and mist netting
observations. I believe these lists represent unbiased and
fairly complete inventories, so additions would have little
influence on the results. The 18 sites were grouped
according to elevational zone and type of land use. Within
each comparison I arranged sites from those covered mostly by
forest to those covered by open vegetation (Table 6-1).
This work relies on two assumptions. First, there is a
common pool of species for elevational zones and sites, but
historical, geographical, and climatic events, and more
recently human activities, have determined the set of species
found today at any one site. Second, within elevational
zones the less disturbed sites represent the conditions that
existed at the other sites prior to human intervention.

Table 6-1. Sites included in this study.
site
Coordinates Elevation Life Zone-*- Land Use2 References
Holdriqe
Upper Lowland (600/800 to 1,400/1,600 m)
Anchicaya-Alto Yunda, PNN
332 1N
7648 W
1,050
TP-rf
F/F, SG
7,
29
Farallones de Cali (AN)
Reserva Forestal Yotoco (YO)
352 1N
7633 'W
1,500
TP-df/mf
F/P, CT
14,
31
Represa San Carlos (SC)
613 N
7451W
750
TP-mf/wf
SGa/F,P
23
Finca La Esmeralda (LE)
1,250
TP-wf
CT/CT, P
4
Finca El Ocaso (EO)
1,000
TP-wf
P/Fb
4
Lower Montane (1,400-1,600 to 2,300/2,600
m)
Reserva Natural La Planada (LP)
110 N
7800'W
1,800
TP-rf
F/SG
6,
15,
19,27
Parque Regional Ucumari,
447 1N
7532 W
1,850
TLM-wf
F/SG, TPEC>
e 6,
13,
16, 26,
28
Ucumar Bajo (UB)
Bosque Protector San Antonio
329 'N
7638 W
2,000
TLM-wf
F/SG, U
6,
9,
10, 26,
27,
(SA)
30
Empresas Publicas de Manizales
528 1N
7532 W
2,400
SG,TPN^, P
22,
25
, 28
(RB)
Finca Merenberg (ME)
214 1N
7608 W
2,300
TLM-wf
P/F
6,
17,
20, 32
Represa Miraflores (MI)
645 1N
7520 W
2,130
TLM-wf
SG/P
5
Finca Rancho Grande (RG)
436 N
7420'W
1,700
P/CT,SG,U
12
Finca Mirador, Munchique (MU)
230'N
76591W
2,300
TLM-wf
TPEe/SG, P
11
Piedras Blancas (PB)
618 N
7530 W
2,350
TLM-wf
TPEe
8
Upper Montane (2,300/2,600 to 3,100/3,400
m)
Reserva Natural Carpanta-
434 'N
7341'W
2,700
F/SG
2
Estacin Sietecuerales (CA)
Reserva Natural Alto Quindio
437 'N
7520 W
2,800
TM-wf
SG/TPNd, F,
3,
18
Acaime (AC)
P
Parque Regional Ucumari,
447 N
7532 W
2,500
SG/TPNd, F,
6,
13,
16, 21,
Ucumar Alto (UA)
P
26,
28
Santuario de Flora y Fauna
540 'N
7330'W
2,600
TLM/TM-mf
SG/P, Fa
1,
21,
24,25
Iguaque-Caon Mamarramos (IG)
Paramo (3,100/3,400 to 4,800 m)
119

Table 6-1. continued
References: (1) Acevedo (1987), (2) Andrade (1993), (3) Arango (1994), (4) Corredor (1989), (5)
Cuadros (1988), (6) Gentry (1992a), (7) Hilty (1980), (8) Johnels and Cuadros (1986), (9) Rattan et al.
(1984), (10) Rattan et al. (1994), (11) Mondragn (1989), (12) Munves (1975), (13) Naranjo (1994), (14)
Orejuela et al. (1979), (15) Orejuela and Cantillo (1990), (16) Rangel (1994), (17) Rangel and Espejo (1989),
(18) Renjifo (1988), (19) Restrepo (1990), (20) Ridgely and Gaulin (1980), (21) Rosas (1986), (22) Uribe (1986),
(23) Velasquez (1992), (24) Velez (1987), (25) N. Arango, personal communication, (26) G. Rattan, personal
communication, (27) C. Restrepo, personal observation, (28) L. M. Renjifo, personal observation, (29)
S. Hilty, unpublished list, (30) G. Rattan, H. Alvarez,and M. Giraldo,unpublished list, (31) H. Alvarez,
unpublished list, (32) G. Rattan, H. Alvarez, and E. Buttkus, unpublished list.
-^Life zone: Tropical premontane dry forest (TP-df), Tropical premontane moist forest (TP-mf),
Tropical premontane wet forest (TP-wf), Tropical premontane rain forest (TP-rf)
Tropical lower montane wet forest (TLM-wf), tropical lower montane moist forest (TLM-mf),
tropical montane wet forest (TM-wf).
2 Land Use: (F) native forest, (SG) second growth, (P) pasture, (CT) shaded coffee plantation,
(TPE) tree plantations with exotic species, (TPN) tree plantations with native species, (U)
weekend cottages. a selectively logged native forest, b native forest dominated by the Giant
Bamboo, Bgmbus^ guadua c plantations with Fraxinus sinensis d plantations with Alnus acuminata.
e Plantations with Pjpus Bakula and Cuoressus lusitanicus.
120

121
Data Analysis
I used the Lump Analysis (LAqri) technique to analyze the
data (GaP Detector 2.1). This technique is being developed
by P. Marples, Arthur Marshall, Jr. Lab, Department of
Zoology, University of Florida. LAqri tests whether breaks in
an observed frequency distribution of an attribute occur by
chance alone due to sampling error (Fig. 6-2a). It relies on
the generation of a continuous, unimodal distribution from
input data (in this case, body mass) and the detection of gap
rarity indexes, or GRI-values within the data set. The
unimodal distributions are derived using the smallest normal
kernel estimate (h) that smoothes a frequency distribution
into an unimodal continuous distribution (Silverman 1986).
GRI-values are derived by calculating the absolute
differences between contiguous rank size-ordered data points
for the observed and expected distributions. The GRI-values
for a given data set are compared against a critical value
(GRIcrit) and those that exceed that value define a
significant break (gap) in the observed frequency
distribution at the indicated alpha level (Fig. 6-2b), the
size of the observed data set (N), and the ratio between the
size of the observed data set and the mean size of all the
data sets that are compared simultanously (rj). All body
masses between two contiguous gaps define an aggregate of
species or lump. Calculations were based on logio-transformed

122
Figure 6-2. Lump Analysis of body mass for Colombian upper
lowland tropical frugivorous birds (a) body mass distribution
vs. rank order and (b) rank size-ordered body mass
distribution vs. gap rarity indexes (GRI-values). In(b)
potential gaps between the lumps are represented by GRI-
values that exceed the criterion lines (alpha values).

123
1.8 2.3 2.8 3.3
0.8
1.3
Log Body Mass
3.8

124
body masses.
Sample size strongly influences lump structure. In
data sets with a large number of observations a small value
of alpha might reveal a strong pattern of gaps whereas a
large value might reveal a weak pattern of lumps especially
at the lower end of the body mass range (Fig.2b). In this
situation, reducing alpha reduces the probability of
detecting gaps that might not exist (reduction of Type I
error). In small data sets a small value of alpha might not
reveal any pattern whereas a large alpha value might reveal
strong pattern. In this situation, increasing alpha reduces
the probability of not detecting gaps that exist (reduction
of Type II error). Thus, the gap/lump structure of a given
data set is determined by the chosen alpha level (Fig. 6-2b).
This interplay between sample size and the two types of
statistical error should be taken into account when comparing
multiple data sets. Lipsey (1990) gives an excellent
discussion of the importance of using different values of
alpha when detection of pattern is important. In this paper
and for simplicity I kept the alpha level constant within
each comparison. Depending on the data sets I used alpha
levels of 0.05 (within comparison average sample size >81,
range 30-395) and 0.1 (within comparison average sample size
<81, range 30-141).
I set up nested comparisons and derived four continuous
unimodal distributions. The four elevational zones were
compared using a null distribution generated from the

125
complete data base on frugivorous montane birds. The sites
within a elevational zone were compared using a null
distribution generated from the database on frugivorous birds
for the corresponding zone.
How to interpret a lump analysis
The main results of the lump analysis are summarized in
two related figures, depicting the distribution of gaps and
lumps. The first illustrates the distribution of GRI-values
against body mass (Fig. 6-2b), while the second, which is
derived from the first, depicts the distribution of lumps and
gaps for the various data sets that are being compared
simultaneously (e.g., Fig. 6-3). In this paper I report only
the latter. Lump structure for a given data set can be
described in terms of the number and size of the lumps and
the proportion of species falling within specific lumps.
Lump structure for multiple data sets can be described in
terms of the correspondence between the position of lumps and
gaps. Presently, P. Marples is developing a procedure that
will allow quantitative testing of the lump and gap
correspondence. In this paper the comparison will be
qualitative.
The size of the lumps represents the ranges of body mass
that do not exhibit discontinuities at a given alpha level.
The number of species that fall within a lump are represented
as the proportion of the total number of species of the
corresponding data set. In the figures depicting the
distribution of lumps the proportions are represented by

126
different shades of gray (e.g., Fig. 6-3). Thus the size of
the lump does not represent the proportion of species falling
within them, but rather the size range of birds involved.
Rgsyilti?
Elevational Zones and Body Mass Distribution
I established the lump structure of bird assemblages for
the four elevational zones using an alpha = 0.05 and found
that the number of lumps decreased from the upper lowland (24
lumps) to the pramo zone (5 lumps) (Fig. 6-3). Most lumps
were lost from the upper range (>316 g) and few from the
lower range (<10 g) of body mass. Lumps at both extremes
contained the lowest proportion of species for the upper
lowland, lower montane, and upper montane zones but not for
the pramo.
The lump structure of the upper lowland and lower
montane zones showed striking similarities in regard to the
position of gaps and lumps in the body mass range of 12-575
g. However, the proportion of species in these lumps
differed between the two zones. Lump structure of the upper
montane zone resembled that of the upper lowland and lower
montane zones in the body mass range of 83-316 g but not
above or below these figures. Finally, the pramo zone
shared only one lump (>301 g) with the other three zones.

Figure 6-3. Lump structure of Colombian montane frugivorous birds according to elevational
zone from forest (bottom) to paramo (top). Upper lowland (UL), lower montane (LM), upper
montane (UM), and pramo (P). Each box represents a lump and the space between the boxes
represent gaps in the distribution of body mass. The different shades indicate the
proportion of species falling within lumps: (1) 0-5, (2) 5-10, (3) 10-20, (4) 20-30, (5) SO
IS, (6) 45-60, and (7) 60-100 % of species. Vertical lines represent 0-5% of species.
Numbers on the right side represent number of species for the corresponding data set.
1 2 3 4 5 6 7

Paramo
(P)
Upper
Montane
(UM)
Lower
Montane
(LM)
Upper
Lowland
(UL)
6.3 15.8 39.8 100 251
Body Mass (g)
62
192
351
395
i i i i i i i i i i i
631
1585
3981
128

129
Sites and Body Mass Distribution
Upper lowland zone
The number of lumps detected at an alpha = 0.05
decreased from the site covered extensively by forest,
Anchicaya (AN, 8 lumps), to that site dominated by pastures,
El Ocaso (EO, 3 lumps) (Fig. 6-4). The lumps that were lost
were at the upper end of the range of body mass (>316 g);
none was lost at the lower end.
The lump structure of two pairs of sites, Anchicaya
(AN)/Yotoco (YO) and San Carlos (SC)/La Esmeralda (LE),
exhibited the closest similarities in regard to the position
of gaps and lumps, even though AN and SC have twice as many
species as YO and LE, respectively (Fig. 6-4) The
similarities occurred almost over the entire range of body
mass. On the other hand the lump structure of LE and EO
showed important differences, even though these two sites
have a similar number of species and are separated only by 7
km.
Lower montane zone
The number of lumps decreased from sites covered
extensively by forest (La Planada (LP), Ucumar Bajo (UB),
and San Antonio (SA), average = 8 lumps) to sites in which
the original forest has been replaced by orchards (Rio
Grande, RG), and forestry plantations of exotic species
(Munchique, MU and Piedras Blancas, PB) (average 4 lumps)

Figure 6-4. Lump structure of Colombian frugivorous birds from sites covered mostly by
forest (bottom) to sites highly transformed by human activities (top) within the upper
lowland zone. Anchicaya (AC), Yotoco (YO), San Carlos (SC), La Esmeralda (LE), El Ocaso
(EO). Each box represents a lump and the space between the boxes represent gaps in the
distribution of body mass. The different shades indicate the proportion of species falling
within lumps: (1) 0-5, (2) 5-10, (3) 10-20, (4) 20-30, (5) 30-45, (6) 45-60, and (7) 60-100
% of species. Vertical lines represent 0-5% of species.
Numbers on the right side represent number of species for the corresponding data set.
1 2 3 4 5 6 7

100
I I I
63
141
n
631
1585
3981

132
(alpha = 0.1) (Fig. 6-5).
I grouped the lower montane sites according to major
types of land use to describe patterns in lump structure.
Three sites are covered extensively by forest in which second
growth (LP), second growth and forestry plantations (UB), and
second growth, pastures, and weekend cottages (SA) cover less
than 50% of land (Table 6-1). The lump structure of LP and
UB, two sites differing in the number of species, is similar
almost over the entire range of body mass. The lump
structure of SA showed similarities with UB in the body mass
range of 63-313 g, but also differences below 63 g. This
occurred, even though both sites had a similar numbers of
species (Table 6-1).
Three sites are covered half by native forests and half
by pastures (Merenberg, ME), second growth and selectively
logged forests (Miraflores, MI), and tree plantations
established for watershed restoration (Rio Blanco, RB). The
lump structure of RB and ME showed similarities below 39 g
and above 301 g and they differed greatly from that of MI
almost over the entire range of body mass. In the last three
sites the native forest has been replaced almost entirely by
coffee plantations and orchards (Rio Grande, RG) and tree
plantations of exotic species for wood production (Munchique,
MU and Piedras Blancas, PB). RG and MU showed a different
lump structure even though they have the same number of
species. The lump stucture of MU and PB was very similar,
but the lump representing birds >398 g in PB was much smaller

Figure 6-5. Lump structure of Colombian frugivorous birds from sites covered mostly by
forest (bottom) to sites highly transformed by human activities (top) within the lower
montane zone. La Planada (LP), Ucumar Bajo (UB), San Antonio (SA), Miraflores (MI),
Merenberg (ME), Rio Blanco (RB), Rancho Grande (RG), Munchique (MU), and Piedras Blancas
(PB). Each box represents a lump and the space between the boxes represent gaps in the
distribution of body mass. The different shades indicate the proportion of species falling
within lumps:(1) 0-5, (2) 5-10, (3) 10-20, (4) 20-30, (5)30-45, (6) 45-60, and (7) 60-100 %
of species. Vertical lines represent 0-5% of species. Numbers on the right side represent
number of species for the corresponding data set.
1 2 3 4 5 6 7

30
6-3 15.8 39.8
100 251
631 1585
Body Mass (g)
134

135
than the same one in MU.
Upper montane zone
I did not find marked variation in the number of lumps
among the four sites when using an alpha = 0.1. Iguaque (IG)
the most disturbed site, showed the smallest number of lumps
(4 lumps) compared to the remaining three sites (5 lumps)
(Fig. 6-6). Lumps at the upper end of the body mass range
for two sites, Carpanta (CA) and Ucumar Alto (UA), showed a
high proportion of species but the contrary was true for the
other two sites, Acaime (AC) and IG. More interesting,
however, was the pattern of the lumps at the lower end of the
body mass range (<12 g). They consistently decreased in size
and in the proportion of species falling within them.
Carpanta (CA), the least disturbed site, is covered by native
forest and second growth and its lump structure differed from
that of AC and UA particularly for species with a body mass
>25 g. It is noteworthy that the highest proportion of
species in CA was found in the lump representing species >63
g. For little disturbed sites at the other elevational zones
those lumps representing the largest birds within the
observed range of body mass always contained the smallest
proportion of species. The lump structure of AC and UA was
very similar. These two sites were planted with native trees
in an effort to restore land previously used for cattle
ranching (Table 6-1). Finally, IG showed the most
conspicuous differences in regard to the position of lumps
and gaps when compared regard to the position of lumps

Figure 6-6. Lump structure of Colombian montane frugivorous birds from sites covered mostly
by forest (bottom) to sites highly transformed by human activities (top) within the upper
montane zone. Carpanta (CA), Acaime (AC), Ucumar Alto (UA) Iguaque (IG). Each box
represents a lump and the space between the boxes represent gaps in the distribution of body
mass. The different shades indicate the proportion of species falling within lumps:(1) 0-5,
(2) 5-10, (3) 10-20, (4) 20-30, (5) 30-45, (6) 45-60, and (7) 60-100 % of species. Vertical
lines also represent 0-5% of species. Numbers on the right side represent number of species
for the corresponding data set.
1 2 3 4 5 6 7

AC
70
CA
I
1 0
i i r
J- I i i i i i i i i it
100 316
63
31.6
Body Mass (g)
794
137

138
and gaps when compared to the other sites.
Discussion
In part because of the exploratory nature of this work,
and in part because of the early stage of development of the
techniques to test ecosystem "lumpiness", my interpretation
of the results are intended as hypotheses rather than
conclusions. By exploratory I mean that I made use of
information that was already available and thus could not
control for many factors that might confound the results,
including size of the area surveyed, hunting, and differences
in vegetation types. However, the repetition of some
patterns among my four analyses suggests that local
differences in assemblages of frugivorous birds might be
overridden by general processes that impose structure on the
landscape.
Patterns in Lump Structure
In general, the number of lumps, i.e., aggregates of
species having a similar body mass, decreased from areas
covered by continuous native forest to areas where forest has
been replaced by simpler vegetation types. It can be argued
that this trend simply reflects a decrease in the number of
species which in turn may reflect a decrease in habitat
complexity (e.g., Karr and Roth 1971, Terborgh 1977). That
is, lump structure reflects biases resulting from sampling
procedures. Several of my data sets, however, did not

139
exhibit this linear relationship. For instance, some pairs
of sites (e.g., Merenberg/Ucumar Bajo and Rio
Grande/Munchique) had a similar number of species and yet
showed a different number of lumps. Others (e.g.,
Anchicaya/Yotoco, San Carlos/La Esmeralda, and
Carpanta/Ucumar Alto) had a different number of species and
yet showed a similar number of lumps. A close examination of
the information available for the sites revealed important
structural differences between members of a pair of sites for
the former whereas the same was not true for the latter
(Table 6-1) This indicates that lump structure of body mass
in frugivorous birds reflects to some degree the structure of
landscapes.
Major changes in lump structure in terms of the
persistence of lumps occurred at the upper and lower ends of
the range of body mass. Lumps representing the largest bird
species were lost when moving from areas covered mostly by
native forest to areas where the forest has been replaced by
pramo (elevation zones) and pasture (sites within the upper
lowland zone). The same trend was found for small birds in
the upper montane zone. In the lower montane zone there was
no clear trend regarding the lumps that disappeared along the
gradient of land-use.
The pattern described above may be related to the
proportion of species falling within lumps. Lumps
representing the largest birds contained the smallest
proportion of species for the observed range of body mass for

140
any elevational zone and sites within the upper lowland zone.
Conversely, lumps representing the smallest birds contanined
the smallest proportion of species for the observed range of
body mass for sites within the upper montante zone. Lump
structure in terms of the persistance of lumps and proportion
of species falling within them was variable for sites in the
lower montane zone. This indicates that lumps representing
the smallest and/or largest species of frugivorous birds are
the ones most prone to disappear as the native forest is
fragmented and replaced by more simple ecosystems. Examples
of more simple ecosystems are the paramo, along the
elevational gradient, and managed ecosystems such as forestry
plantations and pastures, along the gradient generated by
human disturbance (Table 6-1).
I found that gaps and lumps persisted among sites
representing similar landscapes but that lumps fused, broke
down, or disappeared among sites representing different
landscapes. This indicates that gaps in the distribution of
body mass are being closed and opened probably as a result of
the replacement of species (elevational zones) and local
extinctions and invasions by species that were formerly
absent or rare in a site. It has been well documented that
species distribution changes over different spatial and
temporal scales (Terborgh 1971, Hooghiemstra 1984, Gentry
1992) and particularly relevant to this study are those
changes occurring in the distribution of species as a

141
consequence of human activities on the landscape (Pacheco et
al. 1994, Lynam, in press).
Causes of Lump Structure
The comparisons that I set up among elevational zones
and among sites within elevational zones represent two
different scales of inquiry, yet produced similar results.
Elevational zones and sites within elevational zones were
arranged from those covered mostly by native forest to those
in which the native forest has been replaced by open
vegetation (pramo) or managed ecosystems (e.g., pastures and
forestry tree plantations) in which scattered fragments of
native forest remain. Thus, changes in lump structure seem
to reflect a common causality best explained by the
complexity of landscapes in terms of the vertical and
horizontal structure of the vegetation. My proposition is
supported by work done in other regions (Thiollay 1992,
Lescourret and Genard 1994).
Vertical structure of the forest, including height of
the vegetation and diversity of growth forms, is simpler as
one moves from the lowlands to the pramos or from forest to
pastures. Along the elevational gradient, fog, air
temperature, and radiation are proximate factors that explain
changes in the vegetation structure (Leigh 1975, Grubb 1977).
Along the gradient of land-use, soil and rainfall
distribution are proximate factors that explain changes in
the structure of the vegetation (e.g., Holdrige et al. 1971).

142
In addition, within forest fragments wind shear-forces,
atmospheric humidity, and soil moisture can alter the
vertical structure of vegetation (Esseen 1994, Laurance 1994,
Kapos et al., in press). Horizontal structure refers to the
spatial array of vegetation types that results from changes
in abiotic conditions or disturbance (Wiens et al. 1985). In
Colombia, horizontal structure of the vegetation is simpler
as one moves from the lowlands to the pramo or from little
to highly modified landscapes. At lower and middle
elevations, landslides generate spatial heterogeneity locally
(Gardwood et al. 1979, Mejia et al. 1994, Velsquez et al.
1994) At higher altitudes, however, changes in climate and
soil conditions determine the presence of forest and pramo
(Cuatrecasas 1958, Espinal et al. 1977). Along the gradient
of land-use, abiotic and socioeconomic factors determine not
only rates of deforestation but also the matrix in which
forest fragments are embedded.
Examination of the results for two sites in the lower
montane zone, Rio Grande and Munchique, may help understand
how vertical and horizontal structure interplay and how they
relate to lump structure. Rio Grande and Munchique represent
highly modified landscapes where fragments of degraded native
forest are interspersed with orchards and pastures (RG) and
pine plantations (MU) (Munves 1975, Mondragn 1989). In
these two sites horizontal structure of the vegetation is
similar. The same is not true, however, for vertical
structure which if sampled over several points along

143
transects would show a greater variability at RG than at MU.
Even though the two sites have a similar number of species,
RG has more lumps, and the species are more evenly
distributed among lumps, than in MU (Fig. 6-5).
I do not know yet, on a quantitative basis, how vertical
and horizontal structure interact to produce changes in lump
structure of animal and plant assemblages or if they entrain
some other feature of landscapes to which frugivorous birds
are responding. Other features include size of fruit patches
and of seeds, the latter representing a measure of both the
dispersal and regeneration mode of plants (Salisbury 1974,
Hughes et al. 1994, Osunkoya et al. 1994). To my knowledge
there is no published account relating changes in seed size
to altitude in tropical ecosystems. However, in a lowland
neotropical site Martin (1975) found that mean size of seeds
was smaller in second growth areas, compared to mature
forest.
A Model Linking Lumps and Species Diversity in Landscapes
My results showed relationships between the number of
lumps and the number of species along a gradient of
structural complexity of landscapes. In addition, they hint
at a relationship between lump structure and the resilience
of ecosystems. Resilience as defined by Holling (1973) is a
measure of the amount of disturbance and/or change that an
ecosystem can absorb before turning into a different one.

144
These findings are summarized in a model (Fig. 6-7)
where the triangles represent different ecosystems arranged
from the most complex (top) to the simplest (bottom), such as
elevational zones along an altitudinal gradient or types of
land use along a gradient of human disturbance. The
arrangement of the triangles along a diagonal line reflects
the general trend showing a decrease in the number of lumps
with a decrease in the number of species. Ecosystems were
depicted as triangles to reflect the same general trend but
this time historical, climatic, edaphic factors, and recently
human activities, contribute to the within ecosystem
variability.
In a given ecosystem, lump structure (vertical dimension
of each triangle), defined by the number of lumps, can remain
relatively unchanged in spite of differences in the number of
species among sites (horizontal dimension of each triangle).
In my data sets, such changes were associated with changes in
species numbers within a given type of land use. However, as
one approaches the apex of each triangle the probability that
dramatic changes in lump structure will take place increases
even though species numbers can remain relatively unchanged
(vertical lines) (Fig. 6-7). In my data sets, such changes
were associated with changes in land use. Each jump into
successive ecosystems (from top to bottom) is accompanied by
a reduction both in the number of species and the number of
lumps.

145
Number of Species
Figure 6-7. Relationship between species richness and lump
structure in landscapes of variable complexity. Triangles
represent ecosystems from the most complex (El) to the
simplest (E4). Points represent changes in the number of
species within and between ecosystems.

146
The above model generates a set of testable hypotheses
that could contribute to our understanding of how landscape
pattern, biodiversity, and ecosystem processes interact at
large scales. First, there is a threshold in species numbers
below which lump structure changes dramatically, as indicated
by a decrease in the number of lumps. Second, lump structure
is maintained by the persistance of some species that might
function as attractors. Third, lump structure of plant and
animal assemblages reflects the resilience of a given
ecosystem. The removal of species in ecosystems depicted by
the top triangles may have a lesser impact on lump structure
than the removal of species in ecosystems depicted by the the
bottom ones.
Natural and human disturbances, either alone or in
concert, can affect landscapes from hundreds of meters to
hundreds of kilometers. The inherent complexity of
ecological systems defined by this spatial domain has called
for new approaches and methods. Rather than concentrating on
individual parts, these new lines of inquiry concentrate on
aggregates of parts and key processes that structure
ecosystems (Turner et al. 1995, Holling et al. 1995). The
"lump" approach represents one of these new lines of inquiry.
General Implications
In the mountains of Colombia, changes in landscape
structure have dramatic consequences on assemblages of
frugivorous birds. Big changes in land use result in the

147
disappearence of lumps or complete suites of species with
similar mass. The disappearence of particular lumps in body-
size of frugivorous birds in neotropical montane ecosystems
may reflect important changes in seed dispersal and thus
regeneration trajectories of vegetation after disturbance.
There is some indication that assemblages of neotropical
montane frugivorous birds, depending on degree of habitat
modification, are robust to human disturbance. This is based
on the fact lump structure varied little between similar
sites that differ in the number of species. The
fragmentation and transformation of neotropical montane
ecosystems does not seem to generate the same patterns in
assemblages of frugivorous birds in low to middle high
altitudes and high altitudes. This may have important
consequences for the conservation and management of
ecosystems along the altitudinal gradient.

CHAPTER 7
CONCLUSIONS
Studies designed to evaluate "edge effects" (sensu
Harris 1984) have emphasized the maximum distance at which
changes induced by edge creation are apparent within stands
(e.g., Williams-Linera 1990, Blanchard 1992, Chen et al.
1992). Variation among studies is enormous in terms of
"depth" of edge effects due to variation in edge features
(e.g., Kroodsma 1984, Quintela 1986, Noss 1991, DeGraaf 1992)
and to the fact that abiotic factors and organisms show
different responses to the creation of edges.
I have avoided in my work the use of the term "edge
effects" (Harris 1984). This is partially due to the
complexity of the responses that fruits and birds showed to
the presence of edges at my study site. Equally important
was the realization that "edges" are not isolated and fixed
elements in landscapes. They seem to regulate what happens
between the forest and the nearby disturbed area, and at the
same time they connect different elements of landscapes.
This is particularly true in the La Planada region, where
transient corn fields, pastures, and second growth areas of
various ages are embedded in a forest matrix. "Edge
dynamics" reflects the influence of edges on plant and animal
assemblages more accurately than "edge effects".
148

149
The different responses shown by the four variables
describing fruit abundance in the understory and by the
different plant species suggest differences in edge
"penetrability" to the factors regulating fruit production
and seed movement. These differences were manifested across
the four distances within the edges but also among new and
old edges. Steep and permanent gradients from edge towards
forest interior may suggest low penetrability. Conversely,
shallow and more variable gradients may suggest high
penetrability.
My results showed that fruit abundance is influenced by
edges but that the exclusion of certain groups of plants
produces different results. This provided insight regarding
the scales at which edges influence fruit abundance. Future
studies aimed at understanding the role of edges in
landscapes should take the "scale" issue into consideration.
One important finding is that related to the
distribution of sparse species of plants and birds in the
immediate vicinity of pasture-forest edges. These species
may persist and take advantage of changes taking place at
either side of the edge depending on habitat preferences. As
more forest is felled, species characteristic of large
disturbed areas may establish in the recently disturbed area.
On the other hand, as pastures and fields are left abandoned
species characteristic of forest may establish there. Thus,
edges might function as "stepping stones" to recolonization
at both sides and as elements that connect, rather than

150
separate, different elements of the landscape in the La
Planada region.
I argue that a "lump" analysis (Chapter 6) can help
understand the role of edges in landscapes. Edges when
viewed on a scale larger than that of a forest stand and the
nearby disturbed area show a tremendous variability. A
single measure of landscape complexity as proposed in Chapter
(6), which combines a horizontal and vertical component of
landscapes, can summarize edge features in a given area. By
trying to recognize patterns at large scales, such as those
defined by the impact of humans on landscapes, we might be
able to find a relationship between landscape pattern,
processes affecting landscape pattern, and the organisms that
live in them. This might be particularly true for the study
of fruit-frugivore interactions, since one outcome of such
interactions is seed dispersal. Landscape pattern is not
only changed as a consequence of disturbance; it is also
changed in fundamental ways by the process of seed dispersal.
In areas that are highly diverse such an approach might be
the best to understand the magnitude of the impact that
humans have on the biota.

APPENDIX A
PLANT SPECIES FRUITING IN THE UNDERSTORY
OF THE RESERVA NATURAL LA PLANADA

152
APPENDIX A
PLANT SPECIES FRUITING IN THE UNDERSTORY
OF THE RESERVA NATURAL LA PLANADA (SEPTEMBER 1992-AUGUST 1993)
Habitat (H): forest (F), treefall gaps (G), large-disturbed areas, including second
growth, road sides (LD). Abundance of fruiting individuals at the edges included in this
study (AB): very abundant (VA), abundant (A), sparse (S), very sparse (VS), extremely
sparse (ES). Seeds found in bird droppings (BD). Letters followed by numbers indicate
collection number.
FAMILY SPECIES H AB BD
Old and New Edges
Araceae
Anthurium
andinum Encrl .
F
S
BD
Anthurium
carchiense Croat
LD
A
Anthurium
cf. marmoratum Sodiro
F
A
Anthurium
cf. melampyi Croat
F
VA
BD
Anthurium
cf. oulverulentum Sodiro
F
A
Anthurium
lancea Sodiro
F
VS
Anthurium
loncricaudatum Encrl.
F
S
BD
Anthurium
membranaceum Sodiro
LD
VA
BD
Anthurium
mindense Sodiro
S
Anthurium
ovatifolium Encrl.
F
S
BD
Anthurium
sp.
LD
VS
BD
Anthurium
terracolum Croat
F
S
Anthurium
trinerve Mig.
F
vs
Anthurium
umbraculum Sodiro
F-LD
VA
BD
Anthurium
umbricolum Engl.
F
A
BD
Anthurium
versicolor Sodiro
F
VA
CR 556,
CR 583, CR 670, CR 791

Araliaceae
Mgngtera sp.
F VS
Scheffiera aff. lasioavne Harms
Scheffiera lasioavne Harms
Arecaceae
VS
LD VS BD
Aiphanes sp.
Chamaedorea polvchlada
Chamaedorea sp.
CR 466, CR 467
Geonoma weberbaueri
Prestoea aff. purpurea
Boraginaceae
Tournefortia aiaantifolia Killip
Bromeliaceae
F S BD
F A
F S BD
F VA BD
F S
F VS
Ronnberaia aff. deleani L.B. Smith
Campanulaceae
Burmeistera aff. lonaifolia Gleason
Burmeistera carnosa Gleason
Burmeistera sp.
Burmeistera sp. nov.
CR 543, CR 716, CR 734, CR 502
Clusiaceae
Clusia sect. Anandrogyne, "A. multiflora
H.B.K.group"CR 796
Cyclanthaceae
Asolundia sp.1
Asplundia stenophvlla
CR 553, CR 680, CR 579, CR 766
Soheraedenia sp.
F VS
F S
LD A
LD-F S BD
LD-F A BD
LD-F VS BD
F A
F S
F
S
153

Ericaceae
CR 789
Spheraedenia stevermarkii
Cavendishia enaleriana Hoer.
Cavendishia tarapotana (Benth.) Meisner
CR 666, CR 657, CR 590
Macleania stricta A.A. Smith
Psammisia aff. debilis Sleumer sp. nov.
Psammisia cf. dolichcpoda A.A. Smith
Psammisia cf. ulbrichiana Hoerold.
Psammisia ferruainea A.A. Smith
Psammisia sodiroi Hoerold.
Gesneriaceae
Alloplectus sp.
CR 475, CR 761
Alloplectus sp.1
CR 790, CR 654
Alloplectus tenuis Benth.
Alloplectus tetraaonus (Hanst.) Hanst.
Alloplectus teuscheri (Raymond) Wiehler
B^sleria solanoides
Besleria sp.
CR 759
Columnea bvrnsina (Wiehler) L.P. Kvist &
L.E. Skog
Columnea cinerea
Columnea ebrnea (Wiehler) L.P. Kvist &
L.E. Skog
Columnea eubracteata
Columnea qiaantifolia
Columnea minor
Drvmonia sp.
CR 559, CR 688, CR 540, CR 400
LD-F A BD
F S BD
G VS BD
LD VS
LD A
F S
G S
G VS
F S
G S
F VS
G S
G S
F VA
LD VA BD
LD-G S
LD S BD
LD A BD
F VS
LD S BD
F S BD
LD S
F S BD
154

PrymQrtya warszewicziana Hanst.
G
S
Gasteranthus aff. wendladianus (Hanst.)
LD-F
S
Wiehler
Unknown
F
S
CR 582, CR 647, CR 541, CR 499
Marcgraviaceae
Melastornataceae
Meliaceae
Marcaravia eichleriana Wittmaok
F
VS
BD
Marcgraviastrum subssesilis (Benth)
F
VS
Bedell
Blakea cf. stiDulacea Wurdack
G
S
BD
Blakea punctulata (Triana) Wurdack
G
VS
Clidemia sp.l
LD-F
VA
BD
Clidemia sp.2
F
S
BD
Miconia aff. neurotricha
F
s
Miconia loreyoides Triana
LD
vs
BD
Miconia pseudoradula Coan. & Gleason ex
LD
vs
BD
Gleason
Miconia smaraqdina Naudin
LD
vs
BD
Miconia sp.5
LD
vs
BD
CR 533, CR 745, CR 602
Miconia theaezans (Bonpl.) Coan.
LD
vs
BD
Ossaea micrantha (Sw.) Macf. ex Coan.
F
vs
BD
Tooobea oittieri Coan.
F
s
BD
Tooobea sp.
F
vs
CR 263
Unknown
F
vs
CR 676, CR 415, CR 433
Ruaaea qlabra
F
vs
155

Monnimiaceae
Myrsinaceae
Siparuna sp.
CR 694
Myrtaceae
Cvbianthus simplex (Hook.f.)Aaost
Onagraceae
ELuqenia anastomosus DC
Rubiaceae
Fuchsia macrosticrma Bentham
Solanaceae
Faramea affinis
Faramea killioii Standi.
Palicourea qibbosa
Palicourea sp.l
CR 461, CR 695
Palicourea sp.2
CR 430
P-SYCOthria aubletiana Steyerm.
Psvcothria hazenii Standi.
Psvcothria panamensis Standi.
PSYCOthria solitudinum Standi.
Cestrum sp.
CR 485, CR 535, CR 779
PhYsaJlis sp.
Solanum evolulifolium
CR 487, CR 684
Solanum lepidotum
CR 570, CR 711, CR 489, CR 799
Solanum sp.2
LD
S
F
S
F
VS
LD
VS
F-LD
VA
BD
F
S
BD
F-LD
VA
BD
LD
S
F
VS
F
VA
BD
F
S
BD
LD-F
S
BD
LD
S
F
S
G
S
G
S
S
LD
s
BD
G VS
156

CR 753, CR 627
Solanum sp.5
CR 483, CR 722, CR 794, CR 785
Solanum sp.7
CR 537, CR 735, CR 532?
Solanum sp.8
CR 486, CR 750, CR 762
Zingiberaceae
New Edges
Renealmia aff. concinna Standley sp. nov.
CR 638
Acanthaceae
Araceae
Araliaceae
Mendocia orbicularis Turrill
ftnthubium cf. chamberlainii Masters
Anthurium sp. nov.
CR 556
Ehllodepdipn oligospermum Engl
Stenpspermatium lonqipetiolatum Engl.
Stenosoermatium longisoadix Croat
Stenosoermatium soarrei Croat
Xanthosoma subandinum Schott
Scheffiera cf. violcea Cuatr.
Campanulaceae
.CentPQPQgon aff. solanifolius Benth
Cyclanthaceae
Agplundia stenoohvlla
Dicranopygium sp.
F A
G-LD VA
F S
F VS
F VS
F ES
G ES
LD VS
LD VS BD
LD VS BD
LD ES
F ES
F ES
LD ES
F VS
F A BD
157

Ericaceae
CR 552, CR 771, CR 551, CR 506
Psammisia montana Luteyn sp. nov.
Gesneriaceae
F VS
Alloolectus bolivianus (Britton) Wiehler
Columnea cf. oicta Karsten
Drvmonia turrialvae Hanst.
Gasteranthus aff. oncoaastrus (Hanst.)
Gasteranthus oncoaastrus (Hanst.) Wiehler
Kohleria villosa (Fritsch) Weihler
Heliconiaceae
F VS
G ES
G VS
G S
G S
G ES
Heliconia impdica Abalo & Morales
Melastomataceae
G ES
Blakea cf. ouadriflora Gleason F ES
Miconia hvmenanthera Triana VS
Meliaceae
Unknown
CR 751
Phytolacaceae
Phytolacca rivinoides Kunth & Bouche
Piperaceae
Piper qutierrezii T.&J.
Rubiaceae
Palicanrea sp.4
CR 592
Palicourea standlevana A.M. Taylor
Psvcothria allgnij Standi.
Psvcothria braulioi A.M. Taylor sp. nov.
F ES
G ES
ES
G ES
G S
F ES
G ES
158

Solanaceae
Unknown
Unknown
Old Edges
Actinidaceae
Amaralidaceae
Clusiaceae
Ericaceae
Euphorbiaceae
Gesneriaceae
CR 749, CR 637, CR 544
Psvcothria dukei Dwyer
Solanum sp.3
CR 754
Solanum sp.6
CR 536, CR 658, CR 484
Unknown
CR 640, CR 661
CR 665
CR 769
Saurauia parviflora Tr. & Pi.
Bomarea pardina Herbert
Clusia venusta Little
Macleania bullata
Sphvrospermum cordifoliium Bentham
Hveronvma sp.
CR 706
F S
ES
F VS
LD ES
F ES
G ES
LD ES BD
LD ES
F ES
LD S BD
F ES BD
LD-F ES BD
159

Loranthaceae
Alloolectus schultzei Mansf.
Moraceae
Struthanthus aecruatoris Kuiit
Myrsinaceae
Ficus cf. aDollinaris
Ficus aarcia-barriaae
Cvbianthus sorucei (Hook.f.)Aaos
Rubiaceae
Loranthaceae
Haffinania sp.
Lauraceae
Aetanthus sd.
Unknown
F ES
LD-F ES
G ES
G ES
F ES
LD ES
F ES
F ES
BD
160

APPENDIX B
BIRDS CAPTURED IN THE UNDERSTORY
OF THE RESERVA NATURAL LA PLANADA

162
APPENDIX B
BIRDS CAPTURED IN THE UNDERSTORY
OF THE RESERVA NATURAL LA PLANADA (JUNE 1992-AUGUST 1993)
Birds mist netted
in the understory of the Reserva Natural La Planada.
Abundance (AB):
extremely sparse (ES), very sparse (VS), sparse
(S), abundant (A),
and very abundant (VS)
Feeding guild (FG)
: carnivores (A), frugivores
(F), insectivores
(I) ,
and nectarivores
(N) .
FAMILY
SPECIES
AB
FG
Mass
(g)
Accipitridae
Accioiter collaris
VS
A
166.3
Columbidae
Colwnba cavannensis
VS
F
Strigidae
Glaucidium iardinii
ES
A
75.0
Otus columbianus
VS
A
165.0
Trochilidae
Adelomvia melanoqenvs
ES
N
5.22
Aglaiocercus coelestis
VA
N
fenasilia franciae
ES
N
5.02
Bdissonneava iardini
VS
N
9.8
Chlprostilbon mellisuqus
ES
N
4.0
CQeliqena torcruata
VS
N
8.2
Coeliqena wilsoni
VA
N
7.1
Colibri thalassinus
ES
N
Dorvfera ludoviciae
ES
N
6.0
Eutoxeres aauila
ES
N
9.5
Haplophaedia lugens
VA
N
6.0

HeligdQxa imperatrix
LafrggnaY9 lafresnavi
Ocreatus underwoodii
Phaetornis svrmatophorus
Schistes aeoffrovi
Urosticte beniamini
Trogonidae
Troqon personatus
Capitonidae
Semnornis ramphastinus
Ramphastidae
Andiaena laminirostris
Picidae
Campephilus pollens
Dendrocolaptidae
Dendrocincla tvrannina
Glvphorvnchus spirurus
bepidbCQlapte? affinis
Xiphocolaptes promeropirhvnchus
Furnariidae
Anabacerthia varieqaticeps
Cranioleuca ervthrops?
Marqarornis stellatus
Premnoplex brunnescens
Premnornis quttuliqera
Schizoeaca fuliginosa
Sclerurus mexicanus
Svndactvla subalaris
Thripadectes iqnobilis
Thripadectes virqaticeps
Formicariidae
VS
N
8.2
vs
N
5.0
s
N
2.9
VA
N
6.0
VS
N
4.0
ES
N
4.4
S
F
58.6
VS
F
92.1
VS
F
349.5
ES
I
206.0
VS
I
53.4
VA
I
14.3
VS
I
29.0
ES
I
161.0
ES
I
25.0
ES
I
14.0
S
I
20.7
VA
I
15.9
VA
I
14.8
ES
I
15.5
S
I
22.3
A
I
32.9
S
I
45.5
S
I
58.9
163

Drvmophila caudata
Formicarius rufipectus
Grallaricula flavirostris
Thamnophilus unicolor
Cotingidae
Pachvramphus alboariseus
Pachvramphus versicolor
Pipreola riefferii
Lipangug qryptolpphus
Pipridae
Allacotopterus deliciosus
Masius chrvsopterus
Tyrannidae
Mionectes striaticollis
Mviobius barbatus
Mviophobus flavicans
Mviotriccus ornatus
Mviodvnastes chrvsocephalus
Ochtoeca cinnamomeiventris
Poaonotriccus ophtalmicus
Pseudotriccus pelzelni
Pseudotriccus ruficeps
Zimmerius viridiflavus
Troglodytidae
Henicorhina leucophrvs
Turdidae
Catharus ustulatus
Entomodestes coracinus
Mvadestes ralloides
Turdus serranus
Vireonidae
VS I 10.5
VS I 68.5
1 I 20.4
S I 23.3
ES F 16.1
ES F 15.0
A F 51.5
S F 80.4
A F 12.4
VA F 11.5
VA F 13.8
ES I 12.0
A I 12.9
A I 10.4
VS I 38.8
VS I 12.2
VS I 9.5
VA I 11.8
ES I 9.5
VS F 10.2
VA I 15.6
ES F 29.5
VS F 14.5
VA F 28.5
ES F 28.41
164

y.iroo leucoohrvs
Parulidae
Basileuterus coronatus
Basileuterus tristriatus
Myiobprug miniatus
Thraupidae
Anisoanathus flavinucha
Chlorochrvsa ohoenicotis
ghlorospingys semifuscus
DalasSa albilatera
Diqlogga indiqotica
Euohonia xanthoaaster
Iridosornis oorohvroceohala
Tachvohonus luctuosus
Tangara arthus
Tangara labradorides
Tanqara niqroviridis
Catamblyrhynchidae
Catamblvrhvnchus diadema
Fringillidae
Amaurogpiza concolor
Afc.lapetes brunneinucha
1 Arango (1993); 2 Miller (1963)
ES I 22.0
S I 17.1
VA I 13.1
VS I 8.2
VS F 44.4
S F 20.6
A F 28.1
S N 9.9
ES N 12.0
VA F 13.6
ES F 26.0
ES F 35.0
S F 20.8
VS F 14.2
S F 16.9
ES I 18.0
VS F 16.0
A F 45.6
165

LITERATURE CITED
Acevedo, C. I. 1987. Contribucin al conocimiento de las
relaciones entre colibrs y flores con observaciones de
sus ciclos anuales en el bosque altoandino de Iguaque,
Boyac. Tesis, Departamento de Biologa, Pontificia
Universidad Javeriana, Bogot, Colombia.
gren, J. 1988. Seed size and number in Rubus chamaemorus:
beween-habitat variation, and effects of defoliation and
supplemental pollination. Journal of Ecology 77:1080-
1092.
gren, J. 1989. Sexual differences in biomass and nutrient
allocation in the dioecious Rubus chamaemorus. Ecology
69:962-974.
Aizen, M. A. and P. Feinsinger. 1994. Forest fragmentation,
pollination, and plant reproduction in a Chaco Dry
Forest, Argentina. Ecology 75:330-351.
Andrade, G. I., editor. 1993. Carpanta Selva Nublada y
Pramo. Fundacin Natura, Santa F de Bogot, Colombia.
Andrade, G. I., H. Rubio-T. 1994. Sustainable use of the
tropical rainforest: evidence from the avifauna in a
shifting-cultivation habitat mosaic in the Colombian
Amazon. Conservation Biology 8:545-554.
Andrade, G. I., M. L. Rosas and A. Repizzo. 1993. Notas
preliminares sobre la avifauna y la integridad biolgica
de Carpanta. Pages 207-228, in G. I. Andrade, editor.
Carpanta Selva Nublada y Pramo. Fundacin Natura,
Santa F de Bogot, Colombia.
Andrn, H. and P. Angelstam. 1988. Elevated predation rates
as an edge effect in habitat islands: experimental
evidence. Ecology 69:54-547.
Arango, N. 1991. Depredacin de nido y su relacin con la
fragmentacin de habitat en un bosque nublado tropical.
Tesis B.Sc., Departamento de Biologa, Universidad del
Valle, Cali, Colombia.
Arango, S. 1993. Morfologa y comportamiento alimenticio de
las aves frugvoras de Carpanta. Pages 127-140, in G.I.
166

167
Andrade, editor. Carpanta Selva Nublada y Pramo.
Fundacin Natura, Santa F de Bogot, Colombia.
Arango, S. 1994. El papel de las aves dispersoras de semillas
en la regeneracin de pastizales en el Alto Quindo,
Andes Centrales, Colombia. Fundacin Herencia Verde,
Cali, Colombia, unpublished report.
Arias, J. C. 1993. Polinizacin y biologa reproductiva de
Palicourea aibbosa Dwyer (Rubiaceae) en bosques de
estados sucesionales diferentes. Tesis Biologa,
Universidad de Antioquia, Colombia.
Auchter, E. C., A. Lee Schrader, F. S. Lagasse, and W. W.
Aldrich. 1926. The effect of shade on the growth, fruit
bud formation and chemical composition of apple trees.
Proceedings of the American Society for Horticultural
Science 23:368-382.
Auclair, A. N. and G. Cottam. 1971. Dynamics of black
cherry (Prunus sertina Ehrh.) in southern Wisconsin oak
forests. Ecological Monographs 41:153-177.
Banguero, H. 1993. La poblacin de Colombia 1938-2025. Una
visin retrospectiva y prospectiva para el pas, los
departamentos y sus municipios. Coleccin de Edicin
Previa, Editorial Universidad del Valle, Cali, Colombia.
Beltrn, W. 1991. Historia natural de Andiaena laminirostris
(Aves: Ramphastidae). Informe Final, Fundacin para la
Promocin de la Investigacin y la Tecnologa, Banco de
la Repblica, Bogot, Colombia.
Bierregaard, R. O. 1990. Avian communities in the understory
of Amazonia forest fragments. Pages 333-343, in A.
Keast, editor. Biogeography and Ecology of Forest Bird
Communities. SPB Academic Publishing bv, Hague, The
Netherlands.
Bierregaard, R. 0., Jr. and T. E. Lovejoy. 1989. Effects of
forest fragmentation on amazonian understory bid
communities. Acta Amaznica 19:215-241.
Blake, J. G. and W. G. Hoppes. 1986. Influence of resource
abundance on use of tree-fall gaps by birds in an
isolated woodlot. Auk 103:328-340.
Blake, J. G. and B. A. Loiselle. 1991. Variation in resource
abundance affects capture rates of birds in three
lowland habitats in Costa Rica. Auk 108:114-130.
Blake, J. G., B. A. Loiselle, T. C. Moermond, D. J. Levey,
and J.S. Denslow. 1990. Quantifying abundance of

168
fruits for birds in tropical habitats. Studies in Avian
Biology 13:73-79.
Blanchard, J. D. 1992. Light, vegetation structure, and
fruit production on edges of clearcut sand pine scrub in
Ocala National Forest, Florida. M.Sc. Thesis,
University of Florida, Gainesville, FL.
Brittingahm, M. C. and S. A. Temple. 1983. Have cowbirds
caused forest songbirds to decline? BioScience 33:31-35.
Brokaw, N. V. L. 1982. The definition of treefall gap and
its effect on measures of forest dynamics. Biotropica
11:158-160.
Brown, J. H. 1995. Macroecology. The University of Chicago
Press, Chicago, IL.
Brown, N. 1993. The implications of climate and gap
microclimate for seedling growth conditions in a Bornean
lowland rain forest. Journal of Tropical Ecology 9:133-
168.
Burkey, T. V. 1993. Edge effects in seed and egg predation
at two neotropical rainforest sites. Biological
Conservation 66:139-143.
Bush, B. M. 1994. Amazonian speciation: a necessary complex
model. Journal of Biogeography 21:5-17.
Cavelier, J. and A. Etter. Deforestation of mountain forest
in Colombia as a result of illegal plantation of opium
(Papaver somniferum) in Neotropical Mountain Forest.
Proceedings of the Neotropical Mountain Forest
Biodiverstiy and Conservation Symposium, New York
Botanical Garden, New York, August 1992, in press.
Chaikiattiyos, S., C. M. Menzel, and T. S. Rasmussen. 1994.
Floral induction in tropical fruit trees: Effects of
temperature and water supply. Journal of Horticultural
Science 69:397-415.
Chapman, F. M. 1917. The distribution of bird life in
Colombia. Bulletin of the American Museum of Natural
History 33:167-192.
Chason, J. F., D. D. Baldocchi and M. A. Huston. 1991. A
comparison of direct and indirect methods for estimating
forest canopy leaf area. Agricultural and Forest
Meteorology 57:107-128.

169
Chen, J., J. F. Franklin, and T. A. Spies. 1992. Vegetation
responses to edge environments in Old-growth Douglas-Fir
forests. Ecological Applications 2:387-396.
Clark, D. A. and D. B. Clark. 1987. Temporal and
environmental patterns of reproduction in Zamia
skinneri. a tropical rain forest cycad. Journal of
Ecology 75:135-149.
Clark, W. C., D. D. Jones, and C. S. Holling. 1979. Lessons
for ecological policy design: A case study of ecosystems
management. Ecological Modelling 7:1-53.
Compton, S. G., S. J. Ross, and I. W. R. Thornton. 1994.
Pollinator limitation of fig tree reproduction on the
island of Anak Krakatau (Indonesia). Biotropica 26:180-
186.
Connell, J. H. 1978. Diversity in tropical rain forests and
coral reefs. Science 199:1302-1309.
Corredor, G. L. 1989. Estudio comparataivo entre la avifauna
de un bosque natural y un cafetal tradicional en el
Quindio. Trabajo de Grado, Departamento de Biologa,
Universidad del Valle, Cali, Colombia.
Correll, D. L. 1991. Human impact on the functioning of
landscape boundaries. Pages 90-109, in M. J. Holland,
P. G. Risser, and R. J. Naiman, editors. Ecotones: The
Role of Landscape Boundaries in the Management and
Restoration of Changing Environments. Chapman Hall, New
York, USA.
Crist, T. 0. and J. A. MacMahon. 1991. Individual foraging
components of harvester ants: movement patterns and seed
patch fidelity. Insect Socieux 38:379-396.
Crist, T. O., D. S. Guertin, J. A. Wiens, and B. T. Milne.
1992. Animal movement in heterogeneous landscapes: an
experiment with Eleodes beetles in shortgrass prairie.
Functional Ecology 6:536-544.
Cronquist, A. 1981. An Integrated System of Classification
of Flowering Plants. Columbia University Press, New
York, NY.
Cuadros, T. 1988. Aspectos ecolgicos de la comunidad de aves
en un bosque nativo en la Cordillera central en
Antioquia, Colombia. Hornero 13:8-20.
Cuatrecasas, J. 1958. Aspectos de la vegetacin natural.
Revista de la Academia Colombiana de Ciencias Exactas
Fsicas y Naturales 10:221-264.

170
Dalling, J. W. and E. V. J. Tanner. 1995. An experimental
study of regeneration on landslides in montane rain
forest in Jamaica. Journal of Ecology 83:55-64.
De las Salas, G. and M. Ballesteros. 1986. Tipos de
comunidades forestales de la Reserva Natural La Planada,
Nario, Colombia. Informe Final FES, Cali, Colombia
DeGraaf, R. M. 1992. Effects of even-aged management on
forest birds at northern hardwood stand interfaces.
Forest Ecology and Management 46:95-110.
del Hoyo, J., A. Elliott, and I. Sargatal, editors. 1992.
Handbook of the Birds of the World. Lynx Ediciones,
Barcelona, Espaa.
Denslow, J. S. and G. S. Hartshorn. 1994. Tree-fall gap
environments and forest dynamic processes. Pages 120-
127, in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and
G. S. Hartshorn, editors. La Selva: Ecology and Natural
History of a Neotropical Rain Forest. The University of
Chicago Press, Chicago, IL.
Denslow, J. S. and A. E. Gomez-Diaz. 1990. Seed rain to
treefall gaps in a neotropical forest. Canadian Journal
of Forest Researach 20:642-648.
Didham, R. The influence of edge effects and forest
fragmentation on leaf litter invertebrates in central
Amazonia. in W.F. Laurance and R.O. Bierregaard, Jr.,
editors. Tropical Forest Remnants: Ecology, Management
and Conservation of Fragmented Communities, in press.
Draiijers, G. P. J., W. P. M. F. Ivens, and W. Bleuten. 1988.
Atmospheric deposition in forest edges measured by
monitoring canopy throughfall. Water, Air, and Soil
Pollution 42:129-136.
Duellman, W. E. 1979. The herpetofauna of the Andes:
patterns of distribution, origin, differentiation, and
present communities. Pages 371-45 in W.E. Duellman,
editors. The South American Herpetofauna: Its Origin,
Evolution, and Dispersal. Museum of Natural History,
Monograph 7, University of Kansas, Lawrence, Kansas.
Dunning, J. B., Jr. 1993. Handbook of Avian Body Masses.
CRC Press, Boca Raton, FL.
Dwyer, J. D. 1980. Family 179. Rubiaceae, Part II. Annals
of the Missourit Botanical Garden 67:257-522.

171
Ellison, A. M., J. S. Denslow, B. A. Loiselle, and D. Brens.
1993. Seed and seedling ecology of neotropaical
Melastomataceae. Ecology 74:1733-1749.
Escobar, F. 1994. Excremento, coprfagos y deforestacin en
bosques de montaa al suroccidente de Colombia. Tesis,
Universidad del Valle, Cali, Colombia.
Espinal, L. S., J. Tos Jr., E. Montenegro, G. Toro and D.
Daz. 1977. Zonas de vida o formaciones vegetales de
Colombia. Instituto Geogrfico Agustn Codazzi, Bogot,
Colombia.
Esseen, P. A. 1994. Tree mortality patterns after
experimental fragmentation of an old-growth conifer
forest. Biological Conservation 68:19-29.
Estrada, A., R. Coates-Estrada, D. Meritt, S. Montiel, and D.
Curiel. 1993. Patterns of frugivore species richness
and abundance in forest islands and in agricultural
habitats at Los Tuxtlas, Mexico. Vegetatio 107/108:245-
257.
Ewel, J. and A. Madriz. 1968. Zonas de vida de Venezuela.
Memoria Explicativa sobre el Mapa Ecolgico. Editorial
Sucre, Venezuela.
Feinsinger, P., J. H. Beach, Y. B. Linhart, W. H. Busby, and
K. G. Murray. 1987. Disturbance, pollinator
predictability, and pollination success among Costa
Rican Cloud forest plants. Ecology 68:1294-1305.
Feinsinger, P., H. M. Tiebout III, and B. E. Young. 1991. Do
tropical bird-pollinated plants exhibit densigy-
dependent interactions? Field experiments. Ecology
72:1953-1963.
Fitzpatrick, J. W. 1980. Foraging behavior of neotropical
tyrant flycatchers. Condor 82:43-57.
Fjelds, J. and N. Krabe. 1990. Birds of the high Andes.
Zoological Museum, University of Copenhagen and Apollo
Books, Svendborg, Denmark.
Fleming, T. H. 1988. The Short-tailed Fruit Bat: A Study in
Plant-Animal Interactions. Wildlife Behavior and
Ecology Series, Chicago Press, Chicago, IL.
Gagnon, J., K. A. Haycock, J. R. Roth, D. S. Feldman, Jr., W.
F. Finzer, R. Hofman, and J. Simpson. 1989. SuperANOVA.
Accesible General Linear Modeling. Abacus Concepts,
Berkely, CA.

172
Garwood, N., D. P. Janos, and N. Brokaw. 1979. Earthquake-
caused landslides: a major disturbance to tropical
forests. Science 205:997-999.
Gaston, K. J. and T. M. Blackburn. 1995. Birds, body size
and the threat of extinction. Philosophical
Transactions of the Royal Society of London B 347:205-
212.
Gates, J. E. and L. W. Gysel. 1978. Avian nest dipsersion
and fledgling success in field-forest ecotones. Ecology
59:871-883.
Geiger, R. 1965. The Climate Near the Ground. Harvard
University Press, Cambridge, MA.
Gentry, A. H. 1983. Dispersal ecology and diversity in
neotropical forest communities. Sonderband
naturwisseschaften Ver. Hamburg 7:303-314.
Gentry, A. H. 1986. Endemism in tropical versus temperate
plant communities. Pages 153-181, in M. Soul (editor).
Conservation Biology: The Science of Scaracity and
Diversity. Sunderland, MA.
Gentry, A. H. 1988. Changes in plant communty diversity and
floristic composition on environmental and geographical
gradients. Annals of the Missouri Botanical Garden
75:1-34.
Gentry, A. H. 1992. Diversity and composition of Andean
forests of Peru and adjacent countries: implications for
their conservation. Memorias del Museo de Historia
Natural U.N.M.S.M. (Lima) 21:11-29.
Gentry, A. H. 1992b. Tropical forest biodiversity:
distributional patterns and their conservational
significances. Oikos 63:19-28.
Gentry, A. H. and L. H. Emmons. 1987. Geographical variation
in fertility, phenology, and composition of understory
neotropical forests. Biotropica 19:216-227.
George, A. P., R. J. Nissen, and C. Howitt. 1990. Effects
of environmental variables and cropping on leaf
conductance of custard apple (Annona cherimola x Annona
squamosa) 'African Pride'. Scientia Horticulturae
45:137-147.
Gibbs, J. P. 1991. Avian est predation in tropical wet
forest: an experimental study. Oikos 60:155-161.

173
Girden, E. R. 1992. ANOVA: Repeated Measures. Sage
University Paper Series on Quantitative Applications in
the Social Sciences, 07-84, Newbury Park, CA.
Gmez, M. E. and C. Palau. 1994. El campesino de San
Isidro, Nario: Condiciones de vida y participacin en
programas de desarrollo en la zona de influencia de la
Reserva Natural La Planada. Tesis Pregrado, Universidad
del Valle, Cali, Colombia.
Goodwin, D. 1976. Crows of the World. Comstock Publishers
Associate, Ithaca, New York, USA.
Gosz, J. R. 1991. Fundamental ecological characteristics of
landscape boundaries. Pages 8-30 in M. J. Holland, P.
G. Risser, and R. J. Naiman, editors. Ecotones: The
Role of Landscape Boundaries in the Management and
Restoration of Changing Environments. Chapman Hall, New
York, USA.
Grubb, P. J. 1977. Control of forest growth and
distribution on wet tropical mountains: with special
reference to mineral nutrition. Annual Review of
Ecology and Systematics 8:83-107.
Guariguata, M. R. Landslide disturbance and forest
regeneration in the upper Luquillo mountains of Puerto
Rico. Journal of Ecology 78:814-832.
Guimares, I. C., C. Uhl, and D. Nepstad. 1994. The role of
the shrub Cordia multispicata Cham as a succcession
facilitator in an abandoned pasture, Paragominas,
Amazonia. Vegetatio 115:91-99.
Gunderson, L. H. 1992. Spatial and temporal dynamics in the
Everglades ecosystem with implications for water
deliveries to Everglades National Park. Ph.D.
Dissertation, University of Florida, Gainesville, FL.
Haffer, J. 1969. Speciation in Amazonian birds. Science
165:131-137 .
Halls, L. K. 1973. Flowering and fruiting of Southern browse
species. U.S. Forest Service Research Paper S0-90:l-10.
Hannan, N. P. and A. Bgu. 1995. A method to estimate
instantaneous and daily intercepted photosynthetically
active radiation using a hemispherical sensor.
Agricultural and forest Meteorology 74:155-168.
Hansen, A. J., D. L. Urban, and B. Marks. 1992. Avian
community dynamics: the interplay of landscape
trajectories and species life histories. Pages 170-195,
in A.J. Hansen and F. di Castri, editors. Landscape

174
Boundaries: Consequences for Biotic Diversity and
Ecological Flows. Springer-Verlag, New York, NY.
Hansson, L. 1983. Bird numbers across edges between mature
conifer forest and clear cuts in Central Sweden. Ornis
Scandinavica 14:97-103.
Harper, J. L. 1994. Population Biology of Plants. Academic
Press, San Diego, Californa, USA.
Harris, G. P. 1980. Temporal and spatial scales in
phytoplankton ecology. Mechanisms, methods, models, and
management. Canadian Journal of Fisheries and
Aquaculture 37:877-900.
Harris, L. D. 1984. The Fragmented Forest. University of
Chicago Press, Chicago, IL.
Haslett, J. A. 1994. Community structure and the fractal
dimensions of mountain habitats. Journal of Theoretical
Biology 167:407-411.
Heideman, P. D. 1989. Temporal and spatial variation in the
phenology of flowering and fruiting in a tropical
rainforest. Journal of Ecology 77:1059-1079.
Henderson, A., S. P. Churchill and J. Luteyn. 1991.
Neotropical plant diversity. Nature 351:21-22.
Hilty, S. L. 1980. Flowering and fruiting periodicity in a
premontane rain forest in Pacific Colombia. Biotropica
12:292-306.
Hilty, S. L. and W. L. Brown. 1986. A Guide to the Birds of
Colombia. Princeton University Press, Princeton, NJ.
Holdridge, L. R. 1967. Life Zone Ecology. Tropical Science
Center, San Jos, Costa Rica.
Holdridge, L. R., W. C. Grenke, W. H. Hateway, T. Liang, and
J. A. Tosi, Jr. 1971. Forest Environments in Tropical
Life Zones, A Pilot Study. Pergamon Press, Oxford, UK.
Holling, C. S. 1973. Resilience and stability of ecological
systems. Annual Review of Ecology and Systematics 4:1-
23 .
Holling, C. S. 1992. Cross-scale morphology, geometry, and
dynamics of ecosystems. Ecological Monographs 62:447-
502.
Holling, C. S., G. Peterson, P. Marples, J. Sendzimir, K.
Redford, L. Gunderson, and D. Lambert. 1995. Self
organization in ecosystems: lump geometries,

175
periodicities, and morphologies. in B. H. Walker and W.
L. Steffen, editors. Global Change and Terrestrial
Ecosystems. Cambridge University Press, Cambridge,
Massachusetts, USA, in press.
Holloway, J. D., A. H. Kirk-Spriggs and C. von Khen. 1992.
The response of some rain forest insect groups to
logging and conversion to plantation. Philosophical
Transactions of the Royal Society of London B
335:425:436.
Hooghiemstra, H. 1984. Vegetational and climatic history of
the high plain of Bogota, Colombia: A continuous record
of the last 3.5 million years. Dissertationes Botanicae
79. Cramer, Vaduz.
Hubbell, S. G. and R. B. Foster. 1987. Biology, chance, and
history and the structure of tropical rain forest tree
communities. Pages 314-329, in J. Diamond and T. J.
Case, editors. Community Ecology. Harper and Row, New
York, NY.
Hughes, L., M. Dunlop, K. French, M. R. Leishman, B. Rice, L.
Rodgerson, and M. Westoby. Predicting dispersal
spectra: a minimal set of hypotheses based on plant
attributes. Journal of Ecology 82:933-950.
Irving, E. M. 1975. Structural evolution of the
northernmost Andes, Colombia. U.S. Geological Survey
Professional Paper 846, Washington, DC, USA.
Isler, M. L. and P. R. Isler. 1987. The Tanagers: Natural
History, Distribution and Identification. Smithsonian
Institution Press, Washington, DC, USA.
Jackson, J. E. and J. W. Palmer. 1977. Effects of shade on
the growth and cropping of apple trees. II. Effects on
components of yield. Journal of Horticultural Science
52:253-266.
Janzen, D. H. 1983. No park is an island: Increase in
interference from outside as park size decreases. Oikos
41:402-410.
Johnels, S. A. and T. C. Cuadros. 1986. Species composition
and abundance of bird fauna in a disturbed forest in the
Central Andes of Colombia. Hornero 12:235-241.
Johns, A. D. 1992. Vertebrate responses to selective
logging: implications for the design of logging systems.
Philosophical Transactions of the Royal Society of
London B 335:437-442.

176
Johnson, A. R., B. T. Milne, and J. A. Wiens. 1992.
Diffusion in fractal landscapes: simulations and
experimental studies of Tenebronid beetle movements.
Ecology 73:1968-1983.
Rapos, V. 1989. Effects of isolation on the water status of
forest patches in the Brazilian Amazon. Journal of
Tropical Ecology 5:173-185.
Rapos, V., E. Wandelli, J. L. Camargo, and G. Ganade. Edge-
related changes in environment and plant responses due
to forest fragmentation in Central Amazonia. in W. F.
Laurance and R. 0. Bierregaard, Jr., editors. Tropical
Forest Remnants: Ecology, Management and Conservation of
Fragmented Communities, in press.
Rapos, V. G. Ganade, E. Matsui, and R. L. Victoria. 1993.
313C as an indicator of edge effects in tropical
rainforest reserves. Journal of Ecology 81:425-432.
Karr, J. R. 1971. Structure of avian communities in selected
Panama and Illinois habitats. Ecological Monographs
41:207-233 .
Karr, J. R. 1981a. Surveying birds in the tropics. Studies
in Avian Biology 6:548-553.
Karr, J. R. 1981b. Surveying birds with mist nets. Studies
in Avian Biology 6:62-67.
Karr, J. R. and R. R. Roth. 1971. Vegetation structure and
avian diversity in several new world areas. American
Naturalist 105:423-435.
Kattan, G. 1992. Rarity and vulnerability: The Birds of the
Cordillera Central of Colombia. Conservation Biology
6:64-70.
Kattan, G., C. Restrepo, and M. Giraldo. 1984. Estructura de
un bosque de niebla en la Cordillera Occidental, Valle
del Cauca, Colombia. Cespedesia 13:23-43.
Kattan, G. H., H. Alvarez-L. and M. Giraldo. 1994. Forest
fragmentation and bird extinctions: San Antonio eighty
years later. Conservation Biology 8:138-146.
Kellman, M. and M. Kading. 1992. Faciliation of tree
establishment in a sand dune succession. Journal of
Vegetation Science 3:679-688.
King, D. A. 1991. Allometry of saplings and understory trees
of a Panamanian forest. Oecologia 51:351-356.

177
Klein, B. C. 1989. Effects of forest fragmentation on dung
and carrion beetle communities in Central Amazonia.
Ecology 70:1715-1725.
Kroodsma, R. L. 1984. Effect of edge on breeding forest bird
species. Wilson Bulletin 96:426-436.
Kruess, A. and T. Tscharntke. 1994. Habitat fragmentation,
species loss, and biological control. Science 264:1581-
1584.
Kuitunen, M. and M. Mkinen. 1993. An experiment on nest
site choice of the Common Treecreeper in fragmented
boreal forest. Ornis Fennica 70:163-167.
Lambert, F. R. 1992. The consequences of selective logging
for Bornean lowland forest birds. Philosophical
Transactions of the Royal Society of London B 335:443-
457.
Laurance, W. F. 1990. Comparative responses of five arboreal
marsupials to tropical forest fragmentation. Journal of
Mammalogy 71:641-653.
Laurance, W. F. 1991. Ecological correlates of extinction
proneness in Australian tropical rain forest mammals.
Conservation Biology 5:79-89.
Laurance, W. F. 1993. Avian nest predation in modified and
natural habitats in tropical Queensland: an experimental
study. Wildlife Research 20:711-723.
Laurance, W. F. 1994. Rainforest fragmentation and the
structure of small mammal communities in tropical
Queensland. Biological Conservation 69:23-32.
Lawton, J. H., S. Nee, A. J. Letcher, and P. H Harvey. 1994.
Animal distributions: patterns and processes. Pages 41-
58 in P. J. Edwards, R. M. May, and N. R. Webb, editors.
Large-scale Ecology and Conservation Biology. Blackwell
Scientific Publications, Oxford, UK.
Lawton, R. and V. Dryer. 1980. The vegetation of the
Monteverde Cloud Forest Reserve. Brenesia 18:101-116.
Lawton, R. 0. and F. E. Putz. 1988. Natural disturbance and
gap-phase regeneration in a wind-exposed tropical cloud
forest. Ecology 69:764-777.
Leigh, E. G., Jr. 1975. Structure and climate in tropical
rain forest. Annual Review of Ecology and Systematics.
6:67-86.

178
Lescourret, F. and M. Genard. 1994. Habitat, landscape and
bird composition in mountain forest fragments. Journal
of Environmental Management 40:317-328.
Levey, D. J. 1986. The role of seed size and bird species on
seed deposition patterns, Pages 147-158, in A. Estrada
and T.H. Fleming, editors. Frugivores and Seed
Dispersal. Junk, Dordrecht, The Netherlands.
Levey, D. J. 1987. Seed size and fruit-handling techniques of
avian frugivores. American Naturalist 129:471-485.
Levey, D. J. 1988a. Spatial and temporal variation in Costa
Rican fruit and fruit-eating bird abundance. Ecological
Monographs 58:251-269.
Levey, D. J. 1988b. Tropical wet forest treefall gaps and
distributions of understory birds and plants. Ecology
69:1076-1089.
Levey, D. J. 1990. Habitat-dependent fruiting behaviour of an
understorey tree, Miconia centrodesma, and tropical
treefall gaps as keystone habitats for frugivores in
Costa Rica. Journal of Tropical Ecology 6:409-420.
Li-Cor, Inc. 1992. LAI-2000 Plant Canopy Analyzer Operating
Manual. Li-Cor, Inc., Lincoln, NB.
Lipsey, M. W. 1990. Design Sensitivity: Statistical Power
for Experimental Research. SAGE Publications, Inc.,
Newbury Park, California, USA.
Loiselle, B. A. and J. G. Blake. 1991. Temporal variation in
birds and fruits along an elevational gradient in Costa
Rica. Ecology 72:180-193.
Loiselle, B. A. and J. G. Blake. 1993. Spatial distribution
of understory fruit-eating birds and fruiting plants in
a neotropical lowland wet forest. Vegetatio
107/108:177-189.
Lopez de Casenave, J., J. P. Pelotto, S. Caziani, M. Mermiz,
and J. Protomastro. Edge effects in avian assemblages
of a Chaco semiarid forest, Argentina. Unpublished
manuscript.
Lovejoy, T. E., R. O. Bierregaard, A. B. Rylands, J. R.
Malcolm, C. E. Quinetela, L. H. Harper, K. S. Browns, A.
H. Powell, G. V. N. Powell, H. O. R. Schubart, and M. B.
Hays. 1986. Edge and other effects of isolation on
Amazon forest fragments. Pages 257-285 in M. E. Soul,
editor. Conservation Biology: The Science of Scarcity

179
and Diversity. Sinauer Associates, Sunderland,
Massachusetts, USA.
Loye, J. and S. Carroll. 1995. Birds, bugs and blood: avian
parasitism and conservation. TREE 10:232-235.
Lozano, I. E. 1993. Diversidad y organizacin en gremios de
la comunidad de aves del sotobosque de bosque primario y
vegetacin secundaria. Pages 141-164, in G.I. Andrade,
editor. Carpanta Selva Nublada y Pramo. Fundacin
Natura, Colombia.
Lugo, A. E. and J. L. Frangi. 1993. Fruit-fall in the
Luquillo Experimental Forest, Puerto Rico. Biotropica
25:73-84 .
Lynam, A. J. Processes of local extinction of small mammals
in a recently fragmented tropical monsoon forest in
Thailand. in W.F. Laurance and R.O. Bierregaard, Jr.,
editors. Tropical Forest Remnants: Ecology, Management
and Conservation of Fragmented Communities, in press.
Lynch, J. F. 1989. Distribution of overwintering nearctic
migrants in the Yucatan Peninsula, I: General patterns
of occurrence. Condor 91:515-544.
MacDougall, A. and M. Reliman. 1992. The understory light
regime and patterns of tree seedlings in tropical
riparian forest patches. Journal of Biogeography
19:667-675.
Malcolm, J. R. 1991. The small mammals of Amazonian Forest
fragments: Pattern and process. Ph.D. Dissertation,
University of Florida, Gainesville, FL.
Malcolm, J. R. 1994. Edge effects in Central Amazonian
forest fragments. Ecology 75:2438-2445.
Manly, B. F. 1992. The Design and Analysis of Research
Studies. Cambridge University Press, Cambridge, UK.
Margalef, R. 1968. Perspectives in Ecological Theory. The
University of Chicago Press, Chicago, IL.
Marshall, C. and J. Grace, editors. 1992. Fruit and Seed
Production: Aspects of Development, environmental
physiology and ecology. Cambridge University Press,
Cambridge, UK.
Martin, T. E. 1985. Selection of second-growth woodlands by
frugivorous migrating birds in Panama: an effect of
fruit size and plant density? Journal of Tropical
Ecology 1:157-170.

180
Martnez-Ramos, M. and A. Soto-Castro. 1993. Seed rain and
advanced regeneration in a tropical rain forest.
Vegetatio 107/108:299-318.
Mathai, C. K. and K. S. K. Sastry. 1988. Productivity of
black pepper vines (Piper niarum L.) as influences by
the light availability during pre-flowering stage.
Comparative Physiological Ecology 13:97-102.
Matlack, G. R. 1993. Microenvironment variation within and
among forest edge sites in the Eastern United States.
Biological Conservation 66:185-194.
May, P. and A. J. Antcliff. 1963. The effect of shading on
fruitfulness and yield in the sultana. Journal of
Horticultural Science 38:85-94.
McDiarmid, R. W., R. E. Ricklefs, and M. S. Foster. 1977.
Dispersal of Stemmadenia dQnnelljl-smlthii (Apocynaceae)
by birds. Biotropica 9:9-25.
Mejia, J., H. Meyer, A. Campos, A. Gallego, and A. Velsquez.
1994. El terremoto de Paez, anatoma de un sismo, Parte
I: generalidades y movimientos fuertes. Revista de la
Asociacin de Ingenieros del Valle (472):14-15.
Miller, A. H. 1963. Seasonal activity and ecology of the
avifauna of an american equatorial cloud forest.
University of California Publications in Zoology 66:1-
78.
Moermond, T. C. and J. S. Denslow. 1985. Neotropical avian
frugivores: patterns of behavior, morphology, and
nutrition, with consequences for fruit selection.
Neotropical Monographs 36:865-897.
Moller, A. P. 1989. Nest site selection across field-
woodland ecotones: the effect of nest predation. Oikos
56:240-246.
Mondragn, M. L. 1989. Estructura de la comunidad aviaria
en bosques de coniferas y en bosques aledaos de
vegetacin nativa. Trabajo de Grado, Departamento de
Biologa, Universidad del Valle, Cali, Colombia.
Munves, J. 1975. Birds of highland clearing in
Cundinamarca, Colombia. Auk 92:307-321.
Murcia, C. 1993. Edge Effects on the Pollination of Tropical
Cloud Forest Plants. Ph.D. Dissertation, University of
Florida, Gainesville, FL.

181
Murray, K. G. 1987. Selection for optimal fruit-crop size in
bird-dispersed plants. American Naturalist 129:18-31.
Murray, K. G. 1988. Avian seed dispersal of three neotropical
gap-dependent plants. Ecological Monographs 58:271-298.
Naranjo, L. G. 1994. Composicin y estructura de la avifauna
del Parque Regional Natural Ucumari. Pages 305-328 in
J.O. Rangel, editors. Ucumari: Un caso tpico de la
diversidad bitica Andina, Corporacin Autnoma Regional
de Risaralda, Colombia.
Newmark, W.D. 1991. Tropical forest fragmentation and the
local extinction of understory birds in the eastern
Usambara mountains, Tanzania. Conservation Biology
5:67-78.
Noble, I. A. 1993. A model of the responses of ecotones to
climate change. Ecological Monographs 3:396-403.
Norman, J. M. and G. S. Campbell. 1991. Canopy structure,
Pages 301-326, in R. W. Pearcy, J. Ehleringer, H. A.
Mooney, and P. W. Rundel,editors. Plant Physiological
Ecology. Chapman and Hall, London, UK.
Noss, R. F. 1991. Effects of edge and internal patchiness on
avian habitat use in an Old-Growth Florida Hammock.
Natural Areas Journal 11:34-47.
Orejuela, J. E. 1987. La Reserva Natural La Planada y la
biogeografa andina. Humboldtia 1:117-148.
Orejuela, J. E. and G. Cantillo, compilers. 1980. Aves de la
Reserva Natural La Planada. Programa Alegra de Ensear,
Fundacin para la Educacin Superior-FES, Cali,
Colombia.
Orejuela, J. E., R. S. Raitt and H. Alvarez. 1979.
Relaciones ecolgicas de las aves en la Reserva Forestal
de Yotoco, Valle del Cauca. Cespedesia 29-30:7-28.
Osunkoya, O., J. E. Ash, M. S. Hopkins, and A. W. Graham.
1994. Influence of seed size and seedling ecological
attributes on shade-tolerance of rain-forest tree
species in northern Queensland. Journal of Ecology
82:149-163.
Pacheco, S., N. Florencio da Silva, R. Ribon, J. E. Simon and
R. Torres. 1994. Efeito do manejo do cerrado sobre as
populagoes de algus Tinamidae em Tres Marias, Estado de
Minas Gerais. Revista Brasilera Biologia 54:435-441.
Pimm, S. L. and A. M. Sugden. 1994. Tropical diversity and
global change. Science 263:933-934.

182
Piero, D. and J. Sarukhan. 1982. Reproductive behaviour and
its individual variability in a tropical palm,
Astrocarvum mexicanum. Journal of Ecology 70:461-472.
Potvin, C. 1993. Experiments in controlled environments.
Pages 46-68, in S.M. Scheiner and J. Gurevitch, editors.
Design and Analysis of Ecological Experiments, Chapman
Hall, New York, NY.
Poulin, B., G. Lefebvre, and. R. McNeil. 1992. Tropical
avian phenology in relation to abundance and
exploitation of food resources. Ecology 73:2295-2309.
Quintela, C. 1986. Forest fragmentation and differential use
of natural and man-made edges by understory birds in
Central Amazonia. M.Sc. Thesis, University of Chicago,
Chicago, IL.
Rangel, J. 0., editor. 1994. Ucumar: Un caso tpico de la
diversidad bitica Andina, Corporacin Autnoma Regional
de Risaralda, Colombia.
Ranney, J. W., M. C. Bruner, and J. B. Levenson. 1981. The
importance of edge in the structure and dynamics of
forest islands. Pages 67-95, in R. L. Burguess and D. M.
Sharpe, editors. Forest Island Dynamics in Man-
Dominated Landscapes. Springer-Verlag, New York, NY.
Remsen, J.V., Jr. 1994. Use and misuse of bird lists in
community ecology and conservation. Auk 111:225-227.
Remsen, J. V. Jr. and T. A. Parker. 1983. Contribution of
river-created habitats to bird species richness in
Amazonia. Biotropica 15:223-231.
Renjifo, L. M. 1988. Composicin y estructura de la comunidad
aviaria de bosque andino primario y secundario en la
Reserva del Alto Quindo Acaime, Colombia. Tesis,
Departamento de Biologa, Pontificia Universidad
Javeriana, Santa F de Bogot, Colombia.
Renjifo, L. M., G. P. Servat, J. M. Goerck, B. A. Loiselle
and J. G. Blake. Patterns of species composition and
endemism in the northern Neotropics: a case for
conservation of montane avifaunas, in J. V. Remsen,
editor. Natural History and Conservation of Neotropical
Birds, Ted Parker Memorial Volume, Ornithological
Monographs, in press.
Restrepo, C. 1990. Ecology and cooperative breeding in a
frugivorous bird, Semnornis ramohastinus. M.Sc. Thesis,
University of Florida, Gainesville, FL.

183
Ridgely, R. S. and G. Tudor. 1989. The birds of South
America. VI. The Oscine Passerines. University of
Texas Press, Austin, TX.
Ridgely, R. S. and G. Tudor. 1994. The birds of South
America. V 2. The Suboscine Passerines. University of
Texas Press, Austin, TX.
Ridgely, R. S. and S. J. C. Gaulin. 1980. The birds of finca
Merenberg, Huila department, Colombia. Condor 82:379-
391.
Romano, G. B. 1990. Invasibility of a Mixed Hardwood Forest
by Euoatorium caoillifolium and E. comoositifolium.
M.Sc. Thesis, Universtiy of Florida, Gainesville, FL.
Rosas, M. L. 1986. Estudio de la estructura de la comunidad
de frugvoros en sotobosques del Can de Mamarramos en
el Santuario de Fauna y Flora de Iguaque, Boyac.Tesis,
Departamento de Biologa, Pontificia Universidad
Javeriana, Santa F de Bogot, Colombia.
Ryszkowski, L. 1992. Energy and material flows across
boundaries in agricultural landscapes. Pages 270-284 in
A.J. Hansen and F. di Castri, editors. Landscapes
Boundaries: Consequences for Biotic Diversity and
Ecological Flows. Springer-Verlag, New York, NY.
Salisbury, E. 1974. Seed size and mass in relation to
environment. Proceedings of the Royal Society, London B
186:83-88.
Salo, J., R. Kallioloa, R. Hkkinen, Y. Mkinen, P. Niemela,
M. Puhakka, and P. D. Coley. 1986. River dynamics and
the diversity of Amazon lowland forest. Nature 322:254-
258.
Samper, C. 1992. Natural disturbance and plant establishment
in an Andean cloud forest. Ph.D. Dissertation,
University of Harvard, Cambridge, MA.
Scheiner, S. M. 1993. Introduction: Theories, hypotheses,
and statistics. Pages 1-13, in S. M. Scheiner and J.
Gurevitch, editors. Design and Analysis of Ecological
Experiments, Chapman Hall, New York, NY.
Schupp, E. W. 1988. Seed and early seedling predation in the
forest understory and in treefall gaps. Oikos 51:71-78.
Schupp, E. W. and E. J. Frost. 1989. Differential predation
o Welfia qeorqii seeds in treefall gaps and the forest
understory. Biotropica 21:200-203.

184
Schupp, E. W., H. F. Howe, C. K. Augspurger, and D. J. Levey.
1989. Arrival and survival in tropical treefall gaps.
Ecology 30:562-564.
Seghieri, J., Ch. Floret, and R. Pontanier. 1995. Plant
phenology in relation to water availability: herbaceous
and woody species in the savannas of northern Cameroon.
Journal of Tropical Ecology 11:237-254.
Seizer, N. G. 1992. The impact of edge formation on
regeneration and litterfall in a tropical rain forest
fragment in Amazonia. Ph.D. Dissertation, University of
Cambridge, Cambridge, UK.
Shoji, S., M. Nanzyo, and R. A. Dahlgren. 1993. Volcanic ash
soils: Development, properties, and utilization.
Developments in Soil Science 21. Elsevier Science
Publishers B.V., Amsterdam, The Netherlands.
Siegel, S. and N. J. Castellan. 1988. Nonparametric
Statistics for the Behavioral Sciences. McGraw-Hill Book
Company, New York, NY.
Silverman, B. W. 1986. Density Estimation for Statistics and
Data Analysis. Monographs on Statistics and Applied
Probability 26, Chapman Hall, London, UK.
Sokal, R. R. and F. J. Rohlf. 1981. Biometry. Freeman, New
York, NY.
Sork, V. L. 1983. Distribution of pignut hickory (Carva
glabra) along a forest to edge transect, and factors
affecting seedling recruitment. Bulletin Torrey
Botanical Club 110:494-506.
Stephenson, A. G. 1981. Flower and fruit abortion: proximate
causes and ultimate functions. Annual Review of Ecology
and Systematics 12:253-279.
Stephenson, A. G. 1992. The regulation of maternal
investment in plants. Pages 151-172, in C. Marshall and
J. Grace, editors. Fruit and Seed Production: Aspects
of Development, Environmental Physiology and Ecology.
Cambridge University Press, Cambridge, UK.
Stiles, F. G. 1985. On the role of birds in the dynamics of
neotropical forests. ICBP Technical Publication 4:49-
59.
Stiles, F. G. and A. Skutch. 1989. A Guide to the Birds of
Costa Rica. Cornell University Press, Ithaca, New York,
USA.

185
Taylor, Ch. M. 1989. Revision of Palicourea (Rubiaceae) in
Mexico and Central America. Systematic Botany
Monographs 26.
Terborgh, J. 1971. Distribution on environmental gradients:
theory and a preliminary interpretation of
distributional patterns in the avifauna of the
Cordillera Vilcabamba, Peru. Ecology 52:23-40.
Terborgh, J. 1977. Bird species diversity on an Andean
elevational gradient. Ecology 58:1007-1019.
Terborgh, J. 1985. The role of ecotones in the distribution
of Andean birds. Ecology 66:1237-1246.
Terborgh, J. and B. Winter. 1983. A method of setting parks
and reserves with special reference to Colombia and
Ecuador. Biological Conservation 27:45-58.
Terborgh, J. and J. S. Weske. 1969. Colonization of
secondary habitats by Peruvian birds. Ecology 50:765-
782 .
Thiollay, J. M. 1992. Influence of selective logging on bird
species diversity in a Guianan rain forest.
Conservation Biology 6:47-63.
Tilman, D. and J. A. Downing. 1994. Biodiversity and
stability in grasslands. Nature 367:363-365.
Tombesi, A., E. Antognozii and A. Pallioti. 1994. Optimum
leaf area in T-bar trained kiwifruit vines. Journal of
Horticultural Science 69:339-350.
Turner, M. G., R. H. Gardner, and R. V. O'Neill. 1995.
Ecological dynamics at broad scales. BioScience
(Supplement):29-35.
Tutin, C. E. G. and M. Fernandez. 1993. Relationships
between minimum temperature and fruit production in some
tropical forest trees. Journal of Tropical Ecology
9:241-248.
Uribe, D. A. 1986. Contribucin al conocimiento de la
avifauna del bosque muy hmedo montano bajo en las
cercanas de Manizales. Tesis, Facultad de Medicina
Veterinaria y Zootectnia, Universidad de Caldas,
Manizales, Colombia.
van Wambeke, A. 1992. Soils of Tropics: Properties and
Appraisal. McGraw-Hill, Inc., New York, NY.

186
Velsquez, A., H. Meyer, W. Marin, F. Ramrez, A. David, A.
Campos, M. Hermeln, S. 0. Bender, M. Arango and J.
Serje. 1994. Planificacin regional del occidente
colombiano bajo consideracin de las restricciones por
amenazas. Memorias Conferencia Interamericana sobre
Reduccin de los Desastres Naturales, Cartagena de
Indias, Colombia, Marzo 1994. Casa Impresora Pacfico,
Santa F de Bogot, Colombia.
Velsquez, M. P. 1992. Aves frugvoras y su relacin con la
flora en un bosque hmedo, en el municipio de San
Carlos, Antioquia, Colombia. Trabajo de Investigacin,
Departamento de Biologa, Universidad de Antioquia,
Medellin, Colombia.
Velez, B. E. 1987. Contribucin al estudio avifaunstico del
Santuario de Fauna y Flora de Iguaque, Boyac.Tesis,
Departamento de Biologa, Pontificia Universidad
Javeriana, Santa F de Bogot, Colombia.
Vitousek, P. M. 1990. Biological invasions and ecosystem
processes: towards an integration of population biology
and ecosystem studies. Oikos 57:7-13.
von Ende, C. N. 1993. Repeated-measures analysis: Growth and
other time-dependent measures. Pages 113-137, in S.M.
Scheiner and J. Gurevitch, editors. Design and Analysis
of Ecological Experiments, Chapman Hall, New York, NY.
Walker, L. R. and L. E. Neris. 1993. Posthurricane seed rain
dynamics in Puerto Rico. Biotropica 25:408-418.
Warburton, N. Structure and conservation of forest avifauna
in isolated rainforest remnants in tropical Australia.
in W. F. Laurance and R. O. Bierregaard, Jr., editors.
Tropical Forest Remnants: Ecology, Management and
Conservation of Fragmented Communities, in press.
Westoby, M., E. Jurado, and M. Leishman. 1992. Comparative
evolutionary ecology of seed size. TREE 7:368-372.
Wiens, J. A. 1992. Ecological flows across landscape
boundaries: a conceptual overview. Pages 217-235, in
A. J. Hansen and F. di Castri, editors. Landscape
Boundaries: Consequences for Biotic Diversity and
Ecological Flows. Springer-Verlag, New York, NW.
Wiens, J. A., C. S. Crawford and J. R. Gosz. 1985. Boundary
dynamics: a conceptual framework for studying landscape
ecosystems. Oikos 45:421-427.
Wilcove, D. S., C. H. McLellan, and A. P. Dobson. 1986.
Habitat fragmentation in the temperate zone. Pages 237-
256, in M.E. Soul, editor. Conservation Biology: The

187
Science of Scarcity and Diversity. Sinauer, Sunderland,
Massachusetts, USA.
Williams-Linera, G. 1990. Vegetation structure and
environment of forest edges in Panama. Journal of
Ecology 78:356-373.
Winer, B. J., D. R. Brown and K. M. Michaels. 1991.
Statistical Principles in Experimental Design. McGraw-
Hill, New York, NY.
Wong, M. 1986. Trophic organization of understory birds in a
Malasyan dipterocarp forest. Auk 103:100-116.
Young, K. R. 1993. Woody and scandent plants on the edges of
an Andean timberline. Bulletin of the Torrey Botanical
Club 120:1-18.
Zolman, J. F. 1993. Biostatistics: Experimental Design and
Statistical Inference. Oxford University Press, New
York, NY.

BIOGRAPHICAL SKETCH
I was born in the midst of a summer in Los Angeles, CA,
to a Colombian couple. At that time my father was pursuing
his Ph.D. degree in Mathematics and my mother was taking care
of their two daughters. I got to spend a lot of time with my
mother and she would tell me stories about her family,
Colombia, and the activities of my father as a researcher.
She also took the time to answer my questions, many of which
had to do with the whys and hows of animal and plant life.
Later on it was my mother's turn to complete her degree in
the Social Sciences and for my father to take care of us. He
would tell us stories of his own invention. It was from one
of these that I learned that birds migrate and that not all
individuals succeed. My compassion for other organisms was
partially motivated by the little swallow that could never
make it through its long journey.
By the time my parents moved back to Colombia, and after
living in Mexico and Puerto Rico, I was conscious of my
fascination for the living world. I knew that I wanted to
unravel some of its misteries and one way to do so was to
study Biology. At the Universidad del Valle I had plenty of
opportunities to learn about biological principles, natural
history, and politics. I found myself editing environmental
188

189
newsletters, participating in human right groups, and
discovering the pleasure of observing birds. For this I have
to give credit to my former advisor, Humberto Alvarez-L.
Birds and a two-month field season at the Sierra Nevada de
Santa Marta were my gate into Ecology, which I have not
abandoned since then.
The snowball has grown bigger, with lumps here and
there, reminding me that my existence has not been plain.
The time I spent at the University of Florida pursuing my
Ph.D degree certainly shaped this snowball. The core remains
pretty much unaltered, perhaps strengthened, and that is why
some influential people in my life roll their eyes when I
tell them my whereabouts. I do not know if this behavior is
a warning signal "Carla you are getting into trouble" or a
trusting signal "Carla you are going to make it". Perhaps it
is a combination of both.
It seems that the snowball will continue to grow, as my
eyes have widened so has the scale I want to look at things.
I will spend the next two years doing post-doc work at
Stanford University and at the University of Florida, adding
further lumps to my life.

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor~of Philosophy.
Douglas*'#. Levey J Chair
Associate Profesor of
Zoology
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
H^/Jane Brockmann
Professor of Zoology
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor jaf Philosophy.
C.S. Hollirrg
Arthur R. Marshall, Jr.,
Professor of Ecological
Sciences
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
)hn Ewel
¡sor of Botany
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Frrk Slansky
Professor of Ento:
Nematology
ogy and

This dissertation was submitted to the Graduate Faculty
of the Department of Zoology in the College of Liberal Arts
and Sciences and to the Graduate School and was accepted as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
December, 1995
Dean, Graduate School



4-2 Results of Mixed Factorial ANOVAs on fruit
abundance across pasture-forest edge 59
4-3 Results of Goodness of Fit Test on the number of
fruiting individuals of abundant species across
pasture-forest edge 68
4-4 Results of Replicated Goodness of Fit Test on the
number of fruiting individuals across pasture-forest
edge in old and new edges 69
4-5 Distribution of number of fruiting individuals
across pasture-forest edge based on species abundance... 70
4-6 Summary of results of changes in fruit abundance
across the pasture-forest edge 74
5-1 Results of ANOVAs for Mixed Factorial Design on
capture rates of understory birds 92
5-2 Results of Goodness of Fit Test on the number of
bird captures of abundant species across pasture-
forest edge 101
5-3 Results of the Replicated Goodness of Fit Test on
the number of bird captures across pasture-forest
edge 103
5-4 Distribution of bird captures across pasture-
forest edge based on species abundance 104
6-1 Description of sites included in lump analyses of
body mass of frugivorous birds 119
vi 1


Drvmophila caudata
Formicarius rufipectus
Grallaricula flavirostris
Thamnophilus unicolor
Cotingidae
Pachvramphus alboariseus
Pachvramphus versicolor
Pipreola riefferii
Lipangug qryptolpphus
Pipridae
Allacotopterus deliciosus
Masius chrvsopterus
Tyrannidae
Mionectes striaticollis
Mviobius barbatus
Mviophobus flavicans
Mviotriccus ornatus
Mviodvnastes chrvsocephalus
Ochtoeca cinnamomeiventris
Poaonotriccus ophtalmicus
Pseudotriccus pelzelni
Pseudotriccus ruficeps
Zimmerius viridiflavus
Troglodytidae
Henicorhina leucophrvs
Turdidae
Catharus ustulatus
Entomodestes coracinus
Mvadestes ralloides
Turdus serranus
Vireonidae
VS I 10.5
VS I 68.5
1 I 20.4
S I 23.3
ES F 16.1
ES F 15.0
A F 51.5
S F 80.4
A F 12.4
VA F 11.5
VA F 13.8
ES I 12.0
A I 12.9
A I 10.4
VS I 38.8
VS I 12.2
VS I 9.5
VA I 11.8
ES I 9.5
VS F 10.2
VA I 15.6
ES F 29.5
VS F 14.5
VA F 28.5
ES F 28.41
164


94
interaction (ANOVA, F4,i6 = 3.5, P = 0.05; Table 5-1, Fig. 5-
2). Capture rates remained very similar at DI, D2, and D3
for old and new edges but differed at D4, where they were
higher at new edges (mean = 24.2) than at old edges (mean =
14.1) (Contrast of Mean Differences, Fi,i2 = 12.0, P = 0.005).
The effect of month on capture rates for all birds was
significant (ANOVA, Fi5,5g = 2.9, P = 0.008). However,
changes in bird captures over the year depended on edge age
as shown by the significant edge age x month interaction
(ANOVA, F = 15,59 = 2.1, P = 0.09; Table 5-1, Fig. 5-3). This
suggests that edge features, such as edge age, can influence
the distribution of birds, depending on season. Old and new
edges did not differ significantly in regard to bird captures
during the dry months (Contrast of Mean Differences, Fi,4o =
1.5, P = 0.2) but they did differ during the wet months
(Contrast of Mean Differences, Fi,4o = 13.1, P = 0.0008).
Feeding Guild Responses to Edges
The abundance of frugivores and insectivores differed
significantly with distance (ANOVA, F3,i2 = 4.9, P = 0.02 and
F3,i2 = 2.1, P = 0.08, respectively, Table 5-1, Fig. 5-1).
Capture rates for frugivores were significantly higher at D1
(mean = 5.6) and D4 (mean = 7.0) compared to D2 and D3
combined (mean = 3.5) (Contrast of Mean Differences, Fi,i2 =
7.0, P = 0.02 and Fi,i2 = 12.0, P = 0.005,


73
expressed as total number of fruits (TF) and as total number
of ripe fruits (RF), was higher in new than in old edges at
D1 but these differences were not evident at the other
distances. At the scale defined by the edges and by the
length of this study, treefall gaps at D1 generated spatial
heterogeneity in fruit abundance within edges. Fruit
abundance, expressed as total number of fruits (TF) and as
total number of ripe fruits (RF) was higher in treefall gaps
than in forest interior at D1 but, again, these differences
were not found at the other distances.
When I looked at total number of ripe fruits (RF), and
in addition examined the total number of fruits excluding the
Arecaceae (TF-A) and the number of fruiting individuals (TI),
a different picture emerged. At the scale of the study area
and month, RF differed between old and new edges but depended
on month. At the scale of edges and month, TF-A and TI
differed among the four distances but also depended on month.
These results suggest that changes in fruit abundance
and the magnitude of these changes across pasture-forest edge
are related to the size of fruit "patches". In my study
area, large patches of fruit were generated by understory
palms, which produced large fruit crops that persisted for a
long time. These large fruit patches seemed to generate a
steep gradient in fruit abundance from the edge towards the
forest interior on a yearly basis. Conversely, small patches
of fruit seemed to generate gradients that varied in their
magnitude depending on month. I will discuss these results


141
consequence of human activities on the landscape (Pacheco et
al. 1994, Lynam, in press).
Causes of Lump Structure
The comparisons that I set up among elevational zones
and among sites within elevational zones represent two
different scales of inquiry, yet produced similar results.
Elevational zones and sites within elevational zones were
arranged from those covered mostly by native forest to those
in which the native forest has been replaced by open
vegetation (pramo) or managed ecosystems (e.g., pastures and
forestry tree plantations) in which scattered fragments of
native forest remain. Thus, changes in lump structure seem
to reflect a common causality best explained by the
complexity of landscapes in terms of the vertical and
horizontal structure of the vegetation. My proposition is
supported by work done in other regions (Thiollay 1992,
Lescourret and Genard 1994).
Vertical structure of the forest, including height of
the vegetation and diversity of growth forms, is simpler as
one moves from the lowlands to the pramos or from forest to
pastures. Along the elevational gradient, fog, air
temperature, and radiation are proximate factors that explain
changes in the vegetation structure (Leigh 1975, Grubb 1977).
Along the gradient of land-use, soil and rainfall
distribution are proximate factors that explain changes in
the structure of the vegetation (e.g., Holdrige et al. 1971).


185
Taylor, Ch. M. 1989. Revision of Palicourea (Rubiaceae) in
Mexico and Central America. Systematic Botany
Monographs 26.
Terborgh, J. 1971. Distribution on environmental gradients:
theory and a preliminary interpretation of
distributional patterns in the avifauna of the
Cordillera Vilcabamba, Peru. Ecology 52:23-40.
Terborgh, J. 1977. Bird species diversity on an Andean
elevational gradient. Ecology 58:1007-1019.
Terborgh, J. 1985. The role of ecotones in the distribution
of Andean birds. Ecology 66:1237-1246.
Terborgh, J. and B. Winter. 1983. A method of setting parks
and reserves with special reference to Colombia and
Ecuador. Biological Conservation 27:45-58.
Terborgh, J. and J. S. Weske. 1969. Colonization of
secondary habitats by Peruvian birds. Ecology 50:765-
782 .
Thiollay, J. M. 1992. Influence of selective logging on bird
species diversity in a Guianan rain forest.
Conservation Biology 6:47-63.
Tilman, D. and J. A. Downing. 1994. Biodiversity and
stability in grasslands. Nature 367:363-365.
Tombesi, A., E. Antognozii and A. Pallioti. 1994. Optimum
leaf area in T-bar trained kiwifruit vines. Journal of
Horticultural Science 69:339-350.
Turner, M. G., R. H. Gardner, and R. V. O'Neill. 1995.
Ecological dynamics at broad scales. BioScience
(Supplement):29-35.
Tutin, C. E. G. and M. Fernandez. 1993. Relationships
between minimum temperature and fruit production in some
tropical forest trees. Journal of Tropical Ecology
9:241-248.
Uribe, D. A. 1986. Contribucin al conocimiento de la
avifauna del bosque muy hmedo montano bajo en las
cercanas de Manizales. Tesis, Facultad de Medicina
Veterinaria y Zootectnia, Universidad de Caldas,
Manizales, Colombia.
van Wambeke, A. 1992. Soils of Tropics: Properties and
Appraisal. McGraw-Hill, Inc., New York, NY.


113
(Vitousek 1990, Kruess and Tschnarntke 1994, Pimm and Sugden
1994, Tilman and Downing 1994, Turner et al. 1995).
Holling et al. (1995) provided a novel conceptual
framework that may help integrate these issues. First, long
term research has shown that ecosystems are structured by a
few processes operating at various spatial and temporal
scales (e.g., Clark et al. 1979, Harris 1980, Gunderson
1992). Second, the behavior and morphology of organisms
reflect the discontinuous nature of ecosystems. In
particular, body mass of boreal birds and mammals was found
to be discontinuously distributed, such that species of
similar mass tend to aggregate or lump together (Holling
1992) By focusing on lumps, rather than species, it may be
possible to detect patterns at larger spatial scales such as
those defined by the impact of humans on landscapes. Third,
the morphology of organisms reflects differences in landscape
pattern. In particular, the distribution of body mass in
boreal forest and boreal prairie birds differed (Holling
1992). By focusing on the distribution of body mass, i.e.,
lump structure, it could be possible to investigate the
interaction between landscape pattern and the structure of
animal and plant assemblages.
In this paper I ask how changes in landscape pattern
affect assemblages of frugivorous birds in neotropical
mountains. I concentrate on frugivorous birds because seed
dispersal by birds is especially important in neotropical
mountains compared to the lowlands (Terborgh 1977, Stiles


139
exhibit this linear relationship. For instance, some pairs
of sites (e.g., Merenberg/Ucumar Bajo and Rio
Grande/Munchique) had a similar number of species and yet
showed a different number of lumps. Others (e.g.,
Anchicaya/Yotoco, San Carlos/La Esmeralda, and
Carpanta/Ucumar Alto) had a different number of species and
yet showed a similar number of lumps. A close examination of
the information available for the sites revealed important
structural differences between members of a pair of sites for
the former whereas the same was not true for the latter
(Table 6-1) This indicates that lump structure of body mass
in frugivorous birds reflects to some degree the structure of
landscapes.
Major changes in lump structure in terms of the
persistence of lumps occurred at the upper and lower ends of
the range of body mass. Lumps representing the largest bird
species were lost when moving from areas covered mostly by
native forest to areas where the forest has been replaced by
pramo (elevation zones) and pasture (sites within the upper
lowland zone). The same trend was found for small birds in
the upper montane zone. In the lower montane zone there was
no clear trend regarding the lumps that disappeared along the
gradient of land-use.
The pattern described above may be related to the
proportion of species falling within lumps. Lumps
representing the largest birds contained the smallest
proportion of species for the observed range of body mass for


20
fruits in relation to the number of flower buds and as the
percentage of ripe fruits in relation to the number of unripe
fruits counted over the entire study period for each
individual. I present results for P. cribbosa only.
Fruit Damage by Insects
At the same time I monitored infructescences for unripe
and ripe fruits, I recorded two types of fruit damage by
insects: damage to seeds by wasps (Hymenoptera: Chalcidoidea)
and removal of pulp by ants (Hymenoptera: Formicidae:
Ponerinae). The former could be recognized by exit holes
left by newly emerged adults and the latter by bites taken
from fruits. These two types of fruit damage were the most
common ones for these two understory plants. I expressed
seed and fruit damage as the proportion of unripe fruits
exhibiting one of the two types of damage in relation to the
total number of unripe fruits produced by an individual over
the entire study period. I present results for P. cribbosa
only.
Seedling Growth and Leaf Production
I monitored seedlings of Palicourea aibbosa and Faramea
affinis at each of three distances (0-10, 30-40, 60-70 m) at
three old edges (Climo I, Climo II, and Pialapi) to
establish the combined effect of distance from forest edge
and treefall gaps on seedlings growth and leaf production
(Fig. 2-1, Table 2-2, Chapter 2). In May 1992 I located


SEEDLING GROWTH SEEDS LEFT
(mm/month) (number of seeds/tray)
OJ
cn
Figure 3-2. Seed germination (a), seed predation (b), seedling growth (c), and leaf
production in Palicourea aibbosa and Faramea affinis as influenced by distance from forest
edge. Points represent means and bars standard errors.


Table 3-4. Results of Mixed Factorial ANOVAs (Split-splot, 1 repeated measure) on mean
numbers of seeds remaining in trays (predation experiment) and mean number of seeds
germinated in trays (seed germination experiment) for Palicourea aibbosa (PG) and Faramea
affinis (FA). Significant at 10% (*),5% (**), 1% (***), 0.1% (****)
Distance
df F
D
df
i x E
SS
Habitat
df F
H
df
x D
F
TxHx(DxE)]
Error
df SS
FA
(Predation)
3
3.5
6
75.4
1
1
3
2.3*
80
1068.9
PG
(Predation)
3
0.3
6
182.7
1
1.5
3
1.9
80
2721.1
FA
(Germination)
3
0.1
6
19.9
1
2.3
3
0.5
78
366.8
PG
(Germination)
3
5.0**
6
24.2
1
7.3***
3
1.6
80
910.6
Week x D
W X D X E
Week
W x H
W xH x D
W x T x[Hx(DxE)]
df F
df SS
df F
df F
df F
df SS
FA
(Predation)
51
1.9***
102
50.9
17
44.2****
17
1.7
51
1.3
1360
769.2
PG
(Predation)
15
0.3
30
60.6
5
163.7****
5
1.4
15
1.6
400
845.4
FA
(Germination)
70
2.7***
210
71.4
35
304.4****
35
1.2
105
0.6
2730
907.2
PG
(Germination)
42
2.0**
84
87.8
14
835.0****
14
4.3***
42
0.9
1120
1300.
u>
cn


144
These findings are summarized in a model (Fig. 6-7)
where the triangles represent different ecosystems arranged
from the most complex (top) to the simplest (bottom), such as
elevational zones along an altitudinal gradient or types of
land use along a gradient of human disturbance. The
arrangement of the triangles along a diagonal line reflects
the general trend showing a decrease in the number of lumps
with a decrease in the number of species. Ecosystems were
depicted as triangles to reflect the same general trend but
this time historical, climatic, edaphic factors, and recently
human activities, contribute to the within ecosystem
variability.
In a given ecosystem, lump structure (vertical dimension
of each triangle), defined by the number of lumps, can remain
relatively unchanged in spite of differences in the number of
species among sites (horizontal dimension of each triangle).
In my data sets, such changes were associated with changes in
species numbers within a given type of land use. However, as
one approaches the apex of each triangle the probability that
dramatic changes in lump structure will take place increases
even though species numbers can remain relatively unchanged
(vertical lines) (Fig. 6-7). In my data sets, such changes
were associated with changes in land use. Each jump into
successive ecosystems (from top to bottom) is accompanied by
a reduction both in the number of species and the number of
lumps.


68
Table 4-3. Distribution of fruiting individuals of
abundant (>21 individuals) plant species in the understory of
La Planada in relation to distance from the edge. Numbers
represent number of individuals. Distances are 0-10 m (Dl) ,
30-40 m (D2), 60-70 m (D3), and 190-200 m (D4) from forest
edge. P < 0.1 (*), P < 0.05
(****)
Species
Uniform Distribution
Burmeistera carnosa
Solanum sp.7
Anthurium membranaceum
Burmeistera sp. nov.
Anthurium umbraculum
columnea cinerea
Spheraedenia stCYermarkii
Anthurium cf. pulverulentum
Anthurium versicolor
Geonoma weberbaueri
Non-Uniform Distribution
Psammisia aff. debilis
Besleria solanoides
Psvcothria aubletiana
Clidemia sp.l
Solanum sp.5
Chamaedorea polvchlada
Alloplectus tenuis
Asnlundia sp.1
Anthurium umbricclum
Anthurium cf. marmoratum
Faramea affinis
AllQBl.ectUS teuscheri
Anthurium carchiense
Anthurium cf. melampvi
Palicourea aibbosa
Allpplechus tetraqcnus
(**) P < 0.01 (***), P < 0.001
Dl
D2
D3
D4
G-stat
10
11
13
14
0.83 ns
14
15
11
11
0.99 ns
32
30
27
24
1.31 ns
7
4
5
9
2.35 ns
17
17
13
23
2.87 ns
10
5
4
5
3.34 ns
7
2
6
6
3.37 ns
8
6
5
2
3.99 ns
18
8
16
14
4.36 ns
19
20
9
14
5.29 ns
20
9
4
8
12.9
***
39
16
14
4
35.5

45
14
12
5
40.3

54
59
27
10
48.7

3
14
14
17
12.2

5
7
15
18
10.8

2
2
7
9
8.0

2
10
5
8
6.6

9
7
22
7
12.2

10
9
17
5
7.2

104
146
127
117
7.6

22
41
18
30
10.9

3
15
5
12
11.7
* *
20
10
19
32
12.3
*
91
65
91
43
23.6

10
2
7
1
11.7



CHAPTER 2
DESCRIPTION OF STUDY AREA
Study Areft
I conducted this study at Reserva Natural La Planada and
Finca El Bosque, located in the municipality of Ricaurte,
department of Nario, SW Colombia (7800'W and 110'N) (Fig.
2-1). Both localities lie on the western slope of the Andes
at 1,800 m. The biota of La Planada and its surroundings is
one of the most diverse of the northern Andes (Terborgh and
Winter 1983; Orejuela 1987) and twenty percent of the plant
and animal species reported for the area are endemic
(geographical range <50,000 km^, Terborgh and Winter 1983).
Observations were concentrated in the NW and W portions
of La Planada and El Bosque, respectively. These two areas
lie on the watershed division of the Miraflores and Pialapi
rivers. Colonization of this area started in the early 1940s
and proceeded from the bottom of the valleys (1,200 m) to the
top of the mountains (1,800 m) which are still mostly
covered by forest. Small sugarcane plantations, transient
corn fields, pastures, fallows, and second-growth vegetation
are embedded in the forest matrix. These disturbed areas are
concentrated at the bottom of the valleys and range from 1-9
5


56
First, individuals growing at Dl (0-10 m) produced more
fruits than those growing at other distances. Support for
this comes from the fact that distance did not have an effect
on the total number of fruiting individuals (TI) (Table 4-2)
and the observation that some species that were heavily
represented at Dl (e.g., Clidemia sp. 1 and Palicourea
aibbosa) produced larger crops here than at the other
distances. Recall that TI eliminates the variability
associated with crop size because it considers only the
number of fruiting individuals. Second, palm fruits made a
disproportionate contribution to overall fruit production at
Dl. Support for this comes from the fact that distance did
not have an effect on total number of fruits excluding the
Arecaeae (TF-A) (Table 4-2) Third, some species found only
at Dl (0-10 m from the forest edge), including Marcaravia
eichleriana. Marcqraviastrum subssesilis. HidPhia
pseudoradula. M. theaezans. p¡?ammiia ferryiginga, Sghgfflgra
lasiocrvne. Phytolacca rivinoides. produced large fruit crops.
Even though the effect of distance alone was not significant
for total number of fruits excluding palms (TF-A), the trend
was similar to that for the previous two variables, i.e.,
sharp decrease from forest edge towards forest interior.
The effect of distance on fruit abundance was modified
by edge age and habitat, but neither factor alone had an
effect on any of the variables describing fruit abundance
(Table 4-2). The interaction between edge age and distance
from the forest edge was significant for total number of


CHAPTER 7
CONCLUSIONS
Studies designed to evaluate "edge effects" (sensu
Harris 1984) have emphasized the maximum distance at which
changes induced by edge creation are apparent within stands
(e.g., Williams-Linera 1990, Blanchard 1992, Chen et al.
1992). Variation among studies is enormous in terms of
"depth" of edge effects due to variation in edge features
(e.g., Kroodsma 1984, Quintela 1986, Noss 1991, DeGraaf 1992)
and to the fact that abiotic factors and organisms show
different responses to the creation of edges.
I have avoided in my work the use of the term "edge
effects" (Harris 1984). This is partially due to the
complexity of the responses that fruits and birds showed to
the presence of edges at my study site. Equally important
was the realization that "edges" are not isolated and fixed
elements in landscapes. They seem to regulate what happens
between the forest and the nearby disturbed area, and at the
same time they connect different elements of landscapes.
This is particularly true in the La Planada region, where
transient corn fields, pastures, and second growth areas of
various ages are embedded in a forest matrix. "Edge
dynamics" reflects the influence of edges on plant and animal
assemblages more accurately than "edge effects".
148


117
area has been transformed into pastures, cultivated fields,
coffee and tree plantations, and urban areas. A recent surge
in demand for opium derivatives has prompted forest clear-
cutting at higher altitudes to grow poppy (Papaver
somniferum: Cavelier and Etter, in press).
Fruaivorous Birds
I included all species reported to consume fruits and/or
seeds to any degree (Fitzpatrick 1980, Hilty and Brown 1986,
Isler and Isler 1987, Renjifo 1988, Ridgely and Tudor 1989,
Stiles and Skutch 1989, Fjelds and Krabe 1990, Velsquez
1992, Arango 1993, 1994, Ridgely and Tudor 1994, L.M. Renjifo
and C. Restrepo personal observations) and found at an
elevation >800 m. Thus, my data combine seed dispersers and
seed consumers.
Body mass data were obtained from published records
(Goodwin 1976, Isler and Isler 1987, Stiles and Skutch 1989,
Dunning 1993, del Hoyo et al. 1992, Arango 1993), museum
specimens (Coleccin de Ornitologa, Universidad del Valle,
Cali, Colombia), and my own field observations. For bird
species that I could not obtain mass measurements, I averaged
the available mass for congeners of the same length. I could
not estimate body mass for a small fraction (2%) of the
species and these were not included in the analyses.
To explore the relationship between body mass and
landscape pattern along the elevational gradient, I
classified birds into four groups based on their elevational


72
were independent of each other. Plants and droppings were
classified as (1) "edge" if the number of observations at D1
(0-10 m) and D2 (30-40 m) combined together was >0, (2)
"interior" if the number of observations at D3 (60-70 m) and
D4 (190-200 m) was >0, and (3) "edge=interior" if the number
of observations at D1 and D2 = D3 and D4. I found that the
proportion of droppings containing seeds of sparse species
was independent of their abundance in edge and interior (Test
for Independence, X2 = 2.4, df= 4, P = 0.6). Even though
sparse species were found more often at forest edge, their
seeds are potentially reaching forest interior.
DiSPUggipp
Distance from the edge towards the forest interior had a
major effect on fruit abundance but only close to the forest
edge. Nevertheless, the facts that (1) the effect of
distance was modified by edge age, habitat, and month of the
year, and (2) results for the four variables describing fruit
abundance differed, indicate that there are complex
interactions between edges and fruiting plants (summarized in
Table 4-6). I interpret these complex interactions in terms
of scales at which fruit abundance changes in relation to the
creation of edges.
At the scale defined by the pastures and the forest
matrix and by the length of the study (1 yr), new edges
generated spatial heterogeneity in fruit abundance within the
study area at D1 (0-10 m from forest edge). Fruit abundance,


140
any elevational zone and sites within the upper lowland zone.
Conversely, lumps representing the smallest birds contanined
the smallest proportion of species for the observed range of
body mass for sites within the upper montante zone. Lump
structure in terms of the persistance of lumps and proportion
of species falling within them was variable for sites in the
lower montane zone. This indicates that lumps representing
the smallest and/or largest species of frugivorous birds are
the ones most prone to disappear as the native forest is
fragmented and replaced by more simple ecosystems. Examples
of more simple ecosystems are the paramo, along the
elevational gradient, and managed ecosystems such as forestry
plantations and pastures, along the gradient generated by
human disturbance (Table 6-1).
I found that gaps and lumps persisted among sites
representing similar landscapes but that lumps fused, broke
down, or disappeared among sites representing different
landscapes. This indicates that gaps in the distribution of
body mass are being closed and opened probably as a result of
the replacement of species (elevational zones) and local
extinctions and invasions by species that were formerly
absent or rare in a site. It has been well documented that
species distribution changes over different spatial and
temporal scales (Terborgh 1971, Hooghiemstra 1984, Gentry
1992) and particularly relevant to this study are those
changes occurring in the distribution of species as a


32
by ants (7% 0.01%, mean SE) compared to intact forest (3%
0.01%). This trend was reversed at old edges.
Fruit Removal
Even though on average more fruits of Palicourea aibbosa
fruits were removed from the artificial infructescences at D2
(0.4 0.2, mean SE) than at the other distances (Dl, 0.3
0.09, D3, 0.3 0.9, and D4, 0.1 0.08), this difference was
not significant (ANOVA, F3,3 = 0.4, P = 0.7). The same was
true for habitat where on average more fruits were removed
from intact forest (0.3 0.1) than from gaps (0.2 0.06)
(ANOVA, Fi(52 = 0.6, P = 0.4).
Seed Predation
The number of seeds remaining in the trays averaged over
time did not differ among the four distances in Palicourea
qibbcsa but they did differ in Faramea affinis (ANOVA, F3,6 =
3.5, P = 0.09; Table 3-4, Fig. 3-2a). In E- affinis the
number of seeds remaining in the trays decreased from Dl to
D4, indicating higher removal rates at the interior. Habitat
alone did not have an effect on the number of seeds remaining
in the trays for either species (Table 3-4, Fig. 3-3a).
In £. affinis the distance effect was modified by
habitat and by week as shown by the significant distance x
habitat and distance x week interactions (ANOVA, F3(so = 2.3,
P = 0.08 and Fsi/io2 =2.0, P = 0.002, respectively, Table 3-
4). The number of E. affinis seeds remaining in the trays


150
separate, different elements of the landscape in the La
Planada region.
I argue that a "lump" analysis (Chapter 6) can help
understand the role of edges in landscapes. Edges when
viewed on a scale larger than that of a forest stand and the
nearby disturbed area show a tremendous variability. A
single measure of landscape complexity as proposed in Chapter
(6), which combines a horizontal and vertical component of
landscapes, can summarize edge features in a given area. By
trying to recognize patterns at large scales, such as those
defined by the impact of humans on landscapes, we might be
able to find a relationship between landscape pattern,
processes affecting landscape pattern, and the organisms that
live in them. This might be particularly true for the study
of fruit-frugivore interactions, since one outcome of such
interactions is seed dispersal. Landscape pattern is not
only changed as a consequence of disturbance; it is also
changed in fundamental ways by the process of seed dispersal.
In areas that are highly diverse such an approach might be
the best to understand the magnitude of the impact that
humans have on the biota.


109
predation and parasitism of adults (Wiens et al. 1985, Loye
and Carroll 1995). In addition, edges can directly influence
the distribution of organisms by affecting their
physiological condition (Wiens et al. 1985), probability of
establishment in a given area depending on home range size
(Kuitunen and Makirn 1993), and dispersal abilities.
In tropical areas there has been little effort to
elucidate factors underlying changes in the distribution of
animals across edges (Malcolm 1991). Some studies using
artificial nests have shown that nest predation increases
towards forest edges where second growth vegetation abuts
undisturbed forest (Gibbs 1991, Burkey 1993). However, nest
predation rates did not change across pasture-forest edges at
three other sites, including La Planada (Arango 1991,
Laurance 1993, C. Restrepo and C. Samper, unpublished data).
In an Amazonian site, Malcolm (1991) found that the
distribution of small mammals changed between "interior" and
"edge" in continuous tracts of forest. These changes
mirrored those in forest structure and abundance of insects,
and he concluded that changes in resource levels could
explain changes in small-mammal assemblages in the Amazon
across pasture-forest edges.
At La Planada the distribution of frugivores might be
partially influenced by the distribution of resources across
the pasture-forest edge. Fruit abundance (measured as total
number of fruits and as number of ripe fruits), decreased
significantly from the edge towards the forest interior


22
different ages (Table 2-1) and edge age is known to influence
the effect of distance on vegetation (e.g., Williams-Linera
1990), I chose them to conduct this work because they were
close enough to allow frequent monitoring of seeds and
fruits. At each of four distances from forest edge (0-10,
30-40, 60-70, and 190-200 m) (Fig. 2-3, Chapter 2) I mapped
the treefall gaps and randomly chose 4 of them. At each
distance I paired each treefall gap location with an intact
forest location.
Sged._pr.edation an In the seed predation and seed germination experiments I
placed an aluminum tray (15 x 7 cm) in each gap and interior
site. I punctured the trays to prevent water from
accumulating, filled them with soil, and positioned them
flush with ground level. I placed 10 seeds of £. cribbosa and
5 seeds of £. affinis in different trays. Seeds were
obtained from ripe fruits, and those showing damage by
insects were discarded. In total I used 1,920 seeds of £.
gibbosa and 960 of F. affinis. I placed 92 trays containing
seeds for each edge/species/experiment (32 trays) and
simultaneously ran the germination and predation experiments
for each species within each edge. The trays containing the
seeds sown to evaluate changes in germination rates were
covered with galvanized mesh (5x5 mm) to protect seeds from
vertebrates.
I checked trays on a weekly basis and counted the number
of seeds remaining and the number of seeds germinated, i.e.,


179
and Diversity. Sinauer Associates, Sunderland,
Massachusetts, USA.
Loye, J. and S. Carroll. 1995. Birds, bugs and blood: avian
parasitism and conservation. TREE 10:232-235.
Lozano, I. E. 1993. Diversidad y organizacin en gremios de
la comunidad de aves del sotobosque de bosque primario y
vegetacin secundaria. Pages 141-164, in G.I. Andrade,
editor. Carpanta Selva Nublada y Pramo. Fundacin
Natura, Colombia.
Lugo, A. E. and J. L. Frangi. 1993. Fruit-fall in the
Luquillo Experimental Forest, Puerto Rico. Biotropica
25:73-84 .
Lynam, A. J. Processes of local extinction of small mammals
in a recently fragmented tropical monsoon forest in
Thailand. in W.F. Laurance and R.O. Bierregaard, Jr.,
editors. Tropical Forest Remnants: Ecology, Management
and Conservation of Fragmented Communities, in press.
Lynch, J. F. 1989. Distribution of overwintering nearctic
migrants in the Yucatan Peninsula, I: General patterns
of occurrence. Condor 91:515-544.
MacDougall, A. and M. Reliman. 1992. The understory light
regime and patterns of tree seedlings in tropical
riparian forest patches. Journal of Biogeography
19:667-675.
Malcolm, J. R. 1991. The small mammals of Amazonian Forest
fragments: Pattern and process. Ph.D. Dissertation,
University of Florida, Gainesville, FL.
Malcolm, J. R. 1994. Edge effects in Central Amazonian
forest fragments. Ecology 75:2438-2445.
Manly, B. F. 1992. The Design and Analysis of Research
Studies. Cambridge University Press, Cambridge, UK.
Margalef, R. 1968. Perspectives in Ecological Theory. The
University of Chicago Press, Chicago, IL.
Marshall, C. and J. Grace, editors. 1992. Fruit and Seed
Production: Aspects of Development, environmental
physiology and ecology. Cambridge University Press,
Cambridge, UK.
Martin, T. E. 1985. Selection of second-growth woodlands by
frugivorous migrating birds in Panama: an effect of
fruit size and plant density? Journal of Tropical
Ecology 1:157-170.


125
complete data base on frugivorous montane birds. The sites
within a elevational zone were compared using a null
distribution generated from the database on frugivorous birds
for the corresponding zone.
How to interpret a lump analysis
The main results of the lump analysis are summarized in
two related figures, depicting the distribution of gaps and
lumps. The first illustrates the distribution of GRI-values
against body mass (Fig. 6-2b), while the second, which is
derived from the first, depicts the distribution of lumps and
gaps for the various data sets that are being compared
simultaneously (e.g., Fig. 6-3). In this paper I report only
the latter. Lump structure for a given data set can be
described in terms of the number and size of the lumps and
the proportion of species falling within specific lumps.
Lump structure for multiple data sets can be described in
terms of the correspondence between the position of lumps and
gaps. Presently, P. Marples is developing a procedure that
will allow quantitative testing of the lump and gap
correspondence. In this paper the comparison will be
qualitative.
The size of the lumps represents the ranges of body mass
that do not exhibit discontinuities at a given alpha level.
The number of species that fall within a lump are represented
as the proportion of the total number of species of the
corresponding data set. In the figures depicting the
distribution of lumps the proportions are represented by


175
periodicities, and morphologies. in B. H. Walker and W.
L. Steffen, editors. Global Change and Terrestrial
Ecosystems. Cambridge University Press, Cambridge,
Massachusetts, USA, in press.
Holloway, J. D., A. H. Kirk-Spriggs and C. von Khen. 1992.
The response of some rain forest insect groups to
logging and conversion to plantation. Philosophical
Transactions of the Royal Society of London B
335:425:436.
Hooghiemstra, H. 1984. Vegetational and climatic history of
the high plain of Bogota, Colombia: A continuous record
of the last 3.5 million years. Dissertationes Botanicae
79. Cramer, Vaduz.
Hubbell, S. G. and R. B. Foster. 1987. Biology, chance, and
history and the structure of tropical rain forest tree
communities. Pages 314-329, in J. Diamond and T. J.
Case, editors. Community Ecology. Harper and Row, New
York, NY.
Hughes, L., M. Dunlop, K. French, M. R. Leishman, B. Rice, L.
Rodgerson, and M. Westoby. Predicting dispersal
spectra: a minimal set of hypotheses based on plant
attributes. Journal of Ecology 82:933-950.
Irving, E. M. 1975. Structural evolution of the
northernmost Andes, Colombia. U.S. Geological Survey
Professional Paper 846, Washington, DC, USA.
Isler, M. L. and P. R. Isler. 1987. The Tanagers: Natural
History, Distribution and Identification. Smithsonian
Institution Press, Washington, DC, USA.
Jackson, J. E. and J. W. Palmer. 1977. Effects of shade on
the growth and cropping of apple trees. II. Effects on
components of yield. Journal of Horticultural Science
52:253-266.
Janzen, D. H. 1983. No park is an island: Increase in
interference from outside as park size decreases. Oikos
41:402-410.
Johnels, S. A. and T. C. Cuadros. 1986. Species composition
and abundance of bird fauna in a disturbed forest in the
Central Andes of Colombia. Hornero 12:235-241.
Johns, A. D. 1992. Vertebrate responses to selective
logging: implications for the design of logging systems.
Philosophical Transactions of the Royal Society of
London B 335:437-442.


98
MONTH
Figure 5-4. Variation in the distribution of understory birds
at La Planada in relation to distance from forest edge and month.
Points represent means and bars standard errors.


105
contribution of extremely sparse and very sparse species at
D1 was disproportionally high. There was also a significant
association between species abundance and distance from the
edge for frugivores and nectarivores (x2 = 32.8, df = 12, P =
0.001 and X2 = 27.0, df = 9, P = 0.01, respectively; Table
5-4). An examination of the residuals for frugivores showed
that (1) extremely sparse species were captured more
frequently than expected at D1 than at the other three
distances, (2) very sparse and abundant species were captured
more frequently than expected at D3 than at the other
distances, and (3) very abundant species were captured more
frequently than expected at D4 than at the other distances.
Among nectarivores the residuals showed that (1) very sparse
and sparse species were captured more frequently than
expected at D1 than to the other distances and (2) very
abundant species were captured more frequently than expected
at D4 than to the other distances.
Discussion
The distribution of understory birds at La Planada
varied across pasture-forest edges in complex ways, not
always reflecting changes due to the presence of edges. This
complexity is demonstrated by (1) differences among the
response variables and (2) significant interactions between
distance from edge, edge age, and month. Capture rates for
all birds, frugivores, and insectivores changed across the
pasture-forest edge, but the same was not true for


114
1985, Gentry 1988, Renjifo et al., in press), offering an
opportunity to relate the process of seed dispersal to
landscape pattern. I focus on mountains because natural and
human disturbances have generated complex landscapes over
small areas (Haslett 1994), offering an ideal opportunity to
assess how changes in landscape pattern influence animal
assemblages.
I make four comparisons representing two different
scales of inquiry and address how the distribution of body
mass, i.e., lump structure, in frugivorous birds changes from
areas covered mostly by forest to areas covered by open
vegetation. The first scale is defined by elevational zones
within the mountains of Colombia, the second by sites within
elevational zones that have been differently affected by
human activities. I use body mass as an attribute that
reflects information not only on life-history traits, such as
dispersal (Laurance 1991, Lawton et al. 1994, Brown 1995,
Gaston and Blackburn 1995), but also on foraging behavior,
such as size of seeds being dispersed (Moermond and Denslow
1985).
MethQd?
Study Area
My study focused on the Andes of Colombia, South
America. This system consists of three mountain ranges of
different geological origin, each running in a S-N direction


70
Table 4-5. Distribution of fruiting individuals in the
understory of the Reserva Natural La Planada across pasture-
forest edge based on species abundance. Numbers represent
the number of fruiting individuals.
Number of Distance
Species D1 D2 D3 D4
Extremely sparse
(1 individual)
33 13
Very sparse
(2-5 individuals)
40 64 21 28 25
Sparse
(6-20 individuals)
50 167 144 138 120
Abundant
(21-50 individuals)
13 94 99 115 137
Very abundant
(>51 individuals)
12 475 441 384 327


118
ranges: upper lowland (from ca. 800 m to 1,500 m) lower
montane (from ca. 1,500 m to 2,400 m) upper montane (from
ca. 2,400 m to 3,400 m) and pramo (from ca. 3,400 to 4,800
m) species (Table 6-1). Note that some species fell in more
than one elevation zone and were entered into the analyses
two or more times. Elevational zones follow Chapman (1917).
Elevational ranges were taken from Hilty and Brown (1986).
To explore the relationship between body mass and
landscape pattern along a gradient of land-use, I obtained
bird inventories for 18 sites (Table 6-1). All inventories
were conducted by experienced ornithologists over periods of
>1 yr and included visual, auditory, and mist netting
observations. I believe these lists represent unbiased and
fairly complete inventories, so additions would have little
influence on the results. The 18 sites were grouped
according to elevational zone and type of land use. Within
each comparison I arranged sites from those covered mostly by
forest to those covered by open vegetation (Table 6-1).
This work relies on two assumptions. First, there is a
common pool of species for elevational zones and sites, but
historical, geographical, and climatic events, and more
recently human activities, have determined the set of species
found today at any one site. Second, within elevational
zones the less disturbed sites represent the conditions that
existed at the other sites prior to human intervention.


174
Boundaries: Consequences for Biotic Diversity and
Ecological Flows. Springer-Verlag, New York, NY.
Hansson, L. 1983. Bird numbers across edges between mature
conifer forest and clear cuts in Central Sweden. Ornis
Scandinavica 14:97-103.
Harper, J. L. 1994. Population Biology of Plants. Academic
Press, San Diego, Californa, USA.
Harris, G. P. 1980. Temporal and spatial scales in
phytoplankton ecology. Mechanisms, methods, models, and
management. Canadian Journal of Fisheries and
Aquaculture 37:877-900.
Harris, L. D. 1984. The Fragmented Forest. University of
Chicago Press, Chicago, IL.
Haslett, J. A. 1994. Community structure and the fractal
dimensions of mountain habitats. Journal of Theoretical
Biology 167:407-411.
Heideman, P. D. 1989. Temporal and spatial variation in the
phenology of flowering and fruiting in a tropical
rainforest. Journal of Ecology 77:1059-1079.
Henderson, A., S. P. Churchill and J. Luteyn. 1991.
Neotropical plant diversity. Nature 351:21-22.
Hilty, S. L. 1980. Flowering and fruiting periodicity in a
premontane rain forest in Pacific Colombia. Biotropica
12:292-306.
Hilty, S. L. and W. L. Brown. 1986. A Guide to the Birds of
Colombia. Princeton University Press, Princeton, NJ.
Holdridge, L. R. 1967. Life Zone Ecology. Tropical Science
Center, San Jos, Costa Rica.
Holdridge, L. R., W. C. Grenke, W. H. Hateway, T. Liang, and
J. A. Tosi, Jr. 1971. Forest Environments in Tropical
Life Zones, A Pilot Study. Pergamon Press, Oxford, UK.
Holling, C. S. 1973. Resilience and stability of ecological
systems. Annual Review of Ecology and Systematics 4:1-
23 .
Holling, C. S. 1992. Cross-scale morphology, geometry, and
dynamics of ecosystems. Ecological Monographs 62:447-
502.
Holling, C. S., G. Peterson, P. Marples, J. Sendzimir, K.
Redford, L. Gunderson, and D. Lambert. 1995. Self
organization in ecosystems: lump geometries,


108
higher at the "edge" (50 m from edge), than at the forest
interior (>50 m from edge) and these differences were more
pronounced in the summer than in the winter (Hansson 1983).
In central Florida, overall bird density was higher near the
forest edge (50 m from edge) than farther away, and this
difference was greater during winter than at other times of
the year at east-facing edges (Noss 1991). Noss also found
variability among years in the response of birds to edges.
At La Planada, capture rates for all birds and for frugivores
were greater at D4 (190-200 m) than at D1 (0-10 m) during dry
months. Capture rates for insectivores were higher at D4
than at D1 and they did not differ between the wet and dry-
season. For nectarivores capture rates were higher at D1
than at D4 only during the dry season. The within-year
variability in bird abundance across forest edges suggests
that edges are dynamic. In addition, this variability is
suggestive of possible mechanisms underlying the observed
patterns.
Factors Influencing the Distribution of Birds across Edges
Environmental factors and intrinsic features of
organisms may influence their distribution across edges
(Wiens 1992). Most of these factors have been examined in
temperate zones, especially in areas where forest fragments
are embedded in an agricultural matrix. They include nest
parasitism and egg predation (Gates and Gysel 1978, Wilcove
et al. 1986, Andrn and Algestam 1988, Moller 1989), and


187
Science of Scarcity and Diversity. Sinauer, Sunderland,
Massachusetts, USA.
Williams-Linera, G. 1990. Vegetation structure and
environment of forest edges in Panama. Journal of
Ecology 78:356-373.
Winer, B. J., D. R. Brown and K. M. Michaels. 1991.
Statistical Principles in Experimental Design. McGraw-
Hill, New York, NY.
Wong, M. 1986. Trophic organization of understory birds in a
Malasyan dipterocarp forest. Auk 103:100-116.
Young, K. R. 1993. Woody and scandent plants on the edges of
an Andean timberline. Bulletin of the Torrey Botanical
Club 120:1-18.
Zolman, J. F. 1993. Biostatistics: Experimental Design and
Statistical Inference. Oxford University Press, New
York, NY.


17
The Species
Palicourea qibbosa Dwyer and Far ame, a f finis belong to
the Rubiaceae, one of the most speciose and common families
of neotropical montane forests (Taylor 1989, Gentry 1992a.) .
Palicourea aibbosa shrubs reach 4 m and are found at middle
elevations from Panama to Ecuador (Dwyer 1980; C. Taylor,
personal communication), growing in second growth and mature
forest (Arias 1993). Faramea affinis treelets reach 9 m and
grow in old second growth and mature forests. In a 0.1 ha
plot at my study site, £. qibbosa and £. affinis were the
most common species (dbh > 2.5 cm) in the understory (A.
Gentry, unpublished data).
Palicourea qibbosa exhibits three flowering periods per
year. Its yellow flowers are visited mostly by hummingbirds,
including Qgrg^tV? underwoodii. Aglaiocercus coelestis. and
Haploohaedia luqens (Arias 1993). Fruits of £. qibbosa are
dark blue to purple, 7 mm long, and are presented in terminal
yellow, erect infructescences containing up to 50 fruits.
They contain 1-2 seeds, 5.0 x 4.9 mm. Palicourea qibbosa
seeds are dispersed by birds, including Mvadestes ralloides.
PiprgQla riefferj, AUepete? brunneinucha. Masius
chrvsopterus. and Tanqara arthus (C. Restrepo and N. Gomez,
unpublished data).
Faramea affinis exhibits two flowering periods per year.
Its tubular, purple flowers are visited by hummingbirds,
including Coeliqena wilsoni (Samper 1992). Fruits are blue


168
fruits for birds in tropical habitats. Studies in Avian
Biology 13:73-79.
Blanchard, J. D. 1992. Light, vegetation structure, and
fruit production on edges of clearcut sand pine scrub in
Ocala National Forest, Florida. M.Sc. Thesis,
University of Florida, Gainesville, FL.
Brittingahm, M. C. and S. A. Temple. 1983. Have cowbirds
caused forest songbirds to decline? BioScience 33:31-35.
Brokaw, N. V. L. 1982. The definition of treefall gap and
its effect on measures of forest dynamics. Biotropica
11:158-160.
Brown, J. H. 1995. Macroecology. The University of Chicago
Press, Chicago, IL.
Brown, N. 1993. The implications of climate and gap
microclimate for seedling growth conditions in a Bornean
lowland rain forest. Journal of Tropical Ecology 9:133-
168.
Burkey, T. V. 1993. Edge effects in seed and egg predation
at two neotropical rainforest sites. Biological
Conservation 66:139-143.
Bush, B. M. 1994. Amazonian speciation: a necessary complex
model. Journal of Biogeography 21:5-17.
Cavelier, J. and A. Etter. Deforestation of mountain forest
in Colombia as a result of illegal plantation of opium
(Papaver somniferum) in Neotropical Mountain Forest.
Proceedings of the Neotropical Mountain Forest
Biodiverstiy and Conservation Symposium, New York
Botanical Garden, New York, August 1992, in press.
Chaikiattiyos, S., C. M. Menzel, and T. S. Rasmussen. 1994.
Floral induction in tropical fruit trees: Effects of
temperature and water supply. Journal of Horticultural
Science 69:397-415.
Chapman, F. M. 1917. The distribution of bird life in
Colombia. Bulletin of the American Museum of Natural
History 33:167-192.
Chason, J. F., D. D. Baldocchi and M. A. Huston. 1991. A
comparison of direct and indirect methods for estimating
forest canopy leaf area. Agricultural and Forest
Meteorology 57:107-128.


34
Seedling Growth Rate
Distance did not influence relative growth rates in
Palicourea gibbPSfr and Fflrgmea ftffjni? (Table 3-5, Fig. 3-
2c). Habitat, however, had a major effect on both species
(Table 3-5). Seedlings showed greater growth rates (ANOVA,
£. gibbp.Sfl, Fi(399 = 38.7, P = 0.0001 and £. gtffipjg, Fi(422 =
8.1, P = 0.005; Fig. 3-3c) in gaps than in intact forest.
Overall seedlings of P. gibbosa grew faster (5.0 0.17
mm/month, mean SE, n = 411) than those of F. affinis (2.4
0.08 mm/month, mean SE, n = 434).
Leaf Production
As with growth rate, habitat and not distance from forest
edge had a significant effect on leaf production (ANOVA, £.
ai&kosa, Fi,408 = 17.1, P = O.OOOl and £. affinis. Fi>424 =
3.4, P = 0.06; Table 3-5, Figs. 3-2d, 3-3d). Seedlings of
both species produced more leaves per month in gaps than in
intact forest (Fig. 3-3d). Overall, leaf production was
greater in seedlings of Palicourea gibbosa (0.8 0.01 pairs
of leaves/month, mean SE, n = 420) than in seedlings of
Faramea affinis (0.38 0.008 pairs of leaves/month).
Discussion
Of the stages of a plant's life cycle, I examined
pollination, fruit set, seed dispersal, seed predation, seed
germination, and seedling growth. An important stage


CR 753, CR 627
Solanum sp.5
CR 483, CR 722, CR 794, CR 785
Solanum sp.7
CR 537, CR 735, CR 532?
Solanum sp.8
CR 486, CR 750, CR 762
Zingiberaceae
New Edges
Renealmia aff. concinna Standley sp. nov.
CR 638
Acanthaceae
Araceae
Araliaceae
Mendocia orbicularis Turrill
ftnthubium cf. chamberlainii Masters
Anthurium sp. nov.
CR 556
Ehllodepdipn oligospermum Engl
Stenpspermatium lonqipetiolatum Engl.
Stenosoermatium longisoadix Croat
Stenosoermatium soarrei Croat
Xanthosoma subandinum Schott
Scheffiera cf. violcea Cuatr.
Campanulaceae
.CentPQPQgon aff. solanifolius Benth
Cyclanthaceae
Agplundia stenoohvlla
Dicranopygium sp.
F A
G-LD VA
F S
F VS
F VS
F ES
G ES
LD VS
LD VS BD
LD VS BD
LD ES
F ES
F ES
LD ES
F VS
F A BD
157


162
APPENDIX B
BIRDS CAPTURED IN THE UNDERSTORY
OF THE RESERVA NATURAL LA PLANADA (JUNE 1992-AUGUST 1993)
Birds mist netted
in the understory of the Reserva Natural La Planada.
Abundance (AB):
extremely sparse (ES), very sparse (VS), sparse
(S), abundant (A),
and very abundant (VS)
Feeding guild (FG)
: carnivores (A), frugivores
(F), insectivores
(I) ,
and nectarivores
(N) .
FAMILY
SPECIES
AB
FG
Mass
(g)
Accipitridae
Accioiter collaris
VS
A
166.3
Columbidae
Colwnba cavannensis
VS
F
Strigidae
Glaucidium iardinii
ES
A
75.0
Otus columbianus
VS
A
165.0
Trochilidae
Adelomvia melanoqenvs
ES
N
5.22
Aglaiocercus coelestis
VA
N
fenasilia franciae
ES
N
5.02
Bdissonneava iardini
VS
N
9.8
Chlprostilbon mellisuqus
ES
N
4.0
CQeliqena torcruata
VS
N
8.2
Coeliqena wilsoni
VA
N
7.1
Colibri thalassinus
ES
N
Dorvfera ludoviciae
ES
N
6.0
Eutoxeres aauila
ES
N
9.5
Haplophaedia lugens
VA
N
6.0


41
affected by habitat (gaps, edge of gaps, and intact forest),
(2) seed germination rates were faster in gaps (mean = 174
days, n = 148) than along edges of treefall gaps (177, n =
132) and intact forest (187, n = 135), and (3) seedling
establishment (i.e., the stage at which seedlings become
independent from food reserves contained in the seeds) was
not affected by habitat. Samper's work and mine show that
the seed and seedling stages in £. affinis are affected
differently by treefall gaps and edges resulting from human
activities.
Once seeds of £. aibbosa and £. affinis arrive and
germinate at any distance from forest edge, treefall gaps
seem to have a major influence on these two species by
increasing growth and leaf production rates in seedlings.
This is not in accordance with results obtained in the
Amazon, where relative growth rates of seedlings was greater
up to 10 m from forest edge towards forest interior (Seizer
1992).
My study shows that edges can influence recruitment
rates of P. qibbosa and F. affinis through their effect on
seed predation and seed germination but not on pollination
and the growth of seedlings. On the other hand, treefall
gaps influence recruitment rates through their effect on
seedling growth.


EDGES
FRUITS, FRUGIVORES, AND SEED DISPERSAL
IN A NEOTROPICAL MONTANE FOREST
By
CARLA RESTREPO
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR IN PHILOSOPHY
UNIVERSITY OF FLORIDA
1995


104
Table 5-4. Distribution of understory birds across the
pasture forest-edge. Birds were classified according to
abundance. Numbers represent number of captures. Carnivores
were excluded because of small number of captures.
Distance
D1 D2 D3 D4
Extremely sparse
(1 capture)
All Birds
12
3
6
4
Frugivores
5
2
0
0
Insectivores
3
1
3
1
Nectarivores
3
0
3
2
Very sparse
(2-5 captures)
All Birds
31
11
12
14
Frugivores
10
3
8
2
Insectivores
9
3
5
8
Nectarivores
11
0
1
4
Sparse
(6-20 captures) All Birds
41
32
24
43
Frugivores
9
13
10
23
Insectivores
17
13
9
19
Nectarivores
15
6
4
1
Abundant
(21-50 captures)
All Birds
70
39
46
75
Frugivores
36
21
28
33
Insectivores
34
18
18
42
Nectarivores
Very abundant
(>.51 captures)
All Birds
423
314
342
557
Frugivores
124
81
66
179
Insectivores
126
122
143
211
Nectarivores
173
111
133
167


FRUIT ABUNDANCE
(mean counts / 50 m 2
66
MONTH
Figure 4-7. Variation in fruit abundance at the Reserva
Natural La Planada in relation to distance from forest
edge and month. Points represent means and bars standard
errors.


APPENDIX A
PLANT SPECIES FRUITING IN THE UNDERSTORY
OF THE RESERVA NATURAL LA PLANADA


171
Ellison, A. M., J. S. Denslow, B. A. Loiselle, and D. Brens.
1993. Seed and seedling ecology of neotropaical
Melastomataceae. Ecology 74:1733-1749.
Escobar, F. 1994. Excremento, coprfagos y deforestacin en
bosques de montaa al suroccidente de Colombia. Tesis,
Universidad del Valle, Cali, Colombia.
Espinal, L. S., J. Tos Jr., E. Montenegro, G. Toro and D.
Daz. 1977. Zonas de vida o formaciones vegetales de
Colombia. Instituto Geogrfico Agustn Codazzi, Bogot,
Colombia.
Esseen, P. A. 1994. Tree mortality patterns after
experimental fragmentation of an old-growth conifer
forest. Biological Conservation 68:19-29.
Estrada, A., R. Coates-Estrada, D. Meritt, S. Montiel, and D.
Curiel. 1993. Patterns of frugivore species richness
and abundance in forest islands and in agricultural
habitats at Los Tuxtlas, Mexico. Vegetatio 107/108:245-
257.
Ewel, J. and A. Madriz. 1968. Zonas de vida de Venezuela.
Memoria Explicativa sobre el Mapa Ecolgico. Editorial
Sucre, Venezuela.
Feinsinger, P., J. H. Beach, Y. B. Linhart, W. H. Busby, and
K. G. Murray. 1987. Disturbance, pollinator
predictability, and pollination success among Costa
Rican Cloud forest plants. Ecology 68:1294-1305.
Feinsinger, P., H. M. Tiebout III, and B. E. Young. 1991. Do
tropical bird-pollinated plants exhibit densigy-
dependent interactions? Field experiments. Ecology
72:1953-1963.
Fitzpatrick, J. W. 1980. Foraging behavior of neotropical
tyrant flycatchers. Condor 82:43-57.
Fjelds, J. and N. Krabe. 1990. Birds of the high Andes.
Zoological Museum, University of Copenhagen and Apollo
Books, Svendborg, Denmark.
Fleming, T. H. 1988. The Short-tailed Fruit Bat: A Study in
Plant-Animal Interactions. Wildlife Behavior and
Ecology Series, Chicago Press, Chicago, IL.
Gagnon, J., K. A. Haycock, J. R. Roth, D. S. Feldman, Jr., W.
F. Finzer, R. Hofman, and J. Simpson. 1989. SuperANOVA.
Accesible General Linear Modeling. Abacus Concepts,
Berkely, CA.


74
Table 4-6. Summary of results of ANOVAs on fruit abundance
across the pasture-forest edge for the different response
variables. Fruit abundance expressed as total number of
fruits (TF), total number of ripe fruits (RF), total number
of fruits excluding the Arecaceae (TF-A), and total number of
fruiting individuals (TI). Significance at 10% (*), 5% (**),
1% (***).
TF RF
TF-A TI
Large Fruit
Small Fruit
Crops
Crops
ANOVA terms
Distance
*
Distance x Age

Distance x Habitat
*
Distance x Month

Age x Month

Habitat x Month



169
Chen, J., J. F. Franklin, and T. A. Spies. 1992. Vegetation
responses to edge environments in Old-growth Douglas-Fir
forests. Ecological Applications 2:387-396.
Clark, D. A. and D. B. Clark. 1987. Temporal and
environmental patterns of reproduction in Zamia
skinneri. a tropical rain forest cycad. Journal of
Ecology 75:135-149.
Clark, W. C., D. D. Jones, and C. S. Holling. 1979. Lessons
for ecological policy design: A case study of ecosystems
management. Ecological Modelling 7:1-53.
Compton, S. G., S. J. Ross, and I. W. R. Thornton. 1994.
Pollinator limitation of fig tree reproduction on the
island of Anak Krakatau (Indonesia). Biotropica 26:180-
186.
Connell, J. H. 1978. Diversity in tropical rain forests and
coral reefs. Science 199:1302-1309.
Corredor, G. L. 1989. Estudio comparataivo entre la avifauna
de un bosque natural y un cafetal tradicional en el
Quindio. Trabajo de Grado, Departamento de Biologa,
Universidad del Valle, Cali, Colombia.
Correll, D. L. 1991. Human impact on the functioning of
landscape boundaries. Pages 90-109, in M. J. Holland,
P. G. Risser, and R. J. Naiman, editors. Ecotones: The
Role of Landscape Boundaries in the Management and
Restoration of Changing Environments. Chapman Hall, New
York, USA.
Crist, T. 0. and J. A. MacMahon. 1991. Individual foraging
components of harvester ants: movement patterns and seed
patch fidelity. Insect Socieux 38:379-396.
Crist, T. O., D. S. Guertin, J. A. Wiens, and B. T. Milne.
1992. Animal movement in heterogeneous landscapes: an
experiment with Eleodes beetles in shortgrass prairie.
Functional Ecology 6:536-544.
Cronquist, A. 1981. An Integrated System of Classification
of Flowering Plants. Columbia University Press, New
York, NY.
Cuadros, T. 1988. Aspectos ecolgicos de la comunidad de aves
en un bosque nativo en la Cordillera central en
Antioquia, Colombia. Hornero 13:8-20.
Cuatrecasas, J. 1958. Aspectos de la vegetacin natural.
Revista de la Academia Colombiana de Ciencias Exactas
Fsicas y Naturales 10:221-264.


CHAPTER 5
EDGES AND UNDERSTORY BIRDS IN
A NEOTROPICAL MONTANE FOREST
Introduction
Natural disturbances play a major role in maintaining
high levels of diversity in tropical ecosystems (e.g.f
Connell 1978, Salo et al. 1986, Bush 1994, Gentry 1986; but
see Haffer 1969, Hubbell and Foster 1987). At regional
scales, one result of disturbance is the creation of
ecotones, which have been postulated to favor speciation
processes (Bush 1994) and high species richness (Terborgh
1977, Bush 1994). Human disturbances, on the other hand,
have resulted in a variety of land uses. Focus has now
shifted towards understanding how human disturbances affect
distributions of species (e.g., Kattan 1992) and how this
might impact ecosystem processes (Vitousek 1990, Kruess and
Tschnarntke 1994, Tilman and Downing 1994).
One consequence of human activities on landscapes is the
creation of sharp edges bounding disturbed areas, such as
pastures, logged forest stands, and agricultural fields.
These edges, whether found in little or highly modified
landscapes, may influence the movement of organisms between
the undisturbed and disturbed areas (Wiens et al. 1985, Wiens
1992) It is very likely that edges, by influencing the
82


Table 6-1. continued
References: (1) Acevedo (1987), (2) Andrade (1993), (3) Arango (1994), (4) Corredor (1989), (5)
Cuadros (1988), (6) Gentry (1992a), (7) Hilty (1980), (8) Johnels and Cuadros (1986), (9) Rattan et al.
(1984), (10) Rattan et al. (1994), (11) Mondragn (1989), (12) Munves (1975), (13) Naranjo (1994), (14)
Orejuela et al. (1979), (15) Orejuela and Cantillo (1990), (16) Rangel (1994), (17) Rangel and Espejo (1989),
(18) Renjifo (1988), (19) Restrepo (1990), (20) Ridgely and Gaulin (1980), (21) Rosas (1986), (22) Uribe (1986),
(23) Velasquez (1992), (24) Velez (1987), (25) N. Arango, personal communication, (26) G. Rattan, personal
communication, (27) C. Restrepo, personal observation, (28) L. M. Renjifo, personal observation, (29)
S. Hilty, unpublished list, (30) G. Rattan, H. Alvarez,and M. Giraldo,unpublished list, (31) H. Alvarez,
unpublished list, (32) G. Rattan, H. Alvarez, and E. Buttkus, unpublished list.
-^Life zone: Tropical premontane dry forest (TP-df), Tropical premontane moist forest (TP-mf),
Tropical premontane wet forest (TP-wf), Tropical premontane rain forest (TP-rf)
Tropical lower montane wet forest (TLM-wf), tropical lower montane moist forest (TLM-mf),
tropical montane wet forest (TM-wf).
2 Land Use: (F) native forest, (SG) second growth, (P) pasture, (CT) shaded coffee plantation,
(TPE) tree plantations with exotic species, (TPN) tree plantations with native species, (U)
weekend cottages. a selectively logged native forest, b native forest dominated by the Giant
Bamboo, Bgmbus^ guadua c plantations with Fraxinus sinensis d plantations with Alnus acuminata.
e Plantations with Pjpus Bakula and Cuoressus lusitanicus.
120


83
movement of animals, might indirectly affect ecological
processes mediated through plant-animal interactions such as
pollination and seed dispersal. In tropical systems many
organisms are involved in plant-animal interactions. For
the most part, work done in the tropics has focused on how
edges affect animal distributions (Quintela 1986, Laurance
1990, Malcolm 1994). Less emphasis has been given to how
edges influence the distribution of animals mediating
ecological processes.
The extent to which edges can affect the distribution of
organisms varies with edge age (e.g., Williams-Linera 1990),
and land use (e.g., DeGraaf 1992) may determine the degree to
which edges can affect organisms. Equally important is the
variation among organisms in their response to edges (e.g.,
Kroodsma 1984, Noss 1991). Such variation can be used to
tease apart the mechanisms underlying such responses.
Possible mechanisms include changes in the resource base
(Malcolm 1991), parasites and predators (e.g., Gates and
Gysel 1978, Brittingham and Temple 1983, Loye and Carroll
1995), physiological condition of organisms (Wiens et al.
1985), dispersal, and home range size (Kuitunen and Makirn
1993) .
Here I report on how edges influence the distribution of
understory birds in a neotropical montane forest. I looked
at the effects of distance from the edge towards the forest
interior and time since edge creation on birds classified by
feeding guilds. I concentrated on frugivores and


Figure 6-5. Lump structure of Colombian frugivorous birds from sites covered mostly by
forest (bottom) to sites highly transformed by human activities (top) within the lower
montane zone. La Planada (LP), Ucumar Bajo (UB), San Antonio (SA), Miraflores (MI),
Merenberg (ME), Rio Blanco (RB), Rancho Grande (RG), Munchique (MU), and Piedras Blancas
(PB). Each box represents a lump and the space between the boxes represent gaps in the
distribution of body mass. The different shades indicate the proportion of species falling
within lumps:(1) 0-5, (2) 5-10, (3) 10-20, (4) 20-30, (5)30-45, (6) 45-60, and (7) 60-100 %
of species. Vertical lines represent 0-5% of species. Numbers on the right side represent
number of species for the corresponding data set.
1 2 3 4 5 6 7


48
I included unripe fruits because they constitute a food
resource for frugivorous insects. I excluded the Araceae
from variables 2-3 because it was difficult to estimate fruit
numbers for each infructescence. I excluded the Arecaceae
from variable 4 because their high productivity and prolonged
fruiting season could mask patterns of fruit production among
shrubs producing fewer fruits and fruiting over shorter
periods of time.
To explore fruit abundance responses to edges at the
species level I looked at the number of individuals bearing
unripe and/or ripe fruits. For each species I pooled this
information for all subquadrats and months to obtain a single
value for each edge age and distance.
I collected most plant species and deposited voucher
specimens at Botany Department Herbarium, Arizona State
University (ASU), Herbario Nacional de Colombia (COL), Botany
Department Herbarium, Field Museum of Natural History,
Chicaco (F), Herbario de la Universidad de Antioquia (HA),
Kew Botanical Garden (K), Missouri Botanical Garden (MO), New
York Botanical Garden (NY), Herbario Universidad de Nario
(PSO), Utrecht Herbarium (U), Smithsonian Institution (US),
and Department of Botany Herbarium, University of Wisconsin
(WIS). Family names follow Cronquist (1981).
Seed Movement
To evaluate seed movement across edges I counted and
identified seeds contained in bird droppings retrieved from


CHAPTER 4
UNDERSTORY FRUIT ABUNDANCE IN A NEOTROPICAL MONTANE FOREST:
THE INFLUENCE OF EDGES AND TREEFALL GAPS
Introduction
Fruit abundance can be influenced by disturbances
occurring at various scales. In general, fruit abundance
increases in small, natural disturbances, such as treefall
gaps (Blake and Hoppes 1986; Levey 1988&,k), in large,
natural disturbances, such as patches affected by hurricanes
(Walker and Neris 1993) or fire (Fleming 1988), and in large,
human-disturbed areas, such as abandoned fields and pastures
(Martin 1985; Levey 1988a,Blake and Loiselle 1991; Lugo
and Frangi 1993; but see Wong 1986). In treefall gaps high
fruit production is the result of an increase in the number
of fruits produced by individuals growing in the disturbed
area compared to conspecifics growing in intact forest and to
an increase in the number of fruiting individuals (Piero and
Sarukhan 1982; Clark and Clark 1987; Levey 1990). In large
disturbed areas, high fruit production has been related to
the same two factors (Auclair and Cottam 1971; Halls 1973;
McDiarmid et al. 1977; Fleming 1988), and to the appearance
of pioneer species that typically produce more fruits than
late successional or mature-forest species (Martin 1985).
42


93
Figure 5-1. Variation in the distribution of understory birds
at La Planada in relation to distance from the forest edge.
Points represent means and bars standard errors.


107
abundance of birds in forest stands increased from the edge
towards forest interior (Quintela 1986; but see Kroodsma 1984
and Lopez de Casenave et al. ms for opposite results). Edge
age and land use can affect the magnitude of such effects
(DeGraaf 1992). At larger scales, landscape configuration
determines the overall effect of edges on the distribution of
organisms (Hansen et al. 1992).
I reanalyzed Quintela's data on bird captures and found
that, in central Amazon, understory insectivores (Goodness of
Fit Test, G = 49.6, P < 0.001) and frugivores increased
(Goodness of Fit Test, G = 5.8, P <0.1) from edge towards the
forest interior (500 m from forest edge), whereas
nectarivores decreased (Goodness of Fit Test, G = 7.1, P
<0.05). In the Chaco Argentino, bird captures of terrestrial
and arboreal granivores, i.e., doves and parrots, was higher
at forest edge than at forest interior (Lopez de Casenave et
al. ms). Mist netting did not reveal differences among the
other guilds. The Amazon site and La Planada differ markedly
in their understory avifauna (Bierregard and Lovejoy 1989).
Nevertheless, results of both studies show that (1) abundance
of all birds, frugivores, and insectivores is highest in the
forest interior and (2) abundance of nectarivores is highest
at the forest edge. In both areas, pastures and second
growth areas are embedded in a forest matrix (Lovejoy et al.
1986, Chapter 2).
In central Sweden, bird density across clear cut-forest
edges changed over a year (Hansson 1983). Bird density was


44
Seed movement and fruit abundance are related in at
least two ways. First, fruit numbers determine the
availability of seeds. Second, fruit and seed availability
affect the behavior of the dispersers (Murray 1987; Loiselle
and Blake 1993) In spite of these well known relationships
between fruit abundance and seed movement, few studies have
addressed how edges influence fruit abundance (Blanchard
1992). I explored this question in the understory of a
neotropical montane forest using edges resulting from forest
clear-cutting. In particular, I documented how fruit
abundance and seed movement were affected by distance from
the forest edge towards the forest interior for the
assemblage of understory shrubs and for individual species.
Since edge age (Williams-Linera 1990; Blanchard 1992) and
treefall gaps (Janzen 1983; Lovejoy et al. 1986; Noss 1991)
can modify the steepness of such a response, I examined how
they interacted with distance. By looking at four different
scales, edges of different age within the forest matrix,
distance from forest edge within edges, treefall gaps and
foliage density (LAI) within each distance, I could examine
changes in the fruit resource base along edges resulting from
large-scale disturbances.


Table 5-3. Results of Replicated Goodness of Fit Test on the number of captures for
understory birds at the Reserva Natural La Planada. Gneterogeneity (Gh) GTotal (Gt) #
Gold edges (G0), GNew edges (GN) P< 0.1 (*), P< 0.05 (**), P< 0.01 (***), and P< 0.001
(****)
df
gh
df
Gt
df
Go
df
Gn
MiQnectes striaticollis
3
19.89****
6
61.53****
3
6.43*
3
55.07****
Mvadestes ralloides
3
8.76**
6
29.31****
3
17.95****
3
13.16***
PremnQPlex brunnescens
3
13.14***
6
37.08****
3
9.15**
3
27.90****
Aglaiocercus coelestis
3
6.56*
6
18.37***
3
9.74**
3
8.6**
Euphonia xanthoaaster
3
13.50***
6
17.74***
3
1.22
3
16.51****
AllocotoDterus deliciosus
3
1.46
6
1.88
3
0.44
3
1.43
Pasiieuteru? tristriatus
3
2.66
6
6.19
3
1.42
3
4.77
Srallaricula flavirostris
3
0.30
6
6.42
3
3.72
3
2.71
Premnprnis auttuliaera
3
1.50
6
3.02
3
1.18
3
1.83
Pseudotriccus pelzelni
3
2.12
6
23.70****
3
11.64***
3
12.05***
Phastornis svrmatODhorus
3
4.24
6
27.5****
3
13.97***
3
13.56***
Masius chrvsoDterus
3
4.22
6
21.73***
3
7.23*
3
14.42***
Henicorhina leucoDhrvs
3
3.72
6
11.92*
3
1.48
3
10.41**
Cp.eliqpna wilsoni
3
0.79
6
11.26*
3
8.26**
3
3.0
103


135
than the same one in MU.
Upper montane zone
I did not find marked variation in the number of lumps
among the four sites when using an alpha = 0.1. Iguaque (IG)
the most disturbed site, showed the smallest number of lumps
(4 lumps) compared to the remaining three sites (5 lumps)
(Fig. 6-6). Lumps at the upper end of the body mass range
for two sites, Carpanta (CA) and Ucumar Alto (UA), showed a
high proportion of species but the contrary was true for the
other two sites, Acaime (AC) and IG. More interesting,
however, was the pattern of the lumps at the lower end of the
body mass range (<12 g). They consistently decreased in size
and in the proportion of species falling within them.
Carpanta (CA), the least disturbed site, is covered by native
forest and second growth and its lump structure differed from
that of AC and UA particularly for species with a body mass
>25 g. It is noteworthy that the highest proportion of
species in CA was found in the lump representing species >63
g. For little disturbed sites at the other elevational zones
those lumps representing the largest birds within the
observed range of body mass always contained the smallest
proportion of species. The lump structure of AC and UA was
very similar. These two sites were planted with native trees
in an effort to restore land previously used for cattle
ranching (Table 6-1). Finally, IG showed the most
conspicuous differences in regard to the position of lumps
and gaps when compared regard to the position of lumps


64
were two Rubiaceae, Faramea affinis and Palicourea aibbosa.
One hundred species were found both at new and old edges and
49 were exclusive to new (35 species) and old edges (14
species) (Appendix A). In this respect, new edges presented
proportionally more species than old edges (Goodness of Fit
test, G = 9.3, P < 0.01).
Fruit abundance on a species by species level, expressed
as the number of fruiting individuals, varied depending on
distance from the edge and edge age. The distribution of
fruiting individuals for 16 of the 26 abundant species
departed significantly from a uniform distribution across the
four distances (Table 4-3). Given a 10% probability of
obtaining a species that shows a non-uniform distribution it
is very unlikely that 16 or more species out of 26 would have
shown a non-uniform distribution by chance alone (Binomial
test, P = 2.0 x 10-10). Clearly, the distribution of fruiting
individuals of some species is affected by the creation of
edges.
For the 16 species showing a non-uniform distribution I
used residuals to further determine if they were more
abundant at any particular distance from the forest edge
(i.e., if at any given distance the observed frequency was
greater than the expected frequency). Four species were more
abundant at D1 (0-10 m) three species were more abundant at
D4 (190-200 m) four species were more abundant at D2 (30-40
m) or D3 (60-70 m), and five species were more abundant at
two different distances (e.g., Alloolectus tetraaonus and


several people were instrumental in helping to put together
little pieces of my puzzle. J. H. Cock and A. P. Hernandez
from CENICANA lend me the LAI-Canopy Analizer. J. Luteyn, T.
G. Lammers, P. E. Berry, D. Froding, B. Hammel, J. J. Wurdak,
J. Kress, C. Taylor, L. E. Skog, L. R. Landrum, A. M. W.
Mennega, J. S. Miller, and T. Croat kindly identified the
plant material I collected at La Planada. P. Kubilis and C.
Steible provided statistical advice when most needed. L.
Walz prepared maps. P. Amezquita counted pollen tubes. I am
grateful for their valuable help.
This project was crafted some years ago with the input
provided by P. Feinsinger, my former advisor. He presented
me with alternative routes that certainly proved fruitful. I
am particularly grateful for this.
This study was funded by the Fundacin para la Promocin
de la Investigacin y la Tecnologa, Banco de la Repblica,
Colombia and the Wildlife Conservation Society (WCS).
iii


LITERATURE CITED
Acevedo, C. I. 1987. Contribucin al conocimiento de las
relaciones entre colibrs y flores con observaciones de
sus ciclos anuales en el bosque altoandino de Iguaque,
Boyac. Tesis, Departamento de Biologa, Pontificia
Universidad Javeriana, Bogot, Colombia.
gren, J. 1988. Seed size and number in Rubus chamaemorus:
beween-habitat variation, and effects of defoliation and
supplemental pollination. Journal of Ecology 77:1080-
1092.
gren, J. 1989. Sexual differences in biomass and nutrient
allocation in the dioecious Rubus chamaemorus. Ecology
69:962-974.
Aizen, M. A. and P. Feinsinger. 1994. Forest fragmentation,
pollination, and plant reproduction in a Chaco Dry
Forest, Argentina. Ecology 75:330-351.
Andrade, G. I., editor. 1993. Carpanta Selva Nublada y
Pramo. Fundacin Natura, Santa F de Bogot, Colombia.
Andrade, G. I., H. Rubio-T. 1994. Sustainable use of the
tropical rainforest: evidence from the avifauna in a
shifting-cultivation habitat mosaic in the Colombian
Amazon. Conservation Biology 8:545-554.
Andrade, G. I., M. L. Rosas and A. Repizzo. 1993. Notas
preliminares sobre la avifauna y la integridad biolgica
de Carpanta. Pages 207-228, in G. I. Andrade, editor.
Carpanta Selva Nublada y Pramo. Fundacin Natura,
Santa F de Bogot, Colombia.
Andrn, H. and P. Angelstam. 1988. Elevated predation rates
as an edge effect in habitat islands: experimental
evidence. Ecology 69:54-547.
Arango, N. 1991. Depredacin de nido y su relacin con la
fragmentacin de habitat en un bosque nublado tropical.
Tesis B.Sc., Departamento de Biologa, Universidad del
Valle, Cali, Colombia.
Arango, S. 1993. Morfologa y comportamiento alimenticio de
las aves frugvoras de Carpanta. Pages 127-140, in G.I.
166


126
different shades of gray (e.g., Fig. 6-3). Thus the size of
the lump does not represent the proportion of species falling
within them, but rather the size range of birds involved.
Rgsyilti?
Elevational Zones and Body Mass Distribution
I established the lump structure of bird assemblages for
the four elevational zones using an alpha = 0.05 and found
that the number of lumps decreased from the upper lowland (24
lumps) to the pramo zone (5 lumps) (Fig. 6-3). Most lumps
were lost from the upper range (>316 g) and few from the
lower range (<10 g) of body mass. Lumps at both extremes
contained the lowest proportion of species for the upper
lowland, lower montane, and upper montane zones but not for
the pramo.
The lump structure of the upper lowland and lower
montane zones showed striking similarities in regard to the
position of gaps and lumps in the body mass range of 12-575
g. However, the proportion of species in these lumps
differed between the two zones. Lump structure of the upper
montane zone resembled that of the upper lowland and lower
montane zones in the body mass range of 83-316 g but not
above or below these figures. Finally, the pramo zone
shared only one lump (>301 g) with the other three zones.


23
seeds in which the hypocotyyl was visible (ca. 3 mm long).
The seed predation experiment for £. aibbosa lasted for 5
days (July 1993-August 1993) and the seed germination
experiment for 105 days (July 1993-November 1993). The seed
predation experiment for £. affinis lasted for 105 days
(August 1993-December 1993) and the seed germination
experiment for 252 days (August 1993-April 1994) I
concluded the seed germination experiments when 90% of the
seeds had germinated and the seed predation experiments when
no more seeds were being removed. I assumed that seeds
removed from the trays were taken by vertebrates and that
this constituted predation.
Fruit rempv^l
In this experiment I placed eight artificial shrubs per
distance, 4 at each gap and interior site, for a total of 32
artificial shrubs per edge. Each shrub consisted of a 1.5-m-
tall bamboo stick to which I attached an artificial
infructescence resembling that of £. aibbosa. The
artificial infructescences consisted of a 15-cm-long wooden
rod from which four pairs of tooth picks extended. The rods
and tooth picks were dyed bright yellow, and at the end of
each tooth pick I inserted a recently collected ripe fruit of
£ aibbosa. I ran the fruit removal experiment at each edge
for four consecutive days. On the morning of the first day
(0700) I inserted fresh fruits of £. aibbosa and 24 hours
later recorded the number of fruits missing and bitten by
ants. All fruits were changed every 24 hours to start a new


96
MONTH
Figure 5-3. Variation in the distribution of understory birds
at La Planada in relation to edge age and month. Points
represent means and bars standard errors.


This dissertation was submitted to the Graduate Faculty
of the Department of Zoology in the College of Liberal Arts
and Sciences and to the Graduate School and was accepted as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
December, 1995
Dean, Graduate School


86
The use of mist nets versus acoustic and/or visual
censuses to carry out studies on bird assemblages in the
tropics has been widely discussed because of the biases
inherent in any sampling method (e.g., Terborgh and Weske
1969, Terborgh 1971, Karr 1971, Remsen and Parker 1983, Karr
1981a,t, Lynch 1989, Remsen 1994) It is accepted that mist
nets only sample a proportion of bird species found in an
area (Terborgh and Weske 1969, Karr 1981&); that if used over
a prolonged period birds learn the position of nets (Terborgh
1977, Bierregaard 1990); and that figures on bird abundance
might overestimate the abundance of many species (Karr 1981£,
Remsen and Parker 1983, Lynch 1989). In addition, when used
to compare habitats that differ markedly in structure,
capture rates can be misleading in regard to the presence and
abundance of many species (Terborgh 1971, Lynch 1989, Blake
et al. 1990). My study, then, only reflects what happens to
those birds that are effectively sampled by mist nets in the
understory (see also Wong 1986, Levey 1988,b, Blake and
Loiselle 1991, Loiselle and Blake 1991, Poulin et al. 1992).
I stress that mist netting took place only inside the forest,
and the aim of this study was to compare changes in bird
abundance from the edge towards the forest interior. Thus,
problems associated with habitat biases are either minimized
or held constant.
One possible problem for interpreting the results,
however, relates to the timing of mist netting in old and new
edges. In old edges mist netting began in June 1992 and in


2
invasions by an unknown number of species (Henderson et al.
1991, Rattan 1992) .
Even though edges are a prominent feature of neotropical
mountains, little is known about how they influence landscape
pattern. My dissertation focuses on edges and how they
influence seed dispersal through fruit-frugivore
interactions. Changes in the nature of fruit-frugivore
interactions can provide information on the persistence of
edges through time, depending on the ability of plants to
produce and disperse their seeds. On the other hand, changes
in the composition and number of individuals across edges can
provide information on edge structure and productivity.
A common thread among studies focusing on edges is the
lack of a common pattern, without which it is difficult to
propose underlying causes or even consequences. Each new
study has added details at the cost of finding generalities
from which testable predictions can be made. In my
dissertation I followed two different approaches to the study
of edges. Both had as a central theme that of fruit-
frugivore interactions. The first generated detailed
information on fruit-frugivore interactions at the level of
forest stands. Nevertheless, this information precludes
generalizations about the role of edges in neotropical
montane forests due to the intrinsic characteristics of my
study site. The second generated consistent patterns at the
level of entire landscapes that differed in degree of
transformation by human activities. This portion of the


80
fields are maintained as such, the observed distributions
will persist and even become more pronounced.
Sparse species constituted an important component of the
assemblage of understory plants of La Planada. Little is
known about habitat preferences of these species, but clearly
they represent species typical of large disturbed areas,
forest gaps, and forest (Appendix A). Sparse species were
most abundant close to the pasture-forest interface (D1 and
D2) within edges and at new edges within the study area (34
species were found exclusively at new edges compared with 14
at old edges) (Appendix A). In addition, for the few sparse
species from which I recovered seeds in bird droppings, I
found that the proportion of fruiting individuals at the
"edge" and "interior" was independent of the proportion of
bird droppings that contained their seeds and were recovered
in these two zones. Thus, forest edges are being colonized
by sparse species and factors other than seed dispersal might
be influencing recruitment rates across the pasture-forest
edge.
Changes in fruit abundance across the pasture-forest
edge partially paralleled that of bird captures (Chapter 5).
Increased fruit abundance, expressed as total fruits (TF) and
total ripe fruits (RF), was mirrored by an increase in
frugivore capture rates only at D1 (Chapter 5). The opposite
was true at D4 where fruit abundance reached the lowest
values but frugivore capture rates were the highest (Chapter
5). The high fruit production at Dl was mainly due to palms.


Figure 6-4. Lump structure of Colombian frugivorous birds from sites covered mostly by
forest (bottom) to sites highly transformed by human activities (top) within the upper
lowland zone. Anchicaya (AC), Yotoco (YO), San Carlos (SC), La Esmeralda (LE), El Ocaso
(EO). Each box represents a lump and the space between the boxes represent gaps in the
distribution of body mass. The different shades indicate the proportion of species falling
within lumps: (1) 0-5, (2) 5-10, (3) 10-20, (4) 20-30, (5) 30-45, (6) 45-60, and (7) 60-100
% of species. Vertical lines represent 0-5% of species.
Numbers on the right side represent number of species for the corresponding data set.
1 2 3 4 5 6 7


99
The distance effect was also affected by month for
frugivores as shown by the significant interaction between
these two factors (ANOVA, F45fi8i =1.93, P = 0.007; Table 5-
1, Fig. 5-4). I found that capture rate of frugivores was
significantly higher at D4 (mean = 8.4) than at D1 (mean =
4.7) during the dry months (Contrast of Mean
Differences, Fifi20 = 7.3, df = 1, P < 0.008) but not during
the wet months. The distance x month interaction in the
omnibus test was not significant for insectivores and
nectarivores (Table 5-1, Fig. 5-4). However, when I tested
specific hypotheses I found that capture rates of
insectivores were higher at D4 (mean = 8.0 and 8.4 for dry
and wet season, respectively) than at D1 (mean = 4.2 and 6.6
for dry and wet season, respectively) in both seasons
(Contrast of Mean Differences, Fi>i20 = 6.7, P = 0.01). For
nectarivores I found that capture rates were higher at D1
(mean = 5.6) than at D4 (mean = 3.2) only during the dry
season (Contrast of Mean Differences, Fifi20 = 5.7, P = 0.02).
The abundance of insectivores changed in old and new
edges depending on the month of the year, as shown by the
significant interaction between month and edge age (ANOVA,
Fl6, 64 = 3.24, P = 0.004; Table 5-1, Fig. 5-3). Insectivores
were more abundant at new edges (mean = 7.9) than at old
edges (mean = 5.4) only during the wet season (Contrast of
Mean Differences, Fi(4q = 7.0, P = 0.01).


101
Table 5-2. Distribution of common (>21 captures) understory
bird species of La Planada in relation to distance from edge.
Numbers are number of captures.
P< 0.01 (***), P < 0.001 (****)
P <
0.1
(*) ,
P < 0.
05 (**),
Distance
G-stat
P
D1
D2
D3
D4
Uniform Distribution
Allocotopterus deliciosus
13
10
11
12
0.42
ns
Eunhonia xanthocraster
32
22
17
24
4.16
ns
Pipreola riefferii
7
5
4
5
0.89
ns
Tanqara spp.
8
4
4
11
5.08
ns
Basileuterus tristriatus
23
14
25
20
3.54
ns
Grallaricula flavirostris
9
13
19
21
6.14
ns
Mviphobus flavicans
9
6
3
5
3.27
ns
Preumgrnis guttulioera
25
33
25
27
1.52
ns
Non-Uniform Distribution
Increase from Edge to Interior
Atlaoetes brunneinucha
3
4
11
10
7.52

Lipauqus crvptolophus
1
2
5
8
7.78

Mionectes striaticollis
30
16
15
61
41.63

Glvohorrvnchus soirurus
10
9
10
30
18.17
****
Mviotriccus ornatus
11
6
9
23
12.51
*
Premnoplex brunnescens
8
20
25
40
23.94

Pseudotriccus oelzelni
13
14
15
40
21.58

Aalaiocercus coelestis
50
53
68
85
11.82
*
Decrease from Edge to Interior
Chlorosoinaus semifuscus
13
2
3
5
13.64

Cgeliqena wilsoni
42
25
18
28
10.49

Haoloohaedia luaens
20
11
13
7
6.87

Ocreatus underwoodii
10
3
4
1
9.6

Increase at Edge and Interior
Mas,US chrvsopterus
37
27
25
58
17.47

Mvadestes ralloides
25
16
9
37
20.58

Henicorhna leucoohrvs
38
19
24
34
8.19
* *
Svndactvla subalaris
14
6
6
14
6.59

Phaetgrnis svrmatophorus
62
21
34
47
23.28



Monnimiaceae
Myrsinaceae
Siparuna sp.
CR 694
Myrtaceae
Cvbianthus simplex (Hook.f.)Aaost
Onagraceae
ELuqenia anastomosus DC
Rubiaceae
Fuchsia macrosticrma Bentham
Solanaceae
Faramea affinis
Faramea killioii Standi.
Palicourea qibbosa
Palicourea sp.l
CR 461, CR 695
Palicourea sp.2
CR 430
P-SYCOthria aubletiana Steyerm.
Psvcothria hazenii Standi.
Psvcothria panamensis Standi.
PSYCOthria solitudinum Standi.
Cestrum sp.
CR 485, CR 535, CR 779
PhYsaJlis sp.
Solanum evolulifolium
CR 487, CR 684
Solanum lepidotum
CR 570, CR 711, CR 489, CR 799
Solanum sp.2
LD
S
F
S
F
VS
LD
VS
F-LD
VA
BD
F
S
BD
F-LD
VA
BD
LD
S
F
VS
F
VA
BD
F
S
BD
LD-F
S
BD
LD
S
F
S
G
S
G
S
S
LD
s
BD
G VS
156


181
Murray, K. G. 1987. Selection for optimal fruit-crop size in
bird-dispersed plants. American Naturalist 129:18-31.
Murray, K. G. 1988. Avian seed dispersal of three neotropical
gap-dependent plants. Ecological Monographs 58:271-298.
Naranjo, L. G. 1994. Composicin y estructura de la avifauna
del Parque Regional Natural Ucumari. Pages 305-328 in
J.O. Rangel, editors. Ucumari: Un caso tpico de la
diversidad bitica Andina, Corporacin Autnoma Regional
de Risaralda, Colombia.
Newmark, W.D. 1991. Tropical forest fragmentation and the
local extinction of understory birds in the eastern
Usambara mountains, Tanzania. Conservation Biology
5:67-78.
Noble, I. A. 1993. A model of the responses of ecotones to
climate change. Ecological Monographs 3:396-403.
Norman, J. M. and G. S. Campbell. 1991. Canopy structure,
Pages 301-326, in R. W. Pearcy, J. Ehleringer, H. A.
Mooney, and P. W. Rundel,editors. Plant Physiological
Ecology. Chapman and Hall, London, UK.
Noss, R. F. 1991. Effects of edge and internal patchiness on
avian habitat use in an Old-Growth Florida Hammock.
Natural Areas Journal 11:34-47.
Orejuela, J. E. 1987. La Reserva Natural La Planada y la
biogeografa andina. Humboldtia 1:117-148.
Orejuela, J. E. and G. Cantillo, compilers. 1980. Aves de la
Reserva Natural La Planada. Programa Alegra de Ensear,
Fundacin para la Educacin Superior-FES, Cali,
Colombia.
Orejuela, J. E., R. S. Raitt and H. Alvarez. 1979.
Relaciones ecolgicas de las aves en la Reserva Forestal
de Yotoco, Valle del Cauca. Cespedesia 29-30:7-28.
Osunkoya, O., J. E. Ash, M. S. Hopkins, and A. W. Graham.
1994. Influence of seed size and seedling ecological
attributes on shade-tolerance of rain-forest tree
species in northern Queensland. Journal of Ecology
82:149-163.
Pacheco, S., N. Florencio da Silva, R. Ribon, J. E. Simon and
R. Torres. 1994. Efeito do manejo do cerrado sobre as
populagoes de algus Tinamidae em Tres Marias, Estado de
Minas Gerais. Revista Brasilera Biologia 54:435-441.
Pimm, S. L. and A. M. Sugden. 1994. Tropical diversity and
global change. Science 263:933-934.


y.iroo leucoohrvs
Parulidae
Basileuterus coronatus
Basileuterus tristriatus
Myiobprug miniatus
Thraupidae
Anisoanathus flavinucha
Chlorochrvsa ohoenicotis
ghlorospingys semifuscus
DalasSa albilatera
Diqlogga indiqotica
Euohonia xanthoaaster
Iridosornis oorohvroceohala
Tachvohonus luctuosus
Tangara arthus
Tangara labradorides
Tanqara niqroviridis
Catamblyrhynchidae
Catamblvrhvnchus diadema
Fringillidae
Amaurogpiza concolor
Afc.lapetes brunneinucha
1 Arango (1993); 2 Miller (1963)
ES I 22.0
S I 17.1
VA I 13.1
VS I 8.2
VS F 44.4
S F 20.6
A F 28.1
S N 9.9
ES N 12.0
VA F 13.6
ES F 26.0
ES F 35.0
S F 20.8
VS F 14.2
S F 16.9
ES I 18.0
VS F 16.0
A F 45.6
165


7
ha (mode 3 ha) (Gmez and Palau 1994) The patches of forest
are connected by strips of forest that have been left along
streams, steep sloptes, and mountain ridges.
The natural disturbance regime is varied. At small
scales, treefall gaps, which occur mostly during the rainy
season, are common (Samper 1992). At larger scales
landslides, ash rain (e.g., January 1993 volcanic eruption),
and strong winds, the latter resulting in the defoliation of
large areas (e.g., August 1993), affect whole landscapes.
Unpublished climatological records of La Planada (1985-
1994) show a mean annual rainfall and temperature of 4,437 mm
and 19.2C, respectively (Fig. 2-2). Rainfall is distributed
in two wet seasons, interrupted by a mild dry (February-
March) and a strong dry (June-August) season (Fig. 2-2).
Based on these data, La Planada can be classified as a
transitional life zone between tropical premontane rain and
wet forest (Holdridge 1967). An important climatological
feature of La Planada and its surroundings is the presence of
afternoon mist during most of the year.
The water balance for La Planada shows that on average
every month has a surplus of water which is lost as soil
runoff (Table 2-1). Nevertheless, in some years there might
be months in which there is a water deficit. This can
explain why during my study period, in particular during the
months of July-August of 1993, many plants lost their leaves.


178
Lescourret, F. and M. Genard. 1994. Habitat, landscape and
bird composition in mountain forest fragments. Journal
of Environmental Management 40:317-328.
Levey, D. J. 1986. The role of seed size and bird species on
seed deposition patterns, Pages 147-158, in A. Estrada
and T.H. Fleming, editors. Frugivores and Seed
Dispersal. Junk, Dordrecht, The Netherlands.
Levey, D. J. 1987. Seed size and fruit-handling techniques of
avian frugivores. American Naturalist 129:471-485.
Levey, D. J. 1988a. Spatial and temporal variation in Costa
Rican fruit and fruit-eating bird abundance. Ecological
Monographs 58:251-269.
Levey, D. J. 1988b. Tropical wet forest treefall gaps and
distributions of understory birds and plants. Ecology
69:1076-1089.
Levey, D. J. 1990. Habitat-dependent fruiting behaviour of an
understorey tree, Miconia centrodesma, and tropical
treefall gaps as keystone habitats for frugivores in
Costa Rica. Journal of Tropical Ecology 6:409-420.
Li-Cor, Inc. 1992. LAI-2000 Plant Canopy Analyzer Operating
Manual. Li-Cor, Inc., Lincoln, NB.
Lipsey, M. W. 1990. Design Sensitivity: Statistical Power
for Experimental Research. SAGE Publications, Inc.,
Newbury Park, California, USA.
Loiselle, B. A. and J. G. Blake. 1991. Temporal variation in
birds and fruits along an elevational gradient in Costa
Rica. Ecology 72:180-193.
Loiselle, B. A. and J. G. Blake. 1993. Spatial distribution
of understory fruit-eating birds and fruiting plants in
a neotropical lowland wet forest. Vegetatio
107/108:177-189.
Lopez de Casenave, J., J. P. Pelotto, S. Caziani, M. Mermiz,
and J. Protomastro. Edge effects in avian assemblages
of a Chaco semiarid forest, Argentina. Unpublished
manuscript.
Lovejoy, T. E., R. O. Bierregaard, A. B. Rylands, J. R.
Malcolm, C. E. Quinetela, L. H. Harper, K. S. Browns, A.
H. Powell, G. V. N. Powell, H. O. R. Schubart, and M. B.
Hays. 1986. Edge and other effects of isolation on
Amazon forest fragments. Pages 257-285 in M. E. Soul,
editor. Conservation Biology: The Science of Scarcity


90
SuperANOVA (Gagnon et al. 1989).
To determine if individual species were affected by the
presence of edges, I performed two related tests based on
data collected between June 1992 and August 1993. First, I
used a Goodness of Fit test (G-statistic) to determine if the
number of captures across the four distances departed
significantly from a uniform distribution. Second, I used a
Replicated Test of Goodness of Fit (G-statistic) (Sokal and
Rohlf 1981) to establish, in addition to the distance effect,
an edge age effect on the number of captures. For both
tests, monthly captures for each species were pooled, keeping
separate the information on old and young edges. For these
analyses I chose those species in which at least 80 percent
of the expected cell frequencies were greater than 5, since
the statistic G departs from the X2 distribution if this is
not the case (Siegel and Castellan 1988). Five species
(Lipaugus cryptolophus. Ocreatus underwoodii. Tanqara rthys,
T. labradorides, and T. nigroviridis) did not meet this
criterion but were included in the analyses. The first two
were included because of the clear trends they exhibited.
The last three species were lumped in the Tanaara spp. for
the analyses because of similarities in many features of
their life history (Isler and Isler 1987).
A vast majority of the species did not meet the above
criterion. To evaluate the influence of edges on these
species I classified birds into five categories according to
capture frequency: extremely sparse (1 capture), very sparse


43
It has been suggested that the spatial heterogeneity of
the fruit resource base, regardless of the scale at which
disturbances occur, influences the distribution of organisms
feeding on fruits (e.g., Martin 1985; Wong 1986; Levey
1988,b; Heideman 1989; Blake and Loiselle 1991; Loiselle and
Blake 1991) and the resulting dispersal of seeds (Murray
1988). Nevertheless, depending on the scale of disturbance
changes in seed dispersal might have different consequences
for the plants. For example, small-scale disturbances may
influence recruitment within populations whereas large-scale
disturbances may influence colonization of new areas (Harper
1994).
One immediate consequence of disturbance is the creation
of edges or boundaries. In general, boundaries mediate
fluxes between adjacent ecological systems (Margalef 1968;
Wiens et al. 1985; Gosz 1991). Moreover, because of
differences in their permeability to fluxes of material and
energy, boundaries may influence the dynamics of neighboring
systems (Correll 1991; Ryszkowski 1992). In this context,
edges that bound patches resulting from disturbance might
influence the dynamics and structure of the neighboring
patches by influencing the distribution of resources and/or
movement of organisms (Crist et al. 1992; Johnson et al.
1992; Wiens 1992). In particular, edges resulting from
large-scale disturbances may influence the movement of seeds
and thus the structure of whole landscapes.


116
78 76 74 72 70 65
Figure 6-1. Montane habitats of Colombia (>800 m) showing
the location of sites included in this study.


Table 3-1. Results of ANOVAs for Mixed Factorial Designs (Split-split-plot) on percentage
of flowers with pollen tubes in relation to the total number of flowers (F) and average
number of pollen tubes per flower (P) for Palicourea aibbosa and Faramea affinis.
Significance
at 10%
(*)/5% (**),
and
1% (***).
Variable
Age
A
Edge(A)
Error
Distance
D
D
X A
DxE(A)
Error
df
F
df
ss
df F
df
F
df
SS
£ aibbosa
F
1
9.2**
4
8011.5
3 0.5
3
1.3
12
11229.4
P
1
7.7**
4
35.6
3 0.3
3
2.5
12
49.9
£. affinis
F
1
3.9
4
2408.8
3 0.2
3
0.5
12
5054.8
P
1
1.9
4
1.9
3 0.4
3
0.5
12
3.6
Habitat
H
H x
D
H x A
H :
< D X A
Hx[E(A)xD]
Error
df
F
df
F
df F
df
F
df
SS
£. aibbosa
F
1
1.0
3
0.5
1 0.4
3
0.17
16
8968.1
P
1
4.1
3
1.4
1 0.3
3
0.3
16
65.04
£ affinis
F
1
8.0**
3
1.2
1 0.09
3
2.6
16
1239.6
P
1
11.8***
3
0.9
1 2.4
3
4.3**
16
0.9


Ericaceae
CR 552, CR 771, CR 551, CR 506
Psammisia montana Luteyn sp. nov.
Gesneriaceae
F VS
Alloolectus bolivianus (Britton) Wiehler
Columnea cf. oicta Karsten
Drvmonia turrialvae Hanst.
Gasteranthus aff. oncoaastrus (Hanst.)
Gasteranthus oncoaastrus (Hanst.) Wiehler
Kohleria villosa (Fritsch) Weihler
Heliconiaceae
F VS
G ES
G VS
G S
G S
G ES
Heliconia impdica Abalo & Morales
Melastomataceae
G ES
Blakea cf. ouadriflora Gleason F ES
Miconia hvmenanthera Triana VS
Meliaceae
Unknown
CR 751
Phytolacaceae
Phytolacca rivinoides Kunth & Bouche
Piperaceae
Piper qutierrezii T.&J.
Rubiaceae
Palicanrea sp.4
CR 592
Palicourea standlevana A.M. Taylor
Psvcothria allgnij Standi.
Psvcothria braulioi A.M. Taylor sp. nov.
F ES
G ES
ES
G ES
G S
F ES
G ES
158


121
Data Analysis
I used the Lump Analysis (LAqri) technique to analyze the
data (GaP Detector 2.1). This technique is being developed
by P. Marples, Arthur Marshall, Jr. Lab, Department of
Zoology, University of Florida. LAqri tests whether breaks in
an observed frequency distribution of an attribute occur by
chance alone due to sampling error (Fig. 6-2a). It relies on
the generation of a continuous, unimodal distribution from
input data (in this case, body mass) and the detection of gap
rarity indexes, or GRI-values within the data set. The
unimodal distributions are derived using the smallest normal
kernel estimate (h) that smoothes a frequency distribution
into an unimodal continuous distribution (Silverman 1986).
GRI-values are derived by calculating the absolute
differences between contiguous rank size-ordered data points
for the observed and expected distributions. The GRI-values
for a given data set are compared against a critical value
(GRIcrit) and those that exceed that value define a
significant break (gap) in the observed frequency
distribution at the indicated alpha level (Fig. 6-2b), the
size of the observed data set (N), and the ratio between the
size of the observed data set and the mean size of all the
data sets that are compared simultanously (rj). All body
masses between two contiguous gaps define an aggregate of
species or lump. Calculations were based on logio-transformed


46
pasture. I always kept the lens and LAI-2050 optical sensor
pointing in the same direction and 1.5m above the ground. I
covered the optical sensor with a 45 view cap to block my
image and direct beam radiation during clear days (Li-Cor
1992) .
Estimates obtained with this instrument often
underestimate true LAI (Chason et al. 1991; Hannan and Bgu
1995) Nevertheless, in the context of this study these
values are useful to describe relative changes in canopy
structure across the pasture-forest edge.
Fruit Abundance
To establish the influence of edges and treefall gaps on
fruit production by understory plants I subdivided each of
the four 100 x 10 m strips within each edge into five 20 x 10
m quadrats (Chapter 2, Fig. 2-3). In turn each quadrat was
subdivided into four 10 x 5 m subquadrats and for each
quadrat I chose at random two subquadrats in which to monitor
fruit production (Fig. 2-3).
I used Brokaw's (1982) definition of treefall gap to
classify each subquadrat as gap or interior habitat. A
subquadrat was classified as gap if it was within a gap or
located <5 m from a gap edge, and as interior if it
was located >5 m from the edge of the nearest treefall gap at
the time the study began. These two categories do not
reflect the environmental continuum from the center of the
treefall to the intact forest nor do they take into account


Table 4-4. Results of Replicated Goodness of Fit Test on the number of individuals in fruit
for understory plants at the Reserva Natural La Planada. Gneterogeneity (Gh) / GTotal (Gt)
Gold edges (G0), and GNew edges (GN) P < 0.1 (*), P < 0.05 (**), P < 0.01 (***), P < 0.001
(****)#
Species
df
gh
df
Gt
df
Go
df
Gn
AnLlmrium versicolor
3
6.64

6
11 *
3
1.48 ns
3
9.49 **
Anthurium cf. marmoratum
3
7.88

6
15.08 **
3
10.72 **
3
4>3 ns
Anthurium umbricolum
3
28.28

6
40.48 ****
3
17.96 ****
3
3.22 ns
Alloolectus teuscheri
3
7.72
*
6
18.6 ***
3
5.06 ns
3
13.52 ***
Gegnoma weberbaueri
3
17.56

6
22.86 ****
3
3.93 ns
3
18.92 ****
Solanum sp.7
3
6.84
*
6
7.83 **
3
2.65 ns
3
5.18 ns
Anthurium cf. melampvi
3
21.76
*
6
34.02 ****
3
17.38 ****
3
16.6 ****
Palicourea aibbosa
3
6.32

6
29.93 ****
3
16.77 ****
3
13.14 ***
Solanum sp.5
3
6.24
*
18.38 ****
7.16 *
11.19 **
Anthurium membranaceum
3
0.94
ns
6
0.94 ns
3
1.74 ns
3
0.49 ns
Burmesitera carnosa
3
3.47
ns
6
4.32 ns
3
2.39 ns
3
1.89 ns
Clidemia sn.l
3
2.42
ns
6
51.11 ****
3
32.79 ****
3
18.12 ****
Psvcothria aubletiana
3
3.84
ns
6
44.15 ****
3
24.29 ****
3
19.83 ****
Faramea eleaans
3
2.24
ns
6
2.25 ns
3
3.35 ns
3
6.48 *
Besleria solanoides
3
4.04
ns
6
39.5 ****
3
35.13 ****
3
4.35 ns
Chamaedorea Dolvchlada
3
3.32
ns
6
14.14 **
3
12.45 ***
3
1.69 ns


18
and are presented in terminal, pendant, green
infructescences, containing a maximum of 3 fruits. They
measure 20 x 18.2 mm and contain a single seed, 10.4 x 7.6 mm
(Samper 1992). Seeds of F. affinis are dispersed by a
different set of birds, Andiqena laminirostris, pipauqus
crvotolophus. Pipreola riefferi, Semnornis p^mph^gtings, and
Troaon personatus (Restrepo 1990, Beltrn 1991, Samper 1992).
Methods
I evaluated the combined influence of edges and treefall
gaps on several stages of the life cycle of Palicourea
aibbosa and Faramea affinis by sampling individuals and
conducting experiments at six edges (three old and three new)
and at four distances from forest edge towards forest
interior (0-10 m, 30-40 m, 60-70 m, and 190-200 m) (Chapter
2). Depending on the stage of the life cycle I was
examining, I modified the basic sampling design described in
Chapter 2. This was due to logistic constraints, including
accessibility of the edges.
Beginning in March 1992 I tagged all individuals <. 2 m
tall in flower and/or in fruit, and classified them as being
in gap or interior. An individual was classified as in gap
if it was within a gap (sensu Brokaw 1982) or located < 2 m
from a gap edge, and as interior if it was located > 2 m from
the edge of the nearest treefall gap at the time the study
began. I continued tagging individuals throughout the study
period as new individuals flowered. For each individual I


Table 2-2. Characteristics of the edges included in this study.
Edge
T1
O2
A3
C4
PS5
G6
o
G
G
00
Climo I
S
40 NE
1953
H
6.0
4.2
Cattle ranching;
pasture
Extraction of
poles; cattle
palm hearts and
grazing
Climo II
S
19 NE
1953
M
7.6
7.6
Cattle ranching;
pasture
Extraction of
poles; cattle
palm hearts and
grazing
Pialapi
F
8 NE
1950
L
9.9
Trail to Pialapi;
second growth
Selective logging 40 yr ago
Acantayac
F
24 NE
1981
M
3.0
11.0
Cattle ranching;
pasture
Extraction of
poles; cattle
graves
palm hearts and
grazing; ancient
Hermgenes
M
68 NE
1982
H
3.0
10.0
Cattle ranching;
pasture
Extraction of
palm hearts
Marcos
S
59 NE
1982
H
1.0
7.5
Cattle ranching;
pasture
Extraction of
poles; cattle
to Pialapi
palm hearts and
grazing; old path
1 T = Topography; S = steep, M = moderate, F = flat
2 0 = Orientation (position of edges regarding the cardinal points)
3 A = Age (year the forest was clear cut)
* C = Edge contrast; H = high, M = moderate, L = low
5 PS = Size of disturbed area (ha)
6 G = Percent of sampling area covered by gaps
7 CU = Use of clear cut area
FU = Use of forested area


51
factor and used a corrected F-ratio (H-F) to interpret the
analyses (Girden 1992; von Ende 1993).
I used an alpha of 10%. I set alpha at this level
because my design could lead to increases in Type II errors
(reduced power of my tests) (Zolman 1993). Concomitantly I
increased the probability of committing Type I errors.
First, the scale at which I worked precluded inclusion of
more replicates, which is often the case when dealing with
large-scale ecological phenomena (Scheiner 1993). The area
encompassed by the 6 edges was equivalent to 12 ha and the
access to them was difficult due to steep terrain. Second,
in a mixed factorial design the number of degrees of freedom
is reduced compared to a factorial design because of multiple
nesting (Zolman 1993). In the field, I was limited by the
number of edges I could reach within walking distance from
the field station and thus I had to set up the design as was
described above. Lastly, the use of Type III SS to analyze
unbalanced data sets may lead to Type II errors (Potvin
1993).
I analyzed fruit abundance data for understory species
using a Replicated Goodness of Fit Test (G-statistic) (Sokal
and Rohlf 1981) to establish whether distance and edge age
affected the number of fruiting individuals. First, I pooled
the data for old and new edges and calculated Gp (G-Pooled)
to determine if the number of individuals across the four
distances departed significantly from a uniform distribution.
Second, I compared old and new edges and calculated Gh (G-


45
Methpflg
Canopy Structure
I estimated leaf area index (LAI) (m^ foliage area/m^
ground area) to (1) characterize the structure of the forest
canopy across the pasture-forest edge and to (2) relate LAI
to fruit abundance. Fruit production is strongly influenced
by light environment (e.g., May and Antcliff 1963; Jackson
and Palmer 1977), which in the subcanopy and on the forest
floor is influenced by canopy structure (Norman and Campbell
1991) Thus LAI estimates provide a fine-scale description
of the light environment of each point where I sampled
fruits.
I used a LAI-2000 Plant Canopy Analyzer (Li-Cor, Inc.)
in October 1993 to estimate LAI at the three old edges.
Estimates of LAI obtained with this instrument are based on
the transmitted fraction of incident radiation on the canopy.
At each of the old edges (Climo I, Climo II, and Pialap) I
established three transects running perpendicular to the
forest edge and extending 10 m into the pastures and 210 m
into the forest (Chapter 2, Fig. 2-3). I made readings at
intervals of 5 m in the first 50 m of the transect, 10 m in
the next 130 m, and 20 m in the next 40 m. For each point
along a transect I made four consecutive below-canopy
readings that together were paired with a single reading
taken in an area devoid of trees and shrubs in the nearby


78
deposition rates decrease from the forest edge towards the
forest interior (Geiger 1965; Draiijers et al. 1988). In an
area like La Planada, which is characterized by nutrient-poor
soils, the addition of nutrients could affect the production
of fruits.
A combination of several factors then may account for
the changes I observed in fruit abundance for individual
species and for the entire understory plant assemblage. I
propose that changes in TF-A and TI across the four distances
depending on month, are the result of within-year variability
in environmental conditions. On the other hand, changes in
TF and RF over the year may be the result of increased
irradiance at the forest edge due to direct exposure of the
pasture-forest interface to sunlight. Recall that TF and RF,
which included counts of palm fruits, were most influenced by
distance from forest edge. This is consistent with
Blanchard's (1992) results showing that increased abundance
of palm fruits at the forest edge is correlated with
increased light levels.
Consequences for Plants and Fruaivores
Studies looking at the effect of edges on plants have
focused more on the vegetative (e.g., Ranney et al. 1981;
Chen et al. 1992; Seizer 1992; Young 1993; Matlack 1993) than
on the reproductive stage (Romano 1990; Willimas-Linera 1990;
Blanchard 1992; Murcia 1993) of their life cycle. The former
provides information on the structure and productivity of the


129
Sites and Body Mass Distribution
Upper lowland zone
The number of lumps detected at an alpha = 0.05
decreased from the site covered extensively by forest,
Anchicaya (AN, 8 lumps), to that site dominated by pastures,
El Ocaso (EO, 3 lumps) (Fig. 6-4). The lumps that were lost
were at the upper end of the range of body mass (>316 g);
none was lost at the lower end.
The lump structure of two pairs of sites, Anchicaya
(AN)/Yotoco (YO) and San Carlos (SC)/La Esmeralda (LE),
exhibited the closest similarities in regard to the position
of gaps and lumps, even though AN and SC have twice as many
species as YO and LE, respectively (Fig. 6-4) The
similarities occurred almost over the entire range of body
mass. On the other hand the lump structure of LE and EO
showed important differences, even though these two sites
have a similar number of species and are separated only by 7
km.
Lower montane zone
The number of lumps decreased from sites covered
extensively by forest (La Planada (LP), Ucumar Bajo (UB),
and San Antonio (SA), average = 8 lumps) to sites in which
the original forest has been replaced by orchards (Rio
Grande, RG), and forestry plantations of exotic species
(Munchique, MU and Piedras Blancas, PB) (average 4 lumps)


21
seedlings of P. aibbosa and F. affinis. I placed by their
side a stick with a piece of flagging tape with a distinctive
number for each seedling. I recorded whether seedlings were
growing in treefall gaps (n = 201 and n = 213, £. aibbosa and
F. affinis. respectively) or intact forest
(n = 210 and n = 221) To standardize measurements I marked
the stems of each seedling with yellow vinyl paint (ca. 1.5
cm above soil surface) and the youngest pair of leaves with
threads of flagging tape tied around the petioles. With
calipers I took a first measurement of the seedling's height
from the yellow mark to the base of the meristem and I
repeated this procedure five times between May 1992 and
October 1993. I also recorded and marked new pairs of
leaves.
Seedling growth rate (GR) is expressed as the increment
in height between the first (hn) and the last measurement
(hn+l) [GR = (hn+i hn/tn+i tn)*(30 days/month)] (Seizer
1992). Leaf production rate (LPR) is expressed as the number
of new leaves produced between the first (ln) and last (ln+l)
period [LPR = (ln+l ln/tn+l tn)*(30 days/month)].
Field Experiments
Experiments on seed predation and seed germination of £.
aibbosa and £. affinis were performed at three edges
(Hermgenes, Climo I, and Climo II) and that on fruit
removal (£. aibbosa) at two edges (Hermgenes and Climo I)
(Fig. 2-1, Chapter 2). Even though these edges represent two


/
Table 3-5. Results of Mixed Factorial ANOVAs (Split-splot, 1 repeated measure) on relative
growth (iran/month) and relative leaf production (pairs of leaves/month) rates in Palicourea
gibbosa (PG) and Faramea affinis (FA). Significant at 10% (*),5% (**), 1% (***), 0.1%
C
) .
Distance
df F
D
df
x E
SS
Edge
df F
Habitat
df F
H x D
df F
Residual
df SS
FA
(Growth)
2
3.3
4
19.4
2
0.9
1
8.1***
2
2
422
1053.7
PG
(Growth)
2
0.2
4
421
2
1.7
1
38.7****
2
1.1
399
4101.1
FA
(Leaves)
2
1.1
4
0.2
2
5.2***
1
3.4*
2
0.7
424
10.3
PG
(Leaves)
2
0.5
4
1.4
2
2.1
1
17.1****
2
1.2
408
16
U>
00


106
nectarivores. In my study area, and averaging over time,
capture rates for all birds and frugivores showed bimodal
distributions peaking, at D1 (0-10 m) and D4 (190-200 m).
Capture rates for insectivores showed an abrupt increase at
D4. Time since edge creation, however, had a major effect on
capture rates for all birds and frugivores across the
pasture-forest edge, especially at D3 (60-70 m) and D4 (Fig.
5-2). Within edges, capture rates of frugivores varied among
the four distances on a monthly basis. Between edges,
capture rates of insectivores varied between old and new
edges on a monthly basis. In discussing these findings I
will (1) compare them against results from studies conducted
in other tropical areas and (2) look at possible factors that
might influence the distribution of birds across pasture-
forest edges.
Patterns of Bird Distribution across Edges
My results agree with those of other studies on edges in
that (1) abundance (or density) of organisms changes from
edge towards forest interior (Quintela 1986, Laurance 1990,
Malcolm 1994), (2) functional or taxonomic groups of
organisms, as well as individual species, respond in
different ways to the creation of edges (e.g., Noss 1991),
and (3) the magnitude and direction of the responses to the
creation of edges varies depending on time since edge
creation and season.
A study conducted in the Amazon showed that overall


3
study, therefore, might be useful in generating testable
hypotheses that could guide future studies.
My study area and general sampling procedure are
described in detail in chapter 2. I often refer to this
chapter because the studies I describe in chapters 3 to 5
were based on the same sampling procedure.
In chapter 3 I focus on two understory plants, Palicoura
aibbosa and Faramea affinis (Rubiaceae) and ask how edges may
influence different components of the life cycle of these two
species and how this may influence recruitment. These two
species are the most common understory species of my study
site and are bird-dispersed. In particular, I looked at how
the combined effect of distance from forest edge and presence
of treefall gaps affect pollination, fruit set, seed
predation, seed germination, and seedling growth.
In chapter 4 I take a broader approach to the study of
edges and their influence on fruit-frugivore interactions, by
focusing on the assemblage of understory plants. I ask
whether fruit abundance changes across pasture-forest edges
and how treefall gaps and season modify such effects. I look
at changes in fruit abundance for the entire assemblage of
understory plants and for individual species. Changes in
fruit abundance can affect the distribution of birds feeding
on them and thus seed dispersal across edges.
In chapter 5 I keep the previous approach to the study
of edges and their influence on fruit-frugivore interactions
but ask how edges influence the distribution of understory


61
fruits excluding Arecaceae (TF-A) eliminates part of the
variability associated with crop size because (1) TI
considers only the number of fruiting individuals and (2) TF-
A includes plant species producing smaller and less
persistent fruit crops, i.e., number of fruits per
reproductive season, than those of palms and plant species
producing similar number of fruits.
Lastly, fruit abundance expressed as the total number of
fruits (TF) and total number of ripe fruits (RF) differed
significantly between gaps and interior, but depended on
month, as shown by the significant interaction between
habitat and month (ANOVA, Fio,i60 = 2.0, P = 0.03 and Fio,i60 =
2.0, P = 0.03, respectively; Table 4-2). The Arecaceae again
seemed to be mostly responsible for this result as indicated
by the fact that this interaction was not significant for the
total number of fruits when the Arecaceae were excluded (TF-
A) (Table 4-2).
Species Lgygi Responses
I recorded 149 plant species fruiting in the understory
of the edges included in this study and classified them in
five categories: extremely sparse (1 individual), very sparse
(2-5 individuals), sparse (6-20 individuals), abundant (21-50
individuals), and very abundant (>51 individuals). Of 149
species, 26 (17%) were abundant to very abundant (>21
individuals) and 125 (83%) were sparse to extremely sparse
(<21 individuals) (Appendix A). The most abundant species


FRUIT ABUNDANCE
(mean counts / 50 m 2)
65
Figure 4-6. Variation in fruit abundance at the Reserva
Natural La Planada in relation to edge age and month. Points
represent means and bars standard errors.


29
habitat, distance, and edge age (ANOVA, F3(7 = 4.3, P = 0.05,
Table 3-1).
Fruit Set
In Palicourea qibbosa fruit set was influenced by
distance from forest edge, but the effect depended on edge
age, as shown by the significant interaction between distance
and edge age (ANOVA, F2,2 = 5.8, P = 0.02, Table 3-3). At Dl
and D2 the percentage of developing fruits was greater at new
than at old edges but this trend was reversed at D3 where 28%
0.02% (mean SE) of the flower buds resulted in fruits in
old edges as compared to 20.0% 0.02% in new edges (Fig. 3-
1). The percentage of ripe fruits was also influenced by
distance and similarly depended on edge age (ANOVA F2,2 = 3.9,
P = 0.06, Table 3-3). At Dl and D3 the percentage of ripe
fruits was greater at new edges but the trend was reversed at
D2. At D2 in old edges, 51.0% 0.05% of fruits ripened
compared to 42.0 0.04% at new edges (Fig. 3-1).
Seed and Fruit Damage
The percentage of Palicourea aibbosa fruits damaged by
ants and wasps did not differ among the four distances (Table
3-3). Edge age, however, influenced the percentage of fruits
eaten by ants but depended on habitat as shown by the
significant interaction between edge age and habitat (ANOVA,
Fi,i = 4.0, P = 0.07, Table 3-3). At new edges, individuals
growing in gaps had a higher percentage of fruits damaged